RBF神经网络用于家用空调系统匹配特性的研究  

Simulation of RBF Neural Network in Domestic Air-Conditioner Matching

作　　者：林用满[1,2] 王惠龄[1] 杨九铭 贾庆贤 

机构地区：[1]华中科技大学制冷与低温工程系,武汉430071 [2]广东美的制冷设备股份有限公司,广东佛山528311

出　　处：《机械制造》2009年第9期5-8,共4页Machinery

基　　金：国家自然科学基金资助项目(编号:604720067);广东省自然科学基金资助项目(编号:04205783)

摘　　要：RBF神经网络是目前应用较多的一种神经网络。它能以任意精度逼近任意非线性函数,具有良好的逼近性能,并且结构简单,是一种性能优良的神经网络。因此,将RBF神经网络应用于家用空调匹配仿真研究时具有独特的优势。提出采用RBF神经网络估算制冷量和压力来优化研发过程,仿真结果表明,RBF神经网络运用于家用空调匹配仿真,能够精确仿真空调制冷量和低压力等参数,并预测制冷量和压力,能有效地减少家用空调匹配时间,提高研究效率。The RBF neural network is one of the networks widely used nowadays. It can approximate any non - linear functions with any precision, featuring good approximation performance and simple structure. The RBF neural network has a unique advantage when it is applied in the simulation of domestic air - conditioner matching. It is presented that the RBF neural network is adopted to estimate the refrigerating capacity and low pressure so as to optimize the process of research and development. The simulation results show that the refrigerating capacity and low pressure of air conditioners can be simulated precisely when the RBF neural network is applied in the simulation of domestic air - conditioner matching and the refrigerating capacity and pressure can be predicted, thus cutting down the matching time and increasing research work efficiently.

关 键 词：RBF神经网络 匹配仿真 制冷剂充灌量 

分 类 号：TB611[一般工业技术—制冷工程]