Search

The field of minimally invasive surgery (MIS) is continuously changing with the advancement of surgical instrumentation. To pursue further minimisation of skin incision and access trauma, the MIS procedures are evolving from multi-port laparoscopic surgery to single-incision laparoscopic surgery (SILS) and even to scarless natural orifice transluminal endoscopic surgery (NOTES). These new surgical procedures require the surgical instruments, either rigid or flexible, operating in an anatomical area that is not directly aligned with the access site. Significant problems are raised in these procedures, such as reduced dexterity, inadequate instrument triangulation and counterintuitive hand-eye coordination. Suturing -a fundamental surgical technique for wound closure using stitches made by needles and threads - becomes very problematic in these new surgical procedures. In recent decades, one of the most astounding examples of bringing technology into clinical practice is the adoption of robotic surgical systems in MIS. This research aims to investigate the use of robotic techniques to extend the usability of suturing, boost suturing performance and eventually push the field towards autonomous suturing in medical and surgical applications. To avoid complex needle manipulation and facilitate knot tying in MIS, two types of smart and miniaturised suturing instruments have been developed. The rigid suturing instrument has demonstrated its efficacy for fast and reliable tacking stitches in the personalised stent graft sewing task, while the flexible suturing instrument makes it possible for suturing and knot tying in single-port transanal endoscopic microsurgery. In addition to developing these suturing instruments, the author has proposed and given the method of 3using metal printing to monolithically produce the flexible joints - 3D compliant mechanisms - for various surgical instruments, for example, the flexible suturing device and the S-bending flexible endoscope for single-port surgery. Besides needle manipulation, thread manipulation is another crucial aspect in robotic suturing. Towards automated thread manipulation, a deep-learning based method has been proposed for parsing the structural information of a thread in the suturing task. Finally, a fully automated suturing system integrating the rigid suturing instrument, a dual arm industrial robot, RGBD vision and force sensing has been demonstrated for stent graft manufacturing. The central part of the thesis is laid in developing smart instrumentation for robotic sewing and suturing. However, the methods discovered, and the mechanisms invented during the developing process also provide valuable insight into developing various future surgical instruments.

The iris has been one of the most reliable biometric traits for automatic human authentication due to its highly stable and distinctive patterns. Traditional iris recognition algorithms have achieved remarkable performance in strictly constrained environments, with the subject standing still and with the iris captured at a close distance. This enables the wide deployment of iris recognition systems in applications such as border control and access control. However, in less constrained environments with the subject at-a-distance and on-the-move, the iris recognition performance is significantly deteriorated, since such environments induce noise and degradations in iris captures. This restricts the applicability and practicality of iris recognition technology for some real-world applications with more open capturing conditions, such as surveillance, forensic and mobile device security applications. Therefore, robust algorithms for less constrained iris recognition are desirable for the wider deployment of iris recognition systems. This thesis focuses on improving less constrained iris recognition. Five methods are proposed to improve the performance of different stages in less constrained iris recognition. First, a robust iris segmentation algorithm is developed using l1-norm regression and model selection. This algorithm formulates iris segmentation as robust l1-norm regression problems. To further enhance the robustness, multiple segmentation results are produced by applying l1-norm regression to different models, and a model selection technique is used to select the most reliable result. Second, an iris liveness detection method using regional features is investigated. This method seeks not only low level features, but also high level feature distributions for more accurate and robust iris liveness detection. Third, a signal-level information fusion algorithm is presented to mitigate the noise in less constrained iris captures. With multiple noisy iris captures, this algorithm proposes a sparse-error low rank matrix factorization model to separate noiseless iris structures and noise. The noiseless structures are preserved and emphasised during the fusion process, while the noise is suppressed, in order to obtain more reliable signals for recognition. Fourth, a method to generate optimal iris codes is proposed. This method considers iris code generation from the perspective of optimization. It formulates traditional iris code generation method as an optimization problem; an additional objective term modelling the spatial correlations in iris codes is applied to this optimization problem to produce more effective iris codes. Fifth, an iris weight map method is studied for robust iris matching. This method considers both intra-class bit stability and inter-class bit discriminability in iris codes. It emphasises highly stable and discriminative bits for iris matching, enhancing the robustness of iris matching. Comprehensive experimental analysis are performed on benchmark datasets for each of the above methods. The results indicate that the presented methods are effective for less constrained iris recognition, generally improving state-of-the-art performance.

In this dissertation, the author presents two projects regarding teaching strategies that apply to an intelligent tutoring system (ITS). The author applied multiple-solution teaching methods to the ITSs. The first project is an ITS that aims to help college students learn how to use Computer-Aided-Design (CAD) software, FreeCAD ITS. The second project is named SHiB ITS, and it seeks to help seniors learn the principles of finding the right spots to install a smart-home kit in their own house. In the two projects, the author experimented three teaching strategies, trial-and-error, practicing worked examples, and a combination of the two. The results of the post-test survey of the FreeCAD ITS show that college student participants like to learn by practicing multiple-solutions. Also, students in the combination teaching strategy group performed better in post-tests than the ones in the other two groups. The results of the post-test survey of the SHiB ITS shows that both older and younger adults like to learn in having exercises of multiple solutions. The older adults benefit more from the trial-and-error teaching strategy, while the younger adults benefit more from the combination teaching strategy. The experimental results agree to each other. The author then explored machine learning algorithms that uses an AI agent to teach another agent to perform a complicated task. The author proposed two algorithms that require a state importance input, a reinforcement learning (RL) algorithm that trains an RL teacher to teach, and an artificial neural network (ANN) that simulates behaviors of a teacher. Then the author applied the explored RL algorithm to train an RL teacher that knows how to perform tasks of SHiB ITS. The author added the trained teacher to the SHiB ITS and experimented the system with recruited participants. The results suggest that participants taught by the RL teacher performed better in the task that requires alternative solutions. In the post-survey, they significantly indicate that it is quick for them to learn to use the SHiB ITS with the RL teacher. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]

Dimensional measurement plays a critical role in product development and quality control. With the continuously increasing demand for tighter tolerances and more complex workpiece shapes in the industry, dimensional metrology often becomes the bottleneck of taking the quality and performance of manufacturing to the next level. As one kind of the most useful and powerful measuring instruments, coordinate measuring machines (CMMs) are widely employed in manufacturing industries. Since the accuracy and efficiency of a CMM have a vital impact on the product quality, productivity and manufacturing cost, the evaluation and improvement of CMM performance have always been important research topics since the invention of CMM. A novel Advanced Virtual Coordinate Measuring Machine (AVCMM) is proposed against such a background. The proposed AVCMM is a software package that provides an integrated virtual environment, in which user can plan inspection strategy for a given task, carry out virtual measurement, and evaluate the uncertainty associated with the measurement result, all without the need of using a physical machine. The obtained estimate of uncertainty can serve as a rapid feedback for user to optimize the inspection plan in the AVCMM before actual measurement, or as an evaluation of the result of a performed measurement. Without involving a physical CMM in the inspection planning or evaluation of uncertainty, the AVCMM can greatly reduce the time and cost needed for such processes. Furthermore, as the package offers vivid 3D visual representation of the virtual environment and supports operations similar to a physical CMM, it does not only allow the user to easily plan and optimise the inspection strategy, but also provide a cost-effective, risk-free solution for training CMM operators. A modular, multitier architecture has been adopted to develop the AVCMM system, which incorporates a number of functional components covering CMM and workpiece modelling, error simulation, inspection simulation, feature calculation, uncertainty evaluation and 3D representation. A new engine for detecting collision/contact has been developed and utilized, which is suitable for the virtual environment of simulated CMM inspections. A novel approach has been established to calculate errors required for the error simulation, where the data are obtained from FEA simulations in addition to conventional experimental method. Monte Carlo method has been adopted for uncertainty evaluation and has been implemented with multiple options available to meet different requirements. A prototype of the proposed AVCMM system has been developed in this research. Its validity, usability and performance have been verified and evaluated through a set of experiments. The principles for utilising the AVCMM in practical use have also been established and demonstrated. The results have indicated that the proposed AVCMM system has great potentials to improve the functionalities and overall performance of CMMs.

The use of Machine Learning for malware detection is essential to counter the massive growth in malware types compared with the traditional signature-based detection system. However, machine learning models could also be extremely vulnerable and sensible to transferable adversarial example (AE) attacks. The transfer AE attack does not require extra information from the victim model such as gradient information. Researchers explore mainly 2 lines of transfer-based adversarial example attacks: ensemble models and ensemble samples. \\ Although comprehensive innovations and progress have been achieved in transfer AE attacks, few works have investigated how these techniques perform in malware data. Besides, generating adversarial examples on an android APK file is not as easy and convenient as it is on image data since the generated AE of malware should also remain its functionality and executability after perturbation. Therefore, it is urgent to validate whether previous methodologies could still have their effect on malware considering the differences compared to image data. \\ In this thesis, we first have a thorough literature review for the AE attacks on malware data and general transfer AE attacks. Then we design our algorithm for the transfer AE attack. We formulate the optimization problem based on the intuition that the contribution evenness of features towards the final prediction result is highly correlated to the AE transferability. We then solve the optimization problem by gradient descent and evaluate it through extensive experiments. Implementing and experimenting with the state-of-the-art AE algorithms and transferability enhancement techniques, we analyze and summarize the weaknesses and strengths of each method.

Fundamental understanding of the heat generation process of lithium-ion batteries during operations is crucial for securing lifespan and safety by the cost-effective and efficient design of thermal management systems. Heat generation rate (HGR) of lithium-ion batteries varies at different operation conditions, such as charge and discharge rates, state of charge (SOC), temperatures and degradation conditions. In the first part of this work, a low-cost and high-performance multifunctional calorimeter is firstly developed. The calorimeter uses the thermoelectric assemblies (TEAs) as hardware and accompanied with a feedback loop, which enables the dynamic measurement of HGR and the active temperature control. The HGR performance of a large format lithium-ion cells is measured and compared as a function of charge and discharge rates, SOC, temperatures and degradation conditions, and the associated energy efficiency is analyzed. These works are presented as the first and second part of the dissertation. The sources of HGR within lithium-ion cells are predominantly classified as reversible heat and irreversible heat. The reversible heat is generated by a change in entropy during the electrochemical reactions and can be estimated using the entropy coefficient. The irreversible heat is caused by the resistances that represent concentration, activation, and Ohmic polarizations. In the third part, we developed several novel experimental techniques that facilitate the fast and accurate characterization of the two heat source terms, which include (1) accelerated equilibration method, (2) hybridized time-frequency domain analysis (HTFDA) method, and (3) improved frequency-domain calorimetric method, and (4) wavelet-transform based simultaneous and continuous characterization method. The results are compared with those measured by the conventional experimental methods, and show advantages with respect to measurement time and accuracy. In order to further explore the heat generation mechanism within lithium-ion cells, an electrochemical-thermal life model is developed and validated. The electrochemical model describes the cell’s internal reaction mechanisms such as the mass transport, charge conservation, and electrochemical kinetics; the degradation model describes the aging mechanisms including the solid electrolyte interphase (SEI) layer formation and lithium plating; while the associated HGRs are modeled by the coupled thermal model. Based on the developed model, the heat generation and the associated mechanism can be analyzed for both fresh and aged cells. This work is presented in the fourth part of the dissertation. As a closing work to the dissertation, we further proposed an improved battery electrochemical model by considering a SOC-dependent diffusion coefficient lithium ions in cathode. The model has been validated to show a drastic increase of the accuracy in predicting the terminal voltage, while maintaining low computational time. The work may provide guidelines for further improvement and optimization of the battery model.

Grain boundaries are important planar defects which influence a variety of bulk properties, while the structures and behaviors of grain boundaries can be altered by solute atoms to enable the design of beneficial material performance. Different interactions between solutes and grain boundaries exist. One example is that under certain conditions, solutes travel to grain boundaries and change grain boundary chemistry and structure, to vary the so-called “grain boundary complexion” state. Current studies on grain boundary complexions have been primarily limited to binary alloys. While in ternary or quaternary systems, multi-element segregation might be utilized to create thicker amorphous intergranular films (AIFs), a structural feature that helps toughen nanocrystalline alloys. Another example is the migration of grain boundaries at the presence of solutes, a situation in which there are both segregated dopants on boundaries and dopants as solid solution additions in the surrounding crystal. A promising dopant concentration might be found at which grain boundary motion in alloys can be changed and therefore leads to optimal microstructure beneficial for material performance. To provide guidance to experiments as well as explaining the physical mechanisms responsible for the experimental observations, atomistic modeling is chosen to address the challenges mentioned above. In this thesis, certain criteria for selecting appropriate interatomic potentials for modeling are first established. Then, the interfacial segregation and structural transition behavior in ternary alloys have been investigated and the possibility of forming thicker AIFs in certain ternary alloys by controlling the ratio of different dopant elements has been demonstrated, which could open up the opportunity of tuning the structure and properties of materials by co-doping. In the second part of the thesis, the migration of twin boundaries, basal-prismatic interfaces and conjugate twin boundaries (also referred to as twin tips in some cases) in pure Mg and Mg alloys has been studied. By understanding the effect of segregated dopants as well as randomly distributed dopants in the matrix on the motion of a group of special boundaries, it is able to alter the microstructure evolution by alloying and further tune the mechanical properties of Mg alloys.

Over 1000 photovaltaic (PV) systems' lifetime performance changes under real-world operation conditions were studied in pursuit of a better understanding of the impact of external variables on the durability of PV systems. These PV systems are part of the SDLE Research Center's global sunfarm network (GSFN), which has developed to include 787 PV power plants and outdoor test facilities. The analysis of this large PV data set was driven by a non-relational data warehouse for multiple heterogeneous energy data, referred to as Energy-CRADLE. The rate of power change was calculated using a month-by-month approach, explicitly, a linear regression model (the ß model) was fitted to each 30 day increment of data to train a predictive model for the output power of PV systems. The system's power output was then predicted under a selected environmental condition for each 30 day segment. A second linear or piecewise linear regression model (the ¿ model) was fitted to the monthly predicted power output values, with a weight on the inverse of the error of the prediction. For 655 PV systems where the monthly predicted values were well explained by a linear model, the annual rate of change were calculated from the slope of the linear regression fitting. These 655 systems were between two and eight years old, they were distributed in 9 different climate zones according to Koppen-Geiger Climate classification, the PV modules being used were from 20 different PV manufacturers, and the inverters being used were from 9 different manufacturers. A third multiple regression model (the ¿ model) was fitted to the 655 rate of change results. The variables being considered in the third model exhausted all external variables of the PV systems. Through a statistical variable selection method, stepwise Akaike information criterion (AIC), the variables that had statistically significant influence on the system change rate were selected and rank ordered by their contribution follows: Koppen-Geiger climate zones , PV module manufacturer, system age (in months), and the mounting method (roof, ground, and canopy) of the system. From the piecewise fitting of the predicted power output, two competing phenomena were observed, one was an initial performance increase that appeared over a timespan of several months to over a year, the other was the performance degradation that lasted over the system's lifetime. As a result of these two competing phenomena, a PV system's performance exhibits change point behavior, experiencing an initial increase in power at the beginning of its service life, and eventually change to power loss (power degradation) over time.

I will introduce a group lasso algorithm for complex variables and demonstrate its application to a novel time series model called tensor autoregression (T-AR). T-AR utilizes the t-product tensor operation on 3-dimensional tensors and models time series exhibiting seasonality and geometric trend. The tensor structure of T-AR enables historical information from a selection of lag durations to simultaneously affect the prediction of a single series. I will first introduce the topic of tensor computations and motivate the t-product and T-SVD manipulations. Then, I will derive the T-AR model from the t-product definition and discuss its properties and interpretation. Next, I will adapt a group lasso algorithm to complex-valued problems and derive the fast algorithm for lag selection in T-AR. Finally, I will conclude with simulation results.

In pursuit of a higher fidelity understanding of the long-term degradation of long-lived technologies, such as photovoltaic (PV) systems, the framework of Lifetime and Degradation Science (L&DS) goes beyond initial qualification tests and investigates the underpinning mechanisms of degradation. L&DS concerns itself with the complex and multivariate signatures of the degradation process and uncovering the fundamental physical mechanisms contributing to that degradation. In the case of PV modules, this effort requires extensive continuous monitoring of PV modules' power production and climatic conditions. The responses of PV module to the stressors of the real world is cross-correlated to the simulated and accelerated stressors placed on devices in a laboratory setting. A unique, highly instrumented, outdoor test facility for PV materials, components, and systems, the Solar Durability and Lifetime Extension (SDLE) center's SunFarm, was built for the purpose of better understanding the power degradation mechanisms of PV modules and materials. The SDLE SunFarm provides an apparatus for the collection of real-world time series data consisting of output power, weather and insolation metrology. The SunFarm is comprised of 122 individual PV power plants, including 120 module-level plants and 2*8 modules, string-level plants. Output power is monitored through appropriate grid-tied inverters. The metrology package developed at CWRU for the collection of time series data provides a model to be implemented at external sites around the globe. In order to expand the ability of monitoring PV systems' performance under different climatic conditions, a global SunFarm Network was implemented among nine outdoor test facilities around the world in collaboration with academic institutions and industrial partners including commercial power plants. This thesis provides the initial data analytics on the first six months of data from 60 PV modules on the SDLE SunFarm, and serves as a model for the analytics of full dataset from the global SunFarm Network. The data was first validated by characterization of the measurement apparatus, redundancy of measurement, and time-slewing according to minimization of the time cross-correlation function using a free and open-source statistical software language and packages known as "R". Using R (v3.0.1) for clustering data analysis base upon unfiltered AC power time series showed that the data fell into six clusters, which represented the six different electrical sites of SDLE SunFarm. The data were intelligently assembled and subsampled around solar noon time. PV performance ratio (PR), which is a measure of PV modules' output at given incident power from sunlight, was used as a indicator of modules' working effectiveness. Correlations among the filtered sub-set of solar noon time PR data were discerned with hierarchical clustering analysis. K-means clustering was used to confirm the optimum number of clusters for the analytics. The clustering results differentiate modules on different physical sites, pointed out malfunctions of the PV mounting system, and incapacity of certain module brands. These results are useful for correlating different modules' response to stressors and those stressors' effects on overall performance.

An improved synthetic strategy towards the fully functionalized ABC subunit 41 was developed. The optically active intermediate 43, which is expeditiously assembled from 5-oxobornyl pivalate by the sequential implementation of an oxy-Cope rearrangement and an intramolecular ene reaction, proved to be suitably functionalized for ultimate conversion to 41. The synthesis plan relies on two approaches to this targeted intermediate. In the first, the exocyclic double bond introduced during EtAlCl2-promoted closure of aldehyde 44 is cleaved to leave a carbonyl group that is amenable to hydride reduction and elimination of water. Cleavage of the resulting double bond with ruthenium tetroxide provided the seco ketoacid. The same advanced intermediate was obtained by initially positioning the double bond internal to the five-membered ring in advance of transient ring expansion via diketone formation and intramolecular aldolization. Both of these approaches bypass the complications arising from the substantial steric congestion prevailing in these structural networks. The task of covalently positioning an oxygen atom adjacent to the gem-dimethyl-substituted carbon in 41 was properly realized by oxidative decarboxylation.The other optically active building block 40 was synthesized by utilizing CBS asymmetric reduction of cyclopentanone 92, which was easily accessed by aldol cyclization followed by Mukaiyama-type aldol reaction. The stereochemical assignment to 40 was determined by the well-defined transition state of CBS reduction and 2D-NMR analysis.Direct coupling of the ABC and EF-ring fragments was not successful. However, coupling of 133 and 134 was achieved via carbodiimidine coupling. Attempts to perform the Ireland-Claisen rearrangement within 132 resulted in the formation of 139 through an E1cb mechanism.