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.
This paper addresses the issue of network partition used in power system analysis for ancillary services in distribution networks with distributed generation. Simplification and reduction of the computational effort in network calculation for centralized control can be achieved, through a definition of suitable partitioned clusters. The zoning problem is explored using a clustering process based on the k-means algorithm. The proposed method is implemented within a study case of voltage profile calculation in a 20 kV distribution network for steady state voltage control. The results show the performance of the method, with the caveat of the need of preliminary fitting of the parameters.Index Terms--Smart grid, distribution network, Clustering methods, dispersed storage and generation, power distribution, power system simulation.
The paper presents a method to design transaction zones for commercial aggregation at the distribution level for voltage mitigation purpose. The core principle of this zoning is the nodal voltage sensitivity for active power variation. As the transaction zone delimits the elementary area for DER aggregation and separates the distinct roles of the network operator and of the aggregators, a fair nonarbitrary efficient zoning method is needed. The zoning problem is explored using a modified hierarchical clustering method, applied to a set of voltage sensitivity factors. The impact of this method is illustrated on a study case of a balanced MV distribution network.
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.