大温升复叠式高温热泵压缩机输气系数预测模型研究  

Study on prediction model of gas transfer coefficient of high temperature cascade heat pump compressor with large temperature rise

作　　者：胡晓微[1] 王佳文 董胜明 田绅 张蓓 张玉瑾 HU Xiaowei;WANG Jiawen;DONG Shengming;TIAN Shen;ZHANG Be;ZHANG Yujin(Tianjin Key Laboratory of Refrigeration Technology,Tianjin University of Commerce,Tianjin 300134,China)

机构地区：[1]天津商业大学天津市制冷技术重点实验室,天津300134

出　　处：《流体机械》2022年第8期82-88,共7页Fluid Machinery

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(20YDTPJC01150,21YDTPJC00930);天津市科学技术委员会青年项目(18JCQNJC77500);2016年河北省教育厅高等学校科学研究计划项目(QN2016210)。

摘　　要：为了准确获得压缩机输气系数,搭建了复叠式高温热泵系统试验台,对此开展理论及试验研究。建立了压缩机输气系数的机理模型、拟合关联式模型、BP人工神经网络模型和优化后的BAS-BP人工神经网络模型,通过4种模型分别对压缩机输气系数进行预测,结果表明:其经过优化后的BAS-BP人工神经网络模型预测高、低温压缩机输气系数结果平均误差为1.61%和1.02%,较BP人工神经网络模型误差降低了1.01%和0.75%,较机理模型误差降低了4.56%和6.87%,较拟合关联式模型误差降低了1.63%和2.16%;RMSE都为0.0086,R2分别为1.0008,1.0908,在4种模型中拟合性能最佳,表明优化后的BAS-BP人工神经网络模型比机理模型和拟合关联式模型对压缩机输气系数有着更好的测算能力和良好的泛化能力。In order to obtain the compressor gas transmission coefficient more accurately,a test bench of the cascade hightemperature heat pump system was built to carry out theoretical and experimental research on this issue.The mechanistic model of compressor gas transmission coefficient,the fitted correlation model,the BP artificial neural network model and the optimized BAS-BP artificial neural network model were established,and the compressor gas transmission coefficient was predicted by each of the four models.The research results indicate that the average errors of the optimized BAS-BP artificial neural network model in predicting the gas transmission coefficient of high and low temperature compressor are 1.61%and 1.02%,which are 1.01%and 0.75%lower than that of BP artificial neural network model,and 4.56%and 6.87%lower than that of the mechanism model.Compared with the fitting correlation model,the errors are reduced by 1.63%and 2.16%.The RMSE was 0.0086,R2 were 1.0008 and 1.0908.The fitting performance is the best among the four models,indicating that the optimized BAS-BP artificial neural network model has better calculation ability and good generalization ability for gas transmission coefficient of the compressor than the mechanism model and the fitting correlation model.

关 键 词：复叠式高温热泵 压缩机 输气系数 BP人工神经网络 BAS算法 

分 类 号：TH3[机械工程—机械制造及自动化] TK43[动力工程及工程热物理—动力机械及工程]