资源描述:
《基于ann的太阳能集热器热特性仿真的研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、摘要集热器出口气流温度及温升的影响因素很多,有些因素不可控制,且有交互作用,采用传统数学方法模拟较复杂。利用神经网络对集热器出口气流温升建立仿真模型有助于优化集热器参数,为高效集热器的设计提供一定参考依据。主要研究内容如下:(1)通过太阳能集热器热性能试验,分析了太阳能空气集热器出口气流温度与太阳辐射、环境温湿度及其变化量的关系,集热器吸热板表面温度与集热器长度、吸热板材质的关系,确定太阳能空气集热器出口气流温升的主要影响因素。(2)利用BP神经网络建立集热器出口气流温升预测模型,得到拟合度为0.99019,最大绝对误差为8.2013%,平均绝对百分比误差
2、为5.1126%。总而得出,所建立模型预测精度较高的结论,但在某些时刻预测误差比较大。(3)因BP网络由于初始权值选则不当陷入局部极小点,速度慢的缺点,从而过早收敛,导致模型预测精度较低。引入改进BP网络模型进行建模,得到拟合度为0.9945,最大绝对误差为2.9245%,平均绝对百分比误差为1.9062%,结果表明,改进BP网络集热器出口气流温升模型,在试验条件范围内,实测值与模型预测值拟合较好,可以用改进BP网络模型较准确地预测集热器出口气流温升。关键词:太阳能集热器;试验分析;BP神经网络;改进BP网络ModelingAnalysisofTherma
3、lcharacteristicofSolarCollectorsBasedonArtificialNeuralNetworkAbstractThefactorsofsolarcollectorairflowtemperatureandthetemperaturechangingarecomplexanddifficulttocontrol.Meanwhile,traditionalmathematicalmethodsarenotappliedtosimulateeasilybecauseoftheinteractionofthefactors.Howev
4、er,tosimulatethesolarcollectoranditsmoisturecontentwiththeneuralnetworkscanimprovethereferencescollectionfortheautomaticcontrolofthesystem.Thisarticleismainlyfocusedonthefollowingstudies:(1)Theprimaryfactorsofsolarcollector’stemperaturechangingaredeterminedthroughtheexperimentsfor
5、thethermalcharacteristicofsolarcollector,whichanalyzethesolarcollectorairflowtemperatureandthesolarradicalization,therelationshipbetweenthesurroundingstemperatureandhumidityandtherelationshipbetweenthesolarcollector’ssurfacetemperature,thelengthandthematerial.(2)WithBPneuralnetwor
6、ksinModelingAnalysisofthetemperaturechangingofthesolarcollector,theresultsofthebestfittinggoodnessupto0.99019,themaximumabsoluteerrorat8.2013%andtheaverageabsolutepercentageerrorat5.1126%canbeoffered.Accordingtotheseresults,thismodelcanoffermoreprecisepredictions,however,thepredic
7、tingerrorsstillexistconstantly.(3)BecauseoftheslowspeedofBPneuralnetworkwithinappropriateapplicationoftheinitialweightsofneuralnetwork,thepredictingaccuracyofthemodelisreduced.TakingadvantagesofimprovingBPneuralnetworktocontributemodelscangettheresultsofbestfittinggoodness0.9945;t
8、hemaximumabsoluteerror2.9245%andt