基于惩罚似然比的高维空间相关过程EWMA质量监控模型

变量选择控制图是高维统计过程监控的重要方法。针对传统变量选择控制图较少考虑高维过程空间相关性而造成监控效率低的问题,提出一种基于Fused-LASSO的高维空间相关过程监控模型。首先,利用Fused LASSO算法对似然比检验进行改进;然后,推导出基于惩罚似然比的监控统计量;最后,通过仿真模拟和真实案例分析所提监控模型的性能。仿真实验和真实案例均表明:在高维空间相关过程中,当相邻监控变量同时发生异常时,利用所提监控方法能够准确识别潜在异常变量,取得较好的监控效果。

Quality Monitoring EWMA Model Based on Penalized Likelihood Ratio in High-dimensional Correlated Process

In traditional variable selection control chart domain, the spatial correlation problem among high-dimensional process is rarely considered. For solving this problem, a high-dimensional spatially correlated process monitoring model based on Fused LASSO algorithm is proposed. First, the Fused LASSO method is applied to optimize the likelihood ratio test. Then, the control limit of proposed model is obtained from Monte Carlo simulations. Finally, the performance of proposed model is compared with VS-MEWMA control chart through both simulations and real example. The results show that the proposed monitoring model outperforms the alternative method in high-dimensional process when the adjacent variables are spatially correlated, since the potential abnormal variables can be captured accurately by proposed method.