Electricity Price Forecasting Using LSTM Network and K-Means Clustering by Considering the Effect of Wind Power Generation

Abstract

Deregulation of the electricity market offers minimum electricity prices to consumers and benefits utility companies with increased gains. Market participants (suppliers and consumers) depend on electricity price forecasting to reach their requirements. In recent decades, the development of the smart grid has made the electricity market more dynamic with the large-scale integration of wind and solar power generations. This paper mainly aims to analyze the impact of wind power on electricity price forecasting. A hybrid approach of the long short-term memory (LSTM) network and k-means clustering is proposed in this paper to forecast the electricity price in the Austrian electricity market with integrated wind power generation in the smart grid. The forecasting accuracy of the proposed model is compared with a hybrid feedforward neural network—particle swarm optimization (FNN-PSO) model and support vector regression (SVR) model. The evaluation is carried out with the error calculation in terms of mean absolute percentage error (MAPE) and root mean squared error (RMSE). The results show that the proposed model is able to forecast the electricity price with less error as compared to other models, and the integration of wind power generation has resulted in further reduction of error in price forecasting.