基于改进型RBF神经网络的建筑用电能耗预测  

Building Energy Consumption Prediction Based on Improved RBF Neural Network

作　　者：李琳 杨新华[1] 曹磊 韩永军 LI Lin;YANG Xin-hua;CAO Lei;HAN Yong-jun(College of Electrical and Information Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Gansu Electric Power Research Institute,Lanzhou 730070,China)

机构地区：[1]兰州理工大学电气工程与信息工程学院,兰州730050 [2]甘肃省电力科学研究院,兰州730070

出　　处：《建筑节能（中英文）》2021年第1期81-86,139,共7页Building Energy Efficiency

摘　　要：径向基函数(Radial Basis Function,RBF)神经网络由于其网络结构简单、网络适应性好、学习过程收敛速度快等优点被运用于电力负荷预测领域。在将其应用于建筑用电能耗预测的过程中,由于对目前已有的建筑能耗数据和影响能耗的关键因素分析不足,以及网络参数不易确定,将导致预测精度无法满足实际需求。采用粒子群算法(Particle Swarm Optimization,PSO)及列文伯格-马夸尔特算法(Levenberg-Marquard,LM)优化模型参数,并以大型办公建筑为研究对象确定影响能耗的约束条件,将其作为网络输入参数进行学习,以提高预测模型的准确性。实验结果表明,改进后的RBF算法平均绝对误差和最大相对误差分别降低了2.2%和4.76%,误差保持在2%以内,具有更高的预测精度。The Radial Basis Function(RBF)neural network has been applied in the field of power load forecasting because of its simple network structure,good network adaptability and fast convergence of learning process.However,the prediction accuracy of building electricity consumption can not meet the actual demand because of insufficient analysis of the existing building energy consumption data and the key factors affecting energy consumption,as well as the difficult determination of network parameters.Particle Swarm Optimization(PSO)and Levenberg-Marquard(LM)are used to optimize the parameters of the model.And the constraints affecting energy consumption of large office buildings are determined as input parameters of the network to improve the accuracy of the prediction model.The experimental results show that the average absolute error and the maximum relative error of the improved RBF algorithm are reduced by 2.2%and 4.76%respectively,and the error is kept within 2%,which has higher prediction accuracy.

关 键 词：能耗预测 RBF神经网络 粒子群算法 列文伯格-马夸尔特算法 

分 类 号：TU855[建筑科学] TM769[电气工程—电力系统及自动化]