基于GaAs激光器性能退化的可靠性度量方法

摘要

In order to improve the prediction accuracy, the risk of wrong reliability measure as a result of inaccuracy of model existed in traditional reliability measure for high-reliability, long-life laser products. Based on the performance degradation track, the nonparametric local linear regression estimation was proposed to estimate the actual degradation model directly, and then extrapolated to the false failure life time by using failure threshold and consequently made the reliability measure by employing the total life time date. Finally, a reliability test analysis through GaAs laser degradation date was done, it shows that this method improves the prediction accuracy of reliability, high fitted degree and perform robustness. The result based on nonparametric local linear regression estimation method is reasonable and precise.%针对传统可靠性度量方法对高可靠性及长寿命的激光产品进行可靠性度量存在因模型假设不准确而出现可靠性度量错误风险的问题,基于性能退化轨迹提出了利用非参数局部线性回归估计对实际的退化模型进行直接估计的方法.该方法在确定实际模型后,利用失效阈值外推获得伪失效寿命时间,进而采用完全寿命时间数据进行可靠性度量.最后通过对GaAs激光器的退化数据进行可靠性验证分析,结果表明此方法提高了可靠性的预测精度,拟合程度高,稳健性好,采用非参数局部线性回归估计方法得到的结果合理、准确.