基于样本空间排序法的电子器件抗瞬时电离辐射性能评估方法

Survivability Evaluation on Electronic Devices in Transient Ionizing Radiation Based on Ordering Method in Sample Space

针对瞬时电离辐射效应考核中应用经典非参数法存在信息利用率低,评价结果过于保守的问题,本文在识别出瞬时电离辐射效应数据在统计意义上属于Ⅰ型区间删失数据的基础上,利用样本空间排序法,开展电子器件抗瞬时电离辐射性能评估研究。首先,基于无失效数据开展了理论分析和实例计算,并将计算结果与经典非参数法的计算结果进行了对比,结果表明,样本空间排序法提高了信息利用率,不仅可以增大生存概率置信下限,而且将置信下限所对应的吸收剂量率上限从最小值增大到几何平均值。其次,开展了失效分布模型的试验研究,根据瞬时电离辐射效应特点,提出以Ⅱ型区间删失数据为数据源获取经验生存函数,通过最小二乘法拟合经验生存函数用于计算相关系数,根据相关系数及对失效机理的分析,认为商用静态存储器翻转效应服从对数正态分布的可能性最大。最后,基于再抽样技术的蒙特卡罗模拟结果表明,样本空间排序法在高生存概率区间的计算结果具有保守性。

In this paper,the ordering method in sample space is introduced and applied to compute the lower confidence limit of survival probability after classifying the test data of transient ionizing radiation into caseⅠinterval-censored data.The investigation is carried out from three aspects.Firstly,theoretical analysis and case calculation based on no failure data are carried out,and the calculation results of the ordering method are compared with those of the classical non-parametric method.The results show that the ordering method improves the information utilization rate,and not only increases the low confidence limit of survival probability,but also expands the scope of dose rate corresponding to the lower confidence limit.The upper limit of absorption dose rate corresponding to the lower confidence limit is also increased from the minimum value to the geometric mean value.Secondly,based on the characteristics of the transient ionizing radiation effect,the empirical survival function is obtained from the caseⅡinterval-censored data,and the correlation coefficient is calculated by fitting the empirical survival function with the least square method.Based on the correlation coefficient and radiation damage mechanism analysis,it is concluded that the distribution of upset dose rates for commercial SRAM is approximately log-normal.Finally,the Monte Carlo simulation results based on the resampling technique show that the calculation results of the ordering method in sample space in high survival probability interval are conservative.