摘要
在自搭建的液环法节流机构流量特性试验台上,采用R22制冷剂,试验研究节流阀开度(流通面积)、节流前后压差、入口密度、入口过冷度、出口比容、干度以及阀头半锥角和径向间隙对电子膨胀阀制冷剂流量系数的影响;利用Levenberg-Marquart(LM)方法训练BP网络,得到了各影响因素与流量系数的量化关联网络,并进行了试验验证。结果表明,得出的关联网络误差较小,介于±11.0%之间。
A test bench based on liquid ring method（LRM） was introduced. Experiments and analyses were made about the influence of valve＇s opening （flowing area）, pressure drop, inlet density, inlet supercooling, outlet specific volume and valve head linetype to mass flow coefficient utilizing R22. A correlation based on BP network training by Levenberg-Marquart（LM）Method was acquired, and the results demonstrate the relative deviation ranging from -11.0% to ＋ 11.0%.
出处
《上海第二工业大学学报》
2006年第2期147-152,共6页
Journal of Shanghai Polytechnic University
关键词
工程热物理
流量系数
神经网络
电子膨胀阀
engineering thermophysics
mass flow coefficient
neural network
electronic expansion valve
