中文摘要：

为了提高高压脉冲电容器贮存寿命的预测精度,根据其电容量随贮存时间延长而增大并最终导致电容器失效的规律,提出了基于自助方法和回归分析原理的等概率区间虚拟样本回归预测法。对某型400nF复合介质电容器10个产品8个月的测量数据进行分析,得到其贮存寿命服从正态分布,贮存可靠度要求为0.8时,有效贮存寿命为85个月。10个原始样本回归得到的贮存寿命置信区间宽度为27.192,而采用虚拟样本方法（样本数N=100）得到的可靠贮存寿命置信区间宽度为4.444。等概率区间虚拟样本回归方法,可以大大提高贮存寿命的预测精度,并且当虚拟样本数量增至〉1500时,贮存寿命的预测精度基本保持不变。

英文摘要：

In order to improve the prediction precision of high voltage pulse capacitors＇storage life,an equal-probability-interval virtual-samples-regression-analysis method based on Bootstrap and regression was proposed,which considered capacitance degradation rule in storage.By analyzing the data collected for 8months with 10 400nF-composite-medium capacitors,the normal distribution of storage life was got and the 0.8-reliable-life was about 85 months.The interval length of reliability-storage-life was 27.192based on the analysis of original 10products data,while with the proposed virtual sample method（N=100）,the length was improved to be 4.444.Thus,by using the virtual sample method,the prediction precision of storage life can be remarkably improved.When the number of virtual samples added to above 1500,the precision will little change.