模糊神经网络及其在热工建模与控制中的应用研究

《模糊神经网络及其在热工建模与控制中的应用研究》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、东南大学硕士学位论文模糊神经网络及其在热工建模与控制中的应用研究姓名：杨世忠申请学位级别：硕士专业：热能工程指导教师：吕剑虹20070601摘要AbstractWiththeapplicationofDesinautomaticcontrolsystemforthermalpowerplant，advancedcontrolstrategieshavebeenadoptedintIlerTnalprocessescontr01．But∞whatCallbeseerli11manythermalpowerplant，weyetcanno

2、tobtaingoodperformanceinthermalprocessescon订ol，becausethetraditionalPIDcontrolsystemshavebeenwidelyusedinalmosta11thethelTnalpowerplants，whichhasabadeffectonthesafetyandeconomicalefficiencyofpowerplants．111emainrel舶onsare∞follows．First,whenthethermalsystems’statechange,

3、thedynamicperfommceofcontrolledplantsoftenvariedgreatly,whichmeansthatthereisseriousnonlinearityincontrolledplants．SothetraditionalPIDcontrolsystem,whichisdesignedonthebasisofplantscharaeteristieonterrainloadpoint，cannotbeusedtorealizeprocessoptimaleontrolinwholeloadran

4、ge．Inaddition．onthecontrarytothelinearityofPIDeontr01systems，therearenonlinearresponsesincontrolledplants．Asweallknow,nomatterhowweimprovetheeapabilityoflinearPDcontrolsystems．theremustbealimitationintheimprovementofcontroleffects．Soweshouldadoptnewoptimalcontrolsystems

5、whicharebasedonwholenonJinearmodelsforthermalprocessestoimprovethecontrolperformance．Obviously，thewholenonlinearmodelingforthermalprocessesisthebasistooptimalcontrolinwholeloadrage．Thefuzzysystemhasseveraladvantages：itcalldepicttheuncertaintyofplantswell；ithasacapabilit

6、yofuniversalapproximation；itssinglerulecanapproachthelocaldynamicscharacterizationofthenonlinearsystem；itsmodelingresultisrelativelytransparent．AUofthesevirtuesareinfavorofthedesignandapplicationofhi曲．performancemodelcontroller．What’smore，duetothepowerfulabilityofnonlin

7、earfunctionsapproximation，andwiththecharacteristicsofadaptivelearning，parallelanddistributedprocessing，strongrobustnessandfaulttolerance，ArtificialNeuralNetwork(ANN)hasbeenaneriectiveapproachtomodelandcontr01theunknownanduncertainoontrolsysreins．Justonthebasisof出eseappl

8、icailonbackground．thispapercombinesANNwithfuzzysystem,andstudiestouscFuzzyNenralNetwork(FNN)tosolvethecorrelat