基于主成分追踪方法的过程监测

Process monitoring based on principal component pursuit

为了将主成分追踪(PCP)的方法应用在工业系统过程监测中,提出一种基于均值方差的过程监测统计量。首先将标准化后的训练以及测试数据矩阵进行PCP分解,然后分别计算分解得到的稀疏矩阵中每个变量的均值和方差,并进行比较,最后将它应用在数值仿真和TE过程中。研究结果表明:这种统计量能够消除大部分噪声的影响。使用这种统计量能够同时完成模型建立、故障检测、故障识别以及过程重构这4个过程监测的步骤,具有较好的适用性。

To apply principal component pursuit （PCP） method for process monitoring, a mean variance statistic was proposed for process monitoring. First, the standardized training and testing data matrices were decomposed by PCP. Then the mean and variance of each variable were calculated in corresponding sparse matrix. The numerical simulation and TE process were provided. The results show that this statistic can eliminate the influence of almost noise interference in process. It can accomplish the objectives of model building, fault detection, fault isolation, and process reconstruction simultaneously, and illustrate the effectiveness of proposed statistic.