資源描述:
《船舶航向非線性系統(tǒng)模型參考神經(jīng)網(wǎng)絡(luò)自適應(yīng)控制》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、摘要本文對神經(jīng)網(wǎng)絡(luò)模型參考自適應(yīng)控制及其在船舶航向控制中的應(yīng)用進(jìn)行了系絞豹磷究。首先針對船舶線性模型,在討論了模型參考自適應(yīng)控制理論∞基礎(chǔ)上,設(shè)計了模型參考自邋應(yīng)自動舵,并對其進(jìn)行了仿真。贊瓣勰熬靛淘{靼線毪不確定系統(tǒng),箍出了一種凳神經(jīng)兩絡(luò)和模登參考自適應(yīng)控制結(jié)合在一起的新的控制冀法,此算法首先遙過系統(tǒng)的已知動態(tài)特性設(shè)詩一個穩(wěn)定的反饋控制器,然后利用RBF神經(jīng)阿絡(luò)逼近未知非線性,從而消除不確定性的影響。粳重鑫適應(yīng)移歪惑弼是萋于Lyapunov穩(wěn)定性理論癸現(xiàn)的,避免了遞艦訓(xùn)練涎程,保證了自適威控制爨法匏穩(wěn)
2、定憾??貏澠髟O(shè)計畦續(xù)會7餐襻控鍘方法,來消除或減d,#l-部干擾以及神經(jīng)潮絡(luò)逼近誤差,使得系統(tǒng)具有一定的魯棒型。熬個籜法不僅僳諼整個溺環(huán)系統(tǒng)穩(wěn)定,麗鼠能使系統(tǒng)的跟蹤誤麓收斂子零的鄰域內(nèi)。黢蜃恕此算法遴行了仿賓,績囊綴豢令人滿意。將算法在不同外界條件,如襁風(fēng)浪干擾時、在不同載運狀態(tài)下、在不同速度下漣行仿糞,仿真結(jié)采表明該算法的控制效果照著,并與常規(guī)模型參考翻適應(yīng)簿法瓣控割縫果進(jìn)毒亍了魄較,誕弱了毒棗經(jīng)爨絡(luò)模型參考鑫逶痰控制葵法靜援麓捷越于傳統(tǒng)模型參考自適成控制算法的性能。勇了檢驗本文掇出韻船舶航向自動舵的
3、性能,研究了船舶運動數(shù)學(xué)模型,以及毽擐威、渡、浚秘{#線性力穆瘸在艇靛上熬努秀予挽力窩力矩懿詩算穰型。甏翔Matlab的Simulink環(huán)境實現(xiàn)了對各種船舶航向自動舵的大凝仿真試驗。關(guān)鍵詞:模型參考自適應(yīng)、神經(jīng)網(wǎng)絡(luò)、RBF、船舶航向控制Abstract確isthesishassystematicallyresearchedneuralnetworkmodelreferenceadaptivecontrolanditsapplication埝ship'scoul_sesteeringcontr01.Int
4、histhesis,aireedatshiplinearmathematicalmotionmodel,themodelreferenceadaptiveautopilotisfirstdesignedbasedontheanalysisofthemodelreferenceadaptivecontroltheory.琢combiningmodelreferenceadaptivecontrolandneuralnetworkcontrol,anovelmodelreferenceadaptiveneu
5、ralnetworkcontrolschemeisproposedfortheshipsteering'snonlinearmodel+顙theprocedure,theRBFneuratnetworks‘a(chǎn)leusedtoapproximatenonlinearsystems;TheweightscabbeachievedbyUSeofLyapunoveapproach。Tocompensatetheapproximationerrorandattenuatetheexternaldisturbanc
6、e,arobusttechnologyisintroducedt彝theproposedscheme.AsymptoticstabilityoftheclosedloopsystemisestablishedintheLyapunovLheory,andthetrackingeITorsconvergence撼8neighborhoodofzero.Atlast,theschemeisappiliedtOship'scoursesteeringcontrolandthesimutatio馥curvess
7、howthatthedesiredresults曩糙attained.‘Themodelreferenceadaptiveneuralnetworkcontrolautopilotissimulatedunderdifferentconditions,suchasexterndisturbances,loadings,velocities,ere。羽嘛simulationresultsshowtheeffectivenessandthesalientperformmaceoftheproposedcon
8、trolscheme.Andthecomparionbetweenthemodelreferenceadaptiveneuralnetworkschemeandconventionalmodelreferenceadaptiveschemedemonstratesthattheperformanceofthefoixrlerisbetterthanthatofthelatter.Toverifytheperformanceoftheauto