Search

Smart meters implemented at the end-user in the energy sector create the opportunity to develop data-intelligent methods for district heating systems by using a large amount of fine-granular heat consumption time series from end-users. The current state-of-the-art methods for temperature control in district heating systems rely on predefined critical points in the network and a set reference temperature curve that expresses the minimum forward temperature as a function of the outdoor temperature at the end-users. The critical points are used to ensure that the consumers’ supply temperature requirements are met all times. To predefine the critical points at the network, the location of the lowest temperature in the grid needs to be identified at any point in time. Since the lowest temperature often varies over time, one must have a set of critical points in a district heating network. This paper proposes a method to estimate the temperature at an artificial critical point for the network using time-wise quantile estimation using smart meter data at end-users. This novel approach eliminates the need for physical critical points in the net or sensors in wells and creates the possibility of changing the critical point location if needed. The benefits for the provider of using smart meters as feedback, makes the measurement wells redundant and flexibility of the location. The location of low temperature areas in the network can change overtime hence the flexibility of being able to change where the feedback is located. The proposed method to replace the well measurements to provide feedback for temperature control at the production site groups a predefined set of smart meter readings together for each point in time. The grouping is done to have reliable measurements from each smart meter device, excluding some of the meters when a faulty reading occurs. The set of acceptable readings is used to estimate the street pipe’s temperature using the estimated quantile of the forward temperature. The approach is tested on two demo cases. The first demo consists of smart meters to estimate the forward temperature of the main street pipe. The second demo uses three smart meters at large apartment buildings as feedback for the control. Initial results show that the estimated temperature of the network can replace the well-measurements which traditionally are used as feedback for temperature control and give a better and more flexible control.

Combining forecasts from multiple temporal aggregation levels exploits information differences and mitigates model uncertainty, while reconciliation ensures a unified prediction that supports aligned decisions at different horizons. It can be challenging to estimate the full cross-covariance matrix for a temporal hierarchy, which can easily be of very large dimension, yet it is difficult to know a priori which part of the error structure is most important. To address these issues, we propose to use eigendecomposition for dimensionality reduction when reconciling forecasts to extract as much information as possible from the error structure given the data available. We evaluate the proposed estimator in a simulation study and demonstrate its usefulness through applications to short-term electricity load and financial volatility forecasting. We find that accuracy can be improved uniformly across all aggregation levels, as the estimator achieves state-of-the-art accuracy while being applicable to hierarchies of all sizes.

Due to the carbon neutral goals in many countries, a shift from traditional fuels to biomass is currently taking place in the energy sector. In this publication, we are looking at the long-term biomass contracting decisions for combined heat and power plants and power producers. A major share of biomass contracts are long-term contracts with runtimes of around 1 year, so the actual biomass demand is still uncertain when the contracts are negotiated. The operators can select different types of contracts ranging from fixed contracts with fixed amounts and deliveries to more flexible contracts and call options that allow for some flexibility in terms of amount and delivery times. We propose a stochastic program to optimize the contract selection including amounts and deliveries taking the biomass storage and uncertain demand into account. We present results of a case study from industry and show how the model utilizes the contracts for flexibility to adapt to different demand scenarios. Furthermore, the model is used for investigating the tradeoff between storage restrictions and fulfilling the demand in all scenarios. We show why it is important to model this problem as a stochastic program and why considering an expected demand is not enough.

Model predictive control (MPC) has been applied in multiple ways to the load–frequency control problem. In this study, the authors illustrate and compare a target-adjusted MPC to a classical MPC formulation. The target-adjusted approach is also posed as optimal control law. The target-adjusted MPC is an alternative formulation that incorporates the system equilibrium into the control objective. The alternative derived controller can be used as substitute for classical MPCs.

The rapid growth of machine learning (black-box) techniques and computing capacity has started to transform many research domains, including building performance analysis. However, physical interpretation of these models remains a challenge due to their opaque nature. This paper outlines an experiment to unveil analytical expressions from an open-source machine-learning-based algorithm, i.e., symbolic regression. From 241 residential buildings in the Netherlands, 50 unique analytical expressions were produced demonstrating overall better characterization accuracies than an XGBoost baseline, while providing a powerful mean of interpretability from model structures and coefficients. These insights present a starting point for further work towards highly scalable models yielding new characterizations of residential buildings.

Zero-emission ships (ZE-ships) concept has been introduced as a promising solution in reducing the greenhouse gases (GHG) emission in marine shipping industry. Among different solutions, Fuel cells (FCs) are introduced as one of the most efficient technologies for providing the propulsion force of the ZE-ships. Energy storage systems (ESSs) are also used as auxiliary resource to cover the fast dynamics of the loads the the FCs are not able to supply. Design and operation problem of ZE-ships has been investigated in the literature from different viewpoints. This paper provides a survey on available studies in the field of cost effective energy management of FC/ESS based ZE-ships. To this end, first, different studies in the literature are categorized from the viewpoint of energy management strategies (EMSs) and discussed Then other categories of the works such as auxiliary energy resources, problem objectives, and simulation methods are also provided.

Probabilistic forecasting is becoming increasingly important for a wide range of applications, especially for energy systems such as forecasting wind power production. A need for proper evaluation of probabilistic forecasts follows naturally with this, because evaluation is the key to improving the forecasts. Although plenty of excellent reviews and research papers on probabilistic forecast evaluation already exist, we find that there is a need for an introduction with some practical application. In particular, many forecast scenarios in energy systems are inherently multivariate, and while univariate evaluation methods are well understood and documented, only limited and scattered work has been done on their multivariate counterparts. This paper therefore contains a review of a selected set of probabilistic forecast evaluation methods, primarily scoring rules, as well as practical sections that explain how these methods can be calculated and estimated. In three case studies featuring simple autoregressive models, stochastic differential equations and real wind power data, we implement, apply and discuss the logarithmic score, the continuous ranked probability score and the variogram score for forecasting problems of varying dimension. Finally, the advantages and disadvantages of the three scoring rules are highlighted, and this provides a significant step towards deciding on an evaluation method for a given multivariate forecast scenario including forecast scenarios relevant for energy systems.

In the existing building stock, heating, cooling and ventilation usually run on fixed schedules, in many cases, even all day. In particular, ventilation systems often run with a constant air flow rate that is adjusted based on the assumption of maximum occupancy. Hence, reducing the operation to the required extent would offer energy potential. Model-based, demand-controlled heating, ventilation and air-conditioning systems can help to achieve this. Information on the number of occupants present in a room and ventilation-related quantities, such as the room-air change rate, are important parameters to control the ventilation of a building. Hence, an automated estimation of these would help to find optimal model-based control strategies. In this work, the use of a grey-box model based on a carbon dioxide mass balance is explored to estimate room occupancy and ventilation parameters. The main contribution of this study is the employment of stochastic differential equations to describe this mass balance. In contrast to ordinary differential equations, the stochastic framework employed here is able to address measurement errors as well as errors that derive from an inevitably oversimplified description of the physical system. Due to its probabilistic nature, this approach inherently includes a method of parameter estimation using the maximum likelihood approach, which additionally provides a measure of uncertainty for every estimated parameter. The presented model was tested in one naturally ventilated and one mechanically ventilated office room. In both cases, the estimation of occupancy and of the model parameters showed promising results. This leads to the conclusion that the suggested model can be considered as a candidate to be integrated into building control systems.

Demand response (DR) will be an inevitable part of the future power system operation to compensate for stochastic variations of the ever-increasing renewable generation. A solution to achieve DR is to broadcast dynamic prices to customers at the edge of the grid. However, appropriate models are needed to estimate the potential flexibility of different types of consumers for day-ahead and real-time ancillary services provision, while accounting for the rebound effect (RE). In this study, two RE models are presented and compared to investigate the behaviour of flexible electrical consumers and quantify the aggregate flexibility provided. The stochastic nature of consumers' price response is also considered in this study using chance constrained (CC) programming. (C) 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

The energy flexibility associated with energy consumption must be exploited to accommodate more fluctuating renewable energy. The only solution that enables this without violating privacy concerns is penalty-based control, where penalty signals are designed to give incentives for the consumers to adjust their demand according to the needs of the grid. Designing the penalty signals is a challenging task due to different flexibility potential offered by various energy consuming systems. In this paper, it is shown that the best utilisation of energy flexibility requires individual penalty signals tuned towards the energy flexibility of each consumer. Here, we present a very simple yet novel approach for designing such individual penalty signals for each consumer, such that the value of the energy flexibility is increased, for both the grid operators and the consumers. (C) 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

Contrary to electricity smart meter data analysis, little research regarding district heat smart meter data has been published. Previous papers on smart meter data analytics have not investigated autocorrelation in smart meter data. This paper examines district heat smart meter data from the largest district heat supplier in Denmark and autocorrelation is identified in the data. The K-Means algorithm is not able to take autocorrelation into account when clustering. We propose different data transformation methods to enable K-Means to account for this autocorrelation information in the data by using wavelet transformation and autocorrelation features. Our results show that the K-Means yield acceptable clustering results for district heat data when clustering normalized data, inclusion of autocorrelation improves the clustering. The clusters on normalized data are similar to the wavelet transformed clusters, where the autocorrelation has been accounted for. The clustering achieved with the autocorrelation transformation yields finer clusters through accounting for autocorrelation. We are not able to statistically show a difference between the transformations. All transformations result in shadowing clusters, but the autocorrelation transformation generates fewer shadow clusters and reduce the number of dimensions from 744 to 24, resulting in a dramatic reduction in K-Means runtime.

With the continuously increasing size of the wind turbine blades, the complexity of the casting process and the risk of failures has also increased. The IntegralBlades® vacuum assisted resin transfer moulding (VATRM) production process at the Siemens Gamesa Renewable Energy facility in Aalborg, Denmark, does not permit the visual inspection of the process. Hence a sensor system (possibly virtual) for process control and monitoring is highly prized. Furthermore, the effect of material handling, variations in permeability of the casting media and the material (epoxy) properties affect the outcome of the casting process. Therefore, it is necessary to analyse the effect of such variations at an early stage of the design process (e.g. during the simulations) of such a sensor system. Therefore, in this paper, we first describe an effective method to simulate the random changes in the permeability and viscosity in high-dimensional partial differential equations based model of the fluid flow. Next, a low-dimensional grey-box (cyber-physical) spatiotemporal model is proposed to capture the effect of random change in permeability and viscosity during the progression of the flow-front. Finally, a numerical case-study is presented demonstrating the effectiveness of the proposed methodology.

Heating and ventilation strategies in buildings can be improved significantly if information about the current presence and activity level of the occupants is taken into account. Therefore, there is a high demand for inexpensive sensor-based methods to detect the occupancy or occupant activity level. It is shown that the carbon dioxide (CO2) level in a room is dependent on the activity level rather than only on just the number of people. Therefore, this study suggests a new model based on the use of CO2 trajectories to estimate the occupant activity level, trained on measurements both from a school classroom and from a Danish summerhouse. A hidden Markov-switching model was employed to identify the activity level. This modelling approach is a generalization of hidden Markov models, taking autocorrelation in the observations into account. This is done by an additional autoregressive part which models the persistence of the CO2 concentration by relating the current value to past lags. The analysis of one-step prediction residuals shows that this method inherits the dynamics of the CO2 curves much better than an ordinary hidden Markov model, and can therefore be considered a promising candidate for occupant activity estimation. Furthermore, it is shown that the presented model can be used for simulations of activity level and of the accompanying CO2 levels.

In future smart energy systems, consumers are expected to change their load patterns as they become a significant source of flexibility. To ensure reliable load profile forecasts for long-term energy planning, conventional classification approaches will not hold and more advanced solutions are required. In this article, we propose an automatic, data-driven clustering methodology that accounts for heterogeneity in electricity consumers’ load profiles using unsupervised learning. We consider hourly load measurements from 9412 smart-meters from the commercial and industrial sector in Denmark. A wavelet transform is applied to min-max scaled load data, and the extracted wavelet coefficients are used as input to the K-means clustering algorithm. Through cluster validation, eight clearly distinct load profiles are identified and compared to the industry classification of the cluster constituents. Finally, the flexibility potential is traced for each cluster.

A key aspect for achieving a high-accuracy Photovoltaic (PV) power simulation, and reliable digital twins, is a detailed description of the PV system itself. However, such information is not always accurate, complete, or even available. This work presents a novel approach to learn features of unknown PV systems or subsystems using genetic algorithm optimization. Based on measured PV power, this approach learns and optimizes seven PV system parameters: nominal power, tilt and azimuth angles, albedo, irradiance and temperature dependency, and the ratio of nominal module to nominal inverter power (DC/AC ratio). By optimizing these parameters, we create a digital twin that accurately reflects the actual properties and behaviors of the unknown PV systems or subsystems. To develop this approach, on-site measured power, ambient temperature, and satellite-derived irradiance of a PV system located in south-west Germany are used. The approach proposed here achieves a mean bias error of about 10% for nominal power, 3° for azimuth and tilt angles, between 0.01%/C and 0.09%/C for temperature coefficient, and now-casts with an accuracy of around 6%. In summary, we present a new solution to parametrize and simulate PV systems accurately with limited or no previous knowledge of their properties and features.

DSF1305-00027B 8090-00069B EP/S001670/1 If CO2-emissions are to be reduced, the shares of renewable energy sources will have to be significantly increased. However, energy flexibility is required to cope with the increased share of renewable energy. Utilising it necessitates mathematical models of the operational response of energy flexible consumers. In this paper we present an accurate and general dynamic model of energy flexibility based on stochastic differential equations. The intuitive interpretation of the parameters is explained, to show the generality of the proposed model. To validate the approach, the parameters are estimated for three water towers and three buildings controlled by economic model predictive controllers. The model is then used to offer the energy flexibility on the current electricity market of Scandinavia, Nord Pool, using the so called “flexi orders”. Finally, the energy flexibility is used by controlling the demand of the water towers indirectly, through price signals designed based on the proposed model. Compared to having perfect foresight of electricity prices and future demand, between 63% and 98% of the potential savings were obtained in for these case studies. This shows that even without direct control of energy flexible systems, most of the potential can be reached under the current market conditions. publishersversion published

In power systems, electrical consumers can become a significant source of flexibility, by adjusting their consumption according to grid’s needs while respecting their operational constraints. Consumers’ flexibility potential can be exploited through the submission of dynamic electricity prices. Such prices are able to describe the variable condition of the power system and are broadcast to the consumers in order to obtain a certain change in consumption. The formulation of effective dynamic prices requires the development of proper models that describe the price responsiveness of electrical consumers. In this paper, we propose a nonlinear prediction model for the dynamic electricity prices in demand response (DR) programs. Specifically, the nonlinear auto-regressive with exogenous input (NARX) model structure is used to learn from available data to predict appropriate electricity price signals. For the validation of the model (in an aggregate manner) in predicting consumers’ price-response, the data from 10 Danish households is utilised, which has provided by the Danish Transmission Service Operator (TSO) Energinet.

Hidden Markov models are applied in many expert and intelligent systems to detect an underlying sequence of persistent states. When the model is misspecified or misestimated, however, it often leads to unrealistically rapid switching dynamics. To address this issue, we propose a novel estimation approach based on clustering temporal features while penalizing jumps. We compare the approach to spectral clustering and the standard approach of maximizing the likelihood function in an extensive simulation study and an application to financial data. The advantages of the proposed jump estimator include that it learns the hidden state sequence and model parameters simultaneously and faster while providing control over the transition rate, it is less sensitive to initialization, it performs better when the number of states increases, and it is robust to misspecified conditional distributions. The value of estimating the true persistence of the state process is illustrated through a simple trading strategy where improved estimates result in much lower transaction costs. Robustness is particularly critical when the model is part of a system used in production. Therefore, our proposed estimator significantly improves the potential for using hidden Markov models in practical applications.

In Europe, more and more data on building energy use will be collected in the future as a result of the energy performance of buildings directive (EPBD), issued by the European Union. Moreover, both at European level and globally it became evident that the real energy performance of new buildings and the existing building stock needs to be documented better. Such documentation can, for example, be done with data-driven methods based on mathematical and statistical approaches. Even though the methods to extract energy performance characteristics of buildings are numerous, they are of varying reliability and often associated with a significant amount of human labour, making them hard to apply on a large scale. A classical approach to identify certain thermal performance parameters is the energy signature method. In this study, an automatised, nonlinear and smooth approach to the well-known energy signature is proposed, to quantify key thermal building performance parameters. The research specifically aims at describing the linear and nonlinear heat usage dependency on outdoor temperature, wind and solar irradiation. To make the model scalable, we realised it so that it only needs the daily average heat use of buildings, the outdoor temperature, the wind speed and the global solar irradiation. The results of applying the proposed method on heat consumption data from 16 different and randomly selected Danish occupied houses are analysed.

Obtaining reliable, secure and efficient software under optimal resource allocation is an important objective of software engineering science. This work investigates the usage of classical and recent development paradigms of computational intelligence (CI) to fulfill this objective. The main software engineering steps asking for CI tools are: product requirements analysis and precise software specification development, short time development by evolving automatic programming and pattern test generation, increasing dependability by specific design, minimizing software cost by predictive techniques, and optimal maintenance plans. The tasks solved by CI are related to classification, searching, optimization, and prediction. The following CI paradigms were found useful to help software engineers: fuzzy and intuitionistic fuzzy thinking over sets and numbers, nature inspired techniques for searching and optimization, bio inspired strategies for generating scenarios according to genetic algorithms, genetic programming, and immune algorithms. Neutrosophic computational models can help software management when working with imprecise data.

The operation of electric and heat grids alike is complicated due to the dynamic demand, with the increasing penetration of renewable energy sources adding to the problem. In order to improve the integration of variable renewable energy sources, the flexibility of the system needs to be improved. This paper proposed a novel characterization of the short-term energy flexibility, which was further utilized for the district heating capacity extension. The soft-linking of the models includes feedback, but the added computational complexity is kept at a minimum. Compared to the other literature in the field, due to the accurate characterization of the dynamics of the energy flexibility, flexibility is utilized much more frequently. The method was demonstrated for the case of the district heating of Zagreb. Results showed that both capital and operational savings can be achieved by adopting the proposed method. In the best performing scenario, which included the capacity extension planning, the savings of the district heating system were 5.4%. The extensive power exchange in the best performing scenario meant that the flexibility was used to help balancing the power grid as well. Implementing flexibility into energy planning models : Soft-linking of a high- level energy planning model and a short-term operational model. © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/).

Smart grid operation schemes can integrate prosumers by offering economic rewards in exchange for the desired response. In order to activate prosumers appropriately, such operation schemes require models of the dynamic uncertain price-response relationships. In this study, we combine the system identification of nonlinear dynamics with control (SINDyc) algorithm with Bayesian inference techniques based on Markov-chain Monte-Carlo sampling. We demonstrate this combination of two algorithms on an exemplary system in order to obtain parsimonious models alongside parameter uncertainty estimates. The precision of the identified models depends on the identification experiment and the parameterization of the algorithms. Such models may characterize the prosumer response and its uncertainty, thereby facilitating the integration of such entities into smart grid operation schemes.

Portfolio selection involves a trade-off between maximizing expected return and minimizing risk. In practice, useful formulations also include various costs and constraints that regularize the problem and reduce the risk due to estimation errors, resulting in solutions that depend on a number of hyperparameters. As the number of hyperparameters grows, selecting their value becomes increasingly important and difficult. In this article, the authors propose a systematic approach to hyperparameter optimization by leveraging recent advances in automated machine learning and multiobjective optimization. They optimize hyperparameters on a train set to yield the best result subject to market-determined realized costs. In applications to single- and multiperiod portfolio selection, they show that sequential hyperparameter optimization finds solutions with better risk–return trade-offs than manual, grid, and random search over hyperparameters using fewer function evaluations. At the same time, the solutions found are more stable from in-sample training to out-of-sample testing, suggesting they are less likely to be extremities that randomly happened to yield good performance in training. TOPICS:Portfolio theory, portfolio construction, big data/machine learning Key Findings • The growing number of applications of machine-learning approaches to portfolio selection means that hyperparameter optimization becomes increasingly important. We propose a systematic approach to hyperparameter optimization by leveraging recent advances in automated machine learning and multiobjective optimization. • We establish a connection between forecast uncertainty and holding- and trading-cost parameters in portfolio selection. We argue that they should be considered regularization parameters that can be adjusted in training to achieve optimal performance when tested subject to realized costs. • We show that multiobjective optimization can find solutions with better risk–return trade-offs than manual, grid, and random search over hyperparameters for portfolio selection. At the same time, the solutions are more stable across in-sample training and out-of-sample testing.

The increasing proportion of renewable energy sources in power grids leads to challenges concerning balancing production and consumption. One solution to this grid challenge is to utilize demand-side flexibility. To use the full potential of demand-side flexibility, dynamical models and optimal control methods must be used. This paper demonstrates how demand-side flexibility can be enabled for a refrigeration system using an add-on ice-tank module to actively curtail the refrigeration system and thereby leveraging time-varying power prices. The operation of the ice-tank is the solution to an optimal control problem that minimize the integrated electricity costs. This optimal control problem is solved numerically and the performance of the strategy is successfully tested in a real experiment where cost-savings of approximately 20% are observed (compared to not having an ice-tank available). The proposed optimal control strategy is tested in real-life experiments spanning over 24 h. The dynamical relation between the operation of the ice-tank and the power consumption (the compressor capacity) is modelled using stochastic differential equations. This differential equation model is calibrated on 13 h of training data using the continuous–discrete Kalman filter and the maximum likelihood framework.

The large penetration rate of renewable energy sources leads to challenges in planning and controlling the energy production, transmission, and distribution in power systems. A potential solution is found in a paradigm shift from traditional supply control to demand control. To address such changes, a first step lays in a formal and robust characterization of the energy flexibility on the demand side. The most common way to characterize the energy flexibility is by considering it as a static function at every time instant. The validity of this approach is questionable because energy-based systems are never at steady-state. Therefore, in this paper, a novel methodology to characterize the energy flexibility as a dynamic function is proposed, which is titled as the Flexibility Function. The Flexibility Function brings new possibilities for enabling the grid operators or other operators to control the demand through the use of penalty signals (e.g., price, CO2, etc.). For instance, CO2-based controllers can be used to accelerate the transition to a fossil-free society. Contrary to previous static approaches to quantify Energy Flexibility, the dynamic nature of the Flexibility Function enables a Flexibility Index, which describes to which extent a building is able to respond to the grid’s need for flexibility. In order to validate the proposed methodologies, a case study is presented, demonstrating how different Flexibility Functions enable the utilization of the flexibility in different types of buildings, which are integrated with renewable energies. © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/)

In this paper, we present and compare different methods for computing the likelihood function and its gradient. We consider nonlinear continuous-discrete models described by a system of stochastic differential equations (SDEs) with discrete-time measurements. The problem of maximum likelihood estimation (MLE) is formulated as a nonlinear program (NLP) and it is solved numerically using a gradient-based single shooting algorithm. The estimates of the mean and its covariance are computed using a continuous-discrete extended Kalman filter (CDEKF). We derive analytical expressions for the gradient of the likelihood function. We discuss some aspects of the implementation of MLE for non-stiff systems. In particular, we present an efficient way of computing the state covariance matrix and its gradient using explicit Runge-Kutta schemes. We verify our implementation using a numerical example related to type 1 diabetes and demonstrate how to apply it for nonlinear parameter estimation.

We report the closed-loop performance of adaptive model predictive control (MPC) algorithms for a dual-hormone artificial pancreas (AP) intended for patients with type 1 diabetes. The dual-hormone AP measures the interstitial glucose concentration using a subcutaneous continuous glucose monitor (CGM) and administers glucagon and rapid-acting insulin subcutaneously. The discrete-time transfer function models used in the insulin and glucagon MPCs comprise a deterministic part and a stochastic part. The deterministic part of the MPC model is individualized using patient-specific information and describes the glucose-insulin and glucose-glucagon dynamics. The stochastic part of the MPC model describes the uncertainties that are not included in the deterministic part of the MPC model. Using closed-loop simulation of the MPCs, we evaluate the performance obtained using the different deterministic and stochastic models for the MPC on three virtual patients. We simulate a scenario including meals and daily variations in the model parameters for two settings. In the first setting, we try five different models for the deterministic part of the MPC model and use a fixed model for the stochastic part of the MPC model. In the second setting, we use a second-order model for the deterministic part of the MPC model and estimate the stochastic part of the MPC model adaptively. The results show that the controller is robust to daily variations in the model parameters. The numerical results also suggest that the deterministic part of the MPC model does not play a major role in the closed-loop performance of MPC. This is ascribed to the availability of feedback and the poor prediction capability of the model, i.e. the large disturbances and model-patient mismatch. Moreover, a second order adaptive model for the stochastic part of the MPC model offers a marginally better performance in closed-loop, in particular if the model-patient mismatch is large.

This paper presents an individualized model predictive control (MPC) algorithm for overnight blood glucose stabilization in people with type 1 diabetes (T1D). The MPC formulation uses an asymmetric objective function that penalizes low glucose levels more heavily. We compute the model parameters in the MPC in a systematic way based on a priori available patient information. The model used by the MPC algorithm for filtering and prediction is an autoregressive integrated moving average with exogenous input (ARIMAX) model implemented as a linear state space model in innovation form. The control algorithm uses frequent glucose measurements from a continuous glucose monitor (CGM) and its decisions are implemented by a continuous subcutaneous insulin infusion (CSII) pump. We provide guidelines for tuning the control algorithm and computing the Kalman gain in the linear state space model in innovation form. We test the controller on a cohort of 100 randomly generated virtual patients with a representative inter-subject variability. We use the same control algorithm for a feasibility overnight study using 5 real patients. In this study, we compare the performance of this control algorithm with the patient's usual pump setting. We discuss the results of the numerical simulations and the in vivo clinical study from a control engineering perspective. The results demonstrate that the proposed control strategy increases the time spent in euglycemia.

Heat load forecasts are crucial for energy operators in order to optimize the energy production at district heating plants for the coming hours. Furthermore, forecasts of heat load are needed for optimized control of the district heating network since a lower temperature reduces the heat loss, but the required heat supply at the end-users puts a lower limit on the temperature level. Consequently, improving the accuracy of heat load forecasts leads to savings and reduced heat loss by enabling improved control of the network and an optimized production schedule at the plant. This paper proposes the use of temporal hierarchies to enhance the accuracy of heat load forecasts in district heating. Usually, forecasts are only made at the temporal aggregation level that is the most important for the system. However, forecasts for multiple aggregation levels can be reconciled and lead to more accurate forecasts at essentially all aggregation levels. Here it is important that the auto- and cross-covariance between forecast errors at the different aggregation levels are taken into account. This paper suggests a novel framework using temporal hierarchies and adaptive estimation to improve heat load forecast accuracy by optimally combining forecasts from multiple aggregation levels using a reconciliation process. The weights for the reconciliation are computed using an adaptively estimated covariance matrix with a full structure, enabling the process to share time-varying information both within and between aggregation levels. The case study shows that the proposed framework improves the heat load forecast accuracy by 15% compared to commercial state-of-the-art operational forecasts.

The purpose of this study is to develop a method for detecting and compensating the anomalies of continuous glucose monitoring (CGM) sensors as well as detecting unannounced meals. Both features, sensor fault detection/correction and meal detection, are necessary to have a reliable artificial pancreas. The aim is to investigate the best detection results achievable with the proposed detection configuration in a perfect situation, and to have the results as a benchmark against which the imperfect scenarios of the proposed fault detection can be compared. The perfect situation that we set up here is in terms of a patient simulation model, where the model in the detector is the same as the patient simulation model used for evaluation of the detector. The detection module consists of two CGM sensors, two fault detectors, a fault isolator, and an adaptive unscented Kalman filter (UKF). Two types of sensor faults, i.e., drift and pressure induced sensor attenuation (PISA), are simulated by a Gaussian random walk model. Each of the fault detectors has a local UKF that receives the signal from the associated sensor, detects faults, and finally tunes the adaptive UKF. A fault isolator that accepts data from the two fault detectors differentiates between a sensor fault and an unannounced meal appearing as an anomaly in the CGM data. If the fault isolator indicates a sensor fault, a method based on the covariance matching technique tunes the covariance of the measurement noise associated with the faulty sensor. The main UKF uses the tuned noise covariances and fuses the CGM data from the two sensors. The drift detection sensitivity and specificity are 80.9% and 92.6%, respectively. The sensitivity and specificity of PISA detection are 78.1% and 82.7%, respectively. The fault detectors can detect 100 out of 100 simulated drifts and 485 out of 500 simulated PISA events. Compared to a nonadaptive UKF, the adaptive UKF reduces the deviation of the CGM measurements from their paired blood glucose concentrations from 72.0% to 12.5% when CGM is corrupted by drift, and from 10.7% to 6.8% when CGM is corrupted by PISA. The fault isolator can detect 199 out of 200 unannounced meals. The average change in the glucose concentrations between the meals and the detection time points is 46.3 mg/dL.

Thermostatically controlled loads, including heat pumps, electrical space heaters, air-conditioners, electrical boilers, and refrigerators, have a large potential to provide regulation services to energy systems. In this paper, we design a price-responsive control system for aggregating and coordinating a population of thermostatically controlled loads, each with a relatively small power consumption. The goal is to shift electricity consumption of the whole group to the periods with low electricity prices. Model predictive control is an advanced control method, which can naturally include electricity price and weather forecasts as well as the dynamics of the loads into the control problem. The main challenge is to find an accurate aggregate model describing demand response of the population. The proposed solution is a nonlinear modification of bin state transition model, which is further used for formulating nonlinear mixed integer optimization problem. The control objective is to minimize operational cost of the population. As a case study, we consider a population of electrical space heaters. The numerical simulations demonstrate the accuracy of the obtained model and that the proposed control system reduces the operational cost up to 20%, whereas the customer comfort is not compromised.

The challenges to optimally utilize weather dependent renewable energy sources call for powerful tools for forecasting. This paper presents a non-parametric machine learning approach used for multi-site prediction of solar power generation on a forecast horizon of one to six hours. Historical power generation and relevant meteorological variables related to 42 individual PV rooftop installations are used to train a gradient boosted regression tree (GBRT) model. When compared to single-site linear autoregressive and variations of GBRT models the multi-site model shows competitive results in terms of root mean squared error on all forecast horizons. The predictive performance and the simplicity of the model setup make the boosted tree model a simple and attractive compliment to conventional forecasting techniques.

In this paper, we discuss overnight blood glucose stabilization in patients with type 1 diabetes using a Model Predictive Controller (MPC). We compute the model parameters in the MPC using a simple and systematic method based on a priori available patient information. We describe and compare 3 different model structures. The first model structure is an autoregressive integrated moving average with exogenous input (ARIMAX) structure. The second model structure is an autoregressive moving average with exogenous input (ARMAX) model, i.e. a model without an integrator. The third model structure is an adaptive ARMAX model in which we use a recursive extended least squares (RELS) method to estimate parameters of the stochastic part. In addition, we describe some safety layers in the control algorithm that improve the controller robustness and reduce the risk of hypoglycemia. We test and compare our control strategies using a virtual clinic of 100 randomly generated patients with a representative inter-subject variability. This virtual clinic is based on the Hovorka model. We consider the case where only half of the meal bolus is administered at mealtime, and the case where the insulin sensitivity increases during the night. The numerical results suggest that the use of an integrator leads to higher occurrence of hypoglycemia than for the controllers without the integrator. Compared to other control strategies, the adaptive MPC reduces both the time spent in hypoglycemia and the time spent in hyperglycemia.

The deployment of technological pilots is of paramount importance for testing and demonstrating the technical feasibility of the concepts described in this book for two main reasons. On the one hand, there are few real-life experiences in the application of these concepts, as TSO-DSO coordination is a relatively new topic. On the other hand, there may be some implementation difficulties which cannot be anticipated by the scenario analysis and CBA. Whenever possible, more than one pilot should be deployed, so that each of them can focus on different parts of the TSO-DSO coordination value chain. In that sense, it is important to demonstrate different potential TSO-DSO coordination schemes, so that issues arising from each of them can be identified. Moreover, it is also important to demonstrate different types of DER, so that their flexibilities can be better assessed and the advantages and disadvantages for real-life implementation can be properly identified and addressed. As a third complementarity aspect, having different technological pilots allows for focusing on different parts of the value chain, so that one of them may take the vision of the TSO or DSO, while another one can focus on the needs of the aggregator or DER owners. European Comisión H2020, n. 691405, SmartNet

In this paper we present simulations of Economic Model Predictive Control (EMPC) for control of the energy system in a smart home. The energy system in a smart home consists of a stationary battery, photo voltaic solar cells on the roof, a heat pump for heating, and an electrical vehicle battery that is charged. Using weather forecasts, thermal comfort and driving profiles, the EMPC coordinates this energy system and its interaction with the external energy system by purchasing and selling electricity at prices that are announced in advance. The EMPC is a linear program with soft constraints for the output constraints and an objective function that represents the cost of energy. In contrast to existing methods, the key novelties in the present paper is the use of multi-level soft constraints and an objective function that accounts not only for the cost of energy used during the prediction horizon but also for the value of energy stored at the end of the prediction horizon. We demonstrate by simulation, that these novelties are important for well-behaved closed-loop performance of the EMPC.

Demand flexibility will be an inevitable part of the future power system operation to compensate stochastic variations of ever-increasing renewable generation. One way to achieve demand flexibility is to provide time-varying prices to customers at the edge of the grid. However, appropriate models are needed to estimate the potential flexibility of different types of consumers for day-ahead and real-time ancillary services (AS) provision. The proposed method should account for rebound effect and variability of the customers' reaction to the price signals. In this study, an efficient algorithm is developed for consumers' flexibility estimation by the transmission system operator (TSO) based on offline data. No aggregator or real-time communication is involved in the process of flexibility estimation, although real-time communication channels are needed to broadcast price signals to the end-users. Also, the consumers' elasticity and technical differences between various types of loads are taken into account in the formulation. The problem is formulated as a mixed-integer linear programming (MILP) problem, which is then converted to a chance-constrained programming to account for the stochastic behaviour of the consumers. Simulation results show the applicability of the proposed method for the provision of AS from consumers at the TSO level.

We propose a bidding method for the participation of combined heat and power (CHP) units in the day-ahead electricity market. More specifically, we consider a district heating system where heat can be produced by CHP units or heat-only units, e.g., gas or wood chip boilers. We use a mixed-integer linear program to determine the optimal operation of the portfolio of production units and storages on a daily basis. Based on the optimal production of subsets of units, we can derive the bidding prices and amounts of electricity offered by the CHP units for the day-ahead market. The novelty about our approach is that the prices are derived by iteratively replacing the production of heat-only units through CHP production. This results in an algorithm with a robust bidding strategy that does not increase the system costs even if the bids are not won. We analyze our method on a small realistic test case to illustrate our method and compare it with other bidding strategies from literature, which consider CHP units individually. The analysis shows that considering a portfolio of units in a district heating system and determining bids based on replacement of heat production of other units leads to better results.

The EU is committed to reducing greenhouse gas emissions to 80-95% below 1990 levels by 2050. To accomplish this goal the share of renewable energy sources have to increase significantly. Challenges related to keeping a high security of supply have raised while the share of intermittent renewable electricity generation is increasing. The increasing level of RES production leads to higher volatility in market prices. The main challenges lay within the modeling and simulation of the random variables. Therefore, the assessment of the operation of the entire micro-grid should be studied while constructing different test scenarios under uncertainty of the market prices and micro-grid configuration. This paper constructs different tests to investigate the integration of Electric Energy Storage unit technologies in a micro-grid. The assessment of the impact on the micro-grid, like congestion or voltage problem, are analyzed when the EES units are owned by a separate business entity called aggregator or a grid owner.

In Microgrids with uncertain production storages are valuable assets to facilitate system stabilization. Consequently, electric vehicles (EVs) are promising for providing prosumer services. EVs are assets driven by human behavior, consequently they can rarely be directly controlled. However, indirect control approaches are considered promising regarding their integration into system controls. In this paper we consider a hierarchy of optimized system controls including indirect control approaches in order to leverage flexibility potential associated with EVs. (C) 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

This paper develops a two-stage stochastic and dynamically updated multi-period mixed integer linear program (SD-MILP) for optimal coordinated bidding of an electric vehicle (EV) aggregator to maximize its profit from participating in competitive day-ahead, intra-day and real-time markets. The hourly conditional value at risk (T-CVaR) is applied to model the risk of trading in different markets. The objective of two-stage SD-MILP is modeled as a convex combination of the expected profit and the T-CVaR hourly risk measure. When day-ahead, intra-day and real-time market prices and fleet mobility are uncertain, the proposed two-stage SD-MILP model yields optimal EV charging/discharging plans for day-ahead, intra-day and real-time markets at per device level. The degradation costs of EV batteries are precisely modeled. To reflect the continuous clearing nature of the intra-day and real-time markets, rolling planning is applied, which allows re-forecasting and re-dispatching. The proposed two-stage SD-MILP is used to derive a bidding curve of an aggregator managing 1000 EVs. Furthermore, the model statistics and computation time are recorded while simulating the developed algorithm with 5000 EVs.

Real energy performance of new and retrofitted buildings often consistently differs from expectations. While occupants might complain about poor indoor climate, the energy use in such buildings is often higher than expected, leading to the well-known phenomenon called “Energy Performance Gap”. In the past years, monitoring of buildings, both in terms of energy use and indoor climate conditions, was realised mostly for office buildings only, and at high financial costs. However, the exponential growth in availability of IoT devices, over the last years, opens now new scenarios for low-cost monitoring and control solutions of buildings. Yet, modern IoT devices are often only accessible online through the vendors’ software, although some devices make use of open communication protocols and can, therefore, be connected to open platforms such as openHAB. However, the use of open platforms is still connected to a big efforts for many final users. We, therefore, propose climify.org, an open platform for plug and play connection of IoT sensors and actuators, for easy monitoring and controlling of buildings and buildings’ HVAC systems. The platform climify.org offers, at time of writing, three main applications. The first application is an IoT device installation app, to be used on portable devices (e.g. mobile phones or tablets of system administrators): this app allows easily installing and locating a sensor or an actuator, within a building. The second application is an online service for data visualisation and HVAC control: while the monitoring data can be plotted, the service offers several data evaluation methods; moreover, the settings of the connected actuators can be modified and controlled. The third application can be installed on portable devices (mobile phones and tablets of buildings’ occupants) and allows occupants to provide feedback on their perception of the indoor climate through several questionnaires’ formats. Through the three applications developed within climify.org, we aim at providing the best indoor climate and the lowest energy use through a low-cost solution. © The Authors, published by EDP Sciences, 2019 Licence Creative Commons This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

It is widely accepted today that probabilistic forecasts of wind power production constitute valuable information that can allow both wind power producers and power system operators to exploit this form of renewable energy economically, while mitigating the potential adverse effects relating to its variable and uncertain nature. In order to provide reliable wind power forecasts for periods beyond a couple of hours, forecasts of the wind speed are fundamental. In this paper, we propose a modeling framework for wind speed that is based on stochastic differential equations. We show that stochastic differential equations allow us to capture the time dependence structure of wind speed prediction errors naturally (from 1 to 24 h ahead) and, most importantly, to derive point and quantile forecasts, predictive distributions, and time-path trajectories (also referred to as scenarios or ensemble forecasts), all using one single stochastic differential equation model that is characterized by a few parameters.

The ensemble Kalman filter (EnKF) is a popular data assimilation (DA) technique that has been extensively used in environmental sciences for combining complementary information from model predictions and observations. One of the major challenges in EnKF applications is the description of model uncertainty. In most hydrological EnKF applications, an ad hoc model uncertainty is defined with the aim of avoiding a collapse of the filter. The present work provides a systematic assessment of model uncertainty in DA applications based on combinations of forcing, model parameters, and state uncertainties. This is tested in a case where groundwater hydraulic heads are assimilated into a distributed and integrated catchment-scale model of the Karup catchment in Denmark. A series of synthetic data assimilation experiments are carried out to analyse the impact of different model uncertainty assumptions on the feasibility and efficiency of the assimilation. The synthetic data used in the assimilation study makes it possible to diagnose model uncertainty assumptions statistically. Besides the model uncertainty, other factors such as observation error, observation locations, and ensemble size are also analysed with respect to performance and sensitivity. Results show that inappropriate definition of model uncertainty can greatly degrade the assimilation performance, and an appropriate combination of different model uncertainty sources is advised.

In countries with an extended use of district heating (DH), the integrated operation of DH and power systems can increase the flexibility of the power system achieving a higher integration of renewable energy sources (RES). DH operators can not only provide flexibility to the power system by acting on the electricity market, but also profit from the situation to lower the overall system cost. However, the operational planning and bidding includes several uncertain components at the time of planning: electricity prices as well as heat and power production from RES. In this publication, we propose a planning method that supports DH operators by scheduling the production and creating bids for the day-ahead and balancing electricity markets. The method is based on stochastic programming and extends bidding strategies for virtual power plants to the DH application. The uncertain factors are considered explicitly through scenario generation. We apply our solution approach to a real case study in Denmark and perform an extensive analysis of the production and trading behaviour of the DH system. The analysis provides insights on how DH system can provide regulating power as well as the impact of uncertainties and renewable sources on the planning. Furthermore, the case study shows the benefit in terms of cost reductions from considering a portfolio of units and both markets to adapt to RES production and market states.

Integrating flexible consumers in grids with high penetration of renewable energy sources requires a robust power balancing strategy. The methodologies and solutions suggested in this article aim to describe a flexible framework for controlling future electric energy systems by formulating the aggregation problem as a hierarchical robust optimization problem on different aggregation levels. The Aggregator solves a minmax robust optimization problem through a model predictive control framework. With two numercal examples we show how our algorithm controls flexible loads in closed loop, such that consumption follows the stochastic changing production influenced by the penetration of renewables into the power system.

We have developed a versatile Model Predictive Control (MPC) framework, which can handle real-time control of a large variety of water systems. The framework combines a fast-solvable optimisation model (a quadratic program) with evaluation and realignment by a detailed hydrological-hydrodynamic model. The flexibility of the MPC framework is highlighted by two case studies: (1) a large-scale river system with several weeks of travel time, and (2) an urban storm and wastewater system with a concentration time of about half an hour to one hour. Both case studies demonstrate a large potential for improving operations by system-wide real-time optimisation.

Statistics for Finance develops students’ professional skills in statistics with applications in finance. Developed from the authors’ courses at the Technical University of Denmark and Lund University, the text bridges the gap between classical, rigorous treatments of financial mathematics that rarely connect concepts to data and books on econometrics and time series analysis that do not cover specific problems related to option valuation. The book discusses applications of financial derivatives pertaining to risk assessment and elimination. The authors cover various statistical and mathematical techniques, including linear and nonlinear time series analysis, stochastic calculus models, stochastic differential equations, Itō’s formula, the Black–Scholes model, the generalized method-of-moments, and the Kalman filter. They explain how these tools are used to price financial derivatives, identify interest rate models, value bonds, estimate parameters, and much more. This textbook will help students understand and manage empirical research in financial engineering. It includes examples of how the statistical tools can be used to improve value-at-risk calculations and other issues. In addition, end-of-chapter exercises develop students’ financial reasoning skills.