Chapter 4 - The Use of Hybrid Neural Networks, Wavelet Transform and Heuristic Algorithm of WIPSO in Smart Grids to Improve Short-Term Prediction of Load, Solar Power, and Wind Energy

Abstract

Short-term prediction of load, solar power and wind energy is a vital requirement in smart distribution systems to obtain cost-effective, programmable, and reliable operation of the system. This paper suggests a hybrid approach to forecast consumed power, wind and solar power generation output. With this method and data correlation analysis, the number of effective data related to input parameters is determined. The wavelet transform is used for filtering input data of wind, solar, and load power; while, neural network radial basis function is used as the elementary predictor. The main predictor motor comprises of three multilayer perceptron neural networks with learning algorithm of BR, RP, and LM. To improve the accuracy of predictions and escape from local minima, a metaheuristic “Weight Improved PSO” is employed, optimizing the weights of neural networks. Input is realistic data of wind farms & solar plant and load demand in Southern Alberta, Canada. Simulation results reveal the dominant superiority of the proposed method over the other methods; the prediction error is highly reduced.