基于BP神经网络的压缩机性能预测模型的建立  

Model of Compressor Performance Prediction Based on Error Back-propagation Artificial Neural Network

作　　者：王伟[1] 姚杨[2] 马最良 

机构地区：[1]北京工业大学,北京100022 [2]哈尔滨工业大学,黑龙江哈尔滨150090

出　　处：《流体机械》2005年第9期21-24,共4页Fluid Machinery

基　　金：国家自然科学基金项目(50278021)

摘　　要：利用BP网络,以全年运行的空气源热泵冷热水机组性能预测为目标,对半封闭往复式压缩机建立了性能预测模型。以网络的泛化能力作为衡量建模效果的标准,同时指出模型的建立过程就是网络最佳结构参数的搜索过程。计算实例显示,网络的预测效果与训练效果吻合较好,训练代价适中,所建模型可用于压缩机全年运行的性能预测与故障诊断中。An artificial neural network that uses the error back-propagation training algorithm is proposed to predict the whole year performance of the compressor working in the air-source heat pump heater/chiller unit. The generalization ability of the network is used to judge if the model can be safely used. It is pointed out that the modeling process is optimizing the network architecture parameters. The modeling results indicate that the model can be used in the performance prediction and fault diagnosis of the compressor.

关 键 词：BP网络 空气源热泵冷热水机组 性能预测 

分 类 号：TH45[机械工程—机械制造及自动化] TP39[自动化与计算机技术—计算机应用技术]