Comprehensive Study On Methods For Shortening Life Test Time And The Application To An Automotive Component

Modern product’s life is much longer than ever before because of the growing technology and market competition. The improvement of design, manufacturing and material makes the reliability increasingly higher. However, obtaining enough failure data becomes difficult in limited laboratory time, which is not advantage for further product’s update. Two methods are utilized to solve this problem in this dissertation: statistical method and degradation method.In zero-failure statistical analysis section, the author firstly uses Bayesian method to reveal the reliability information contained in zero-failure verification test. The author also shows the relationship of sample quantity and test time for further experiment optimization. Then traditional statistic method and double-level Bayesian method are used respectively for reliability evaluation under exponential and Weibull condition.In degradation analysis section,two ways are introduced: estimation with degradation paths and estimation based on degradation characteristic’s distribution. The author mainly focuses on the later one. Models based on Normal Wiener Process and Gamma Process are introduced and used to solve some examples, showing that they have some shortages. For this problem, the author proposed suggestions to improve the models. The result shows that the improvement is satisfying.Finally, a bending life test example of an automotive sensor harness is expounded to prove feasibility of the methods mentioned above. Comparison of the results and some questions for further study are stated in the end.