資源描述:
《基于BP神經(jīng)網(wǎng)絡(luò)的PID控制器設(shè)計(jì)6.16.doc》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、中文摘要經(jīng)典PID控制算法作為一般工業(yè)過程控制方法應(yīng)用范圍相當(dāng)廣泛,原則上講它并不依賴于被控對象的具體數(shù)學(xué)模型,但算法參數(shù)的整定卻是一件很困難的工作,更為重要的是即使參數(shù)整定完成,由于參數(shù)不具有自適應(yīng)能力,因環(huán)境的變化,PID控制對系統(tǒng)偏差的響應(yīng)變差,參數(shù)需重新整定。針對上述問題,人們一直采用模糊、神經(jīng)網(wǎng)絡(luò)等各種調(diào)整PID參數(shù)的自適應(yīng)方法,力圖克服這一難題。一般情況下,一個(gè)自適應(yīng)控制系統(tǒng)能夠運(yùn)行,其相應(yīng)的參數(shù)要適應(yīng)現(xiàn)場狀況的變化,因此就必須根據(jù)現(xiàn)場的數(shù)據(jù)對相應(yīng)的參數(shù)進(jìn)行在線辨識或估計(jì)。對非時(shí)變參數(shù)可以通過一段時(shí)間的在線辨識
2、確定下來,但對時(shí)變參數(shù)系統(tǒng),必須將這個(gè)過程不斷進(jìn)行下去,因此要求辨識速度快或參數(shù)變化速度相對較慢,極大地限制了自適應(yīng)技術(shù)的應(yīng)用。為克服這種限制,本文利用文獻(xiàn)[1]的思想,將神經(jīng)網(wǎng)絡(luò)的技術(shù)應(yīng)用于參數(shù)辨識過程,結(jié)合經(jīng)典的PID控制算法,形成一種基于BP神經(jīng)網(wǎng)絡(luò)的自適應(yīng)PID控制算法。這一算法的本質(zhì)是應(yīng)用神經(jīng)網(wǎng)絡(luò)建立系統(tǒng)參數(shù)模型,將時(shí)變參數(shù)系統(tǒng)的參數(shù)變化規(guī)律轉(zhuǎn)化為神經(jīng)網(wǎng)絡(luò)參數(shù)模型,反映了參數(shù)隨狀態(tài)而變的規(guī)律,即當(dāng)系統(tǒng)變化后,可直接由模型得到系統(tǒng)的時(shí)變參數(shù),而無需辨識過程。在神經(jīng)網(wǎng)絡(luò)參數(shù)模型的基礎(chǔ)上,結(jié)合文獻(xiàn)[1]已知系統(tǒng)模型下P
3、ID控制參數(shù)的計(jì)算,推導(dǎo)出一種自適應(yīng)PID控制算法。通過在計(jì)算機(jī)上對線性和非線性系統(tǒng)仿真,結(jié)果表明了這種自適應(yīng)PID控制算法的有效性。關(guān)鍵詞自適應(yīng)PID控制算法,PID控制器,參數(shù)模型,神經(jīng)網(wǎng)絡(luò),BP算法AbstractClassicalPIDcontrolalgorithm,asageneralmethodofindustrialprocesscontrol,applicationscopeisbroad-ranged.Inprinciple,itdoesnotdependonthespecificmathematica
4、lmodelofthecontrolledplant,buttuningalgorithmparametersisaverydifficulttask.Tomoreimportant,eveniftuningtheparameteriscompleted,asparametersdonothaveadaptivecapacity,duetoachangeinenvironment,PIDcontroloftheresponseofthesystemdeviationgetworse,parametersneedtobere
5、-tumed.Inresponsetotheseproblems,peoplehavebeenusingtheadaptivemethodoffuzzy,neuralnetworkstoadjustPIDparameters,tryhardtoovercomethisproblem.Undernormalcircumstances,anadaptivecontrolsystemcanbecapableofrunning,andthecorrespondingparametersshouldadapttotllechange
6、instatusofthescene,sothecorrespondingparametersmustbebasedonthedataofthescenetoconductonlineidentificationorestimated.Non-time—varyingparameterscanbeconfirmedforaperiodofon-lineidentification,butthetime-varyingparameterssystemwillbenecessarytocontinuethisongoingpr
7、ocess,sotherequirementoffastidentificationortherelativeslowpaceofchangeofparameters,greatlylimitstheapplicationofadaptivetechnology.Toovercomethislimitation,thispaperusestheideologyofliterature[1],thetechnologyofneuralnetworkwillbeusedintheprocessofparameteridenti
8、fication,combiningclassicalPIDcontrolalgorithm,formsanadaptivePlDcontrolalgorithmbasedonBPneuralnetwork.Theessenceofthisalgorithmappliesneuralnetworktob