小子样装备贮存可靠性评定方法

The storage reliability assessment for small-sample equipment

贮存可靠性评定的关键问题是小子样装备贮存寿命分布参数的估计,结合长贮装备无确切失效时间,只能实施定期检测以及试验数据的小子样等特点,推理得出定期检测随机截尾的区间型数据分布参数的极大似然估计式,该估计函数对定期测试的寿命数据具有普遍适用性。由参数的极大似然估计值确定出贮存寿命分布函数,就可对任一贮存时间的可靠度进行评定。并以威布尔分布和指数分布为例,运用遗传算法进行优化求解,算法只需给定参数的一个大致取值区间,就可快速搜索求出参数的最优估计值,避免了传统迭代求解速度慢、受迭代初值影响大的缺点。

Parameter estimation is very important for storage reliability estimation. Especially for large- scale products which storage life data is small- sample. Due to the time of failure and censor is could not test and the storage life data is small - sample. The paper puts forward Likelihood Estimation function of random censored data. The function is suitable for any periodic testing data. Finding the function＇ s solution can get the MLE of parameters. So storage reliability can estimate. As an example, the MLE equation of Weibull distribution and Exponent distribution is solved by optimum Genetic Algorithm. The Algorithm can quickly search for the optimum value that only needs an approximate interval of parameters. The method avoids the shortcoming such as low speed and strong affect about initial value for conventional iteration.