Evolutionary Short Term Load Forecasting Based on ANN: A Comparison with the Most Well-known Algorithms

Precise prediction of load demand is necessary in power systems which can be used in electricity industry and market. Most of load forecasting methods are based on Artificial Neural Networks (ANNs). However ANN parameters need to be adjusted to have an accurate forecasting. For this purpose evolutionary algorithms have been used to optimize the ANN based forecasting. Th e main obstacle in the use of evolutionary algorithms for training ANNs is the problem of requiring so much time for the training process as well as the weak stability of the evolutionary-based training methods to adapt ANNs weighting factors. In this paper, first several forecasting methods such as AR, ARMA, ANN and SVR are evaluated using a real case study. Th en in the second stage a complete comparison between some of the most well-known evolutionary algorithms such as; GA, PSO, HBMO, and MHBMO is done which are utilized to optimizes the ANN based load forecasting. A real case study is investigated and the simulation results are described, precisely.