江水源热泵的广义生长-剪枝径向基网络预估  

PREDICTING GENERALIZED GROWING-PRUNING RADIAL BASIS NEURAL NETWORK OF RIVER WATER HEAT PUMP

作　　者：周洪煜[1] 高鹏飞[1] 童明伟[1] 陈孜虎[1] 陈小健[1] 

机构地区：[1]重庆大学动力工程学院,重庆400030

出　　处：《太阳能学报》2012年第4期647-652,共6页Acta Energiae Solaris Sinica

基　　金：国家科技支撑计划重点项目(2007BAB21B02-1-1)

摘　　要：基于广义生长-剪枝径向基神经网络(GGAP-RBF)对江水源热泵系统(RWHP)建模,通过模型预测江水源热泵系统相关性能参数。性能预估得到的模型有助于实现热泵系统最优化设计和节能运行。而相应模型由广义生长-剪枝(GGAP)算法确定隐层神经元数量为7。相应的均方根值和变异系数的百分比分别为0.0047和0.1363,决定系数R2值0.9998也较适宜。结果表明GGAP-RBF模型对RWHP系统量化模型具有很好的适用性。The modeling of the GGAP-RBF network in fiver water heat pump （RWHP） was carried out to predic relevant parameters of PWHP, which was propitious to realize the optimization design and efficient operation of the heat pump. And the GGAP-RBF model was determinated by the GGAP algorithm in which the hidden layer number was 7. The corresponding value of the root-mean squared（RMS） and the coefficient of variation（cov） were 0. 0047 and 0. 1363, respectirely, and the value of coefficient of multiple determinations （R2） was 0.9998, which was also satisfying. The results showed that there is applicability of the GGAP-RBF for the quantitative modeling of RWHP system.

关 键 词：江水源热泵系统 神经网络 预测 最优化 

分 类 号：TK529[动力工程及工程热物理—热能工程]