A Neuro-Predictive Controller Scheme for Integration of a Basic Wind Energy Generation Unit with an Electrical Power System

Abstract

Developing control methods that have the ability to preserve the stability and optimum operation of a wind energy generation unit connected to power systems constitutes an essential area of recent research in power systems control. The present work investigates a novel control of a wind energy system connected to a power system through a static VAR compensator (SVC). This advanced control is constructed via integration between the model predictive control (MPC) and an artificial neural network (ANN) to collect all of their advantages. The conventional MPC needs a high computational effort, or it can cause difficulties in implementation. These difficulties can be eliminated by using Laguerre-based MPC (LMPC). The ANN has high performance in optimization and modeling, but it is limited in improving dynamic performance. Conversely, MPC operation improves dynamic performance. The integration between ANN and LMPC increases the ability of the Neuro-MPC (LMPC-ANN) control system to conduct smooth tracking, overshoot reduction, optimization, and modeling. The new control scheme has strong, robust properties. Additionally, it can be applied to uncertainties and disturbances which result from high levels of wind speed variation. For comparison purposes, the performance of the studied system is estimated at different levels of wind speed based on different strategies, which are ANN only, Conventional MPC strategy, MPC-LQG strategy, ANN- LQG strategy, and the proposed control. This comparison proved the superiority of the proposed controller (LMPC-ANN) for improving the dynamic response where it mitigates wind fluctuation effects while maintaining the power generated and generator terminal voltage at optimum values.