Short Term Load Forecasting using Genetically Optimized Radial Basis Function Neural Network

Management and pricing of electricity in power system is largely influenced by Short-Term Load Forecasting (STLF). This paper presents a hybrid algorithm, where Radial Basis Function Neural Network (RBFNN) is optimized using Genetic Algorithm (GA) for STLF, with load and day-type as input parameters. Since, conventional training methods, viz., principle component analysis and least square method, does not provide optimum selection of RBFNN parameters, a novel model is proposed utilizing GA to optimize the center width of radial basis functions and weights of output layer in RBFNN. The performance of the proposed approach is evaluated using Mean of Mean Absolute Percentage Error (MMAPE) on New South Wales (NSW), Australia load data and compared with the existing approaches, i.e., Feed Forward Neural Network (FFNN) and RBFNN models. Simulation results show that, in comparison to the existing approaches, the proposed model results in significant improvement in forecasting accuracy.