Prediction of electricity consumption during epidemic period based on improved particle swarm optimization algorithm

Abstract

A prediction method of electricity consumption is developed in order to address the problems of big change and imbalance in electricity consumption caused by COVID-19. In this method, BP (Back Propagation) neural network and improved particle swarm optimization (IPSO) algorithm are combined and applied. Firstly, Pearson correlation coefficient approach is utilized to conduct data correlation analysis. Then, the BP neural network prediction model is built, and IPSO algorithm is used to optimize the neural network’s initial weights and thresholds. Considering the medical data, public opinion data, policy data and historical data of electricity consumption during epidemic period, the electricity consumption of each industry in the future is predicted. The findings suggest that the proposed model performs well in terms of prediction. The Mean Absolute Percentage Error (MAPE) for each industry’s evaluation index is 1.41%, 1.70 %, and 1.37 %, respectively. Compared with other models, the prediction accuracy is higher. By exploring the predicted results of electricity consumption during epidemic period, it is hoped that a basis prediction method of electricity consumption for power grid companies in the event of a sudden outbreak will be provided.

Keywords

• COVID-19
• Prediction of electricity consumption
• Modeling
• BP neural network
• Improved particle swarm optimization
• Epidemic