• Energy Research

To improve the design of the electricity infrastructure and the efficient deployment of distributed and renewable energy sources, a new paradigm for the energy supply chain is emerging, leading to the development of smart grids. There is a need to add intelligence at all levels in the grid, acting over various time horizons. Predicting the behavior of the energy system is crucial to mitigate potential uncertainties. An accurate energy prediction at the customer level will reflect directly in efficiency improvements in the whole system. However, prediction of building energy consumption is complex due to many influencing factors, such as climate, performance of thermal systems, and occupancy patterns. Therefore, current state-of-the-art methods are not able to confine the uncertainty at the building level due to the many fluctuations in influencing variables. As an evolution of artificial neural network (ANN)-based prediction methods, deep learning techniques are expected to increase the prediction accuracy by allowing higher levels of abstraction. In this paper, we investigate two newly developed stochastic models for time series prediction of energy consumption, namely Conditional Restricted Boltzmann Machine (CRBM) and Factored Conditional Restricted Boltzmann Machine (FCRBM). The assessment is made on a benchmark dataset consisting of almost four years of one minute resolution electric power consumption data collected from an individual residential customer. The results show that for the energy prediction problem solved here, FCRBM outperforms ANN, Support Vector Machine (SVM), Recurrent Neural Networks (RNN) and CRBM.

To realize the goals of energy transition, becoming energy-neutral at the neighborhood level by sharing energy among clusters of heterogeneous buildings with local distributed energy resources (DERs), will play a vital role. However, uncertainties related to demand and renewable sources pose a major operational challenge to schedule the DERs. In this paper, a scenario-based mixed-integer linear programming (MILP) model is proposed for an energy management system (EMS) of a local energy community. The proposed EMS executes a stochastic day-ahead scheduling operation of multi-energy systems (MES). A set of scenarios are generated with the Gaussian mixture model (GMM) to consider uncertainties of demand and renewable sources. Moreover, Monte Carlo simulations (MCS) are performed to assess the effectiveness of the proposed EMS compared to the deterministic one. The proposed method is validated by using a real-world case study of a generic Dutch university medical campus in Amsterdam, the Netherlands. Two types of analysis are performed: one-day analysis and seasonal analysis. In both cases, in an average, the stochastic process outperforms the deterministic process considerably, in terms of cost, CO2 emission, imported electricity from grid and usage of local energy resources.

Energy is a limited resource which has to be managed wisely, taking into account both supply-demand matching and capacity constraints in the distribution grid. One aspect of the smart energy management at the building level is given by the problem of real-time detection of flexible demand available. In this paper we propose the use of energy disaggregation techniques to perform this task. Firstly, we investigate the use of existing classification methods to perform energy disaggregation. A comparison is performed between four classifiers, namely Naive Bayes, k-Nearest Neighbors, Support Vector Machine and AdaBoost. Secondly, we propose the use of Restricted Boltzmann Machine to automatically perform feature extraction. The extracted features are then used as inputs to the four classifiers and consequently shown to improve their accuracy. The efficiency of our approach is demonstrated on a real database consisting of detailed appliance-level measurements with high temporal resolution, which has been used for energy disaggregation in previous studies, namely the REDD. The results show robustness and good generalization capabilities to newly presented buildings with at least 96% accuracy. To appear in IEEE PES General Meeting, 2016, Boston, USA

The increasing penetration of intermittent Renewable Energy Sources (RES) such as PV generators in the residential distribution grid raises technical concerns. Residential energy storage is an enabling technology for mitigating the adverse effects of the increasing RES penetration and provides the necessary flexibility for its owner to participate in electricity markets or in market-based control schemes. This paper presents a model for calculating the "cost of use" for Li-Ion batteries. The cost of use depends on the usage pattern of the battery, indicated by the Depth of Discharge (DoD). Knowledge of the "cost of use" is needed for the economic optimization of the PV-storage system and the bid formulation in the electricity market.

Normal operation of an active distribution network (ADN) requires simultaneous optimization of different objectives of the various involved actors. This results in a multi-objective optimization problem which has not yet been treated completely. This paper considers a particular relationship between commercial and technical coordination, involving capacity management of the distribution network. First, the market-based ADN, its actors and their objectives are described. An agent-based approach is desirable to handle the complexity of this ADN. Then, several technical issues for integrating capacity management within a multi-agent market-based ADN are pointed out. After that, the developed agent architecture and coordination mechanism are further elaborated upon, along with a formulation of the multi-objective optimization problem. Finally, a decentralized approach for integrating capacity management is introduced and demonstrated.

The increasing dependency of society on electricity motivates the research for more reliable and resilient distribution networks. Self-healing of networks is one of the key elements of the Smart Grid concept. Conventional self-healing networks (CSHNs) recover part of the network by rerouting power with automated switchgear. By utilizing microgrids (MGs), the sectionalizing distribution networks (SDNs) concept is an extension of CSHNs which utilizes both automated switchgear and distributed energy resources (DERs) to create both grid-connected and islanded microgrids (MGs). The SDN concept provides more resilience against faults than CSHNs and allows local bottom-up blackout restoration. This paper provides a high-level conceptual and operation states description, a system framework and the operational sequence of SDNs in case of faults. Finally, several future research directions are proposed.

In this paper, the authors apply a surrogate model-based method for probabilistic power flow (PPF) in the power system subject to truncated random variables. Due to a growing number of uncertainty sources are being brought into the modern power system, the traditional deterministic power flow analysis lacks its ability to recognize the realistic states of power systems, and thus turns to PPF for help. However, the PPF analysis is still facing several challenges: the computational effort required by the traditional simulation method is prohibitively expensive; and the modeling of uncertainty sources needs the improvement on both distribution type selection and parameter evaluation. The novelty of this work lies in taking advantage of both general polynomial chaos (gPC) expansion and ordinary least squares (OLS) to deal with PPF in presence of the truncated random variables. The performance of the proposed method is verified on the IEEE 30-Bus test system, considering uncertain factors brought by active power at load buses. In different test scenarios, the proposed method shows sound performances at the cost of less computational effort, compared to the traditional approach.

The operation of a community energy storage system (CESS) is challenging due to the volatility of photovoltaic distributed generation, electricity consumption, and energy prices. Selecting the optimal CESS setpoints during the day is a sequential decision problem under uncertainty, which can be solved using dynamic learning methods. This paper proposes a reinforcement learning (RL) technique based on temporal difference learning with eligibility traces (ET). It aims to minimize the day-ahead energy costs while maintaining the technical limits at the grid coupling point. The performance of the RL is compared against an oracle based on a deterministic mixed-integer second-order constraint program (MISOCP). The use of ET boosts the RL agent learning rate for the CESS operation problem. The ET effectively assigns credit to the action sequences that bring the CESS to a high state of charge before the peak prices, reducing the training time. The case study shows that the proposed method learns to operate the CESS effectively and ten times faster than common RL algorithms applied to energy systems such as Tabular Q-learning and Fitted-Q. Also, the RL agent operates the CESS 94% near the optimal, reducing the energy costs for the end-user up to 12%.