基于人工神经网络的转轮除湿系统建模方法研究  

Study on the modeling method for desiccant wheel system based on Artificial Neural Network

作　　者：丛华[1] 杨永利[1] 江鹏程[1] 冯辅周[1] 张红星[2] 李耀凯 郝晋峰[2] 

机构地区：[1]装甲兵工程学院机械工程系,北京100072 [2]太原卫星发射中心,山西太原030027

出　　处：《可再生能源》2016年第1期125-130,共6页Renewable Energy Resources

基　　金：军队科研支撑项目

摘　　要：转轮除湿系统具有除湿精确度高且能耗低的优点,但其除湿机理相对复杂,除湿过程具有很强的非线性和时变性,对其建模具有一定难度。现有的建模手段以物理方法为主,建模过程复杂且模型误差大。本文引入了神经网络技术,建立了转轮除湿系统的两种神经网络模型——反向传播(BP)网络模型和径向基函数(RBF)网络模型,通过对两种模型进行运算分析,可知两种神经网络对于转轮除湿系统具有良好的适用性,同时RBF网络模型具有较高的精度,应用前景较好。Desiccant wheel system（DWS） has the merits of high dehumidifying accuracy and low energy consumption, whereas its dehumidifying principle is relatively complex and there exists strong nonlinear and time-varying in dehumidifying process, which brings in many difficulties to its modeling.This paper introduces Artificial Neural Network（ANN） to model desiccant wheel system, and two related ANN models were built for DWS- BP model and RBF model. By computing and analyzing with the two models, it can be concluded that the two models are proper to model DWS, and the RBF model is of higher precision and has better application prospect.

关 键 词：转轮除湿 人工神经网络 建模 BP网络 RBF网络 

分 类 号：TK83[动力工程及工程热物理—流体机械及工程]