基于GA-BP的压缩机数据驱动建模研究  

Research on Compressor Data Drive Modeling Based on GA-BP

作　　者：王佳文 胡晓微[1] WANG Jiawen;HU Xiaowei(School of Mechanical Engineering,Tianjin University of Commerce,Tianjin 300134,China)

机构地区：[1]天津商业大学机械工程学院,天津300134

出　　处：《湖北民族大学学报（自然科学版）》2022年第1期115-120,共6页Journal of Hubei Minzu University:Natural Science Edition

基　　金：天津市自然科学基金项目(18JCYBJC90500);天津市技术创新引导专项基金项目(20YDTPJC01150)。

摘　　要：准确计算压缩机热力性能,对于压缩制冷热泵系统的设计和优化有重要作用,压缩机机理模型中存在经验系数的选择,这对于复杂工况下压缩机运行时模型的准确度提出挑战.应用数据驱动模型对压缩机进行建模,以压缩机进出口压力和进口温度为输入参数,以压缩机功率、输气量及出口温度为输出能数,利用压缩机机理模型、BP人工神经网络模型和GA-BP模型对压缩机性能进行预测.结果表明,数据驱动模型预测压缩机性能结果相对误差均在4%以内,GA-BP模型预测压缩机性能参数结果相对误差均在1%以内,这表明基于GA-BP的压缩机模型对压缩机性能参数有较好的预测能力.Accurate calculation of the thermodynamic performance of the compressor plays an important role in the design and optimization of the compressor refrigeration heat pump system.The selection of the empirical coefficient in the compressor mechanism model is a challenge to the accuracy of the compressor operation model under complex working conditions.In this paper, a data-driven model is proposed to model the compressor.The compressor inlet and outlet pressure and inlet temperature are taken as the input parameters, and the compressor power, air volume and outlet temperature are taken as the output parameters.The compressor mechanism model, BP artificial neural network model and GA-BP model are used to predict the compressor performance.The results show that the relative errors of the compressor performance prediction results of the data-driven model are all within 4%,and the relative errors of the compressor performance parameter prediction results of the GA-BP model are all within 1%,which indicates that the compressor model based on GA-BP has a good ability to predict the compressor performance parameters.

关 键 词：压缩机 GA算法 BP神经网络 数据驱动建模 

分 类 号：TH45[机械工程—机械制造及自动化] TP183[自动化与计算机技术—控制理论与控制工程]