Research On Ejection Fault Diagnosis Of Automobile Engine Based On Bayesian Classified For Waveform Analysis

Injection faults have been troubled car engines for a long time. There is a new requirement that how to diagnosis fast and accurate. In the ejection fault diagnosis fields of automobile engine, old diagnose methods are still based on the traditional determination of threshold and few of public papers involved solving the uncertain problems. The paper contrives a rapid diagnose method based on waveform analysis via building a special Bayesian Classified. In order to get the better correct rate, this model improved the store solution of leaning samples for Bayesian Classified by using an expert system to optimize the selections of feature vectors. The example shows that this model could be well for rapid ejection fault diagnosis, solve uncertain problems and raise correct rate. Compared with expert system and such only focus on the improvement of theory itself, this model has a better result and the value of engineering.The system has achieved the results in the following areas:1. Analysis the applied field of the Bayesian theory, explore the differences between the traditional expert system and the intelligent algorithm and conduct a feasibility analysis for car jet engine fault diagnosis based on Bayesian theory;2. According to the waveform characteristics of automotive engine injection process, this paper analysis feature extraction rules from available experimental data for researching the feature option programs;3. Write a suite of software management system of car engine fault diagnosis platform to manage the expansion modules, businesses and the sample data. The system consists of three parts: Bayesian classifier which writing in the firmware (i.e. intelligent diagnosis algorithm model), online and offline version of the diagnostic management system platform.In this paper, there is a deep analysis to existing vehicle diagnostic system of CAERI. System is based on Visual Studio platform, mixed developed by C and VB program language. At present, the diagnostic model has been initially try to be in the automobile engine fault diagnosis system platform of China Automotive Engineering Research Institute, stable and good operation, diagnostic error rate, to achieve the artificial intelligence engine fault diagnosis in automotive applications, to a certain extent, reduce human intervention, so that more objective and reliable diagnostic results, more importantly, because the introduction of the original system is not a result of the complexity of the algorithm performance decreased significantly. The diagnostic system improves the functionality of the original platform and the diagnostic accuracy, breaking the old platform to rely heavily on manual intervention to complete to determine limitations.