论文部分内容阅读
针对目前基于加速试验产品可靠性分析方法忽略外场环境的差异性及未考虑外场数据等问题,从数据融合的角度,考虑试验与外场环境的差异性,综合加速寿命试验的试验数据和现场数据2种信息,建立更为准确的模型,并利用马尔科夫链蒙特卡洛(MCMC)进行贝叶斯分析,解决模型复杂难于计算的问题,通过Gibbs抽样获得模型后验参数估计,从而对产品在外场工作状况下的寿命及可靠性进行评估,最后通过实例对比分析,说明综合2种数据来源不仅扩大了信息量,同时评估的精度更高,结果更加可信。
According to the current method of product reliability analysis based on accelerated test, ignoring the differences of the external environment and not considering the external data, we consider the difference of the test and the external environment from the data fusion point of view, and combine the experimental data and field data of accelerated life test To establish a more accurate model and Bayesian analysis using the Monte Carlo chain of Markov chains (MCMC) to solve the problem of complex and difficult to calculate the model. Gibbs sampling is used to obtain the model posterior parameter estimation, so that the product is out Field work conditions of the life and reliability of the assessment, and finally by case comparison analysis shows that the combination of two kinds of data sources not only expanded the amount of information, while the assessment of higher accuracy, the result is more credible.