Title: Electricity consumption scenario prediction based on factor analysis and least squares support vector machine optimised by fruit fly algorithm
Authors: Siwei Wei; Ting Wang
Addresses: School of Economics and Management, North China Electric Power University, Baoding 071000, China ' School of Economics and Management, North China Electric Power University, Baoding 071000, China
Abstract: Electricity consumption forecasting is the basis and premise of power grid planning through the analysis of historical electricity consumption data and related factors. For future electricity consumption accurate prediction and influencing factors analysis, we use both scenario analysis and econometric methods comprehensively. Firstly, this paper analyses the effects of GDP, population, energy consumption and many other factors of electricity consumption in depth and then extracts the key influencing factors of electricity consumption. Secondly, electricity consumption scenario prediction model is established based on factor analysis and least squares support vector machine optimised by fruit fly algorithm. Thirdly, the performance of proposed model is tested through the comparison of different models and we get the forecast results for further analysis. The proposed model is proven to have good prediction accuracy and we provide more than one research perspective about future development of electricity consumption for decision-makers by scenario analysis.
Keywords: factor analysis; fruit fly algorithm; least squares support vector machine; LSSVM; electricity consumption prediction; scenario analysis.
International Journal of Applied Decision Sciences, 2017 Vol.10 No.3, pp.254 - 273
Received: 27 Jul 2016
Accepted: 18 Feb 2017
Published online: 05 Jun 2017 *