Substation short term load forecasting using neural network with genetic algorithm

This research describes an innovative load forecasting scheme employing a neural network (NN) with a genetic algorithm (GA). The new load forecasting technique is compared with the conventional NN approaches. which can suffer from the local minima problem. Employing GA to search for the initial weights and biases of NNs allows the NN weights and biases to be easily optimized. The proposed NNs with GA load forecasting scheme (NNGA) has been tested with data obtained from a case study. The experimental evaluations have demonstrated the accuracy and effectiveness of the scheme to support distribution operation. Forecast results, when compared with the actual historical load data, show that the load prediction has an average error of 7.31 % which is lower than the conventional NN by 0.77 %.