資源描述:
《基于動(dòng)態(tài)神經(jīng)模糊模型的船舶運(yùn)動(dòng)智能控制》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(kù)。
1、中文摘要摘要水面船舶運(yùn)動(dòng)控制一直是備受關(guān)注的研究領(lǐng)域。船舶動(dòng)態(tài)具有大慣性、大時(shí)滯、非線性的特點(diǎn),由于受航速、風(fēng)浪流等外界干擾以及量測(cè)誤差等因素的影響,建模參數(shù)甚至結(jié)構(gòu)的攝動(dòng)引起的不確定性,使得對(duì)船舶運(yùn)動(dòng)的控制具有相當(dāng)難度。同時(shí),它也是迄今國(guó)內(nèi)外一般基于線性控制理論和傳統(tǒng)智能控制設(shè)計(jì)的控制算法的實(shí)際效果與研究者的預(yù)期相差甚遠(yuǎn)的一個(gè)主要原因。為此,本文提出一種基于動(dòng)態(tài)神經(jīng)模糊模型的船舶航向智能控制方法。本文主要考慮航速變化引起的不確定性因素對(duì)船舶航向跟蹤控制的影響,通過(guò)將RBF神經(jīng)網(wǎng)絡(luò)與T-S模糊模型有機(jī)結(jié)合,構(gòu)造
2、結(jié)構(gòu)與參數(shù)在學(xué)習(xí)中能同時(shí)調(diào)整的動(dòng)態(tài)神經(jīng)模糊模型(DMM)。DNFM用于充分逼近建模參數(shù)時(shí)變的Norrbin非線性船舶模型的逆動(dòng)力學(xué)。訓(xùn)練好的DNFM與PID控制器并聯(lián)構(gòu)造自適應(yīng)控制器用于航向跟蹤控制,且以PID控制器的輸出作為自適應(yīng)律的自變量,進(jìn)一步在線調(diào)整DNFM的權(quán)值,設(shè)計(jì)的智能控制器能有效解決具有建模參數(shù)不確定性的船舶運(yùn)動(dòng)控制問(wèn)題。文中以中遠(yuǎn)集團(tuán)5446TEU集裝箱船為對(duì)象進(jìn)行仿真,通過(guò)設(shè)定兩種不同的航速變化工況,完成了對(duì)船舶逆動(dòng)力學(xué)的逼近;隨后通過(guò)設(shè)計(jì)基于DNFM的自適應(yīng)智能控制器,實(shí)現(xiàn)船舶航向的快速跟
3、蹤,有效地克服了建模參數(shù)不確定性的影響。關(guān)鍵詞:不確定性;動(dòng)態(tài)神經(jīng)模糊模型;航向跟蹤;智能控制英文摘要AbstractShippingmovementcontrolinwatersurfaceiscatchingmoreandmoreattentionintheseyears.Withthefeaturesofbiginertia,largetime—delayandnonlinear,themodelparameterperturbationcausedbyshipspeed,storm—flowinterf
4、erence,measurementerrorandSOon,thecomplexuncertaintyproblemshavebeenexistinginthisfieldalongtime.Allthesemaketheshipcourse-keepingcontrolproblemconsiderabledifficulty.Sofaritisalsothemainreasonforthediscrepancyoftraditionaldesignofintelligentcontrolalgorithm
5、sbasedonlinearcontroltheoryandtheactualresultsresearchersexpected.Therefore,inordertoresolvetheuncertaintiesandtherealizationofthedynamiccontrolofnonlinearsystems,thispaperpresentsamethodfortheshipcourseintelligentcontrolthatbasedondynamicneuro—fuzzymodel.Th
6、isarticleconsidertheimpactfortheshiptrackingcontrol,whichcoursedbytheuncertaintiesofthechangingspeed.BycombiningRBFneuralnetworkandT-Sfuzzymodel,settingllpadynamicneuro-fuzzymodel,networkparametersandstructureoftheruleslayercanregulatethemselvesinlearningpro
7、cess.DNFMfullyapproximationtheinversedynamicsofnonlinearshipmodelNorrbin,whosemodelingparameterstime-varyingwiththechangespeed.AnadaptivecontrollerbygoodtrainedDNFMinparallelwiththePIDcontrollerareusedtocoursetrackingcontr01.TakingtheoutputofPIDcontrollerasi
8、ndependentvariableoftheadaptivelaw,theDNFMweightWasadjustedfurtheronlineandshipcourseintelligentcontrollerWasdesignedforuncertaintyshipmotionmodel.Inthispaper,thesimulationtakestheCOSCO5446TEUla