论文部分内容阅读
现代高可靠元器件在寿命试验时会出现失效数据很少的小子样情形,而传统的可靠性评估方法需要大量的失效数据,针对此情况,从工程实践的实际需求出发,提出了基于最小二乘支持向量机的小子样元器件寿命预测方法。该方法通过建立最小二乘支持向量机模型,从而可根据已知元器件的失效时间去直接预测同一批未失效元器件的失效时间。将该方法应用于热载流子效应引起MOS管退化失效的加速寿命试验中进行MOS管失效时间的预测,结果表明基于最小二乘支持向量机的寿命预测方法在进行小子样元器件的寿命预测时具有很高的精确度。
In the life test of modern high reliability components, there will be a small subsample of failure data. However, the traditional reliability assessment method requires a large amount of failure data. In view of this situation, based on the actual needs of engineering practice, The Life Prediction Method of Small Samples by Support Vector Machine. This method establishes the least square support vector machine model, which can directly predict the failure time of the same batch of non-failed components according to the failure time of the known components. The proposed method is applied to the prediction of MOS tube failure time in the accelerated life test of MOS tube caused by hot carrier effect. The results show that the life prediction method based on least square support vector machine When with high accuracy.