Research On Reliability Data Analysis For Diesel Engine

With the rapid development of automobile industry, and the improvement of science and technology, the government adopted a series of state macro-control policy. As the environmental problems outstanding constantly, demand of customers become more diversified and personalized, so that the diesel engine increasingly towards the upsizing,intelligent and high precision, and the working performance and automatic have been constantly improved. Improvement of the structure and the upsizing and complication also make the reliability of diesel engine play an increasingly prominent role. Research on the reliability of the engine has great significance to promote structural optimization and performance improvement, improve the operation reliability, security and maintainability.Engine reliability data is the real reflection of the actual quality condition, which is rich in information, is an important carrier of the operating status and faulty conditions of engine. It’s one of the effective ways to make an authentic assessment of engine reliability by analyzing these reliability data. This paper regards the diesel engine as the object of research, reliability data analysis is the master line. Analyzed the reliability data in detail with the Weibull sectional model, Weibull mixture model and D-S evidence theory, and research on engine reliability test. The presented method provided the theoretical and technological evidence for life prediction, technological improvement, fault diagnosis and prevention of engine.The main contents of this paper are as follows:(1) Reliability data analysis method of diesel engine based on Weibull distribution combination model is studied in chapter 2. The Weibull distribution combination model was used for engine reliability data analysis and processing in order to describe the distribution of engine components failure data during the life cycle. Weibull plotting paper(WPP) is not sensitive to inflection point, to solve this problem, we utilized the non-linear least-square estimation model to the mixture Weibull modeling, and adapted BFGS algorithm to resolve it.The optimization model enhanced failure curve fitting accuracy, and the results more truthful and reliable, the MAPE is 50% fewer than WPP method.(2) Research on the multi-data fusion based on improving Dempster-Shafer evidence theory(D-S evidence theory). The reliability data of diesel has the characteristic of multi-sources and uncertainty, so it is necessary to introduce the data fusion method to the fault diagnosis of diesel engine. Although D-S evidence theory is one of the effective ways to deal with problems on multi-sources and uncertainty, the method in dealing with high conflictevidences may cause misjudgment. Aiming at this problem, the similarity coefficient is introduced to D-S evidence theory to measure the weight of each evidence, thus reduce the interference of conflict evidences and improve the accuracy and efficiency of data fusion, so that provide more accurate information for fault diagnosis of diesel engine.(3) Research on the 200 hours over-speed and over-load reliability test of diesel engine.The corresponding test process is given, and the reliability data are obtained through the over-speed and over-load reliability test of a type of diesel engine. Analysis of failure mechanism of the faulty parts and propose improving measures. Observe the reliability of the new material valve seat ring by analysis of the test results, which can check the new parts and evaluate the reliability of diesel engine.