国产日韩欧美一区二区三区三州_亚洲少妇熟女av_久久久久亚洲av国产精品_波多野结衣网站一区二区_亚洲欧美色片在线91_国产亚洲精品精品国产优播av_日本一区二区三区波多野结衣 _久久国产av不卡

?

一種基于高階內(nèi)模的新型自適應(yīng)迭代學(xué)習(xí)算法

2022-03-09 03:30張國山李思祺
關(guān)鍵詞:高階誤差因子

張國山,李思祺

一種基于高階內(nèi)模的新型自適應(yīng)迭代學(xué)習(xí)算法

張國山,李思祺

(天津大學(xué)電氣自動化與信息工程學(xué)院,天津 300072)

本文針對一類含多個(gè)時(shí)間迭代變化參數(shù)控制方向未知的非線性離散時(shí)間系統(tǒng)的輸出跟蹤問題,提出了一種基于高階內(nèi)模的新型自適應(yīng)迭代學(xué)習(xí)算法.假設(shè)多個(gè)時(shí)間迭代變化參數(shù)由不同的高階內(nèi)模所生成,本文所提出的算法借鑒了模型預(yù)測控制的思想,通過構(gòu)建預(yù)測輸入,將獲得的當(dāng)次迭代預(yù)測跟蹤誤差作為先驗(yàn)知識,應(yīng)用到系統(tǒng)輸入的控制律的設(shè)計(jì)中,從而在預(yù)測跟蹤誤差的基礎(chǔ)上進(jìn)一步縮小系統(tǒng)的跟蹤誤差.相較于基于高階內(nèi)模的傳統(tǒng)迭代學(xué)習(xí)算法,大幅度縮減了系統(tǒng)的輸出跟蹤誤差,明顯地提高了跟蹤精度.此外,由于預(yù)測跟蹤誤差作為先驗(yàn)知識參與了系統(tǒng)輸入控制律的設(shè)計(jì),該方法對于系統(tǒng)擾動和輸出噪聲具有較強(qiáng)的魯棒性.通過Lyapunov穩(wěn)定性理論,證明了該方法下系統(tǒng)跟蹤誤差的收斂性和所提算法的優(yōu)越性.通過兩組仿真算例,考慮在控制方向已知和未知兩種情況下,和兩種基于高階內(nèi)模的已有迭代學(xué)習(xí)算法進(jìn)行了對比,驗(yàn)證了理論結(jié)果.

自適應(yīng)迭代學(xué)習(xí)控制;高階內(nèi)模;非線性離散時(shí)間系統(tǒng);時(shí)間迭代變化參數(shù)

迭代學(xué)習(xí)控制是一種解決重復(fù)環(huán)境下動態(tài)系統(tǒng)跟蹤問題的有效方法,當(dāng)系統(tǒng)輸出重復(fù)跟蹤參考軌跡時(shí),利用上一次迭代的跟蹤誤差信息更新當(dāng)前迭代控制輸入[1-5].近年來,迭代學(xué)習(xí)控制在工業(yè)過程中有很多應(yīng)用,如高速公路交通控制[6]、機(jī)器人操縱器[7]、高速火車[8]、微型飛行器(MAV)[9]和多智能體系統(tǒng)等[10-11].

傳統(tǒng)的迭代學(xué)習(xí)控制框架要求過程具有嚴(yán)格的可重復(fù)性,但實(shí)際應(yīng)用中很難滿足這一要求.迭代變化初始狀態(tài)[12]、迭代變化擾動[13]和迭代變化軌跡[14]在工業(yè)過程中會經(jīng)常遇到,有時(shí)還會同時(shí)遇到多個(gè)迭代變化因子[15].實(shí)際上,在工業(yè)過程中所遇到的一些變化方向是已知的迭代變化因子可以沿迭代軸用高階內(nèi)模來進(jìn)行描述;即當(dāng)前迭代中的迭代變化因子是前幾次迭代中對應(yīng)項(xiàng)的線性組合[16].

因此基于高階內(nèi)模的迭代學(xué)習(xí)算法對于含有時(shí)間迭代變化因子的系統(tǒng)的輸出跟蹤問題時(shí)效果極好.文獻(xiàn)[17-18]研究了含迭代變化因子的線性離散時(shí)間系統(tǒng)的輸出跟蹤問題.文獻(xiàn)[19]考慮了含迭代變化因子的非線性離散時(shí)間系統(tǒng)的輸出跟蹤問題.在此基礎(chǔ)上,Liu等[20]針對一類具有二階內(nèi)模參數(shù)不確定性的離散時(shí)間非線性系統(tǒng),提出了一種基于最小二乘法自適應(yīng)迭代學(xué)習(xí)控制算法.雖然考慮了狀態(tài)擾動對系統(tǒng)的影響,但并未將二階內(nèi)模推廣至高階內(nèi)模.Yu等[21]針對一類同時(shí)具有參數(shù)和非參數(shù)不確定性的離散時(shí)間非線性系統(tǒng),提出了一種新的魯棒自適應(yīng)迭代學(xué)習(xí)控制算法,得益于良好的死區(qū)函數(shù)設(shè)計(jì),參數(shù)和非參數(shù)不確定性可以同時(shí)進(jìn)行學(xué)習(xí).文獻(xiàn)[22]討論了與文獻(xiàn)[20]相同的系統(tǒng)在控制增益未知的情況下的迭代學(xué)習(xí)控制,由于控制增益也成為了學(xué)習(xí)對象,系統(tǒng)的狀態(tài)擾動無法有效描述,導(dǎo)致該算法與文獻(xiàn)[20]所提算法均無法有效削減狀態(tài)擾動對系統(tǒng)輸出造成的影響,魯棒性也無法得到保證.

本文針對一類含多個(gè)迭代變化參數(shù)的非線性離散時(shí)間系統(tǒng),提出了一種基于高階內(nèi)模的同時(shí)適用于控制方向已知和未知兩種情況的新型迭代學(xué)習(xí)控制算法.該方法借鑒了預(yù)測控制的思想,通過構(gòu)造一個(gè)預(yù)測輸入,得到當(dāng)次迭代的預(yù)測跟蹤誤差,并將其作為先驗(yàn)知識去構(gòu)建系統(tǒng)輸入控制律,進(jìn)而有效地減小了跟蹤誤差,提升了跟蹤效果.通過理論證明,該方法可以在存在多個(gè)迭代變化參數(shù)且控制方向未知的情況下實(shí)現(xiàn)對非線性離散系統(tǒng)的有效輸出跟蹤.與現(xiàn)有方法相比較,本文所提方法適用于控制增益已知和未知兩種情況,在跟蹤精度和收斂速度上有明顯提升,且在存在外部擾動的情況下具有一定的魯棒性.最后通過兩個(gè)算例驗(yàn)證了理論結(jié)果.

1?問題描述

考慮離散時(shí)間非線性系統(tǒng)

式中為已知連續(xù)函數(shù).

下面是后文中要用到的兩個(gè)假設(shè).

本文所提出的兩個(gè)假設(shè)均為該類問題所需的一般假設(shè).

引理1[23]如果

式(4)的高階內(nèi)??梢悦枋鰹槎囗?xiàng)式

式(8)也可以寫作

將式(9)代入系統(tǒng)(1)得

根據(jù)式(8)可得

進(jìn)而有

式(12)可以表示為

2?算法設(shè)計(jì)

本文所提出的迭代學(xué)習(xí)算法借鑒了模型預(yù)測控制思想,以第次迭代為例,進(jìn)行控制律算法設(shè)計(jì).

首先,構(gòu)建一個(gè)預(yù)測輸入

式中為一正常數(shù).

然后,將得到的預(yù)測跟蹤誤差加入到系統(tǒng)輸入的控制律的設(shè)計(jì)當(dāng)中

3?收斂性分析

定理 1?對于非線性離散系統(tǒng)(1),在假設(shè)1和2的前提下,控制律(15)、(20)和參數(shù)更新律(18)、(19)滿足:

將式(18)代入式(22)右邊第1部分可得

同理可以得到式(22)右邊的第2部分為

則式(22)可以表示為

由式(25)可得

由式(25)可得

根據(jù)學(xué)習(xí)律式(18)和式(19)可得

同理

由式(29)和式(31)可得

因此

由式(34)可得

由式(17)、式(32)和式(37)可得

由式(42)和式(43)可得

將式(41)和(44)代入式(40),可得

由式(11)可以得到

進(jìn)而

綜上可得

將式(48)代入式(45)

由式(10)和式(15)可知

根據(jù)控制律(15)可知

將式(57)代入式(40)可得

4?仿真算例

例1:考慮文獻(xiàn)[20]中的非線性系統(tǒng)

圖1?時(shí)間迭代變化因子和

例2:考慮文獻(xiàn)[22]中的非線性系統(tǒng)

圖4?時(shí)間迭代變化因子

圖5?例2中輸入增益估值

圖6?例2中跟蹤誤差絕對值的最大值

本文所研究的兩個(gè)仿真實(shí)例均為離散時(shí)間系統(tǒng),其采樣周期為0.05s.作為對比,文獻(xiàn)[20]中的方法是在控制方向已知的情況下設(shè)計(jì)的;文獻(xiàn)[22]中的方法是在控制方向未知的情況下設(shè)計(jì)的.

5?結(jié)?語

本文針對一類含多個(gè)迭代變化參數(shù)的非線性離散系統(tǒng),提出了一種新的基于高階內(nèi)模的迭代學(xué)習(xí)控制方法.借鑒預(yù)測控制的思想,通過構(gòu)建預(yù)測輸入并將其代入系統(tǒng)的方式獲得預(yù)測跟蹤誤差作為先驗(yàn)知識,然后將其運(yùn)用到系統(tǒng)輸入的控制律設(shè)計(jì)中.在已有的基于高階內(nèi)模的迭代學(xué)習(xí)控制方法的基礎(chǔ)上,成功引入了當(dāng)次迭代的跟蹤信息,從而進(jìn)一步提高了跟蹤精度,縮小了跟蹤誤差,加快了跟蹤誤差的收斂速度,并分別通過理論證明和仿真對比驗(yàn)證了其有效性和優(yōu)越性.

[1] Zhang B,Wang D W,Ye Y Q. On learning transient,auto-tunings of learnable bandwidth and lead step in iterative learning control[J]. International Journal of Systems Science,2010,41(4):353-363.

[2] Yu M,Ye X D,Qi D L. Repetitive learning control for triangular systems with unknown control directions[J]. IET Control Theory & Applications,2011,5(17):2045-2051.

[3] Dabkowski P,Galkowski K,Cai Z,et al. Iterative learning control based on relaxed 2D systems stability criteria[J]. IEEE Transactions on Control Systems Technology,2013,21(3):1016-1023.

[4] Bu X,Hou Z,Yu F,et al. H∞iterative learning controller design for a class of discrete-time systems with data dropouts[J]. International Journal of Systems Science,2014,45(7/8/9):1902-1912.

[5] Meng D,Moore K L,et al. Robust iterative learning control for nonrepetitive uncertain systems[J]. IEEE Transactions on Automatic Control,2017,62(2):907-913.

[6] Jin X. Adaptive iterative learning control for high-order nonlinear multi-agent-systems consensus tracking[J]. Systems & Control Letters,2016,89:16-23.

[7] Zhang H G,Jiang H,Luo Y H,et al. Data-driven optimal consensus control for discrete-time multi-agent systems with unknown dynamics using reinforcement learning method[J]. IEEE Transactions on Industrial Electronics2017,64(5):4091-4100.

[8] Sun S T,Li X D,Zhong R X. An open-closed-loop iterative learning control approach for nonlinear switched systems with application to freeway traffic control[J]. International Journal of Systems Science,2017,48 (13):2752-2763.

[9] Sun S,Endo T,Matsuno F. Iterative learning control based robust distributed algorithm for non-holonomic mobile robots formation[J]. IEEE Access,2018,6:61904-61917.

[10] Yu Q X,Hou Z S,Xu J X. D-type ILC based dynamic modeling and norm optimal ILC for high-speed trains[J]. IEEE Transactions on Control Systems Technology2018,26(2):652-663.

[11] He W,Meng T T,He X Y,et al. Iterative learning control for a flapping wing micro aerial vehicle under distributed disturbances[J]. IEEE Transactions on Cybernetics2019,49(4):1524-1535.

[12] Xu J X,Yan R. On initial conditions in iterative learning control[J]. IEEE Transactions on Automatic Control,2005,50(9):1349-1354.

[13] Ahn H S,Moore K L,Chen Y Q. Stability analysis of discrete-time iterative learning control systems with interval uncertainty[J]. Automatica,2007,43(5):892-902.

[14] Chi R H,Hou Z S,Xu J X. Adaptive ILC for a class of discrete-time systems with iteration-varying trajectory and random initial condition[J]. Automatica,2008,44(8):2207-2213.

[15] Yin C K,Xu J X,Hou Z S. A high-order internal model based iterative learning control scheme for nonlinear systems with time-iteration-varying parameters[J]. IEEE Transactions on Automatic Control,2010,55(11):2665-2670.

[16] Liu C P,Xu J X,Wu J. On iterative learning control with high-order internal models[J]. International Journal of Adaptive Control & Signal Processing,2010,24(9):731-742.

[17] Zhu Q,Xu J X,Huang D,et al. Iterative learning control design for linear discrete-time systems with multiple high-order internal models[J]. Automatica,2015,62:65-76.

[18] Zhou W,Yu M,Huang D Q. High-order internal model based iterative learning control scheme for discrete-time nonlinear system[J]. International Journal of Automation and Computing,2015(3):104-110.

[19] Liu B B,Zhou W. Projection algorithm based adaptive iterative learning control for a class of discrete-time systems[C]//Chinese Control and Decision Conference. Qingdao,China,2015:2993-2998.

[20] Liu B B,Zhou W. On adaptive iterative learning control algorithm for discrete-time systems with parametric uncertainties subject to second-order internal model[J]. Control and Intelligent Systems,2015,43(4):183-190.

[21] Yu M,Huang D Q,He W. Robust adaptive iterative learning control for discrete-time nonlinear systems with both parametric and nonparametric uncertainties[J]. International Journal of Adaptive Control & Signal Processing,2016,30(7):972-985.

[22] Zhao D,Yang Y. An iterative learning control design method for nonlinear discrete-time systems with unknown iteration-varying parameters and control direction[J]. Mathematical Problems in Engineering,2016,2016:1-7.

[23] Goodwin G C,Ramadge P J. Discrete-time multivariable adaptive control[J]. IEEE Transactions on Automatic control,1980,25(3):335-340.

[24] Yan W,Sun M. Adaptive iterative learning control of discrete-time varying systems with unknown control direction[J]. International Journal of Adaptive Control and Signal Processing,2013,27(4):340-348.

A New Adaptive Iterative Learning Algorithm Based on a High-Order Internal Model

Zhang Guoshan,Li Siqi

(School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China)

In this paper,a new adaptive iterative learning algorithm based on a high-order internal model is proposed. The algorithm is applied to the output tracking problem of a class of nonlinear discrete-time systems with multiple time-iteration-varying parameters and unknown control directions. Assuming that multiple time-iteration-varying parameters are generated by different high-order internal models,the proposed algorithm draws on the idea of model predictive control. Through the construction of predictive input,the predictive tracking error obtained in the current iteration is used as a priori knowledge for system input control law design. This has the effect of further reducing the tracking error of the system on the basis of predictive tracking error. Compared with the traditional iterative learning algorithm based on a high-order internal model,the output tracking error of the system is greatly reduced,and the tracking accuracy is evidently improved. In addition,as predictive tracking error is involved in the design of the input control law as a priori knowledge,the method is robust to system disturbance and output noise. Through the Lyapunov stability theory,the convergence of the system tracking error and the superiority of the proposed method are proved. Through two groups of sample simulations,and considering the two cases of known and unknown control direction,the theoretical results are verified by comparison with two existing iterative learning algorithms based on a high-order internal model.

adaptive iterative learning control;high-order internal model;nonlinear discrete-time system;time-iteration-varying parameters

10.11784/tdxbz202103063

TK13

A

0493-2137(2022)05-0480-09

2021-03-26;

2021-08-18.

張國山(1961—??),男,博士,教授.

張國山,zhanggs@tju.edu.cn

國家自然科學(xué)基金資助項(xiàng)目(62073237).

Supported by the National Natural Science Foundation of China(No. 62073237).

(責(zé)任編輯:孫立華)

猜你喜歡
高階誤差因子
我刊2021年影響因子年報(bào)
我刊2021年影響因子年報(bào)
北斗導(dǎo)航種蘿卜百米誤差僅2厘米
基于高階LADRC的V/STOL飛機(jī)懸停/平移模式魯棒協(xié)調(diào)解耦控制
高階思維介入的高中英語閱讀教學(xué)
山藥被稱“長壽因子”
三個(gè)高階微分方程的解法研究
隧道橫向貫通誤差估算與應(yīng)用
隧道橫向貫通誤差估算與應(yīng)用
高階非線性慣性波模型的精確孤立波和周期波解
保靖县| 洛南县| 贵溪市| 南城县| 牡丹江市| 松滋市| 六盘水市| 新兴县| 综艺| 札达县| 安徽省| 扶风县| 长治县| 明溪县| 天峨县| 清水县| 恩平市| 新宁县| 六盘水市| 灵石县| 介休市| 石柱| 钦州市| 金乡县| 长子县| 华蓥市| 泰安市| 博湖县| 新余市| 阳高县| 浦县| 麻阳| 息烽县| 当涂县| 庆城县| 景谷| 建宁县| 阳西县| 郸城县| 临朐县| 桐乡市|