Computation and sampling time requirements for real-time implementation of observers is studied. A common procedure for state estimation and observer design is to have a system model in continuous time that is converted to sampled time with Euler forward method and then the observer is designed and implemented in sampled time in the real time system. When considering state estimation in real time control systems for production there are often limited computational resources. This becomes especially apparent when designing observers for stiff systems since the discretized implementation requires small step lengths to ensure stability. One way to reduce the computational burden, is to reduce the model stiffness by approximating the fast dynamics with instantaneous relations, transforming an ordinary differential equations (ODE) model into a differential algebraic equation (DAE) model. Performance and sampling frequency limitations for extended Kalman filter (EKF)'s based on both the original ODE model and the reduced DAE model are here analyzed and compared for an industrial system. Furthermore, the effect of using backward Euler instead of forward Euler when discretizing the continuous time model is also analyzed. The ideas are evaluated using measurement data from a diesel engine. The engine is equipped with throttle, exhaust gas recirculation (EGR), and variable geometry turbines (VGT) and the stiff model dynamics arise as a consequence of the throttle between two control volumes in the air intake system. The process of simplifying and modifying the stiff ODE model to a DAE model is also discussed. The analysis of the computational effort shows that even though the ODE, for each time-update, is less computationally demanding than the resulting DAE, an EKF based on the DAE model achieves better estimation performance than one based on the ODE with less computational effort. The main gain with the DAE based EKF is that it allows increased step lengths without degrading the estimation performance compared to the ODE based EKF.
[1] Lino P, Maione B, Amorese C. Modelling and predictive control of a new injection system for compressed natural gas engines. Control Eng Pract, 2008, 16: 1216-1230 CrossRef Google Scholar
[2] García-Nieto S, Martínez M, Blasco X. Nonlinear predictive control based on local model networks for air management in diesel engines. Control Eng Pract, 2008, 16: 1399-1413 CrossRef Google Scholar
[3] Mandela R K, Rengaswamy R, Narasimhan S. Recursive state estimation techniques for nonlinear differential algebraic systems. Chem Eng Sci, 2010, 65: 4548-4556 CrossRef Google Scholar
[4] Friedrich I, Liu C S, Oehlerking D. Coordinated EGR-rate model-based controls of turbocharged diesel engines via an intake throttle and an EGR valve. In: Proceedings of IEEE Vehicle Power and Propulsion Conference, Dearborn, 2009. Google Scholar
[5] Reichhartinger M, Horn M. Application of higher order sliding-mode concepts to a throttle actuator for gasoline engines. IEEE Trans Ind Electron, 2009, 56: 3322-3329 CrossRef Google Scholar
[6] Chauvin J, Petit N, Rouchon P. Air Path Estimation on Diesel HCCI Engine. SAE Technical Paper 2006-01-1085. 2006. Google Scholar
[7] Müller M, Hendricks E, Sorenson S. Mean Value Modelling of Turbocharged Spark Ignition Engines. SAE Technical Paper 980784. 1998. Google Scholar
[8] Eriksson L. Modeling and control of turbocharged SI and DI engines. Oil Gas Sci Technol Rev IFP, 2007, 62: 523-538 CrossRef Google Scholar
[9] Hockerdal E, Eriksson L, Frisk E. Air mass-flow measurement and estimation in diesel engines equipped with GR and VGT. SAE Int J Passeng Cars Electron Electr Syst, 2008, 1: 393-402 CrossRef Google Scholar
[10] Wahlström J, Eriksson L. Modeling of a Diesel Engine with Intake Throttle, VGT, and EGR. Linköping: Linköping University, 2010. Google Scholar
[11] H?ckerdal E, Frisk E, Eriksson L. Observer design and model augmentation for bias compensation with a truck engine application. Control Eng Pract, 2009, 17: 408-417 CrossRef Google Scholar
[12] H?ckerdal E, Frisk E, Eriksson L. EKF-based adaptation of look-up tables with an air mass-flow sensor application. Control Eng Pract, 2011, 19: 442-453 CrossRef Google Scholar
[13] Brenan K E, Campbell S L, Petzold L R. Numerical solution of initial-value problems in differential-algebraic equations. Phys Rev D Part Fields, 1996, 85: 261--268. Google Scholar
[14] Darouach M, Boutat-Baddas L. Observers for a class of nonlinear singular systems. IEEE Trans Autom Control, 2008, 53: 2627-2633 CrossRef Google Scholar
[15] Boutat D, Zheng G, Boutat-Baddas L, et al. Observers design for a class of nonlinear singular systems. In: Proceedings of the 51st Annual Conference on Decision and Control (CDC), Maui, 2012. 7407--7412. Google Scholar
[16] Bejarano F J, Perruquetti W, Floquet T. Observation of Nonlinear Differential-Algebraic Systems With Unknown Inputs. IEEE Trans Autom Control, 2015, 60: 1957-1962 CrossRef Google Scholar
[17] ?slund J, Frisk E. An observer for non-linear differential-algebraic systems. Automatica, 2006, 42: 959-965 CrossRef Google Scholar
[18] Nikoukhah R. A new methodology for observer design and implementation. IEEE Trans Autom Control, 1998, 43: 229-234 CrossRef Google Scholar
[19] Kidane N, Yamashita Y, Nishitani H. Observer based I/O-linearizing control of high index dae systems. In: Proceedings of American Control Conference, Denver, 2003. 3537--3542. Google Scholar
[20] Kalman R E. A new approach to linear filtering and prediction problems. J Basic Eng, 1960, 82: 35-45 CrossRef Google Scholar
[21] Jazwinski A H. Stochastic Processes and Filtering Theory. Cambridge: Academic Press, 1970. Google Scholar
[22] Becerra V M, Roberts P D, Griffiths G W. Applying the extended Kalman filter to systems described by nonlinear differential-algebraic equations. Control Eng Pract, 2001, 9: 267-281 CrossRef Google Scholar
[23] Kaprielian S R, Turi J. An observer for a nonlinear descriptor system. In: Proceedings of the 31st Conference on Decision and Control, Tucson, 1992. 975--976. Google Scholar
[24] Heywood J B. Internal Combustion Engine Fundamentals. New York: McGraw-Hill, 1988. Google Scholar
[25] Kailath T, Sayed A H, Hassibi B. Linear Estimation. Upper Saddle River: Prentice-Hall, 2000. Google Scholar
[26] Gerdin M, Sch?n T B, Glad T. On parameter and state estimation for linear differential-algebraic equations. Automatica, 2007, 43: 416-425 CrossRef Google Scholar
[27] Yip E L, Sincovec R R. Solvability, controllability, and observability of continuous descriptor systems. IEEE Trans Autom Control, 1981, 26: 702-707 CrossRef Google Scholar
[28] Dai L. Singular Control Systems. New York: Springer, 1989. Google Scholar
[29] Mehrmann V L, Stykel T. Descriptor Systems: a General Mathematical Framework for Modeling, Simulation and Control. Technical Report 292-2005, 2005. Google Scholar
[30] Losse P, Mehrmann V. Controllability and observability of second order descriptor systems. SIAM J Control Optim, 2008, 47: 1351-1379 CrossRef Google Scholar
[31] Lee E R, Markus L. Foundations of Optimal Control Theory. New York: John Wiley and Sons, 1968. Google Scholar
[32] Kalman R E, Ho T C, Narendra K S. Controllability of linear dynamical systems. Control Linear Dyn Syst, 1963, 1: 189--213. Google Scholar
[33] Economic Commission for Europe — Inland Transport Committee. Regulation No 49 of the Economic Commission for Europe of the United Nations (UN/ECE). Official Journal of the European Union, 2010. Google Scholar
[34] Wahlström J, Eriksson L, Nielsen L. EGR-VGT control and tuning for pumping work minimization and emission control. IEEE Trans Control Syst Technol, 2010, 18: 993--1003. Google Scholar
Figure 1
Schematic of the diesel engine model
Figure 2
(Color online) DAE simulations with and without control volume adjustment for a step length of 50 ms using backward Euler, compared to high end simulation of the default ODE using the Matlab variable step length solver, ode23t, with absolute and relative tolerances of 10$^{-\text{6}}$. The better dynamic behavior of the model with increased volume is seen in the areas highlighted with solid ellipses, representing open throttle operation. Also, there is only minor loss in simulation accuracy for the larger volume during closed throttle operation, indicated by dotted ellipses. (a) $\dot{p}_{\text{im}}=~\frac{R_{\text{a}} ~~~~T_{\text{im}}}{V_{\text{im}}}(W_{\text{th}}(\cdot)~+~W_{\text{egr}}(\cdot)~-~W_{\text{ei}}(\cdot))$; (b) $\dot{p}_{\text{im}}=~\frac{R_{\text{a}}T_{\text{im}}}{V_{\text{im}}+~V_{\text{ic}}}(W_{\text{th}}(\cdot)~+~W_{\text{egr}}(\cdot)~-~W_{\text{ei}}(\cdot))$.
Figure 3
(Color online) Estimates of EKF observers based on ODE and DAE models with (a) 3 ms and (b) 10 ms step lengths and using forward Euler for the prediction. Using the ODE based EKF and a step size of 10 ms, results in a high frequency estimation error, especially prominent for high pressures corresponding to open throttle operation. It can also be noted that also the ODE based EKF with 10 ms step length would predict the states well if a low pass filter was applied.
Figure 4
(Color online) Estimated PDF's of the estimation errors for $p_{\text{im}}$, $p_{\text{ic}}$, and $W_{\text{c}}$ from observers based on ODE and DAE models with different step lengths. The PDF's of the estimation errors for simulation step lengths of (a) 3 ms are similar for both the DAE and the ODE based EKF's. For step lengths of (b) 10 ms the ODE based EKF estimation error PDF's are wider than those of the DAE based EKF which agrees with the noisy estimates of the ODE based EKF observed in Figure
Figure 5
(Color online) Normalized RMSE as function of step length for DAE and ODE EKF's. Both (a) forward and (b) backward Euler results are presented. The RMSE is normalized by the ODE RMSE with 3 ms step length for each variable. Presented discretization step lengths are $T_{\rm~s}$ = 3, 10, 15, 20, 30, 50, 100, and 125 ms, where the last five only applies to the DAE observers due to stability issues of the ODE observers. Note that the scale is logarithmic.
Copyright 2020 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有
京ICP备18024590号-1