周銅林 譚瑞 張瑜
【摘 要】利用指示扭矩能夠?qū)Πl(fā)動(dòng)機(jī)實(shí)施在線控制,進(jìn)行故障診斷。首先,建立發(fā)動(dòng)機(jī)非線性曲軸動(dòng)力學(xué)模型;其次,將曲軸角度和瞬時(shí)轉(zhuǎn)速作為控制器輸入并進(jìn)行微分運(yùn)算;最后,利用BP神經(jīng)網(wǎng)絡(luò)對(duì)控制器參數(shù)進(jìn)行在線自適應(yīng)設(shè)計(jì)。仿真結(jié)果表明,在穩(wěn)態(tài)工況下,設(shè)計(jì)的控制器對(duì)多缸發(fā)動(dòng)機(jī)的指示扭矩能夠精確地進(jìn)行估計(jì),其對(duì)參數(shù)擾動(dòng)具有較好的魯棒性,可以用于發(fā)動(dòng)機(jī)的失火檢測(cè)和故障診斷。
【關(guān)鍵詞】滑模觀測(cè)器;BP神經(jīng)網(wǎng)絡(luò);發(fā)動(dòng)機(jī);指示扭矩
【Abstract】Indicated torque can be used to control of automotive engine on-line and fault diagnosis.Firstly,establish the non-linear crankshaft dynamic model;secondly,input the crankshaft angle and instantaneous engine speed to the controller and implementation differential operation;finally, take advantage of the BP neural network to design the self-adaption parameters of controller.Estimation results demonstrate that the controller can estimation the indicated torque of multi cylinder engine precisely,which presents the strongly robustness for parameters variation,and is a suitable way for misfire detection and fault diagnosis in the future.
【Key words】Sliding mode observer;BP neural network;Engine;Indicated torque
0 引言
指示扭矩作為表征發(fā)動(dòng)機(jī)動(dòng)力性能的一項(xiàng)重要參數(shù),及時(shí)可靠的獲取指示扭矩不僅可以正確地評(píng)價(jià)發(fā)動(dòng)機(jī)工作性能,實(shí)時(shí)的監(jiān)測(cè)故障發(fā)生[1]。在轉(zhuǎn)速和負(fù)載變化的情況下,可以有效提高汽車(chē)的燃油經(jīng)濟(jì)性、駕駛舒適性和汽車(chē)安全性。
目前,指示扭矩的估計(jì)方法主要有三種:一是,基于速度波動(dòng)的方法[2];二是,基于神經(jīng)網(wǎng)絡(luò)的方法[3];三是,基于觀測(cè)器的方法[4]。
隨著滑模變結(jié)構(gòu)控制技術(shù)的不斷發(fā)展,滑模理論常被用來(lái)設(shè)計(jì)觀測(cè)器估計(jì)指示扭矩[5]。其中,負(fù)載扭矩的獲取是一個(gè)難點(diǎn)[6]。本文通過(guò)引入汽車(chē)動(dòng)力學(xué)方程,解決了這個(gè)問(wèn)題。為了進(jìn)一步提高指示扭矩的估計(jì)精度,針對(duì)系統(tǒng)中存在的不確定性和時(shí)變擾動(dòng),設(shè)計(jì)雙積分滑模觀測(cè)器對(duì)指示扭矩進(jìn)行估計(jì)。
1 非線性曲軸動(dòng)力學(xué)模型
3 控制器參數(shù)自適應(yīng)設(shè)計(jì)
智能雙積分滑模控制相對(duì)傳統(tǒng)的滑??刂?,由于具有Ⅰ控制特性,所以,在穩(wěn)態(tài)誤差調(diào)節(jié)方面效果更好。本文利用BP神經(jīng)網(wǎng)絡(luò)對(duì)參數(shù)k1,k2,k3進(jìn)行了在線自適應(yīng)設(shè)計(jì)。
4 仿真結(jié)果分析
4.1 仿真實(shí)驗(yàn)
利用LMS Imagine.Lab Amesim14平臺(tái)建立發(fā)動(dòng)機(jī)多缸模型,并通過(guò)Matlab/Simulink建立控制器。發(fā)動(dòng)機(jī)的參數(shù)配置如表1所示:
5 結(jié)論
針對(duì)系統(tǒng)中存在的擾動(dòng)問(wèn)題,設(shè)計(jì)了基于非線性曲軸動(dòng)力學(xué)模型的雙積分滑模控制器,將扭矩估計(jì)問(wèn)題轉(zhuǎn)換為轉(zhuǎn)速誤差跟蹤控制問(wèn)題。通過(guò)Lyapunov穩(wěn)定性理論和BP神經(jīng)網(wǎng)絡(luò)設(shè)計(jì)自適應(yīng)控制器增益參數(shù),實(shí)現(xiàn)了指示扭矩的精確估計(jì)。該方法具有精度高、魯棒性好和成本低等特點(diǎn),且避免了利用扭矩傳感器對(duì)負(fù)載扭矩的測(cè)量。仿真結(jié)果表明,在穩(wěn)態(tài)工況下,該方法能夠較精確地估計(jì)多缸發(fā)動(dòng)機(jī)指示扭矩,可用于發(fā)動(dòng)機(jī)故障診斷。
【參考文獻(xiàn)】
[1]F Ponti,M Rinaldi. Analysis of the relationship between mean indicated torque and its waveform for modern common rail diesel and gasoline engines[J].2008 Spring Technical Conference of the ASME Internal Combustion Engine Division,2008:149-159.
[2]Ponti F,Serra G,Lupo S.Powertrain Torsional Model Development or On-Board Indicated Torque Estimation[J].Linear Algebra & Its Applications,2009,1(1):263-264.
[3]Ge Y,Huang Y,Hao D,et al.An indicated torque estimation method based on the Elman neural network for a turbocharged diesel engine[J].Proceedings of the Institution of Mechanical Engineers Part D Journal of Automobile Engineering, 2015,230(10).
[4]Itoh Y,Higashi K,Iwase M.UKF-based estimation of indicated torque for IC engines utilizing nonlinear two-inertia model[C]// Decision and Control. IEEE,2012:4077-4082.
[5]Azzoni P,Moro D,Ponti F,et al.MODEL-BASED INDICATED TORQUE ESTIMATION,USING NON LINEAR OBSERVERS[C]// IEEE Mediterranean Conference.IEEE,2014:160-166.
[6]Higashi K,Itoh Y,Iwase M.Torque estimation for combustion engine using nonlinear model[C]// Sice Conference,2012:1959-1964.
[責(zé)任編輯:田吉捷]