Title: A short-term prediction method of building energy consumption based on gradient progressive regression tree
Authors: Qiuhong Zhao
Addresses: Changchun University of Architecture and Civil Engineering, Changchun 13000, China
Abstract: In order to overcome the large error of traditional methods in predicting building energy consumption, a short-term prediction method of building energy consumption based on gradient progressive regression tree is proposed. Building benchmark model is constructed by using eQuest software to obtain the main parameters affecting building energy consumption, build the impact index system of building energy consumption, and extract the main impact factors. Genetic algorithm is used to extract the characteristics of building energy consumption, combined with gradient progressive regression tree method to build a short-term prediction model of building energy consumption, and complete the short-term prediction of building energy consumption. The experimental results show that the minimum relative error of the proposed method is about 0.1, the absolute error is about 0.2, and the maximum standard deviation is 0.41.
Keywords: gradient regression tree; building benchmark model; influencing factors of energy consumption; genetic algorithm; building energy consumption prediction.
International Journal of Global Energy Issues, 2022 Vol.44 No.2/3, pp.182 - 197
Received: 27 Aug 2020
Accepted: 24 Nov 2020
Published online: 10 Mar 2022 *