Research On Intelligent Control Strategy And Fault Diagnosis Of Electronic Controlled Gasoline Engine

In this paper, according to present situation of engine control technology in our country and priority tasks that achieving electronic control, reducing pollutant and decreasing fault, I search and analyze the related material that the development of research on electronic control system of gasoline engine in domestic and foreign. Then under our actual circumstance, it is regarded as the research object that electronic control system of gasoline engine from Delphi four cylinders minibus and 465Q electronic control gasoline engine made from Shanxi Huaihai machine plant, as well as I make analysis to software and hardware from the control system in detail. Finally, I acquire the basic design thought and method of electronic control software. On this base, intelligent control strategy of electronic control system of gasoline engine is selected as the main research orientation, using fuzzy control, neural networks and predictive control etc intelligent control theory. The control to gasoline engine idle speed, fuel injection and ignition system are researched from theory and made a great lot MATLAB experiment. Analysis and research on that problem is favorable to design high performance electronic control system. In addition, to the characteristic of more fault and high complex, design a new fuzzy neural network intelligent fault diagnosis based on expert idea.At first, the structure, working principle and control strategy of electronic control system of 465Q gasoline engine, and the fault types and fault reason of electronic control gasoline engine is analized in this paper. I design MAP figure measure and test system of 465Q electronic control gasoline engine. I measure and obtain the optimal MAP figure of the fuel injection and ignition control of 465Q gasoline engine on engine test bench by PCHud Testing and Controlling Software of electronic control system of Delphi automobile engine.Second, a fuzzy control system to idle of gasoline engine is designed in this paper by researching the test that fuzzy control theory - control idle of engine, according to the characteristic that non-linear, time-variation, uncertainty and the precise mathematical model not easy to be established. The thought of integrating fuzzy control with PID control is applied in that idle control system. Thereinto, macro-control achieves fast control by fuzzy control, and micro-control achieves accurate control by PID control. So merit of the both is brought into play to the extreme. The result of experiment shows that the method can effectively achieves the control to idle of engine, idle speed smooth and strong anti-interference capability.Third, according to optimal MAP of ignition and fuel injection of 465Q engine, it is established mathematical model for ignition and fuel injection system of 465Q engine in steady condition by neural network. Then it is proposed multiple new method and strategy of ignition and fuel injection control to highly nonlinear system. It is proposed particularly that a self-turning fuel injection control system based on neural network. The system is applicable to the control of engine that gasoline is as fuel, and meets with the diversified the request of air fuel ratio in different working condition. The system is also applicable to the control of engine that gasoline and methanol is compounded as fuel, and is flexible to specify the object air fuel ratio in meeting with the different mixture ratio of gasoline and methanol. It may achieve the request that the object air fuel ratio is arbitrary continuous to be specified in certain scope (5~30) , and also meet with the request of lean combustion technology in direct injection gasoline engines (air fuel ratio is more than 17). The result of experiment indicates that the system has good the characteristic of self-applicability, robustness and speediness. Moreover, the system structure is simple, memory is occupied less, on-line train time is short, computation speed is fast, learn capacity is strong and the object specified value of system is by zero error tracking. The system not only overcome the diversified error from manufacturing, wearing and parameter changing, but also meets with the rapid demand of real time control.Fourth, according to the characteristic of non-linear, time-variation, uncertainty and precise mathematical model not easy to be established, research the predictive control theory in fuel injection and ignition control system. It is achieved that the accurate request on air-fuel ratio and knock control of gasoline engine by multiple effective implicit generalized predictive self-adaptive control. The result of experiment indicates that the implicit generalized predictive self-tuning control algorithm is a feasible and very good effect control method.Fifth, the characteristic of more fault and high complex, developed BP neural network is applied in fault diagnosis to gasoline engine on the basic that introduction, according to the fault of electronic control gasoline engine. The result of diagnosis demonstrates that BP neural network is more practical neural network to the fault diagnosis of electronic control gasoline engine. Moreover, BP neural network has very good ability of pattern recognition and classification. But because fault of electronic control gasoline engine is complex, multiform and fuzzy, it is too rough and inaccurate to apply traditional two-valued logic which based on the Boolean algebra. So fuzzy logic concept is injected on the basic that neural network make fault diagnosis to electronic control gasoline engine. A fuzzy neural network is constructed from integrating fuzzy logic and neural network, exerting respective advantage and applying fuzzy membership functions to depict the fault extent. The result of diagnosis indicates that it is more accurate, more reasonable and higher validity by fuzzy neural network. Additionally, the thought and algorithm of constructing the fault diagnosis expert system of electronic control gasoline engine is researched, associatively the development of expert system and the characteristic of fault diagnosis of electronic control gasoline engine. It is constructed that the fault diagnosis expert system of electronic control gasoline engine based on fuzzy neural network by expert thought being great integrated into fuzzy neural network. That design integrates respective advantage of artificial neural network, fuzzy logic theory and expert system, so there is well ability of fault diagnosis. By applying GUI function of MATLAB, design a new fuzzy neural network intelligent fault diagnosis expert system and man-machine interaction interface. The system has a great improvement in convenient operation, facilitates to use and is simply and intuitively updated.In the end, It is analyze the performance of improved methanol engine in the methanol M85-petrol fuel(85% methanol and 15% gasoline) by using the methanol engine bench test system being power with admixture fuel, and makes the compared experiment where the improved methanol engine and old gasoline engine.