中文摘要：

研究电子器件生命特征、生命信息与潜在缺陷及寿命的相关性，依赖于其生命信息可靠寿命的预测方法。根据电子器件多参数多采样点组成的序列信息矩阵来研究其性能退化规律，其中涉及数据繁多，需进行压缩。结合神经网络，对包含电子器件序列信息的采样本点进行多方式选取，通过比较所得误差来最终确定采样点的选取方法；采用主成分分析法对继电器序列信息矩阵的列向量进行处理，提取出关键性能参数；构造适合的价值函数以获取描述电子器件性能可靠程度的特征量，对电子器件性能退化进行研究。

英文摘要：

The relationship of the potential defect and life with life characteristic and information were studied, and the life prediction depending on life initial information was introduced. The performance degradation rules of the relays were studied based on the sequence information matrix that contains many parameters and sampling points. The mass of data related had to be compressed. Some methods combined with neural network were used to select the sample of sequence information, and the choosing method of samples was determined by comparing the errors. To extract the key performance parameters, the column vectors of sequence information matrix were processed with principal component analysis. The proper value function was constructed to acquire the characteristic quantity which can describe the comprehensive performance reliability of the relays. The performance degradation model of the relays was studied.