Improve Prediction Accuracy of Electrical Consumption Adjusted with Demand Response Programs

Prediction on electrical consumption of clients and their selection for appropriate demand response programs (DRPs), are a major challenge for the stability of the distribution network. Prediction accuracy of patterns and their matching with reality plays an important role in DRPs modeling for better consumers' participation. Existing Approaches studied several aspects of prediction and baselines in DRPs environment. The originality of this work consists in demonstrating a prediction based approach, studying its accuracy and continuously adding new patterns to the training set. The variation versus planning phase is kept minimal and adapted to customers' variations. It is achieved at the individual and aggregated profiles, thus customers' classification and engagement are ensured. The approach is validated through a simulation on Matlab. Objectives include the customer prequalification and the effects of prediction accuracy on the balance generation/demand.