Neural Network‐Based Long‐Term Hydropower Forecasting System

Artificial neural networks are alternatives to stochastic models even if the optimization of their architectures remains a tricky problem. Two different approaches in long‐term forecasting of potential energy inflows using a feedforward neural network (FNN) and a recurrent neural network (RNN) are proposed. The problem of overfitting, particularly critical for limited hydrologic data records, is addressed using a new approach entitled optimal weight estimate procedure (OWEP). The efficiency of the two models using OWEP is assessed through multistep forecasts. The experiment results show that, in general, OWEP improves the models' performance and significantly reduces the training time on the order of 60 percent. The RNN outperforms the FNN but costs about a factor of 2 longer in training time. Furthermore, the neural network‐based models provide more accurate forecasts than traditional stochastic models. Overall, the RNN appears to be the best suited for potential energy inflows forecasting and therefore for hydropower systems management and planning.