两阶段特征选择的冷水机组能耗预测方法  

Two-stage Feature Selection Algorithm for Chiller Energy Consumption Prediction

作　　者：陈静 陈焕新[2] 徐成良[2] CHEN Jing;CHEN Huanxin;XU Chengliang(China-EU Institute for Clean and Renewable Energy,Huazhong University of Science and Technology,Wuhan,Hubei 430074,China;School of Energy and Power Engineering,Huazhong University of Science and Technology,Wuhan,Hubei 430074,China)

机构地区：[1]华中科技大学中欧清洁与可再生能源学院,湖北武汉430074 [2]华中科技大学能源与动力工程学院,湖北武汉430074

出　　处：《制冷技术》2019年第5期22-26,共5页Chinese Journal of Refrigeration Technology

基　　金：国家自然科学基金（No.51876070,No.51576074）

摘　　要：本文针对冷水机组运行参数繁多的特点,提出了一种结合ReliefF和最大相关最小冗余(minimal-Redundancy-Maximal-Relevance,mRMR)算法的两阶段特征选择算法。实验利用筛选出的特征子集建立支持向量回归模型进行能耗预测,并与单一特征选择算法的结果进行比较。结果表明,与单独使用ReliefF和mRMR算法的预测模型相比,ReliefF-mRMR预测模型的最高精度分别提高了2.22%和0.83%,平均预测精度分别提高了3.92%和8.11%;在最优精度的情况下,ReliefF-mRMR预测效率比ReliefF算法仅降低了0.87%,比mRMR算法提高了53.60%。Aiming at the various operating parameters of water chillers,a two-stage feature selection algorithm combined ReliefF and minimal-Redundancy-Maximal-Relevance(m RMR)algorithm is proposed in this paper.The selected feature subset is used to establish a support vector regression model for energy consumption prediction,and the results are compared with those of a single feature selection algorithm.The results show that,compared with the prediction model using ReliefF and mRMR algorithms alone,the maximum accuracy of the ReliefF-mRMR prediction model is increased by 2.22%and 0.83%,respectively,and the average prediction accuracy is improved by 3.92%and 8.11%,respectively.In this case,the ReliefF-mRMR prediction efficiency is only 0.87%lower than that of the ReliefF algorithm,and 53.60%higher than that of the mRMR algorithm.

关 键 词：能耗预测 特征选择 RELIEFF 冷水机组 最大相关最小冗余 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]