Grey Prediction Model and Multivariate Statistical Techniques Forecasting Electrical Energy Consumption in Wenzhou, China

Abstract

Electricity consumption has always been one of the critical economic issues in Wenzhou. This paper presents a combination method of grey prediction models and multivariate statistical techniques to forecast the trend of electrical energy consumption in Wenzhou. Hierarchical cluster analysis and discriminant analysis grouped 18 sampling years into three clusters, i.e., relatively less electrical energy consumption phase (LEECP), medium electrical energy consumption phase (MEECP) and highly electrical energy consumption phase (HEECP). The two grey prediction models established are the first-degree. Mean absolute percentage error (MAPE) criteria are more suitable than traditional accuracy and error test to evaluate grey models accuracy. Grey prediction model has been tested with high precision in a short-term. Using the grey prediction model, electrical energy consumption of Wenzhou will be 44.4719 billion kilowatt hour in 2010.