【正文】 vergence are highly Extended Kalman?lter (EKF) is a stateoftheart state estimation solution that is most mon in every branch of engineering. Although it was developed in the 60’s and there are reported thousands of successful applications, its theoretical properties such as convergence are not very well understood. In fact, it is well known that certain ”tricksofthetrade” are necessary in order to handle numerical and more fundamental divergence phenomena. Recently, general conditions for global convergence and stability of the EKF has been established, see [1] and the references therein. A su?cient condition for convergence of the EKF is boundedness of the timevarying solution to the associated Riccati equation. Although this result maybe difficult to use directly during design, it gives support and a theoretical foundation to standard modi?cations such as covariance matrix more promising approach, which is the main focus of this workpackage, is nonlinear observer design. This eliminates the need for the Riccati equation and greatly simpli?es the analysis of theoretical properties such as stability. In fact,we use nonlinear ?xed or timevarying gain observers instead of the EKF methods where the gains are puted via the putationally expensive Riccati , using the powerful framework of Lyapunov theory, sufficient conditions for stability and convergence can be imposed on the design. The main drawback ofnonlinear observers is that while the EKF is a general purpose algorithm (although it is not pletely ”plug and play” since it requires certain tricks and modi?cations to work in many cases), the nonlinear observer approach requires a detailed design for each application. Several nonlinear observer design approaches based on Lyapunov theory have been considered. There is a number of existing and new automotive control and monitoring applications that rely or bene?t from state estimation. These include antilock brakes (ABS), electronic stability program (ESP), rollover protection, collision avoidance,intelligent cruise control, active body control (ABC). Some of these application require special sensors to achieve the necessary performance or fault 上海理工大學(xué)畢業(yè)設(shè)計(jì)（論文） 33 tolerance. This leads to the fact that di?erent car models will have a di?erent set of onboard sensors, and di?erent requirements for state estimation. For example, lowend cars may only have an ABS system and the associated basic wheel and steering sensors, while highend cars with ESP and ABC may have additional sensors. This paper presents a parison of performances and characteristics of these observers. The criterion for parison is based on the observer tracking errors, both at steady state and during transients, and the robustness of the performance with respect to the uncertainties of plant. To further enhance the performance of NESO in the presence of unknown initial conditions, several nonlinear gain functions are introduced. The simulations conducted assisted in gaining insight of observer behavior in both openloop and closed–loop scenarios. The experimental results are also provided to give realism. The anization of the paper is as follows. Section II gives an introduction to existing observer design techniques. In Section III, simulation results of three observers in both openloop and closedloop forms are presented. A parison of several nonlinear gains for NESO is also performed. The experimental results are presented in Section IV and concluding remarks are given in S。 基于 MATLAB 的控制系統(tǒng)仿真及分析 討論 觀測(cè)器的性能 上海理工大學(xué)畢業(yè)設(shè)計(jì)（論文） 23 配置 觀測(cè)器的極點(diǎn)為： ?? ， ?? 在仿真過(guò)程中，我在 4 秒和 8 秒時(shí)刻，分別給電機(jī)一個(gè)幅值為 10 和 20，分別 持續(xù) 秒的擾動(dòng)電流。再?gòu)膭?dòng)態(tài)性能上看，調(diào)速系統(tǒng)首先需要有較好的抗擾性能，典型Ⅱ型系統(tǒng)恰好能滿足這個(gè)要求。這時(shí)，輸出量與輸入無(wú)關(guān)，而是由它后面環(huán)節(jié)的需要決定的。兩個(gè)放大器可獨(dú)立使用 ,如分別用于反饋穩(wěn)壓和過(guò)流保護(hù)等，此時(shí)腳 3 應(yīng)接 RC 網(wǎng)絡(luò)，提高整個(gè)電路的穩(wěn)定性。 4）、低速平穩(wěn)性好，系統(tǒng)的調(diào)速范圍可達(dá) 1： 20210 左右。例如，雷達(dá)在xCeyyBuAxxji?????????????????????..?上海理工大學(xué)畢業(yè)設(shè)計(jì)（論文） 9 大風(fēng)環(huán)境下，天線執(zhí)行電機(jī)的負(fù)載轉(zhuǎn)矩受風(fēng)阻力矩的影響而改變；機(jī)床加工零件時(shí)，在加工工件的切削過(guò)程中，負(fù)載力矩要發(fā)生變化，并引起轉(zhuǎn)速 的波動(dòng)或加工誤差。 降階觀測(cè)器 如果原系統(tǒng)是狀態(tài)部分可觀測(cè)的，則應(yīng)采用降階觀測(cè)器。 2)、 由于發(fā)生了零極點(diǎn)對(duì)消，因此，觀測(cè)器系統(tǒng)是不可控的，即它總是與系統(tǒng)的狀態(tài)保持一致，而不受控制輸入 u 的影響。 式（ 13）就是式（ 11）的系統(tǒng)狀態(tài)觀測(cè)器， x? 就是重構(gòu)狀態(tài) 。 14 降階狀態(tài)觀測(cè)器的設(shè)計(jì) 但是對(duì) 負(fù)載轉(zhuǎn)矩的測(cè)量是十分困難的， 由此我們可以利用 負(fù)載轉(zhuǎn)矩觀測(cè)器估值，實(shí)現(xiàn)對(duì)轉(zhuǎn)矩的測(cè)量，從而實(shí)現(xiàn)對(duì)轉(zhuǎn)矩變化的擾動(dòng)補(bǔ) 償。 12 轉(zhuǎn)速、負(fù)載轉(zhuǎn)矩調(diào)節(jié)器的設(shè)計(jì) 但是，由于不可避免的存在著各式各樣的 噪聲，模型以及初始 條件不可能做到與實(shí)際系統(tǒng)完全相同。 4）、 觀測(cè)器的動(dòng)態(tài)性能由觀測(cè)器的系統(tǒng)矩陣 F 決定。 3）、嚴(yán)格地講，系統(tǒng)得參數(shù)隨著運(yùn)行情況不同，是變化的。對(duì)新系統(tǒng)（ 13），重新按照以上方法構(gòu)造狀態(tài)觀測(cè)器，即可實(shí)現(xiàn)對(duì)系統(tǒng)（ 11） 的 狀態(tài)觀測(cè)。 圖 22 橋式可逆 PWM 變換器 電動(dòng)機(jī)的正反轉(zhuǎn)則體現(xiàn)在驅(qū)動(dòng)電壓正、負(fù)脈沖的寬窄上。 TL494 主要引腳的功能為： 腳 1 和腳 2 分別為誤差比較放大器的同相輸入端和反相輸入端； 腳 15 和腳 16 分別為控制比較放大器的反相輸入端和同相輸入端； 腳 3 為控制比較放大器和誤差比較放大器的公共輸出端，輸出時(shí)表現(xiàn) 為或輸出控制特性，也就是說(shuō)在兩個(gè)放大器中，輸出幅度大者起作用；當(dāng)腳 3 的電平變高時(shí)， TL494 送出的驅(qū)動(dòng)脈沖寬度變窄，當(dāng)腳 3 電平變低時(shí) ,驅(qū)動(dòng)脈沖寬度變寬； Rt Ct? ?上海理工大學(xué)畢業(yè)設(shè)計(jì)（論文） 13 腳 4 為死區(qū)電平控制端，從腳 4 加入死區(qū)控制電壓可對(duì)驅(qū)動(dòng)脈沖的最大寬度進(jìn)行控制，使其不超過(guò) 180176。下面我們來(lái)設(shè)計(jì)轉(zhuǎn)速、 負(fù)載轉(zhuǎn)矩調(diào)節(jié)器。其閉環(huán)傳遞函數(shù)為： 2( 1 ) 1()11 1( 1 )Iic liI ii IIKs T sWs KTss T s KK?????? ???? （ 24） 忽略高次項(xiàng)， Wcli(s)可降階近似為： 1() 11cliIWs sK? ? （ 25） 接入轉(zhuǎn)速環(huán)內(nèi)， 轉(zhuǎn)矩 環(huán)等效環(huán)節(jié)的輸入量應(yīng)為 Ui*(s)，因此 轉(zhuǎn)矩 環(huán)在轉(zhuǎn)速環(huán)節(jié)中應(yīng)等效為： *1( ) ( )1() 1d c liiII s W sUs sK???? ? （ 26） 這樣原來(lái)雙慣性環(huán)節(jié)的電流環(huán)控制對(duì)象，經(jīng)閉環(huán)控制后，可以近似地等效成只有較小時(shí)間常數(shù)的一階慣性環(huán)節(jié)。系統(tǒng)dI dU d R Id L CedtdnCm Id M l Jdt?? ? ???0dMldt ?20 0 10CC e RN C ALLCARC e C e C e C mL L J L J????????? ? ? ??