基于Bootstrap仿真的性能可靠性评估

Performance Reliability Assessment Based on Bootstrap Simulation

性能可靠性是很多工业产品重要的质量指标。在对产品的性能可靠性进行评估的时候,经典统计方法常常面临数据不足的问题,而Bayes方法容易引起争议,为此将Bootstrap仿真方法引入到产品性能可靠性数据的统计分析中,提出了基于Bootstrap仿真的性能可靠性评估方法。Bootstrap方法又名自助法,是一种通过数字仿真扩大样本量的方法,它完全依赖于样本本身的数据,不需要任何主观假设。首先给出了利用Bootstrap方法计算性能参数的期望与方差的算法,并分析了Bootstrap方法的适用范围,最后通过一个实例说明了本方法的实用性。

Performance reliability is an important quality index for many industrial products. For the credibility assessment, classic method always lacks test data and Bayes method is still in dispute, so this paper introduces Bootstrap method to the performance reliability statistical analysis. Bootstrap method extends the amount of samples by numerical simulation. First, this paper gives an algorithm for calculating the expectation and variance of performance parameter, then it analyses the applicability of Bootstrap method. In the end, an example is given to show the practicality of this method.