資源描述:
《基于模糊觀測器的故障診斷和容錯(cuò)控制研究》由會(huì)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、南京航空航天大學(xué)碩士論文摘要本文針對(duì)非線性系統(tǒng)研究了基于模糊觀測器的故障診斷和故障調(diào)節(jié)問題,并將提出的算法用Matlab實(shí)現(xiàn)仿真驗(yàn)證。文中首先介紹了故障診斷與容錯(cuò)控制的發(fā)展及現(xiàn)狀,同時(shí)對(duì)模糊控制、神經(jīng)網(wǎng)絡(luò)以及自適應(yīng)控制技術(shù)的發(fā)展給出了簡單的介紹。在用T-S模型對(duì)非線性系統(tǒng)進(jìn)行建模的基礎(chǔ)上,分別設(shè)計(jì)了模糊檢測觀測器以及模糊自適應(yīng)診斷觀測器,前者用于檢測故障,后者用于估計(jì)執(zhí)行器的增益損失。利用診斷觀測器估計(jì)出的故障信息設(shè)計(jì)故障調(diào)節(jié)律給執(zhí)行機(jī)構(gòu)以相應(yīng)的補(bǔ)償,恢復(fù)系統(tǒng)的性能。并用倒立擺仿真驗(yàn)證了其有效性。進(jìn)一步,提出了一種基于T-S模型的魯棒故障檢測方法。針對(duì)在對(duì)非線性系統(tǒng)用T-
2、S模糊模型建模時(shí)存在不確定項(xiàng)的系統(tǒng),設(shè)計(jì)魯棒模糊觀測器,給出了系統(tǒng)穩(wěn)定的約束條件,并將其用于故障檢測。仿真結(jié)果證明,設(shè)計(jì)的觀測器對(duì)于系統(tǒng)模型的不確定性有很好的魯棒性,且能夠及時(shí)的檢測出故障的發(fā)生。最后,將神經(jīng)網(wǎng)絡(luò)與T-S模糊模型相結(jié)合,進(jìn)行非線性系統(tǒng)建模,對(duì)于考慮了加性故障的非線性系統(tǒng),用神經(jīng)網(wǎng)絡(luò)在線估計(jì)故障信息;并對(duì)系統(tǒng)設(shè)計(jì)具有容錯(cuò)功能的控制律恢復(fù)系統(tǒng)的性能。將提出的算法用簡化的F-16模型進(jìn)行仿真,仿真結(jié)果表明設(shè)計(jì)的觀測器能夠及時(shí)地估計(jì)出故障信息,在設(shè)計(jì)的容錯(cuò)控制律的調(diào)節(jié)下,系統(tǒng)能夠快速的恢復(fù)原來的性能。關(guān)鍵詞:故障調(diào)節(jié)、故障檢測、模糊觀測器、非線性系統(tǒng)、T-S模型、
3、神經(jīng)網(wǎng)絡(luò)i基于模糊觀測器的故障診斷與容錯(cuò)控制研究AbstractFuzzyobserverbasedfaultdiagnosisandfaultaccommodationproblemsforakindofnonlinearsystemsarestudiedinthisdissertation.TheproposedmethodsaresimulatedbyMatlabsoftware.AtFirst,thehistoryandthedevelopmentoffaultdetectionandisolation(FDI)andfaulttolerantcontrol(FT
4、C)aresurveyed.Also,fuzzycontroltheory,theadaptivecontroltechniqueandneuralnetworkareintroduced.ThenonlinearsystemismodeledbyT-Sfuzzymodel.Afuzzydetectiveobserverandafuzzyadaptivediagnosticobserveraredesignedrespectively.Theformerisusedtodetectthefaultandbasedonthelatter,afaultestimatesche
5、meisproposed.Then,thefaultestimationisusedtoreconfigurethecontrollawtorecoverthesystemperformance.Theefficiencyofthealgorithmisdemonstratedbyaninvertpendulumexample.Furthermore,afaultdetectionmethodofT-Smodelwithuncertaintiesispresented.Arobustobserverisdesignedforaclassofuncertainnonline
6、arsystemsrepresentedbyT-Smodel,whichisusedtodetectthesystemfault.Stabilityconditionsofsuchobserverareexpressedintermsoflinearmatrixinequalities(LMIs).Thesimulationresultvalidatestheproposedobserverhasgoodabilityagainstuncertaintyandcandetectthefaultintime.Finally,afaultaccommodationapproa
7、chbasedonneuralnetworkfuzzyobserverfornonlinearsystemsispresented.Theneuralnetworkisconstructedtoapproximatethefaulton-line.Thefuzzyobservercanestimatetheshapeofthefault,whichcanbeusedtoreconfigurethecontroller.TheLyapunovtheoremisusedtoprovethestabilityofthefaultys