Study On Observer For Air-Fuel Ratio Control Of A Coal-bed Gas Engine

With the increasingly stringent emission regulations and requirement of on-board diagnosis (OBD) as well as appearance of high performance microcontroller, the application of observer theory to engine control is promoted. Observer, which is an algorithm embedded in control strategies, is used for providing the state estimation required by the control low. Without waited for the measurement signals from the sensors, the delay-times originated from the response time of the sensors and the anti-aliasing filters and the signal processing are eliminated by reconstructing the state parameters before delaying. Thus, the measurement results at transient operating conditions will be more close to the actual varying states of the measured parameters, and the conditions for precise air fuel ratio control at different operating conditions are provided. The study of the observers and its related algorithms is important part of electronic control for a coal-bed gas engine. It is greatly significant to improve the engine power performance and economy and emissions behavior by applying the observer theory to predict the related parameters in the air fuel ratio control of the coal-bed gas engine.In this thesis, the observer for the air fuel ratio control of a coal-bed gas engine was studied based on the observer-related basic theories. Main contents as follows:(1) The intake manifold dynamic equation based on mean value engine model (MVEM) was established. The model parameters were identified and an observer state space equation was formed by means of the engine test data at steady state and dynamic operating conditions. The error feedback matrix was determined and the linear manifold pressure observer was constructed by using a pole assignment method. The observation effects of the linear observer were examined and analyzed with multiple sets of test data for positive and negative throttle angle step changes. The results show that the linear manifold pressure observer has good predication and noise suppression ability at certain varying operation condition by choosing proper poles to form appropriate feedback matrix.(2) In order to improve robustness of the observer, a nonlinear sliding mode observer of the manifold pressure was built by using a constant velocity reaching law. The observation effects of the observer were examined with transient manifold pressure data which were the same as the linear observer used, and the nonlinear sliding mode observer was compared with the linear observer. The results show that the performance of nonlinear sliding mode observer is better than linear observer.On the basis of the characteristics of Kalman filter, system noise vector covariance matrix and observation noise vector covariance matrix were selected, and the manifold pressure observer was built based on the extended Kalman filter algorithm. The differences in observation effects between the sliding mode observer and the observer based on the extend Kalman filter were studied. The results show that the smoother and cleaner signals can be obtained by the observer based on extended Kalman filter.(3) Aimed at the features of a pre-mixed coal-bed gas engine with double valves control structure, the feedforward and feedback control strategies for air fuel ratio were adopted. The dynamic characteristics of inlet flows and an exhaust oxygen sensor were considered, and the static feedforward look-up tables and dynamic combined air fuel ratio model were built. A closed-loop sliding mode observer was designed, which was used to simultaneously estimate the feedforward control input (air mass flow rate and fuel gas mass flow rate) and controlled plant output (excess air factor). The observation effects of observer using constant velocity reaching law, exponential velocity reaching law, power exponent reaching law and general rule reaching law were studied, respectively. The results show that the observation effects based on exponential velocity reaching law is best.(4) The above-mentioned control structure and combined air fuel ratio model were still adopted. A certain set of test data at dynamic engine operating condition was selected to calculate noise vector covariance matrix and observation noise vector covariance matrix. An air fuel ratio closed-loop observer based on the extended Kalman filter algorithm was established, and the observation effects of the observer were studied and compared with that of the sliding mode observer. The results show that, the observer based on extend Kalman filter has a better robustness and noise suppression ability since a correcting term of observation noise is introduced and the state estimations were updated using recursive calculation. Compared with the sliding mode observer, the computing complexity of the observer based on extended Kalman filter is smaller, and it is more suitable for on-line estimation of inlet mass flow and exhaust air fuel ratio.