基于混合伽玛分布的Bayes可靠性评估方法

研究指数寿命产品验前信息的融合问题。为了合理地利用验前信息，同时兼顾产品可靠性试验的不确定性，提出运用混合伽玛分布作为验前分布的贝叶斯方法，并将之推广到多源验前信息和变总体的情况。重点解决多源验前信息融合权重的确定问题。通过实例，分析比较这一方法与经典方法和单纯使用伽玛分布作为验前分布的贝叶斯方法的差异，证明该方法的合理性和可行性。

The Method of Bayesian Reliability Evaluation Based on Mixed Gamma Distribution

The prior information fusion of exponential life product is investigated. In order to utilize the prior informa- tion reasonably and consider the uncertainty of the reliability tests, a Bayesian method using mixed Gamma distribution as prior distribution is proposed. This method can also be used in cases of multi--source prior information and dynamic population. The key problem of determining the weights of multi--source prior information is studied and solved. An example is presented, which analyzes and compares the difference among this method, classic method and the Bayesian method using Gamma distribution as prior distribution. The results show that this method is reasonable and feasible.