An intelligent model for predicting the day-ahead deregulated market clearing price: A hybrid NN-PSO-GA approach

Under the restructuring of the electric power industry and transformation of the traditional vertically integrated electric utility structure to the competitive market scheme, market clearing price (mcp) prediction models have become essential for all generation companies (gencos). in this paper, a hybrid model is presented to predict hourly electricity mcp. the proposed model contains a neural network (nn), particle swarm optimization (pso), and genetic algorithm (ga). the nn is used as a major forecasting module for prediction, pso is applied to improve the traditional neural network learning capability, and ga is applied to optimize nn architecture. the main contribution of this paper includes: (a) presentation of a new hybrid intelligent model for market clearing price prediction; (b) application of k-means algorithm to clustering nn’s test set and seasonality pattern detection; and (c) performance evaluation of the proposed model by improved mean absolute error (mae) with a penalty factor for positive error. the proposed method has been tested for the real-world electricity market of iran within one month in each year of 2010-2013; obtained results show that the average weighted mae designed for prediction purposes is equal to 0.12; the prediction accuracy of mcp by the proposed model can be improved by more than 6.7% and 4% in mae , compared to a simple nn by a combination of nn and bat algorithm.