神經(jīng)網(wǎng)絡(luò)pid控制-預(yù)覽頁
【正文】 1)t an h ()?？捎檬街械模虼藢⒈平妮敵鼋?jīng)過適當(dāng)?shù)膶W(xué)習(xí)后，）（的計(jì)算式，即式可導(dǎo)出由)()1(?)()1()1(?9)()()()()()()()()1(?)()1(?)()1(?)6(~)4()2()2(1)3()2()2()2(1)2(kukykukyyyN N Ikkn e tfkkukn e tkn e tkOkOkykukykukyyiniQiiiiiQi i????????????????????????????10 綜上所述，圖一所示的神經(jīng)網(wǎng)絡(luò) PID控制系統(tǒng)的算法步驟： 1. 事先選定 NNI BP神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)，即選定輸入層節(jié)點(diǎn)數(shù) 和隱含層節(jié)點(diǎn)數(shù) ；選定學(xué)習(xí)速率 和動(dòng)量系數(shù) 。的輸出用下列各式前向計(jì)算。式對(duì)式及用14 ? ?? ? ? ?10 1,0 ,1,0 ,2,1 )1( )()()()1(?)1()( )1()()1(?)1()(.9)2()1()2()3()2()3()2()3(???????????????????????????????????uyijjiiijiiinnjQikkOkn e tfkkykykkkOkykykNNI??的權(quán)值進(jìn)行修正。分別為比例、積分、微，式中，控制算式為經(jīng)典DIPDIPKKKkekekeKkeKkekeKkukuP I D)12()]2()1(2)([)()]1()([)1()(???????????19 曲正切函數(shù)。輸入神經(jīng)網(wǎng)絡(luò)是一個(gè)設(shè)SKKKKKKP I DQMBPDIPDIP3??20 ? ?層、輸出層。1()1()1()1()1(.6),()12(.5)15(~)13(.4)1,1,)(()1()()(.3)。取采樣時(shí)間為時(shí)，的正弦信號(hào)當(dāng)輸入信號(hào)為幅值是上的隨機(jī)數(shù)。這時(shí)來代替從而可實(shí)現(xiàn)用預(yù)測輸出為）－＋－＝：出參數(shù)向量用最小二乘法在線估計(jì)25(3,2,1)()()()()1(?)1()()()1()20()()1()()1(?)24()(?)()1(?23()1()1()(1)()1()1()1(1)1()1()()]1(?)1()()[()1(?)(?)(?)3()3()3()3()2()3()3(??????????????????????????????????????????????lkn e tgkOkukukykekkOkkukykukykkkykPkkKkPkkPkkkPkKkkkykKkkklllliilliTTTT?????????????????32 )?！保钡叫阅苤笜?biāo)返回到“置；權(quán)系數(shù)式，計(jì)算修正隱含層的由；權(quán)系數(shù)式，計(jì)算修正輸出層的由；和式，計(jì)算由Jkkkkkukykyijli3,)()21(.10)()25(.9)()1(?)1(?)24(.8)2()3(????????34 圖四 采用非線性預(yù)測模型的神經(jīng)網(wǎng)絡(luò) PID控制系統(tǒng)結(jié)構(gòu)圖 四、方案四：采用非線性預(yù)測模型的神經(jīng)網(wǎng)絡(luò) PID控制器 NNI35 ??? ? ? ?? ??? )3()1(,),1(),(),1(,),1(),()1()()(2)(,),2(),1(),(,),2(),1()(＋＋改寫為為非線性函數(shù)。()()5(iQiiixxinninniijkOkkyxxgeexxffkkuyuy39 ? ?代替。()()()()(.2。1()1()1()1()1(.8)()1(?)64(.7)1(?)34(~)14(.6)2()3()2()3(???????????????
高考資料相關(guān)推薦
鄂ICP備17016276號(hào)-1