基于LLE-SVDD的高维非线性轮廓数据实时监控方法  

A Real-time Monitoring Method for High-dimensional Nonlinear Contour Data Based on LLE-SVDD

作　　者：刘玉敏[1] 梁晓莹 赵哲耘 田光杰 Liu Yumin;Liang Xiaoying;Zhao Zheyun;Tian Guangjie(School of Business,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学商学院,郑州450001

出　　处：《统计与决策》2020年第19期20-24,共5页Statistics & Decision

基　　金：国家自然科学基金资助项目(71672182,U1604262,U1904211)。

摘　　要：针对高维非线性轮廓数据的实时监控问题,文章提出了基于局部线性嵌入(LLE)和支持向量数据描述(SVDD)相结合的高维非线性轮廓监控方法。首先对受控的高维轮廓数据进行局部线性嵌入降维,然后使用降维后的轮廓数据对SVDD算法进行训练,最后用训练好的SVDD算法对高维轮廓数据进行实时监控。并利用蒙特卡洛方法生成仿真数据,以证实所提方法的有效性。结果表明,相较于其他方法,所提方法在失控状态下平均运行链长较小,能够及时发现生产过程中的异常轮廓。Aiming at the real-time monitoring of high-dimensional nonlinear contour data,this paper proposes a high-dimensional nonlinear contour monitoring method based on the combination of locally linear embedding(LLE)and support vector data description(SVDD).Firstly,the controlled high-dimensional contour data is given LLE dimensionality reduction,and then the contour data after dimensionality reduction is used to train the SVDD algorithm.Finally,the trained SVDD algorithm is applied to monitor the high-dimensional contour data in real time.The paper also employs Monte Carlo method to generate simulation data to verify the effectiveness of the proposed method.The results show that the average run length of the proposed method under the out-of-control state is smaller than that of other methods,and it can detect the abnormal contours in time.

关 键 词：局部线性嵌入 非线性轮廓 支持向量数据描述 平均运行链长 

分 类 号：O21[理学—概率论与数理统计]