The Research On The Expert Systems Of Fault Diagnosis Of The Electronic Control Engine's Sensor Based On Neural Network

Electronic control EFI engine has many advantages, such as enhances the power, reduces the fuel oil consumption, and reduces the emission, and so on. With the nationalⅢstandard's execution, electronic control EFI engine will become the future trend in development. Although the fault rate of electronic control engine is low, and once it breaks down, which not only affects the engine's performance, but also difficult to diagnose. Especially sensor's fault, it is closely related with engine's performance. Thus research on the fault diagnosis of the electronic control engine's sensor will extremely essential and necessary.Aiming at the various sensors used by electronically controlled engine currently, this paper stresses to introduce the throttle position sensor and the structure, principle, signal characteristics and common fault of the inlet manifold air pressure sensor. And on basis of the development platform of Visual C ++ 6.0 and MataLab7.0 and the joint of the advantages of neural networks BP and the traditional expert systems ,fault diagnosis neural network expert system is designed. Process waveform data of throttle position sensor and the inlet manifold air pressure sensor with the method of digital filtering technology, and then put them in neural networks BP. After comparing the margin between the output estimated by fault diagnosis neural network and the actual output of the two sensor signals working under idling in every moment of sampling, the fault that the throttle position sensor and the inlet manifold air pressure sensor value holding and the non-calibration is diagnosed by judging this margin whether oversteps error limits. Finally, the diagnosis is converted into expert system to identify the knowledge output to the user interface. After collecting waveform data from the test bench, the fault diagnosis neural network expert system is verified, the results show that the strategy is feasible. At the same time have a high value projects.