基于高噪声数据的VMD-EWMA控制图  

VMD-EWMA control chart based on strong noise data

作　　者：杨亦浏 朱永忠[1] 徐薪苡 YANG Yiliu;ZHU Yongzhong;XU Xinyi(School of Mathematics,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学数学学院,江苏南京211100

出　　处：《甘肃科学学报》2025年第1期33-39,68,共8页Journal of Gansu Sciences

摘　　要：生产过程中的数据因受到复杂随机因素影响,呈现高噪声特点,使得传统的质量控制方法难以准确识别和响应,从而误导生产环节的决策。现有的指数加权移动平均(EWMA)控制图虽然能够在一定程度上减少数据中的噪声干扰,但在面对高噪声数据时效果并不理想。为了提高EWMA控制图在处理高噪声数据时的准确性,将变分模态分解(VMD)与EWMA控制图相结合,提出了一种基于高噪声数据的VMD-EWMA控制图。通过VMD方法分解出高噪声数据中的本征模态函数来还原真实的过程特征信息,进而结合EWMA控制图进行异常监控。最后使用仿真验证所提方法的有效性,将VMD-EWMA控制图与奇异值分解和小波阈值去噪结合EWMA控制图后的结果进行对比分析,结果显示VMD-EWMA控制图能够更好地抑制高噪声对过程监控的影响,减少误报和漏报的风险,有较高的应用价值。The data in the production process is characterized by strong noise due to the influence of complex random factors,which makes it difficult for traditional quality control methods to accurately identify and respond,thus misleading the decision-making in the production process.Existing exponentially weighted moving average(EWMA)control charts can reduce the noise interference in the data to a certain extent,but its effect is not ideal when facing strong noise data.In order to improve the accuracy of the EWMA control chart when facing strong noise data,this paper combines the variational modal decomposition(VMD)with the EWMA control chart to propose a VMD-EWMA control chart based on high-noise data.The intrinsic modal functions in the strong noise data are decomposed by the VMD method to restore the real process characteristic information,which is then combined with the EWMA control chart for anomaly monitoring.The effectiveness of the proposed method is verified by simulation,and the results are compared and analyzed with the results of singular value decomposition and wavelet threshold denoising combined with EWMA control chart,which show that the VMD-EWMA control chart is able to better inhibit the influence of strong noise on process monitoring and reduce the risk of false alarms and omissions,and it has a high application value.

关 键 词：高噪声数据 变分模态分解 指数加权移动平均控制图 质量控制 

分 类 号：O212.1[理学—概率论与数理统计]