贫数据多退化量产品可靠性评估

Product Reliability Evaluation on Multivariate Degradation Measures with Few Data

为了在退化数据比较贫乏时对产品进行可靠性评估,首先从数据扩充方面着手,用灰色理论对数据进行预处理,探求数据变化规律,在保证精度的前提下,用GM(1,1)模型进行一定长度的数据预测;其次,从产品的多个失效模式出发,考虑退化量之间的相关性,建立多退化量产品可靠性评估模型;最后,对于评估模型中的参数,采用线性回归方法进行建模,求出可靠度函数,从而评估任意时刻产品的可靠度。为了验证方法的有效性,进行了仿真分析,用未进行数据扩充的评估方法作为比较,仿真结果表明,所提方法有更高的评估精度。因此,在对贫数据产品进行可靠性评估时,该方法可行且有效。

To estimate the reliability of a product without enough degradation data,we considered to enlarge the degradation data .The grey theory was used to deal with the degradation data and explore the changing nature of data;and the data with certain length was predicted while ensuring the accuracy by the aid of GM （1,1） model.Then,the correlation of degradation measures was discussed and the reliability evaluation model was built up based on multiple failure modes .Finally,linear regression method was applied to model parameters of the reliability evaluation model for obtaining the reliability function,and the reliability of product could be gained at any time .To verify the effectiveness of the method,simulation was made to compare the method with the one without enlarging the degradation data .The result showed that the proposed method has higher estimation accuracy .Therefore,the method is feasible and effective for reliability evaluation of the product without adequate data .