An Emotional Neural Network for Electrical Load Demand Forecast
Emotional neural network (EmNN) is a new approach that implements the virtual emotions to support the learning process of neural networks. The inspiration of EmNN is adopted from neurophysiological studies of the human brain behaviors under emotional circumstances. In this research, EmNN based models are designed and experimented for electrical load forecasting application. The numerical parameters are fine-tuned by applying genetic algorithm as an optimization tool. Two case studies are developed with different data sets for the training and testing of the proposed model. A hybrid input variable selection method is proposed for identifying and implementing the most appropriate input variables in the learning process. A couple of conventional training algorithms of ANN are employed for the same datasets and the outcomes are compared with EmNN model. The results of the proposed model show that the suggested technique performed better as compared to conventional ANN with respect to prediction accuracy and generalization.