基于改进天牛须算法-优化极限学习机的地源热泵能耗预测研究  

Research on Energy Consumption Prediction of Ground Source Heat Pump Based on Improved Beetle Antennae Search Algorithm-extreme Learning Machine

作　　者：刘涛 徐成良[1] 陈焕新[1] LIU Tao;XU Chengliang;CHEN Huanxin(School of Energy and Power Engineering,Huazhong University of Science and Technology,Wuhan,Hubei 430074,China)

机构地区：[1]华中科技大学能源与动力工程学院

出　　处：《制冷技术》2019年第3期6-11,共6页Chinese Journal of Refrigeration Technology

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

摘　　要：针对极限学习机随机产生输入权值和隐层偏置导致模型不稳定的问题,本文提出了一种基于改进天牛须算法-优化极限学习机(BSAS-ELM)的地源热泵能耗预测模型,此模型收敛速度快、泛化能力强。将BSAS-ELM的预测结果与极限学习机(ELM)及支持向量机(SVM)进行比较表明,基于BSAS-ELM的地源热泵能耗预测模型能显著提高能耗预测精度。Aiming at the problems that extreme learning machine generates input layer weights and hidden layer thresholds randomly make the network model unstable,a prediction method for ground source heat pump energy consumption is proposed based on an improved beetle antennae search algorithm-extreme learning machine(BSAS-ELM),and it has the advantages of faster convergence speed and stronger generalization ability.Compared with popular support vector machine(SVM)method and extreme learning machine(ELM),the BSAS-ELM model yields higher accuracy for ground source heat pump energy consumption prediction.

关 键 词：地源热泵系统 能耗预测 天牛须搜索 极限学习机 时间序列预测 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]