针对产品可靠性统计分析中,经常面临的小样本问题或误差项分布不明确的问题,将Bootstrap方法引入到产品可靠性的回归统计分析,提出了基于极大似然-最小二乘估计(ML-LSE)二步法的产品可靠性Bootstrap统计分析方法;同时通过对Bootstrap估计值进行纠偏处理,提高了小样本条件下或误差项分布不明确时产品可靠性的统计精度,并求得某型电连接器在正常应力水平下可靠性特征值的区间估计值。统计模拟的结果表明,经纠偏处理后的Bootstrap回归统计分析方法,所得产品可靠性特征值的估计精度能满足置信度的要求。
Aiming at the problems of small sample and indefinite error distribution confronted in product reliability statistical analysis, Bootstrap method is introduced into regression statistical analysis of product reliability, a product reliability Bootstrap statistical analysis method based on Maximum likelihood-Least square estimation (ML-LSE) two-step method is proposed. At the same time, through correcting the bias of the Bootstrap estimate value, the product reliability statistical precision under the condition of small sample or indefinite error distribution is improved. The confidence interval estimate value of the reliability characteristic at normal working stresses for certain type electrical connector was obtained. Statistical simulation computation result shows that, using the Bootstrap regression statistical analysis method, when the bias is corrected, the estimation precision of product reliability characteristic satisfies the requirement of confidence level.