Short-Term Electricity Price Forecasting Using Wavelet Transform Integrated Generalized Neuron

Abstract

With the advent of deregulation, electricity has become a commodity which is capable of being traded in the deregulated electricity market. In the deregulated environment, accurate electricity price forecasting has become necessity for the generating companies in order to maximize their profits. The existing forecasting models can be broadly classified into statistical models, simulation models, and soft computing models. The soft computing based models have gained popularity among other existing models because of their nonlinear mapping capabilities and ease of implementation. In the presented work, a generalized neuron based electricity price forecasting model has been proposed to forecast the electricity price of New South Wales electricity market. The de-noising capability of the wavelet transform is explored for decomposing the ill-behaved price signal into low- and high-frequency signals for better representation. The low- and high-frequency signals were given as input to the generalized neuron model individually for improving the forecasting accuracy of the model.