Search

This project aims to achieve various goals related to the selection and optimisation of powertrain components in a hybrid powertrain system. To achieve these objectives, automatic generation, calibration, and comparison of algorithms have been implemented. This research offers the opportunity to optimise multiple targets, such as state of charge (SOC), components selection, CO2 emissions, drive modes, and driveability for McLaren's hybrid powertrain system. The robustness of the Equivalent Consumption Minimisation Strategy (ECMS) is generated automatically for different optimisation objectives, and robustness validation is performed. Additionally, automatic calibration of the torque split strategy is applied to optimise another PhD's work on powertrain components selection. All simulations are based on McLaren's powertrain model. Furthermore, the limitations of the McLaren powertrain model are discussed, which leads to the development of a new powertrain model. The research on powertrain modelling is explored, and the process of generating the new powertrain model is presented in this project. Calibration and validation of the model, along with the implementation of the control strategy, are also researched. The new powertrain model is named after me, YR-Sim. For the application of the Dynamic Programming (DP) algorithm, the new powertrain model has been modified and upgraded with more subsystems and features. All the inputs and outputs of these subsystems have been standardised. The calibration and validation of the model with different algorithms guarantees that the simulation results are comparable to McLaren's model with different algorithms and optimisation cases. Besides optimising CO2 emissions and SOC conditions, driveability has been introduced to research deeper into different algorithms. Driveability is still an open question, but in this project, some novel concepts have been introduced. CO2 emissions, SOC conditions, and driveability have been optimised and researched in different drive cycles, including NEDC, WLTC, FTP75, and even the Nürburgring race track. For different views of driveability, the benefits of the application of ECMS and DP have been deeply researched. Because this is a project that focuses on real industry applications, the research on the algorithms shows the pros and cons of these two algorithms in different dimensions. From the view of the industry, the research results bridge the gap between mathematical algorithm research and strategies application. A series of functions for automatic generation, calibration, and optimisation of control strategy, as well as the experiences in complex powertrain modelling, control, and validation, can also be used for future hybrid/electric vehicle platform development. The academic goals for algorithm research include the performance comparison between different algorithms, as well as the calibration and validation of the powertrain model. The industry goals include algorithm application, robustness testing, and multicomponent selection in different drive cycles. Both of these two series of requirements are satisfied by the end of this project. For the novelty part of this project, a new quantitative definition of driveability is introduced and discussed. Based on this new definition and research on the torque margin, the relationship between throttle pedal position and torque margin is studied. Driveability is numerically divided into prediction and linearity, and both dimensions are quantitatively researched and compared between different algorithms. Another novel research is the parameterisation method of the powertrain model, which significantly accelerates the simulation speed of the model and transfers the model to a steady-state model to ensure it generates reliable results on different control calibrations. Regarding the application and contribution of this project, it has helped the industry sponsor, McLaren, to solve the problem of high-performance hybrid powertrain system driveability. During the engagement of the high-torque electric motor, the torque margin can be controlled, and the relationship between driveability and drive modes is discovered. This will be applied as an important reference for future high-performance powertrain system development.

The promise of autonomous vehicles is a popular and exciting concept that has seen considerable investment over the last decade. However, most of this investment is in the automotive sector with far less investment in autonomous maritime. This thesis explores the application of computer vision and robotics to autonomous navigation and docking in a maritime domain. Sensor fusion is a core robotics concept allowing the outputs from multiple complementary sensors to be combined for robust localisation and perception. An essential first step of which is extrinsic sensor calibration, which determines the relative position and orientation of those sensors. However, once they are mounted to a vehicle it is usually difficult and unwieldy to carry out. As the first main contribution, this thesis proposes a method to automatically perform calibration of arbitrary sensor types during normal operation. In addition to enabling a non-specialist to perform extrinsic calibration, it can also provide automatic updates in the case of sensor failure. Furthermore, it allows the relative scale factor to be inferred, facilitating the inclusion of non-metric sensors into the fusion framework without an explicit metric scaling step. Autonomous vehicles can only navigate unseen environments if they are able to build a map from which they can perform localisation. This is a common robotics problem traditionally solved with depth sensors and/or 3D reconstruction. However, the operational range of many depth sensors is insufficient for maritime scenarios. The second contribution of this thesis proposes a bird's-eye-view mapping approach using only monocular cameras. By projecting and accumulating semantic information, a probabilistic map is constructed that represents navigable and non-navigable areas. The advantages of this approach versus depth sensors are demonstrated in the marine, robotic, and automotive domains. Any perception system with a vision-based backbone is vulnerable to adverse lighting conditions, whether that is low light, reflections or fog. These conditions are especially prevalent in the maritime domain. Other spectra, such as infra-red, are less affected but rarely used in automotive. The third contribution solves the stereo correspondence problem between visible, near-infrared, and thermal images. By learning a common representation between unaligned stereo pairs in a self-supervised manner, the proposed approach provides stereo cues between different arbitrary spectra. This also allows annotations, which are often easier to obtain for the visible spectrum, to be transferred to other sensors easing the collection of multispectral datasets. The final chapter brings the contributions in automated sensor calibration, and semantic ground plane mapping together with path planning to demonstrate autonomous functionality. The presented experiments show significant performance improvements over what is currently available in a real-world practical application. This thesis advances the field of robotics and autonomous vehicles by exploring approaches which cater to both the maritime and automotive domains. It also shows the viability of cost effective monocular cameras as a sensor.

Market makers quote bid and ask prices throughout the trading day for assets under their consideration. Consequently, they face an optimisation problem in which they seek to maximise profits based on their bid-ask spread, and to minimise price risk, which they bear from holding inventory. In this thesis, we will explore this problem and apply it to real trading data. We begin with a market making model framework à la Avellaneda-Stoikov, where the objective is to maximise the trader's utility function. We calibrate the model to real limit order book data which we back-test. Additionally, we consider the limit-order priority system, which is extremely important when trading on a limit order book, to identify a range that the model would perform within. Moreover, little literature has addressed this issue, and the problem of applying a model that is continuous in time and space to a real-world discrete market. For instance, we find that there is a significant improvement in the model when using, what we call, the implied parameters. These parameters implicitly take into account the trading algorithm mechanics and not just the data's characteristics. We extend the model by introducing stochastic volatility. Our model describes the intraday price and variance of the asset as a mean-reverting process. Through our choice of ansatz solution, we reduce the problem from 5 to 3 dimensions. Furthermore, we show that our numerical scheme is stable and converges to the viscosity solution. We compare our model to the original model and a simple strategy, which pegs quotes at a fixed distance from the best bid and ask prices. Our back-test results show that the market making models significantly outperform the simple strategy concerning the certainty-equivalent value and inventory control. However, our novel stochastic model framework is most beneficial for a risk-averse trader seeking stable returns. We approximate a limit order book using data that is more easily available to researchers than low-latency data. We found our approximation useful for estimating parameters on strategies that quote prices over larger time intervals, with profits similar to a model calibrated using real limit order book data. However, the approximated data for calibrating high-frequency strategies significantly underperformed.

In the present thesis, I examined exercise calibration using running and high-intensity functional movement exercise (HIFME). In doing so, I analysed and presented results from eight studies across Chapters 2, 3, 4, and 5. In Chapters 2 and 3, I explored the extent to which demographic factors (e.g. expertise, experience, age, and gender) are effective in informing us about running and HIFME calibration. Studies 1 and 2 demonstrated that such factors do exhibit associations with running calibration, though the relatively minor strength and inconsistency of these associations also indicate that we should not overestimate the factors' contributions. Study 3 found a positive role of having a HIFME background in HIFME calibration, but other demographic factors did not exhibit associations with it. Overall, results from Chapters 2 and 3 highlighted the importance of considering demographic factors when assessing athlete calibration. However, they also highlighted the importance of understanding their limitations when doing so. In Chapter 4, I examined whether we can use self-reports of exercise metacognition and cognitive calibration to predict running and HIFME calibration. There was no significant association between any of these measures and exercise calibration in Studies 4, 5, and 6, suggesting that metacognition self-reports and calibration from other modalities are not reliable predictors of exercise calibration. In Chapter 5, I tested whether a minimal metacognitive intervention in the form of prediction guidance would lead to improved exercise calibration when participants received strategic, as opposed to impulsive, instructions. Findings from Studies 7 and 8 illustrated that strategic predictions facilitated prediction precision compared to impulsive predictions, though their effects on bias appeared to be less consistent and more dependent upon instructions. In Chapter 6, I discussed the general implications of the present thesis, and proposed ways in which future research can further explore the field of exercise calibration.

Aim: To test the reliability of the Modified 5-Year-Olds’ Index using assessors with a range of experience, and to determine whether calibration improves reliability. Design: Prospective reliability study. Setting: Bristol Dental School, University of Bristol. Methods: Fifty study models of non-syndromic 5-year-olds with a repaired UCLP were selected from the CCUK archives. Fifteen participants with a range of clinical experience in orthodontics were divided into three groups of equal experience. Each group was given differing amounts of information: Group 1 information sheet; Group 2 information sheet and reference models; Group 3 calibration course, information sheet and reference models. Each participant scored the 50 models using the Modified 5-Year-Olds’ Index on two occasions at least four weeks apart. ICCs calculated from a two-way random effects model were used to calculate intra-rater reliability and inter-rater reliability comparing assessors’ scoring to the experts’ consensus scores (gold standard). Results: Group 2 (ICC 0.80 – 0.93) and Group 3 (ICC 0.80 – 0.91) demonstrated high levels of intra-rater agreement, with lower levels shown by Group 1 (0.68 – 0.93). Inter-rater agreement was high in Group 2 (ICC 0.87 – 0.93) and Group 3 (0.82 – 0.91), with Group 1 showing the lowest levels of agreement (ICC 0.69 – 0.94). The level of training of the assessors in Group 1 appears to influence reliability scores, with high intra- and inter-rater reliability scores of the consultant and post-CCST trainee at a similar level to those of the same grade in Groups 2 and 3. Conclusions: The Modified 5-Year-Olds’ Index is a reliable method of assessing outcomes when model scoring is carried out by consultants or post-CCST trainees. Calibration in use of the Index does not improve reliability.

Surface profile and topography measurements of manufactured goods and components across all sectors of engineering are an integral component of a production process. Control measurements are taken on a regular basis to ensure items fulfil the required quality standards. Recent industry developments are driving towards more in-situ sensory measurement and monitoring systems. Depending on the dimensional layout and integral material properties of a sample, different metrological measurement techniques come into action for example Coherence Scanning Interferometry or focus variation microscopy. These techniques require high environmental stability and cannot be used for in-line inspection due to long mechanical scanning times. Recent developments in Hyperspectral Interferometry have laid the foundation for a potential single-shot absolute distance measurement technique suitable for in-line inspection. In this thesis a new version of Hyperspectral Interferometry is presented based on a pinhole array.

The design of electronics needs to account for the non-ideal characteristics of the device technologies used to realize practical circuits. This is particularly important in mixed analog-digital design since the best device technologies are very different for digital compared to analog circuits. One solution for this problem is to use a calibration correction approach to remove the errors introduced by devices, but this adds complexity and power dissipation, as well as reducing operation speed, and so must be optimised. This thesis addresses such an approach to improve the performance of certain types of analog-to-digital converter (ADC) used in advanced telecommunications, where speed, accuracy and power dissipation currently limit applications. The thesis specifically focuses on the design of compensation circuits for use in successive approximation register (SAR) ADCs. ADCs are crucial building blocks in communication systems, in general, and for mobile networks, in particular. The recently launched fifth generation of mobile networks (5G) has required new ADC circuit techniques to meet the higher speed and lower power dissipation requirements for 5G technology. The SAR has become one of the most favoured architectures for designing high-performance ADCs, but the successive nature of the circuit operation makes it difficult to reach ∼GS/s sampling rates at reasonable power consumption. Here, two calibration techniques for high-performance SAR ADCs are presented. The first uses an on-chip stochastic-based mismatch calibration technique that is able to accurately compute and compensate for the mismatch of a capacitive DAC in a SAR ADC. The stochastic nature of the proposed calibration method enables determination of the mismatch of the CAPDAC with a resolution much better than that of the DAC. This allows the unit capacitor to scale down to as low as 280aF for a 9-bit DAC. Since the CAP-DAC causes a large part of the overall dynamic power consumption and directly determines both the sizes of the driving and sampling switches and the size of the input capacitive load of the ADC and the kT/C noise power, a small CAP-DAC helps the power efficiency. To validate the proposed calibration idea, a 10-bit asynchronous SAR ADC was fabricated in 28-nm CMOS. Measurement results show that the proposed stochastic calibration improves the ADC’s SFDR and SNDR by 14.9 dB, 11.5 dB, respectively. After calibration, the fabricated SAR ADC achieves an ENOB of 9.14 bit at a sampling rate of 85 MS/s, resulting in a Walden FoM of 10.9 fJ/c-s. The second calibration technique is a timing-skew calibration for a time-interleaved (TI) SAR ADC that calibrates/computes the inter-channel timing and offset mismatch simultaneously. Simulation results show the effectiveness of this calibration method. When used together, the proposed mismatch calibration technique and the timing-skew calibration technique enables a TI SAR ADC to be designed that can achieve a sampling rate of ∼GS/s with 10-bit resolution and a power consumption as low as ∼10mW; specifications that satisfy the requirements of 5G technology.

Stochastic models are commonly used in quantitative finance to describe the dynamics of the derivatives market. As the market quotes are constantly changing, the models need to be calibrated to make real-time investment decisions. This can involve the sensitivity calculation to support the calibration process and investment portfolio management. For investment portfolios consisting of thousands of assets and options, the sensitivity calculation and calibration process are computationally expensive. This thesis presents a number of approaches to sensitivity calculation and model calibration utilizing high-performance computing architectures and Automatic Differentiation that improve performance and accuracy in financial modeling when compared to finite differences and pathwise methods. A parallel Monte-Carlo engine has been developed using the Adjoint methods for the first-order sensitivity calculation and model calibration This addresses the sensitivity and calibration problem for general stochastic differential models. The engine supports multi-/many-core CPU, GPU and distributed computing architectures. The work utilizes a graph representation and overloading operators to express general stochastic differential models. The sensitivities for the model calibration are calculated in parallel via a single simulation by the Adjoint method with the gradient computation cost being 1.8x that of function evaluation. The computational experiments consider both the Heston model and Heston with term-structure. These show that the engine improves performance by up to two orders of magnitude when compared to a sequential version. A hardware implementation has been developed for the Heston model calibration via the Adjoint on FPGA. The work also shows performance improvement of up to two orders of magnitude when compared to a sequential implementation. A parallel non-linear least squares optimization framework using Automatic Differentiation has been developed. This utilizes a graph representation and overloaded operator techniques to express the general objective and constraint functions. The framework supports multi-/many-core architectures such as GPUs and Intel Xeon/Xeon-Phi. The computational experiments consider the semi-closed form Heston model with the Gauss-Kronrod integration. These show performance improvement of 8.4x on GPU (OpenCL) and 7x on (CUDA) vs a sequential OMP (OpenMP) implementation. A Xeon-Phi implementation improves performance by 34x when compared to a single-thread implementation.

This thesis describes the optimisation of the calorimeter design for collider experiments at the future Compact Linear Collider (CLIC) and the International Linear Collider (ILC). The detector design of these experiments is built around high-granularity Particle Flow Calorimetry that, in contrast to traditional calorimetry, uses the energy measurements for charged particles from the tracking detectors. This can only be realised if calorimetric energy deposits from charged particles can be separated from those of neutral particles. This is made possible with fine granularity calorimeters and sophisticated pattern recognition software, which is provided by the PandoraPFA algorithm. This thesis presents results on Particle Flow calorimetry performance for a number of detector configurations. To obtain these results a new calibration procedure was developed and applied to the detector simulation and reconstruction to ensure optimal performance was achieved for each detector configuration considered. This thesis also describes the development of a software compensation technique that vastly improves the intrinsic energy resolution of a Particle Flow Calorimetry detector. This technique is implemented within the PandoraPFA framework and demonstrates the gains that can be made by fully exploiting the information provided by the fine granularity calorimeters envisaged at a future linear collider. A study of the sensitivity of the CLIC experiment to anomalous gauge couplings that {affect} vector boson scattering processes is presented. These anomalous couplings provide insight into possible beyond standard model physics. This study, which utilises the excellent jet energy resolution from Particle Flow Calorimetry, was performed at centre-of-mass energies of 1.4 TeV and 3 TeV with integrated luminosities of 1.5$\text{ab}^{-1}$ and 2$\text{ab}^{-1}$ respectively. The precision achievable at CLIC is shown to be approximately one to two orders of magnitude better than that currently offered by the LHC. In addition, a study into various technology options for the CLIC vertex detector is described.

The International Temperature Scale of 1990 (ITS-90) defines that the calibration of standard platinum resistance thermometers (SPRTs) is based upon the use of temperature fixed points. Since residual impurities (even below parts-per-million levels) present in the fixed point cells influence their realisation temperature (in the order of a few millikelvins), the fixed point material employed in the cell must be ≥ 99.9999 % pure. Impurities usually constitute the most substantial contribution to the uncertainty of primary SPRT calibrations. With a view to tackle this matter, the Consultative Committee for Thermometry (CCT) of the Bureau International of Weights and Measures (BIPM) has recommended the use of a specific correction methodology (the sum of individual estimates, SIE) but other methods have also emerged, each being advocated by a particular National Metrology Institute. The study reported in this thesis aims at investigating the application of seven available correction methodologies to the freezing point of aluminium (660.323 °C) to identify the most consistent methods together with any difficulties related to their implementation. In order to achieve this, a suite of five aluminium fixed point cells have been constructed according to a rigorous protocol, each cell using metal samples sourced from a different supplier. Glow discharge mass spectrometry (GDMS) assays have been obtained from three independent laboratories. Besides, for each cell constructed, a set of long duration freezing curves have been measured under nominally identical conditions. They provided the basis for the calculations of the correction methodologies investigated. The most consistent corrections were achieved with a hybrid method that combines the SIE and the overall maximum estimate (OME): the hybrid SIE/modified OME method. Furthermore, the correction methodology based on the fitting of a Scheil solidification model to the measured freezing curves was found to be highly consistent, provided certain constraints are applied.

In this thesis, we study the FX option pricing problem and put forward a 4-factor hybrid stochastic-local volatility model. The model, which describes the dynamics of an exchange rate, its volatility and the domestic and foreign short rates, allows for a perfect calibration to European options and has a good hedging performance. Due to the high-dimensionality of the problem, we propose a Monte Carlo simulation scheme that combines the full truncation Euler scheme for the stochastic volatility component and the stochastic short rates with the log-Euler scheme for the exchange rate. We analyze exponential integrability properties of Euler discretizations for the square-root process driving the stochastic volatility and the short rates, properties which play a key role in establishing the finiteness of moments and the strong convergence of numerical approximations for a large class of stochastic differential equations in finance, including the ones studied in this thesis. Hence, we prove the strong convergence of the exchange rate approximations and the convergence of Monte Carlo estimators for a number of vanilla and exotic options. Then, we calibrate the model to market data and discuss its fitness for pricing FX options. Next, due to the relatively slow convergence of the Monte Carlo method in the number of simulations, we examine a variance reduction technique obtained by mixing Monte Carlo and finite difference methods via conditioning. We consider a purely stochastic version of the model and price vanilla and exotic options by simulating the paths of the volatility and the short rates, and then evaluating the "inner" Black-Scholes-type expectation by means of a partial differential equation. We prove the convergence of numerical approximations and carry out a theoretical variance reduction analysis. Finally, we illustrate the efficiency of the method through a detailed quantitative assessment.

The millimeter-wave (mmWave) scanner is a system built a few years ago by the phased array antenna research and development group (PAARD) in the Advanced Radar Research Center (ARRC) with the purpose of measuring, characterizing, and calibrating mmWave antennas without significantly changing the setup. The most important reason for developing a multitasking system is to utilize a vector network analyzer (VNA), in many applications. This thesis proposes upgrades to this system, starting with the software, which restricts the system with poor motion flow, and addition of new features to the system that allow faster and reliable measurements. A relevant feature that was added is the calibration and measurement of a W-band active phased array antenna in a package that was manufactured for commercial purposes. Due to the nature of the active element in a phased array antenna, calibration of the system is required for precise and correct performance, which is critical for radio frequency systems. Over-the-air (OTA) calibration is a powerful technique used in phased array antennas in real-world environments. This thesis proposes a novel measurement system to calibrate and characterize the antenna. The system is a multi-axis hybrid system with nine degrees of freedom, used to perform the calibration process and also to scan the antenna radiation pattern in both near-field (NF) and far-field (FF) regions. Experimental results of the calibration method show high precision in the element radiation values, and excellent agreement between the antenna array radiation pattern measured in the near-field and far-field is achieved.

Engine Management Systems (EMS) is getting more complicated each year with new functions being introduced due to tighter emission regulations of both air quality and CO2. This directly a ects the calibration process of a powertrain because the number of vehicle parameters has increased about 10 times in the last 10 years. Self-calibrating feature such as proposed in this thesis has the potential to increase the e ciency of calibrating a complex EMS. The feature is intended to adapt itself to the engine behaviour and performance by continuously updating its calibration maps as the engine is being operated. This process will reduce the needs for new calibration data and additional ne-tuning when an EMS is being carried over to a di erent vehicle. The self-calibrating feature automatically adjusts the air path calibration maps of an engine. It adjusts the air path setpoint maps in real-time for steady state operating conditions.

Measurement uncertainty is an indication of the quality of a given measurement and ultimately translates into the confidence with which a decision can be made. In the context of PV, measurement uncertainties propagate into energy yield uncertainty, which in turn culminates into financial risk associated with an investment. This risk increases the cost of a PV installation. The aim of this thesis is to contribute towards the reduction of PV related measurement uncertainties. This is done in two ways. One is via developing and utilising more comprehensive methodologies for uncertainty propagation of complex measurands. The other is via more detailed estimates of the uncertainty contributors. In particular, the areas addressed in this thesis are the uncertainty estimation of the temperature coefficient measurements of modules; the uncertainty estimation of energy rating and module performance ratio measurements; and the uncertainties due to spectral effects on measurements performed with a flash solar simulator. The reported deviation in measurements of the temperature coefficients of P_MAX of modules is in the order of ±10% to ±15%. This is larger than the difference in the temperature coefficients of modules of the same type. The first step to improving the deviation between measurements is to estimate the uncertainty in a robust way. It was identified that there was no accepted approach of doing this. These measurements are strongly correlated, which complicates the uncertainty estimates. For the sake of simplicity, previously correlations have been avoided and conservative estimates used instead. In this work, uncertainties in both temperature and power and their correlations are estimated and propagated into the overall temperature coefficient uncertainty. Furthermore, temperature coefficients were calculated via weighing the measurements with their associated uncertainties. This was done for five different measurement setups that represent the majority of setups used worldwide. The approach was validated with measurement intercomparison of two modules measured on all systems. The approach reduced the overall uncertainty by half compared to the previous conservative estimates. It was demonstrated that uncertainties as low as 3% are achievable. The improved uncertainty estimates enabled the identification of a systematic effect due to a class B spectrum. This work culminated in the lowest reported measurement deviation of ±3.2% for module P_MAXtemperature coefficient measurements that was within the stated measurement uncertainties. The clear benefit of accounting for correlations was extended to measurements at different irradiance conditions and into the calculation of module performance ratio and energy rating. This was done via defining all the correlations between measurements and then propagating them with Monte Carlo simulations. The simulations are done with samples of a multivariate normal distribution with a variance-covariance matrix that corresponds to the estimated measurement correlations. It is demonstrated that both the energy rating and module performance ratio uncertainties strongly depend on the correlation estimates and that they cannot be conservatively overestimated. The module performance ratio uncertainty can be significantly lower than the measurement uncertainty at STC. This is because of the additional knowledge encoded into the selection of the underlying model used for calculating the energy rating. Therefore, the significance of the choice of model in the upcoming standard has been highlighted. It was confirmed that both bilinear interpolation and the proposed climatic datasets could be used for energy rating, but there are some areas that may need further investigation. An alternative way of improving uncertainty estimates and in turn reducing the associated uncertainty is via a more detailed characterisation of the uncertainty sources. A key uncertainty source is due to spectral effects in flash solar simulators. To better quantify this source, a complementary device was built to monitor the spectrum. The device is based on a matrix of photodiodes with commercially available interference filters situated on top and custom designed data acquisition electronics. This device is used in conjunction with the spectroradiometer to estimate the effects of flash-variation on the spectrum, the spectral temporal stability of the flash and spectral uniformity of the simulator and the attenuation masks used for altering the irradiance levels. It was demonstrated that the spectrum changes significantly during the flash and between flashes. While this effect is partially corrected for via the monitoring cell, it introduces additional uncertainty for non c-Si modules. This uncertainty is minimised by changes in the operational procedures. The spectral non-uniformity of the attenuation masks was shown to be significant, i.e. as large as 4%, in the NIR, prompting further investigation of the additional uncertainty for non c-Si modules. In this work, the methodology of estimating and propagating correlations in PV related measurements and the benefits of doing so are demonstrated. It is also highlighted that the uncertainty due to spectral effects goes beyond the uncertainty of spectroradiometer measurements. Finally, it is shown how they can be estimated with a complementary spectral monitor.

Most existing houses in the UK have a single thermostat, a timer and conventional thermostatic radiator valves to control the low pressure, hot water space heating system. A number of companies are now offering a solution for room-by-room temperature and time control in such older houses. These systems comprise of motorised radiator valves with inbuilt thermostats and time control. There is currently no evidence of any rigorous scientific study to support the energy saving claims of these zonal control systems. This thesis quantifies the potential savings of zonal control for a typical UK home. There were three components to the research. Firstly, full-scale experiments were undertaken in a matched pair of instrumented, three bedroom, un-furbished, 1930s, test houses that included equipment to replicate the impacts of an occupant family. Secondly, a dynamic thermal model of the same houses, with the same occupancy pattern, that was calibrated against the measured results. Thirdly, the experimental and model results were assessed to explore how the energy savings might vary in different UK climates or in houses with different levels of insulation. The results of the experiments indicated that over an 8-week winter period, the house with zonal control used 12% less gas for space heating compared with a conventionally controlled system. This was despite the zonal control system resulting in a 2 percentage point lower boiler efficiency. A calibrated dynamic thermal model was able to predict the energy use, indoor air temperatures and energy savings to a reasonable level of accuracy. Wider scale evaluation showed that the annual gas savings for similar houses in different regions of the UK would be between 10 and 14% but the energy savings in better insulated homes would be lower.

Ultra-broadband Monolithic Microwave Integrated Circuit (MMIC) amplifiers find applications in multi-gigabit communication systems for 5G and millimeter wave measurement instrumentation systems. The aim of the research was to achieve maximum bandwidth of operation of the amplifier from the foundry process used and high reverse isolation ( < -25.0 dB) across the whole bandwidth. To achieve this, several design variations of DC - 110 GHzMMIC Cascode TravellingWave Amplifier (TWA) on 100 nm AlGaAs/GaAs pHEMT process were done for application in next generation instrumentation and high data transfer rate (100 Gb/s) optical modulator systems. The foundry service and device models used for the design are of the WINPP10-10 process from WIN Semiconductor Corp., Taiwan, a commercial and highly stable process. The cut-off frequency ft and maximum frequency of oscillation fmax for this process are 135 GHz and 185 GHz respectively. Thus, the design was aimed at pushing the ultimate limits of operation for this process. The design specifications were targeted to have S21 = 9.0 to 10.0 ± 1.0 dB, S11 & S22 ≤ -10.0 dB and S12 ≤ -25.0 dB in the whole frequency range. In order to achieve the targeted RF performance, it is imperative to have accurate transistor models over the frequency range of operation, transistor configuration mode and operating bias points. Using smaller periphery transistors results in lower extrinsic & intrinsic input and output capacitances that lead to achieving very wide band performance. Thus, device sizes as small as 2x10 μm were used for the design. A cascode topology, which is a series connection of a common-source and common-gate field effect transistor (FET), was used to achieve large bandwidth of operation, high reverse isolation and high input and output impedance. Using very small periphery devices at cascode bias points posed limitation in the design in terms of accuracy of transistor models under these conditions, specifically at high frequencies i.e., above 50 GHz. One of the major systemrequirements for the application of MMIC ultra-broadband amplifiers in instrumentation is to achieve and maintain high reverse isolation (≤ -25.0 dB) over the whole frequency range of operation which cannot be achieved alone by the cascode topology and new design techniques have to be devised. These twomajor challenges, namely high frequency small periphery FET model modification & development and design technique to achieve high reverse isolation in ultra-broadband frequency range have been addressed in this research.

Fringe projection is a commonly used optical technique for measuring the shapes of objects with dimensions of up to about 1 m across. There are however many instances in the aerospace and automotive industries where it would be desirable to extend the benefits of the technique (e.g., high temporal and spatial sampling rates, non-contacting measurements) to much larger measurement volumes. This thesis describes a process that has been developed to allow the creation of a large global measurement volume from two or more independent shape measurement systems. A new 3-D large volume calibration artefact, together with a hexapod positioning stage, have been designed and manufactured to allow calibration of volumes of up to 3 x 1 x 1 m3. The artefact was built from carbon fibre composite tubes, chrome steel spheres, and mild steel end caps with rare earth rod magnets. The major advantage over other commonly used artefacts is the dimensionally stable relationship between features spanning multiple individual measurement volumes, thereby allowing calibration of several scanners within a global coordinate system, even when they have non-overlapping fields of view. The calibration artefact is modular, providing the scalability needed to address still larger measurement volumes and volumes of different geometries. Both it and the translation stage are easy to transport and to assemble on site. The artefact also provides traceabitity for calibration through independent measurements on a mechanical CMM. The dimensions of the assembled artefact have been found to be consistent with those of the individual tube lengths, demonstrating that gravitational distortion corrections are not needed for the artefact size considered here. Deformations due to thermal and hygral effects have also been experimentally quantified. The thesis describes the complete calibration procedure: large volume calibration artefact design, manufacture and testing; initial estimation of the sensor geometry parameters; processing of the calibration data from manually selected regions-of-interest (ROI) of the artefact features; artefact pose estimation; automated control point selection, and finally bundle adjustment. An accuracy of one part in 17 000 of the global measurement volume diagonal was achieved and verified.

This thesis concerns methods to enhance current and explore future radiometric imaging systems for security screening. Its main focus is on the test and calibration procedures for the BorderWatch system – an established 33 GHz passive millimetre-wave imager developed by QinetiQ PLC as an outdoor security scanning portal for soft-sided heavy goods vehicles at ports of entry. The effects of the limited size of the sky background reflector of the operational system are addressed and modifications to mitigate these effects are proposed. Systematic diurnal and seasonal variations in the receiver output powers are characterised and strongly linked with variations in the physical temperature of the RF components. The proposed BorderWatch temperature calibration subsystem requires a reliable cold sky temperature reference point; the aim being to reduce the level of fixed pattern noise in present-day imagery and allow for post-processing methods requiring absolute temperature values. Cost considerations rule out independent millimetre-wave radiometers at each site so a proposed alternative is to use infra-red measurements as a proxy. A literature investigation is made into the millimetre-wave and infrared atmospheric opacities for a variety of meteorological conditions. The design, calibration and operation of a 35 GHz switching radiometer is described together with a comparison of the millimetre-wave radiometric brightness temperature measurements against the data from a low cost commercial infra-red sensor and from a nearby meteorological station. The results show an excellent correlation between the zenith sky temperatures in the infrared and millimetre-wave bands but only in clear sky conditions – as anticipated the presence of clouds affects the infra-red brightness distribution much more strongly than that of the millimetre-wave. Future security imagers may well incorporate interferometric arrays. An extensive simulation programme has been carried out to explore appropriate numbers of antennas and whether radio astronomy configurations and post-processing techniques can offer cost-effective routes to high image fidelity. The results of a quantitative analysis are promising and array configurations and techniques appropriate for potential future security imagers are suggested.

Many powders and particulate solids are stored and handled in large quantities across various industries. These solids often encounter handling and storage difficulties that are caused by the material cohesion. The cohesive strength of a bulk material is a function of its past consolidation stress. For example, high material cohesive strength as a result from high storage stresses in a silo can cause ratholing problems during discharge. Therefore, it is essential to consider the stress-history dependence when evaluating such handling behaviour. In recent years the Discrete Element Method (DEM) has been used extensively to study the complex behaviour of granular materials. Whilst extensive DEM studies have been performed on cohesionless solids, much less work exists on modelling of cohesive solids. The commonly used DEM models to model adhesion such as the JKR, DMT and linear cohesion models have been shown to have difficulty in predicting the stress-history dependent behaviour for cohesive solids. DEM modelling of cohesive solid at individual particle level is very challenging. To apply the model at single particle level accurately would require one to determine the model parameters at particle level and consider the enormous complexity of interfacial interaction. Additionally it is computationally prohibitive to model each and every individual particle and cohesion arising from several different phenomena. In this study an adhesive elasto-plastic contact model for the mesoscopic discrete element method (DEM) with three dimensional non-spherical particles is proposed with the aim of achieving quantitative predictions of cohesive powder flowability. Simulations have been performed for uniaxial consolidation followed by unconfined compression to failure using this model. Additionally, the scaling laws necessary to produce scale independent predictions for cohesionless and cohesive solids was also investigated. The influence of DEM input parameters and model implementation have been explored to study the effect of particle (meso-scale) properties on the bulk behaviour in uniaxial test simulation. The DEM model calibration was achieved using the Edinburgh Powder Tester (EPT) – an extended uniaxial tester to measure flowability of bulk solids. The EPT produced highly repeatable flowability measurements and was shown to be a good candidate for DEM model calibration. The implemented contact model has been shown to be capable of predicting the experimental flow function (unconfined compressive strength versus the prior consolidation stress) for a limestone powder which has been selected as a reference solid in the Europe wide PARDEM research network. Contact plasticity in the model is shown to affect the flowability significantly and is thus essential for producing satisfactory computations of the behaviour of a cohesive granular material. The model predicted a linear relationship between a normalized unconfined compressive strength and the product of coordination number and solid fraction. Significantly, it has been found that contribution of adhesive force to the limiting friction has a significant effect on bulk unconfined strength. Failure to include the adhesive contribution in the calculation of the frictional resistance may lead to under-prediction of unconfined strength and incorrect failure mode. The results provide new insights and propose a micromechanical based measure for characterising the strength and flowability of cohesive granular materials. Scaling of DEM input parameters in a 3D simulation of the loading regimes in a uniaxial test indicated that whilst both normal and tangential contact stiffness (loading, unloading, and load dependent) scales linearly with radius of the particle, the adhesive forces scales with the square of the radius of the particles. This is a first step towards a mesoscopic representation of a cohesive powder that is phenomenological based to produce the key bulk characteristics of a granular solid and the results indicate that it has potential to gain considerable computational advantage for large scale DEM simulations. The contact model parameters explored include particle contact normal loading stiffness, tangential stiffness, and contact friction coefficient. The DEM model implementation parameters included numerical time step, strain rate, and boundary condition. Many useful observations have been made with significant implications for the relative importance of the DEM input parameters. Finally the calibration procedure was applied to a spray dried detergent powder and the simulation results are compared to whole spectrum of loading regime in a uniaxial experiment. The experimental and simulation results were found to be in reasonable agreement for the flow function and compression behaviour.

This thesis presents novel solutions for two fundamental problems associated with autonomous road driving. The first is accurate and persistent localisation and the second is automatic extrinsic sensor calibration. We start by describing a stereo Visual Odometry (VO) system, which forms the basis of later chapters. This sparse approach to ego-motion estimation leverages the efficacy and speed of the BRIEF descriptor to measure frame-to-frame correspondences and infer subsequent motion. The system is able to output locally metric trajectory estimates as demonstrated on many kilometres of data. We then present a robust vision only localisation system based on a two-stage approach. Firstly we gather a representative survey in ideal weather and lighting conditions. We then leverage locally accurate VO trajectories to synthesise a high resolution orthographic image strip of the road surface. This road image provides a highly descriptive and stable template against which to match subsequent traversals. During the second phase, localisation, we use the VO to provide high frequency pose updates, but correct for the drift inherent in all locally derived pose estimates with low frequency updates from a dense image matching technique. Here a live image stream is registered against synthesised views of the road image generated form the survey. We use an information theoretic measure, Mutual Information, to determine the alignment of live images and synthesised views. Using this measure we are able to successfully localise subsequent traversals of surveyed routes under even the most intense lighting changes expected in outdoor applications. We demonstrate our system localising in multiple environments with accuracy commensurate to that of an Inertial Navigation System. Finally we present a technique for automatically determining the extrinsic calibration between a camera and Light Detection And Ranging (LIDAR) sensor in natural scenes. Rather than requiring a stationary platform as with prior art, we actually exploit platform motion allowing us to aggregate data and adopt a retrospective approach to calibration. Coupled with accurate timing this retrospective approach allows for sensors with non-overlapping fields of view to be calibrated as long as at some point the observed workspaces overlap. We then show how we can improve the accuracy of our calibration estimates by treating each single shot estimate as a noisy measurement and fusing them together using a recursive Bayes filter. We evaluate the calibration algorithm in multiple environments and demonstrate millimetre precision in translation and deci-degrees in rotation.

In this thesis the jet energy scale uncertainty of small and large radius jets at the ATLAS detector is evaluated in-situ using gamma-jet events. The well calibrated photon in the gamma-jet events is used to probe the energy scale of the jets. The studies of the jet energy scale of small radius jets are performed using 4.7 fb-1 of data collected at sqrt{s} = 7 TeV in 2011. The gamma-jet methods which were developed are then adapted and applied to large radius jets, using 20.3 fb^-1 of data collected at sqrt{s} = 8 TeV in 2012. The new jet energy scale uncertainties are found to be ~1 % for |eta| < 0.8, rising to 2-3 % for |eta| > 0.8. These uncertainties are significantly lower than the 3-6 % precision which has previously been achieved at ATLAS using track jets as a reference object. Due to the increase in precision, uncertainties due to pile-up and the topology of the jet also had to be evaluated. The total energy scale uncertainties for large radius jets are reduced by ~1-2 % (0.5-1 %) for |eta| < 0.8 (> 0.8). This reduction will be beneficial to analyses using large radius jets and it is specifically shown to benefit the t-tbar resonance search in the semi-leptonic channel. The t-tbar search looks for events with two top quarks in the final state, where one decays leptonically and the other hadronically. The hadronically decaying top quark is reconstructed using a large radius jet, and the jet energy scale uncertainty is a dominant source of uncertainty in the analysis. In addition to the studies of the jet energy scale of large radius jets, the first derivation of a calibration, and jet energy scale uncertainties derived with gamma-jet events, are shown for Variable R jets. The Variable R jet algorithm is a new type of jet algorithm with a radius that is inversely proportional to the size of the jet, making it useful for the study of high momentum top quarks. It is shown that similar methods can be used to calibrate and assess the uncertainties of Variable R jets as are used for standard, fixed radius jets at the ATLAS detector, although some adaptations will be necessary. The studies provide a basis for the calibration of Variable R jets in the future.

We study a sequential binary prediction setting where the forecaster is evaluated in terms of the calibration distance, which is defined as the L1 distance between the predicted values and the set of predictions that are perfectly calibrated in hindsight. This is analogous to a calibration measure recently proposed by Błasiok, Gopalan, Hu and Nakkiran (STOC 2023) for the offline setting. The calibration distance is a natural and intuitive measure of deviation from perfect calibration, and satisfies a Lipschitz continuity property which does not hold for many popular calibration measures, such as the L1 calibration error and its variants. We prove that there is a forecasting algorithm that achieves an O(√T) calibration distance in expectation on an adversarially chosen sequence of T binary outcomes. At the core of this upper bound is a structural result showing that the calibration distance is accurately approximated by the lower calibration distance, which is a continuous relaxation of the former. We then show that an O(√T) lower calibration distance can be achieved via a simple minimax argument and a reduction to online learning on a Lipschitz class. On the lower bound side, an Ω(T^(1/3)) calibration distance is shown to be unavoidable, even when the adversary outputs a sequence of independent random bits, and has an additional ability to early stop (i.e., to stop producing random bits and output the same bit in the remaining steps). Interestingly, without this early stopping, the forecaster can achieve a much smaller calibration distance of polylog(T).

Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) is now a well established analytical technique used to sample solid materials and determine their elemental composition. Two areas that are becoming increasingly important, and for which LA-ICP-MS is a key tool, are bio-imaging and the analysis of micro-particulates. However, current instrumental designs limit the practicality of the technique for these applications. This study investigates the development of a high speed, high efficiency LA-ICP-MS interface through modelling of the flow dynamics of a newly designed laser ablation cell and experimental investigation of single laser pulse response. Through this work the Sniffer-Dual Concentric Injector interface was realised. This interface reduced particle residence times within the laser cell and transport tubing. The interface was also used to investigate turbulence related aerosol dispersion within the ICP and potential designs to overcome this. The resulting design yields an interface with improved sensitivity and reduced aerosol dispersion such that a lower limit of detection is achieved, important when considering the mass of analyte in a single cell or micro-particulate, compared to existing designs. Thus the interface can be used to improve image spatial resolution as the ablation spot size, and thus pixel information, can be reduced; and also reduces total analysis time. The calibration technique Laser Ablation of a Sample In Liquid (LASIL) was also investigated as a means of calibration for solid samples. The investigation lead to the development of LASIL in a droplet, a technique that can be used to calibrate solid samples when a matrix matched standard is unavailable. The mechanism of the technique resulted in an improved laser-energy sample coupling efficiency and a reduction in the liquid to ablated mass ratio, thus decreasing sampling time. As the technique captures the ablated particulate in solution, post chemistry techniques can be used to remove analyte interferences.

Background: Studies have reported that preschool children may not be sufficiently active according to guidelines. This is worrying because early childhood is a critical period for the establishment of sedentary behaviour (SB) and physical activity (PA) habits, which have immediate and long term influences on health. The majority of evidence on levels, determinants, and health consequences of SB and PA in young children is, however, based on subjective measures in predominately White children aged three years or older. Aims: To 1) assess the feasibility and acceptability of using three different accelerometers in South Asian and White British 2-3 year olds and their parents; 2) calibrate and validate the accelerometers to measure SB and PA in 2-3 year olds; 3) investigate the influence of 5-, 10-, and 15-second epochs on time spent in SB, light PA, and moderate-to-vigorous PA (MVPA) in 2-3 year olds; and 4) assess the feasibility of measuring the habitual SB and PA with the ActiGraph GT3X+ accelerometer in South Asian and White British 2-3 year olds and their parents. Methods: Focus groups were performed with 17 South Asian and White British mothers and the audio-recordings were transcribed verbatim and analysed with thematic analysis (Aim 1). To calibrate the three accelerometers against direct observation (Aim 2), semi-structured activity sessions were run with 18 South Asian and White British 2-3 year olds. Mixed-effects regression and receiver-operating characteristics (ROC) curve analysis were used to generate prediction equations and 5- and 10-second cut-points to assess children s SB and PA. Validity of the generated cut-points against direct observation (Aim 2) was assessed in a separate sample of 20 White British and Black African 2-3 year olds during free-play, using Cohen s kappa, ROC curve analysis, Bland-Altman plots, and Lin s concordance coefficient. Differences in estimated time spent in SB, light PA and MVPA from 5-, 10- and 15-second epochs were tested with repeated-measures ANOVA and paired t-tests in the calibration sample (Aim 3). A pilot study was conducted with 120 South Asian and White British families from a birth cohort study to address aim 4. Study intake and compliance with an 8-day measurement protocol were calculated, and differences between ethnicities were assessed with the Chi-square test. Results: The ActiGraph GT3X+ was the most widely accepted accelerometer, with the least amount of issues raised by mothers. Practical and software issues with the Actiheart and activPAL3 during the calibration phase resulted in insufficient good quality data collected, which made it unfeasible to calibrate both monitors. The overall 5-second Axis1 cut-points for the ActiGraph GT3X+ provided a valid tool to measure the SB and total PA of 2-3 year olds in free-living conditions. Using 10- and 15-second epochs overestimated children s light PA and underestimated time spent in SB and MVPA. Less South Asian than White British families were recruited into the study, and less South Asian than White British children complied with the 8-day measurement protocol. There were no ethnic differences in the number of children and parents providing enough accelerometry data (i.e. ≥ 3 valid days), or the number of parents complying with the measurement protocol. Conclusions: The results demonstrated that it is feasible to use the ActiGraph GT3X+ to assess the habitual SB and PA of a bi-ethnic sample of 2-3 year old children and their parents. The accurate assessment of SB and PA in 2-3 year olds using the overall 5-second Axis1 cut-points developed and validated in this thesis will enable researchers to investigate the levels, determinants, and health consequences of SB and PA. Such research will inform public health policies and interventions to improve children s health.

The dynamic properties of several rectangular and V-shaped microcantilevers were investigated. Particular attention was paid to the higher flexural eigenmodes of oscillation. The potential of the higher flexural modes was demonstrated through the use of cantilevers as standalone sensors for adsorbed mass. The mass adsorbed on the surface of a cantilever was in the form of a homogeneous water layer measured as a function of relative humidity. The minimum detectable water layer thicknesses were 13.7 Å, 3.2 Å, 1.1 Å, and 0.7 Å for the first four modes of a rectangular cantilever, clearly demonstrating enhanced accuracy for the higher eigenmodes of oscillation. These thicknesses correspond to minimum detectable masses of 33.5 pg, 7.8 pg, 2.7 pg and 1.7 pg for the first four modes. For quantitative applications the spring constants of each cantilever must be determined. Many methods exist but only a small number can calibrate the higher flexural eigenmodes. A method was developed to simultaneously calibrate all flexural modes of microcantilever sensors. The method was demonstrated for the first four eigenmodes of several rectangular and V-shaped cantilevers with nominal fundamental spring constants in the range of 0.03 to 1.75 N/m. The spring constants were determined with accuracies of 5-10 %. Spring constants of the fundamental mode were generally in agreement with those determined using the Sader method. The method is compatible with existing AFM systems. It relies on a flow of gas from a microchannel and as such poses no risk of damage to the cantilever beam, its tip, or any coating. A related method was developed for the torsional modes of oscillation. Preliminary results are shown for the fundamental mode of a rectangular cantilever. The method can be easily extended to the higher torsional modes, V-shaped cantilevers, and potentially, the flapping modes of the legs of V-shaped microcantilevers.

If the analyses of trace components in complex organic samples are to be optimised, then these compounds must be isolated either physically or chemically from surrounding matrices. Ion mobility spectrometry (IMS) is an analytical technique used worldwide for the detection of on-site trace compounds. The technique can be optimised to isolate the target species from complex matrices through both physical separation, based on the mobility of the analyte ions at ambient pressure, and chemical discrimination through preferential ionisation of the target. Optimisation of the latter is commonly achieved through doping the spectrometer with a selective reagent gas, termed a dopant. The chemical processes required to optimise the responses of target analytes are dependent on the identity and concentration of the dopant. As such, a variety of dopants have been successfully implemented in ion mobility spectrometers. The technology for the deliverance of dopants in IMS is commonly through permeation sources, which provide a stable chemical environment in the ion mobility cell. Althoughrelatively inexpensive and durable, these devices are difficult to change and generally deliver a single dopant concentration. As a result, only one type of chemistry is possible and the responses cannot be optimised for a range of analytical applications. Such limitationsbecome increasingly significant when IMS is hyphenated to a chromatograph where a range of different dopant conditions may be sought over the course of a chromatographic run. This thesis sought to overcome these limitations through the development and implementation of piezoelectric dispensers, interfaced directly to the transport gas regions of IMS cells. The study demonstrates for the first time the ability to use piezoelectric dispensing as a dopant introduction methodology in IMS for controlling and calibrating a range of dopant chemistries. 2-butanol, acetone, dichloromethane, 1-chlorohexane, 4-heptanone and 1-bromohexane were the candidate dopants chosen for the studies, covering a wide range of physical and chemical properties. The novel technology was used to dispense the target dopants into IMS cells at concentration ranges over three orders of magnitude. Dopant chemistries were achieved within three seconds from the point of dispensing, administered in drop-ondemand formats, and could be delivered either transiently or at steady-state concentrations. The concept was validated through integrated spectral dopant responses. In transient control, dynamic linear relationships of R2 = 0.991 - 0.998 were achieved between dispensed dopant mass and peak area. Under continuous operation, the RSD of the dopant level was < 18% for all dopants. Clear out times for dopant responses were in the order of 3-5 seconds, indicating negligible hysteresis effects. The study also proved the concept of controlling monomer and dimer ion chemistries from 2-butanol actuations when interfaced to a differential mobility spectrometer at mass fluxes between 21 - 1230 ng m-3 , and the simultaneous control of dopants in negative and positive ionisation modes to RSDs <10%. This thesis describes the techniques used to optimise the piezoelectric dispensing of the full dopant range, as well as the full design protocols required to interface to mobility spectrometers. The outcomes from these studies provide a realisation for piezoelectric dispensers as a future mainstream dopant introduction technique for the analysis of complex samples.

In November 2012, the European Space Agency (ESA) is planning to launch Gaia, a mission designed to measure with microarcsecond accuracy the astrometric properties of over a billion stars. Microarcsecond astrometry requires extremely accurate positional measurements of individual stellar transits on the focal plane, which can be disrupted by radiation-induced Charge Transfer Inefficiency (CTI). Gaia will suffer radiation damage, impacting on the science performance, which has led to a series of Radiation Campaigns (RCs) being carried out by industry to investigate these issues. The goal of this thesis is to rigorously assess these campaigns and facilitate how to deal with CTI in the data processing. We begin in Chapter 1 by giving an overview of astrometry and photometry, introducing the concept of stellar parallax, and establishing why observing from space is paramount for performing global, absolute astrometry. As demonstrated by Hipparcos, the concept is sound. After reviewing the Gaia payload and discussing how astrometric and photometric parameters are determined in practice, we introduce the issue of radiation-induced CTI and how it may be dealt with. The on board mitigating strategies are investigated in detail in Chapter 2. Here we analyse the effects of radiation damage as a function of magnitude with and without a diffuse optical background, charge injection and the use of gates, and also discover a number of calibration issues. Some of these issues are expected to be removed during flight testing, others will have to be dealt with as part of the data processing, e.g. CCD stitches and the charge injection tail. In Chapter 3 we turn to look at the physical properties of a Gaia CCD. Using data from RC2 we probe the density of traps (i.e. damaged sites) in each pixel and, for the first time, measure the Full Well Capacity of the Supplementary Buried Channel, a part of every Gaia pixel that constrains the passage of faint signals away from the bulk of traps throughout the rest of the pixel. The Data Processing and Analysis Consortium (DPAC) is currently adopting a 'forward modelling' approach to calibrate radiation damage in the data processing. This incorporates a Charge Distortion Model (CDM), which is investigated in Chapter 4. We find that although the CDM performs well there are a number of degeneracies in the model parameters, which may be probed further by better experimental data and a more realistic model. Another way of assessing the performance of a CDM is explored in Chapter 5. Using a Monte Carlo approach we test how well the CDM can extract accurate image parameters. It is found that the CDM must be highly robust to achieve a moderate degree of accuracyand that the fitting is limited by assigning finite window sizes to the image shapes. Finally, in Chapter 6 we summarise our findings on the campaign analyses, the on-board mitigating strategies and on how well we are currently able to handle radiation damage in the data processing.

The subject of this thesis is the development of parametric methods for the calibration of array shape errors. Two physical scenarios are considered, the online calibration (self-calibration) using far-field sources and the offline calibration using near-field sources. The maximum likelihood (ML) estimators are employed to estimate the errors. However, the well-known computational complexity in objective function optimization for the ML estimators demands effective and efficient optimization algorithms. A novel space-alternating generalized expectation-maximization (SAGE)-based algorithm is developed to optimize the objective function of the conditional maximum likelihood (CML) estimator for the far-field online calibration. Through data augmentation, joint direction of arrival (DOA) estimation and array calibration can be carried out by a computationally simple search procedure. Numerical experiments show that the proposed method outperforms the existing method for closely located signal sources and is robust to large shape errors. In addition, the accuracy of the proposed procedure attains the Cram´er-Rao bound (CRB). A global optimization algorithm, particle swarm optimization (PSO) is employed to optimize the objective function of the unconditional maximum likelihood (UML) estimator for the farfield online calibration and the near-field offline calibration. A new technique, decaying diagonal loading (DDL) is proposed to enhance the performance of PSO at high signal-to-noise ratio (SNR) by dynamically lowering it, based on the counter-intuitive observation that the global optimum of the UML objective function is more prominent at lower SNR. Numerical simulations demonstrate that the UML estimator optimized by PSO with DDL is optimally accurate, robust to large shape errors, and free of the initialization problem. In addition, the DDL technique is applicable to a wide range of array processing problems where the UML estimator is employed and can be coupled with different global optimization algorithms.

In this thesis we address problems associated with financial modelling from a Bayesian point of view. Specifically, we look at the problem of calibrating financial models, measuring the model uncertainty of a claim and choosing an optimal hedging strategy. Throughout the study, the local volatility model is used as a working example to clarify the proposed methods. This thesis assumes a prior probability density for the unknown parameter in a model we try to calibrate. The prior probability density regularises the ill-posedness of the calibration problem. Further observations of market prices are used to update this prior, using Bayes law, and give a posterior probability density for the unknown model parameter. Resulting Bayes estimators are shown to be consistent for finite-dimensional model parameters. The posterior density is then used to compute the Bayesian model average price. In tests on local volatility models it is shown that this price is closer than the prices of comparable calibration methods to the price given by the true model. The second part of the thesis focuses on quantifying model uncertainty. Using the framework for market risk measures we propose axioms for new classes of model uncertainty measures. Similar to the market risk case, we prove representation theorems for coherent and convex model uncertainty measures. Example measures from the latter class are provided using the Bayesian posterior. These are used to value the model uncertainty for a range of financial contracts priced in the local volatility model. In the final part of the thesis we propose a method for selecting the model, from a set of candidate models, that optimises the hedging of a specified financial contract. In particular we choose the model whose corresponding price and hedge optimises some hedging performance indicator. The selection problem is solved using Bayesian loss functions to encapsulate the loss from using one model to price and hedge when the true model is a different model. Linkages are made with convex model uncertainty measures and traditional utility functions. Numerical experiments on a stochastic volatility model and the local volatility model show that the Bayesian strategy can outperform traditional strategies, especially for exotic options.

The Vanna-Volga (VV) method has been in wide use as one of the major tools for several years among foreign exchange (FX) trading desks. Despite its popularity, the properties of the VV method are not well studied and understood. This thesis attempts to understand better why and when the VV method makes sense, and how to use it better. Often under practical circumstances the state of calibration can be described as being frequent but imperfect. To take advantage of this level of calibration, we studied the properties and benefits of the Karasinski method, and extended this method to a few useful applications. We have found that the Karasinski method, if used with a reasonably calibrated model, can provide significant performance improvement over the VV method.The VV and Karasinski chapters contain most of the original research in this thesis; there are a wealth of discoveries made in these chapters. Novel methods and applications related to the VV and Karasinski methods are proposed, and some of which can be readily applied to the practical trading environment. To make the VV and Karasinski methods work well in practice, the numerical issues for computing the price and Greeks have been carefully addressed with finite difference schemes that are second-order convergent and fast to compute. As an example of easy-to-compute but difficult-to-calibrate model candidates for the Karasinski method, the Multi-Heston model has been discussed too. A sound computational preparation enables the VV and in particular Karasinski methods to enjoy high viability as being fast, efficient and practical. This thesis is tailored to the purpose of making a detailed study on these useful methods whose great potential has not been adequately understood and fully realised.

The International Temperature Scale of 1990 (ITS-90) defines the present S.I.(“System International”) means of measuring temperature. The ITS-90 uses the freezing points of metals to define temperature fixed points. It also uses long-stemplatinum resistance thermometers to interpolate between the fixed points from 660 °Cdown to 84 K (if one includes the Argon triple point). Impurities are a major source of uncertainty in the fixed point temperature (of the order of 1 mK). And a better understanding of the impurity effect is required to improve top-level metrologicalthermometry. Most historical experiments with impurities have worked at a muchhigher levels of impurities – say of the order of 100ppm - and in arrangements that are not used on a day-to-day basis in a metrology laboratory. This thesis describes the deliberate doping of tin and aluminium, each with three different impurities and the effects of these on the temperature of the tin and aluminium liquid-solid phase transitions. The impurities, of the order of 1-30 ppm,were Co, Pb and Sb in the tin and Cu, Si and Ti in the aluminium. The tin and aluminium samples were in the form of ~0.3 kg ingots that would normally be used to realise an ITS-90 fixed point. Measurements were made using equipment normally available in a metrological thermometry laboratory, rather than using specially prepared samples. The samples were chemically analysed (by Glow Discharge Mass Spectrometry(GD-MS)) before and after the doping. Using the amount of dopants introduced,and/or the chemical analysis data, the measured temperature changes were compared with those interpolated from the standard text. The experimental undoped liquid-solid transition curves were also compared against theoretical curves (calculated from atheoretical model MTDATA). The results obtained did not disagree with the Hansen interpolated values (though there was considerable uncertainty in some of the measurements (e.g. a factor of 2 ormore) due to the measurement of small changes. Within these uncertainties it indicatesthat the Sum of Individual Estimates (SIE) method of correcting for, at least, metal impurities in otherwise high purity metals remain valid. However the results also showed considerable discrepancies between the initial measured and calculated temperature shifts (based on the pre-existing impurities prior to doping) suggesting that there may be impurities that are not (separately) detected by the GD-MS method. There was evidence that the thermal history of the metal phase transitions can cause considerable segregation of some impurities, particularly those likely to increase the phase transition temperature through a peritectic (“positive” impurities), and that the effects of this segregation can be clearly seen on the shape of the melting curves of thetin doped with Sb. Some of the aluminium doped with Ti freezing curves may also show evidence of a“concave up” shape at the start of the freezing curve, as previously calculated by MTDATA, though the effect is not as pronounced. All individual phase transition measurements - made over tens of hours – were repeated at least three times and found to be reproducible, hence providing a real dataset that can be used for comparison with theoretical models still under development.

After undergoing a study about current engine modelling and mapping approaches as well as the engine modelling requirements for different applications, a major problem found to be present is the extensive and time consuming mapping procedure that every engine has to go through so that all control parameters can be derived from experimental data. To improve this, a cycle-by-cycle modelling approach has been chosen to mathematically represent reciprocating engines starting by a complete dynamics crankshaft mechanism model which forms the base of the complete engine model. This system is modelled taking into account the possibility of a piston pin offset on the mechanism. The derived Valvetrain model is capable of representing a variable valve lift and phasing Valvetrain which can be used while modelling most modern engines. A butterfly type throttle area model is derived as well as its rate of change which is believed to be a key variable for transient engine control. In addition, an approximation throttle model is formulated aiming at real-time applications. Furthermore, the engine inertia is presented as a mathematical model able to be used for any engine. A spark ignition engine simulation (SIES) framework was developed in MATLAB SIMULINK to form the base of a complete high fidelity cycle-by-cycle simulation model with its major target to provide an environment for virtual engine mapping procedures. Some experimental measurements from an actual engine are still required to parameterise the model, which is the reason an engine mapping (EngMap) framework has been developed in LabVIEW, It is shown that all the moving engine components can be represented by a single cyclic variable which can be used for flow model development.

Logarithmic sensors are capable of capturing the wide dynamic range of intensities available in nature with minimum number of bits and post-processing required. A simple circuit able to perform logarithmic capture is one utilising a MOS device in weak inversion. However, the output of this pixel is crippled due to fixed pattern noise. Technique proposed to reduce this noise fail to produce high quality images on account of unaccounted high gain variations in the pixel. An electronic calibration technique is proposed which is capable of reducing both multiplicative as well as additive FPN. Contrast properties matching that of human eye are reported from these sensors. With reduced FPN, the pixel performance at low intensities becomes concerning. In these regions, the high leakage current of the CMOS process affects the logarithmic pixel. To reduce this current, two different techniques using a modified circuit and another with modified layout are tested. The layout technique is observed to reduce the leakage current. In addition, this layout can be used to linearise the output of logarithmic pixel in low light regions. The unique linear response at low light and logarithmic pixel at high light is further investigated. A new model based on the device physics is derived to represent this response. The fixed pattern noise profile is also investigated. An intelligent iterative scheme is proposed and verified to extract the photocurrent flowing in the pixel and correct the fixed pattern noise utilising the new model. Future research ideas leading to better designs of logarithmic pixels and post-processing of these signals are proposed at the end of the thesis.

The aim of this research is to make a newly constructed Stewart-Gough Platform-based test frame Tiger 66.1 operational by developing control software and estimating the error in its pose accuracy. The accuracy of the platform is affected by one source or multiple sources. The typical error sources are kinematic and structural, some of them originate from manufacturing imperfections, assembly deviations, elastic deformations, thermal deformations, and joint clearances which change the expected kinematic behavior of the manipulator. Also, some non-mechanical errors like transmission error, sensor accuracy, algorithm error, and truncation error in calculation contribute significantly in some cases. Using pose deviations as a foundation, this research further aims to develop a calibration method that enhances pose accuracy, leveraging the pose deviations observed during the initial measurements. This research presents a novel calibration method for a Stewart-Gough platform using photogrammetry, a digital image-based technique for 3D measurement and modeling. As part of this research, a new forward kinematic algorithm has been developed to implement an accurate non-contact calibration method that can improve the accuracy and precision of a general Stewart-Gough platform. The Stewart-Gough platform is one of the most popular Parallel Kinematic Machines (PKM). This mechanism, also known as a parallel manipulator or parallel robot, consists of a fixed base and a movable platform connected by six actuators or struts. The Stewart-Gough platform has remained an interesting machine for research due to its flexibility, structural rigidity, high accuracy, and reliability in motion control. The calibration of a Stewart-Gough platform is one of the essential steps in ensuring the accuracy and stability of the moving platform center. This dissertation investigates a forward kinematic calibration method for improving the accuracy of the Stewart-Gough platform-based test platform “Tiger 66.1” which is intended for use in the characterization of additively manufactured parts, although the method is not restricted to this platform only and can be extended to any similar platforms designed for use in various applications. For calibrating Tiger 66.1, a new forward kinematic algorithm has been developed in this research. The forward kinematics for parallel manipulator generates multiple solutions and they include both feasible and unfeasible solutions. The developed algorithm is a new way of finding unique feasible solution for a platform pose. The proposed algorithm utilizes the high power of modern computing systems and finds a unique solution for a pose through iterations. The advantage of this new algorithm is that the solution obtained from the iterations does not need to be verified manually to check the feasibility in real life and can be directly used as input for any further calculations without stopping the computation process. The proposed calibration methods used photogrammetry which minimizes the need for manual handling of the platform during the calibration process. Photogrammetry uses images of targets taken by one or more cameras to reconstruct 3D positions and orientations. In this research a high-resolution digital camera has been used to take multiple images of the moving platform center from three different angles for each pose and analyze those images through the commercial software package “Photomodeler”, to measure the pose of the platform in three-dimensional space. This method eliminates the need for any additional measurement instrument to be used directly or indirectly interacting with the Stewart-Gough platform. The proposed forward kinematic technique is validated through both simulations and experiments on the physical Stewart-Gough platform-based test frame Tiger 66.1. The vision-based approach is showcased to improve absolute positioning accuracy by up to 25% compared to an uncalibrated platform. The method requires minimal hardware modification and renders it highly suitable for precision application. It provides a practical approach to the calibration of parallel manipulators intended for high-precision tasks. The dissertation concludes by summarizing the contributions of this research, which includes the development of a novel photogrammetry-based calibration method that involves minimal hardware modification and full extrinsic calibration from vision data. The demonstration shows the substantial potential of the proposed method in augmenting the positioning performance of the Stewart-Gough platform tailored for precision applications. Furthermore, the work identifies the areas for further work, such as implementing an online calibration method. Overall, this research demonstrates the capabilities of photogrammetry for parallel manipulator calibration while minimizing hardware modifications, thereby presenting a pragmatic approach to the calibration of such systems in the pursuit of high-precision applications.

Neutrinos hold the promise of untangling many unresolved questions in particle physics. Their unique properties and behaviors offer a distinctive window into understanding the fundamentals of the universe, potentially providing answers to some of the most deep puzzles in modern physics. CEνNS, or Coherent Elastic Neutrino-Nucleus Scattering, is a process where a neutrino interacts with an atomic nucleus and scatters away, leaving the nucleus to recoil. CEνNS is an important area of study for understanding neutrino properties as well as their role in the universe. The COHERENT collaboration was the first to measure CEνNS, using neutrinos from the Spallation Neutron Source (SNS). The direct measurement of the SNS neutrino flux is vital for the precision of CEνNS measurements. This work introduces the latest addition to the COHERENT's armory –a D2O detector specifically designed to measure the SNS neutrino flux. In the present dissertation, the emphasis is made on the steps taken to operationalize COHERENT's D2O detector. This work unfolds the intensive simulation work directed to determine the detector's optimal design, ensuring it stands strong to the demands of neutrino physics experiments. Establishing the detector's calibration is essential to its operational phase. A dedicated calibration system, described in detail in this work, has been developed, utilizing encapsulated LED flashers controlled by a microcontroller unit to ensure the systematic and reliable calibration of the detector. A significant portion of the document is devoted to the calibration analysis, where we use Michel electrons to obtain an energy scale for the detector, thereby ensuring the reliability and accuracy of the future neutrino flux measurements.