Accurate forecasts of electrical substations are mandatory for the efficiency of the Advanced Distribution Automation functions in distribution systems. The paper describes the design of a class of machine-learning models, namely neural networks, for the load forecasts of medium-voltage/low-voltage substations. We focus on the methodology of neural network model design in order to obtain a model that has the best achievable predictive ability given the available data. Variable selection and model selection are applied to electrical load forecasts to ensure an optimal generalization capacity of the neural network model. Real measurements collected in French distribution systems are used to validate our study. The results show that the neural network-based models outperform the time series models and that the design methodology guarantees the best generalization ability of the neural network model for the load forecasting purpose based on the same data.Index Terms-Model design, machine learning, neural network, short-term load forecast, variable selection, virtual leave-one-out.
In the context of smart grid development, this paper considers the problem of interoperability of micro-grid platforms, particularly among research institutions. Various levels of interoperability are introduced with the respective requirements. The primary aim of the paper is to propose a suitable private hybrid cloud based SCADA architecture satisfying various necessities in the framework of interoperability of micro-grid platforms while maintaining security restriction conditions. Due to the limited time restriction of critical SCADA functions in the electrical grid (protection, real time control, etc.), only selected non-critical SCADA functions (back-up, data historian, etc.) are accessible to partners from the private cloud. The critical SCADA tasks functionality remains under control of local server, thus, a hybrid cloud architecture. Common Information Model (IEC 61970 and IEC 61968, CIM/XML/RDF) is proposed to be used as model for information exchange. The communication model is based on PaaS delivery model and OPC Unified Architecture (OPC UA) specifications are considered. OPC gateway is proposed as conversion between the old OPC Distributed Common Object Model (DCOM) protocol and the Simple Object Access Protocol (SOAP) for cloud.
Co-simulation is an emerging method for cyber-physical energy system (CPES) assessment and validation. Combining simulators of different domains into a joint experiment, co-simulation provides a holistic framework to consider the whole CPES at system level. In this paper, we present a systematic structuration of co-simulation based on a conceptual point of view. A co-simulation framework is then considered in its conceptual, semantic, syntactic, dynamic and technical layers. Coupling methods are investigated and classified according to these layers. This paper would serve as a solid theoretical base for specification of future applications of co-simulation and selection of coupling methods in CPES assessment and validation.
Primary frequency control is the automatic mechanism implemented on power systems to regulate the power balance through frequency and hence, its action should be taken into account when modeling any contingency state leading to a modification of the active power balance (e.g. generator failures). This paper presents a fully distributed method to solve the DC security constrained power flow (DC-SCOPF) that takes into account the automatic primary frequency response of generators after an incident. In more detail, we extend existing distributed DC-SCOPF formulations by: (1) introducing a new variable representing the frequency deviation; and (2) enhancing the local problem of each generator to consider how it adjusts its production after each contingency following its primary frequency regulation curve. The computation of the frequency deviations in the DC-SCOPF problem is formulated into a suitable form (i.e. in the form of a general consensus problem) so that smaller problems, corresponding to individual sub-regions or actors, can be solved and coordinated via the alternating direction method of multipliers (ADMM) in a distributed manner. In this way, actors of the system do not need to exchange any confidential information with other actors during the optimization procedure. A salient feature of our approach is that it can consider contingencies that lead to area separation without any prior specification of the topology and thus can adapt to many kinds of situations that are of interest in interconnected systems. Extensive simulation results on several standard IEEE systems show the good performance of the proposed model and algorithm in terms of convergence speed and accuracy as well as its capacity to deal with the disconnection of areas in interconnected systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.