An adaptive local learning-based methodology for voltage regulation in distribution networks with dispersed generation

This paper proposes a computational architecture for the voltage regulation of distribution networks equipped with dispersed generation systems (DGS). The architecture aims to find an effective solution of the optimal regulation problem by combining a conventional nonlinear programming algorithm with an adaptive local learning technique. The rationale for the approach is that a local learning algorithm can rapidly learn on the basis of a limited amount of historical observations the dependency between the current network state and the optimal asset allocation. This approach provides an approximate and fast alternative to an accurate but slow multiobjective optimization procedure. The experimental results obtained by simulating the regulation policy in the case of a medium-voltage network are very promising.