Research On Fault Intelligent Diagnosis System Of Petrol/CNG Dual-fuel Engine Sensor

The use of natural gas vehicles has made a great contribution for the conservation of oil resources and environmental protection. Petrol/CNG dual-fuel engines use the performance of both the natural of petrol and gas vehicles in the environmental protection, energy conservation, economic, and security advantages, so it has good prospects for development. With gasoline/CNG cars, the vehicle fault diagnosis system encourages more and more attention.In this paper, the development of the petrol/CNG dual-fuel engine sensor fault diagnosis system is the intelligent use of intelligent diagnostic techniques of artificial neural network; it can be more effective organization, use of people's knowledge and experience to complete the sensor fault diagnosis. As more and more rapid development of high technology used in automotive structures, the complexity of modern vehicle structures determine its fault status showed diversity and uncertainty of the neural network technology for automotive fault diagnosis can greatly improve the diagnostic speed and diagnostic accuracy. Engine fault diagnosis can become more intuitive, faster, and diagnostic efficiency has been greatly improved and the engine fault diagnosis more scientific.The system is based on Windows XP platform, applies MATLAB6.5 for data processing software, using object-oriented programming language Visual C++6.0 language for the programming language and database Access 2003 to complete the work. This Visual C++ pairs of MATLAB to call through the MATLAB computational engine to complete, and better integration between the full use of Visual C++ friendly user interface and MATLAB data analysis of the advantages of processing operations, making the system more perfect.This article based on the sensor fault diagnosis in the design of neural network structure including the layer selected, the training function selected, select the number of nodes in the hidden layer and so on. To experimental data (HC, CO, NOx, CO2 and O2 emissions) and the corresponding relationship between the sensors for the training sample set, using the sample data for network training. After good neural network training and checking with several sets of data, get the desired results, the specific fault conditions of the engine sensors can be determined which meet the requirements of the sensor fault diagnosis. Intelligent analysis of automobile engine fault diagnosis method is completed. Therefore, the development of gasoline/CNG Engine Intelligent Sensor fault diagnosis system has strong practicality and expansibility.