Study On Air-Fuel Ratio And Engine Speed Control For Automotive Systems

Air-fuel ratio control and engine speed control are the two important problems in automotive control field. It is necessary to maintain the AFR close to its stoichiometric value and control the engine speed to ensure vehicle driveability and low pollutant levels. This is because the stoichiometry represents the maximum point of catalytic converter efficiency. Efficient performance of engine speed control can relieve the driver's weariness and reduce the traffic accidents. Also, it can reduce the fuel consumption to enhance the engine economy. The control systems of engine speed have been widely applied to automobiles.There are three main parts in this thesis, included establishing the air-fuel ratio model using BP neural network of the Matlab toolbox function, designing the air-fuel ratio controller through BP neural network and the engine speed controller through feedback linearization control method.Mathematical model establishment in nonlinear system such as automobile engine is still very hard. Artificial neural network theory brings us a new method in nonlinear system identification. This thesis builds the model of the air-fuel ratio using neural networks. This thesis uses en-DYNA software to get the required air-fuel ratio data. By using the collection data in training the neural network, this thesis establishes the proper engine air-fuel ratio neural network model through the neural network toolbox function of Matlab. Also, the neural network training algorithm, network structure and training result validation problems are discussed here.Neural network control has the virtue of simple algorithm and stronger robust. So, it is suit for controlling the nonlinear system such as engine which has pure time delay and open-loop gradually stabilization characters. To compensate for the wall wetting dynamics and the time lag, this thesis introduces the on-line training algorithm to realize the real-time control, using the weights through off-line training as the initial value in on-line training. Simulation results show that the algorithm is adaptable for the engine air-fuel ratio control.The performance of the engine speed control has effect on the economy and...