Volume 4: Innovative Solutions for Energy Transitions: Part III

A Multi-Step Prediction Model for Household Power Consumption Using PSO, Holt-Winter, and Extreme Learning Machine Che Liu, Chenghui Zhang, Fan Li, Bo Sun

Abstract

With the increase of residential energy consumption, its proportion in primary energy consumption is higher and higher. Accurate prediction of residential electricity consumption is the premise of rational residential energy management. In this paper, a novel multi-step prediction model using particle swarm optimization (PSO), Holt-Winter (HW) method, and extreme learning machine (ELM) network is proposed for forecasting household power consumption. The HW model optimized by PSO is the main predictor and used to deal with the periodicity and seasonality of household electricity load. ELM model is introduced as the correction predictor to predict the prediction error of HW, so as to improve the prediction accuracy. The experimental results show that the PSO-HW-ELM model has higher prediction accuracy and better stability compared with the single HW and ELM model.

Keywords Holt-Winter method, particle swarm optimization, extreme learning machine, household power consumption prediction