Title: Method for predicting comprehensive energy demand in industrial parks based on echo state networks

Authors: Xiaojun Zhu; Yan Li; Decheng Wang; Qun Zhang; Yinzhe Xie; Na Li; Zhu Chen

Addresses: State Grid Jiangsu Electric Power Design Consulting Co., Ltd., Nanjing, Jiangsu Province, 210008, China; State Grid Jiangsu Electric Power Co., Ltd., Economic Research Institute, Nanjing, Jiangsu Province, 210008, China ' State Grid Jiangsu Electric Power Design Consulting Co., Ltd., Nanjing, Jiangsu Province, 210008, China; State Grid Jiangsu Electric Power Co., Ltd., Economic Research Institute, Nanjing, Jiangsu Province, 210008, China ' State Grid Jiangsu Electric Power Design Consulting Co., Ltd., Nanjing, Jiangsu Province, 210008, China; State Grid Jiangsu Electric Power Co., Ltd., Economic Research Institute, Nanjing, Jiangsu Province, 210008, China ' State Grid Jiangsu Electric Power Design Consulting Co., Ltd., Nanjing, Jiangsu Province, 210008, China; State Grid Jiangsu Electric Power Co., Ltd., Economic Research Institute, Nanjing, Jiangsu Province, 210008, China ' East China Electric Power Design Institute of China Power Engineering Consulting Group, Huangpu District, Shanghai, 200001, China ' East China Electric Power Design Institute of China Power Engineering Consulting Group, Huangpu District, Shanghai, 200001, China ' East China Electric Power Design Institute of China Power Engineering Consulting Group, Huangpu District, Shanghai, 200001, China

Abstract: To achieve accurate prediction of energy demand, this study designed a new method for predicting comprehensive energy demand in industrial parks using echo state networks. Firstly, analyse the comprehensive energy structure of the park, then collect and supplement historical comprehensive energy load consumption data. Secondly, select the factors that affect the load demand forecast, and calculate the comprehensive similarity of similar days of historical energy demand according to the mutual information between the influencing factors. Finally, input the calculation results into the optimised echo state network of the crossbar algorithm, and output the predicted comprehensive energy demand of the park. Experiment shows that after applying this method, the predicted values fluctuate between 1.410%-2.384%, RMSE values fluctuate between 176.4 MW-205.3 MW, indicating that the error of the predicted results using this method is relatively small.

Keywords: comprehensive energy system of the park; energy demand; cold/hot/electrical loads; crossover algorithm; echo state network; demand forecast.

DOI: 10.1504/IJETP.2024.138535

International Journal of Energy Technology and Policy, 2024 Vol.19 No.1/2, pp.2 - 16

Received: 18 May 2023
Accepted: 19 Oct 2023
Published online: 10 May 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article