A novel fault diagnostic strategy based on time-varying dynamic PCA model has been presented, which makes use of continuous updating time-lagged data matrix to construct a changeable PCA model to fulfill the task of fault detection and identification.

• 摘要提出利用不断更新的时滞数据矩阵建立变化的动态主元模型对某些动态系统内的传感器故障进行检测，利用变量贡献图的样本平均对故障进行识别的方法。