time series prediction using emotional neural networks

Time series forecasting is important in many fields including energy management, power market, and engineering. therefore, it is vital to introduce new algorithms that can predict time series with high accuracy. emotional networks have recently been introduced based on emotional processes occurring in the mammalian brain. they have shown desirable numerical properties such as fast response, simple structure, learning capability, and the ability to accurately approximate and address time and complexity issues. however, their use in time-series prediction is at the primary stages. therefore, we are inspired to use emotional models in the time-series prediction problems. specifically, we propose to use a continuous radial basis emotional neural network (crbenn) for time-series prediction. the normal rules of the emotional brain are used to update the network weights and the gradient descent algorithm is used to update the radial basis parameters. the proposed method is compared with two neuro and fuzzy methods in three benchmark problems. the results show the lower prediction error of the proposed method.