【摘要】：本文主要研究伺服系統(tǒng)的精確建模問題,通過分析機理建模的復雜與不精確的問題,指出引入神經(jīng)網(wǎng)絡(luò)建模能帶來的快速性、精確性以及簡便性的提升。而目前針對神經(jīng)網(wǎng)絡(luò)辨識的研究雖然有很多改進方案,但是大多都只是在一些特定的仿真模型下效果較好,缺乏實際系統(tǒng)的驗證,有些算法甚至并不適用于實際系統(tǒng)辨識,因此本文研究基于神經(jīng)網(wǎng)絡(luò)的伺服系統(tǒng)精確模型辨識問題,主要的研究成果可歸納為:首先,對一類以永磁同步電機為執(zhí)行元件的位置伺服系統(tǒng),進行了標稱模型分析與詳細的攝動項環(huán)節(jié)分析,分析了不同非線性環(huán)節(jié)以及攝動項會對神經(jīng)網(wǎng)絡(luò)辨識造成哪些影響,為優(yōu)化設(shè)計神經(jīng)網(wǎng)絡(luò)辨識方法提供了理論依據(jù)。其次,對比分析了神經(jīng)網(wǎng)絡(luò)辨識的基本結(jié)構(gòu)、神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)特征與選型依據(jù)、神經(jīng)網(wǎng)絡(luò)訓練的基本方法,通過對比指出選擇徑向基神經(jīng)網(wǎng)絡(luò)辨識的選型依據(jù),通過對比訓練方法的優(yōu)劣為神經(jīng)網(wǎng)絡(luò)的參數(shù)訓練方法提供了改進方向。然后,結(jié)合伺服系統(tǒng)的特點,提出了適用于伺服系統(tǒng)的兩點差分式串-并聯(lián)辨識結(jié)構(gòu),優(yōu)化了神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu),改進了神經(jīng)網(wǎng)絡(luò)參數(shù)的訓練算法,提出將正交最小二乘法(OLS)與梯度下降法(GD)相結(jié)合,能夠有效地改進減少神經(jīng)網(wǎng)絡(luò)中心節(jié)點數(shù)量以及降低對初始位置選取的依賴,然后結(jié)合伺服系統(tǒng)工作的頻段,給出樣本數(shù)據(jù)、測試數(shù)據(jù)的選擇方法以及給出神經(jīng)網(wǎng)絡(luò)模型的評價方法。最后得到一個一步預測的模型結(jié)構(gòu),該結(jié)構(gòu)使用前幾時刻的實際數(shù)據(jù)作為輸入能夠準確預測出下一時刻的輸出,并且通過仿真實驗驗證了改進結(jié)構(gòu)和訓練算法的有效性。最后,結(jié)合提出的針對伺服系統(tǒng)改進的神經(jīng)網(wǎng)絡(luò)辨識方案,在實際轉(zhuǎn)臺伺服系統(tǒng)當中采集開環(huán)的訓練樣本與測試數(shù)據(jù),訓練出其神經(jīng)網(wǎng)絡(luò)模型,再通過與傳統(tǒng)的掃頻方案得到的模型進行對比,驗證了神經(jīng)網(wǎng)絡(luò)用于實際系統(tǒng)建模的可行性。
[Abstract]:In this paper, the exact modeling of servo system is studied. By analyzing the complexity and imprecision of mechanism modeling, the paper points out the improvement of rapidity, accuracy and simplicity brought by the introduction of neural network modeling. However, although there are many improved methods for neural network identification, most of them only work well under some specific simulation models, lacking the verification of the actual system, and some algorithms are not even suitable for the actual system identification. Therefore, in this paper, the exact model identification of servo system based on neural network is studied. The main research results can be summarized as follows: firstly, for a class of position servo system with permanent magnet synchronous motor as the actuator, The nominal model analysis and the detailed analysis of perturbation terms are carried out, and the effects of different nonlinear links and perturbation terms on neural network identification are analyzed, which provides a theoretical basis for the optimization design of neural network identification methods. Secondly, the basic structure of neural network identification, the structural characteristics and selection basis of neural network, the basic training method of neural network, and the selection basis of selecting radial basis function neural network identification are pointed out. By comparing the advantages and disadvantages of the training methods, the improvement direction of the neural network parameter training method is provided. Then, combined with the characteristics of servo system, a two-point differential series-parallel identification structure is proposed for servo system. The structure of neural network is optimized, and the training algorithm of neural network parameters is improved. The combination of orthogonal least square method (OLS) and gradient descent method (GD) can effectively reduce the number of neural network center nodes and reduce the dependence on initial position selection, and then combine with the frequency band of servo system. The sample data, the selection method of test data and the evaluation method of neural network model are given. Finally, a one-step prediction model structure is obtained, which can accurately predict the output of the next moment by using the actual data from the previous time as input, and the effectiveness of the improved structure and the training algorithm is verified by simulation experiments. Finally, combined with the improved neural network identification scheme for servo system, the open-loop training samples and test data are collected in the actual turntable servo system, and its neural network model is trained. Compared with the model obtained by the traditional frequency sweeping scheme, the feasibility of the neural network used in the practical system modeling is verified.
【學位授予單位】：哈爾濱工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP183;TM921.54

【參考文獻】

相關(guān)期刊論文 前10條

1 覃業(yè)梅;彭輝;阮文杰;;基于線性函數(shù)型權(quán)重的RBF-ARX模型的磁懸浮球系統(tǒng)預測控制[J];中南大學學報(自然科學版);2016年08期

2 段艷杰;呂宜生;張杰;趙學亮;王飛躍;;深度學習在控制領(lǐng)域的研究現(xiàn)狀與展望[J];自動化學報;2016年05期

3 鄒友龍;胡法龍;周燦燦;李潮流;李長喜;Keh-Jim Dunn;;徑向基函數(shù)插值方法分析(英文)[J];Applied Geophysics;2013年04期

4 王鵬;關(guān)宇東;沈逢京;杜克;提純利;;一種力矩電機轉(zhuǎn)矩波動系數(shù)自動化檢測方法[J];微電機;2013年11期

5 胡巖;關(guān)朕;吳偉;;異步起動永磁同步電動機的轉(zhuǎn)子槽漏抗計算[J];微特電機;2012年09期

6 陳進東;張相勝;潘豐;;基于Wiener模型的非線性預測函數(shù)控制[J];吉林大學學報(工學版);2011年S1期

7 吳德會;;非線性動態(tài)系統(tǒng)的Wiener神經(jīng)網(wǎng)絡(luò)辨識法[J];控制理論與應用;2009年11期

8 田一鳴;黃友銳;高志安;黃宜慶;;基于GA與CSA-RBF神經(jīng)網(wǎng)絡(luò)辨識的自適應PID控制器[J];系統(tǒng)仿真學報;2008年17期

9 張從鵬;劉強;;直線電機定位平臺的摩擦建模與補償[J];北京航空航天大學學報;2008年01期

10 曲東才;;增強神經(jīng)網(wǎng)絡(luò)辨識模型泛化能力的研究[J];海軍航空工程學院學報;2007年01期

相關(guān)博士學位論文 前2條

1 鄭偉峰;交流伺服系統(tǒng)無時滯反饋高響應驅(qū)動控制研究[D];哈爾濱工業(yè)大學;2010年

2 黃進;含摩擦環(huán)節(jié)伺服系統(tǒng)的分析及控制補償研究[D];西安電子科技大學;1998年

相關(guān)碩士學位論文 前5條

1 陳宇飛;機電伺服系統(tǒng)魯棒控制設(shè)計與實現(xiàn)[D];哈爾濱工業(yè)大學;2014年

2 張斯倫;機電伺服系統(tǒng)低速性能分析與控制設(shè)計[D];哈爾濱工業(yè)大學;2013年

3 楊旭;基于RBF神經(jīng)網(wǎng)絡(luò)的工業(yè)過程建模與優(yōu)化研究[D];哈爾濱理工大學;2009年

4 葛翔;基于重復控制方法的高精度速率伺服控制系統(tǒng)設(shè)計[D];哈爾濱工業(yè)大學;2007年

5 張新良;非線性系統(tǒng)神經(jīng)網(wǎng)絡(luò)辨識與控制的研究[D];南京航空航天大學;2004年

，

本文編號：2415066