基于自适应Lasso阈值的EWMA协方差矩阵控制图
An Adaptive Lasso Thresholding-Based Covariance Matrix EWMA Control Chart

许多工业多元质量控制应用中通常过程失控时偏移只发生在协方差矩阵中的小部分元素中，即协方差矩阵的偏移具有稀疏性。本文针对这一现实提出一种基于自适应lasso阈值的EWMA (E_ALT)控制图，并通过蒙特卡洛模拟得到该控制图在不同偏移量下的平均运行链长指标，实验结果证明其在中小偏移下监控效率较高。之后进一步对控制图中包含的参数进行优化，以提高控制图性能。
In many industrial multivariate quality control applications, when the process is out of control, the offset usually occurs only in a small part of the covariance matrix, which calls that the offset of the covariance matrix is sparse. In view of this reality, this paper proposes an EWMA(E_ALT) control chart based on adaptive lasso threshold, and obtains the average running chain length index of the control chart at different offsets through Monte Carlo simulation, and the experimental results prove that its monitoring efficiency is high under medium and small offsets. The parameters contained in the chart are then further optimized to improve chart performance.