Forecasting diversion type hydropower plant generations using an artificial bee colony based extreme learning machine method

Abstract

ABSTRACT In this study, a hybrid method based on extreme learning machine (ELM) method and artificial bee colony (ABC) algorithm was proposed to forecast small hydropower plant generations. The input weights and biases of ELM were optimized by ABC algorithm to achieve more accurate forecasting results. The forecasting performance of the proposed method was compared with benchmark methods, namely backpropagation-based artificial neural network (ANN), radial basis function-based ANN, and long short-term memory. The experimental results verified that the proposed method significantly outperformed the benchmark methods. Specially, when the proposed method was compared with ELM, the improvement percentages in correlation coefficient, root mean square error, and mean absolute error values were calculated as being 6.20%-29.08%-26.29% for 14 days ahead and 5.47%-24.42%-20.33% for 21 days ahead, respectively.