Day-Ahead Deregulated Electricity Market Price Forecasting Using Recurrent Neural Network

Abstract

This paper proposes a recurrent neural network model for the day ahead deregulated electricity market price forecasting that could be realized using the Elman network. In a deregulated market, electricity price is influenced by many factors and exhibits a very complicated and irregular fluctuation. Both power producers and consumers need a single compact and robust price forecasting tool for maximizing their profits and utilities. In order to validate the chaotic characteristic of electricity price, an Elman network is modeled. The proposed Elman network is a single compact and robust architecture (without hybridizing the various hard and soft computing models). It has been observed that a nearly state of the art Elman network forecasting accuracy can be achieved with less computation time. The proposed Elman network approach is compared with autoregressive integrated moving average (ARIMA), mixed model, neural network, wavelet ARIMA, weighted nearest neighbors, fuzzy neural network, hybrid intelligent system, adaptive wavelet neural network, neural networks with wavelet transform, wavelet transform and a hybrid of neural networks and fuzzy logic, wavelet-ARIMA radial basis function neural networks, cascaded neuro-evolutionary algorithm, and wavelet transform, particle swarm optimization, and adaptive-network-based fuzzy inference system approaches to forecast the electricity market of mainland Spain. Finally, the accuracy of the price forecasting is also applied to the electricity market of New York in 2010, which shows the effectiveness of the proposed approach.