Open AccessPrediction of electricity demand of China based on the analysis of decoupling and driving force
Author(s) -
He Yongxiu,
Guang Fengtao,
Chen Rongjun
Publication year - 2018
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2017.1493
Subject(s) - extreme learning machine , electricity , computer science , decoupling (probability) , particle swarm optimization , genetic algorithm , support vector machine , econometrics , mathematical optimization , economics , artificial neural network , artificial intelligence , engineering , machine learning , mathematics , control engineering , electrical engineering
The emerging complex circumstances caused by the essence of the new economic normal and the low‐carbon development of power industry lead to the demand for electricity from the rapid growth phase to a ‘new normal’ pace for China. To better capture the more and more non‐linear and non‐stationary characteristics of electricity demand, Tapio decoupling model is applied to research the decoupling of electricity demand from economic growth, and T 's correlation degree is used to measure the influencing degree and reveal the role direction of each driver on electricity demand. On these bases, a novel improved particle swarm optimisation‐extreme learning machine (IPSO‐ELM) hybrid forecasting model is proposed to predict electricity demand. In IPSO‐ELM, a mutation operator of genetic algorithm is introduced into standard PSO to ensure the overall convergence and enhance the accuracy of convergence, and the improved PSO is adopted to optimise the input weights and hidden bias of traditional ELM which are randomly generated. Finally, the case study of China demonstrated the efficacy and feasibility of IPSO‐ELM. Simultaneously, the comparisons with ELM, back propagation neural network, support vector machine, GM (1, 1), and logistic model showed that IPSO‐ELM has a better forecasting performance in electricity demand.