Wavelet transform and Kernel-based extreme learning machine for electricity price forecasting

Abstract

In deregulated electricity markets, sophisticated factors, such as the weather, the season, high frequencies, the presence of jumps and the relationship between electricity loads and prices, make electricity prices difficult to predict. To increase the accuracy of electricity price forecasting, this paper investigates a hybrid approach that is based on a combination of the wavelet transform, a kernel-based extreme learning machine and a particle swarm optimization algorithm. The performance and robustness of the proposed method are evaluated by using electricity price data from two Australian districts (New South Wales and Victoria) and Pennsylvania-New Jersey-Maryland (PJM) electricity markets. These case studies show that the proposed method can effectively capture the nonlinearity features from the price data series with a smaller computation time cost and high prediction accuracy compared with other price forecasting methods. The results also demonstrate that the proposed method represents an accurate price forecasting technique for power market price analysis.