Modeling and predicting the electricity production in hydropower using conjunction of wavelet transform, long short-term memory and random forest models

Electricity is an important pillar for the economic growth and the development of societies. Surveying and predicting the electricity production (EP) is a valuable factor in the hands of electricity industry managers to make strategic decisions, especially if electricity is generated by renewable resources for environmental considerations. However, because the EP series is non-stationary and nonlinear, traditional methods are less robust to predict it. In this study, we offer a hybrid model which combines adaptive wavelet transform (AWT), long short-term memory (LSTM) and random forest (RF) algorithm (AWT-LSTM-RF) to predict the EP in hydroelectric power plant. We also apply the exogenous affecting variables on EP in the structure of hybrid model, which were selected by ant colony optimization (ACO) algorithm. To evaluate the predictive power of the AWT-LSTM-RF model, we compared its predictive results with the benchmark models including RF, ARIMA-GARCH, wavelet transform-feed forward neural network (WT-FFNN), wavelet transform-random forest (WT-RF), wavelet transform-LSTM (WT-LSTM), and WT-FFNN-RF. The empirical results indicate that the hybrid model of AWT-LSTM-RF outperforms the benchmark models. The results also suggest that applying the wavelet transform on input data of the RF algorithm (WT-RF) can improve the predictive power of the RF. (c) 2021 Elsevier Ltd. All rights reserved.

Automated summary

This study proposes a hybrid model combining AWT, LSTM, and RF algorithms to predict electricity production in hydroelectric power plants, which outperforms benchmark models. Applying wavelet transform on input data of the RF algorithm can improve predictive power.