节点文献

基于BP神经网络的R32热力性质显式快速计算模型

FAST CALCULATION MODEL OF THERMODYNAMIC PROPERTIES OF R32 REFRIGERANT BASED ON BP NEURAL NETWORK

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 田镇谷波方继华

【Author】 Tian Zhen;Gu Bo;Fang Jihua;Institute of Refrigeration and Cryogenics Engineering,Shanghai Jiaotong University;

【机构】 上海交通大学制冷与低温工程研究所

【摘要】 制冷剂热力性质计算模型的优劣直接影响系统仿真的速度和稳定性。对R32制冷剂的湿蒸汽区和过热蒸汽区(153~423K,0.00469~50 bar)分别以温度、干度和温度、压力为设计变量,通过BP神经网络确定设计变量与热力性质之间的非线性关系,进而建立R32热力性质的显式快速计算模型;该模型不存在迭代,具有较高的计算精度,同时可以保证计算速度和稳定性,满足制冷系统仿真的需求。将模型计算结果与REFPROP7.0计算值进行对比,结果表明,所有快速计算模型的平均相对误差不超过1.7349%,最大相对误差不超过9.8826%;其计算速度比REFPROP7.0提高了2个数量级。

【Abstract】 Calculation models of thermodynamic properties for refrigerants influence the speed and stability of system simulation directly.Temperature,vapor quality and temperature,pressure were used as design variables for the liquid-vapor region and the superheated vapor region(153~423 K,0.00469~50 bar),respectively.The relationship between thermodynamic properties and design variables of refrigerant were determined by BP neural network,and then the fast calculation models of thermodynamic properties of R32 refrigerant were established.These calculation models with no iteration can not only achieve the accuracy,but also guarantee the calculation stability and high speed,which can meet the requirement in refrigeration system simulation.Comparisons were made between the models and REFPROP7.0.The mean relative error and maximum relative error for all fast calculation models are less than 1.7349%and 9.8826%,respectively.The calculation speed of the fast calculation model is two orders faster than those of REFPROP7.0.

  • 【会议录名称】 2013中国制冷学会学术年会论文集
  • 【会议名称】勇于创新,服务发展——2013中国制冷学会学术年会
  • 【会议时间】2013-11-05
  • 【会议地点】中国湖北武汉
  • 【分类号】TP18;TB64
  • 【主办单位】中国制冷学会
节点文献中: 

本文链接的文献网络图示:

本文的引文网络