论文部分内容阅读
研究Phase Ⅰ线性轮廓的局部数据发生变化时的变点识别问题。通过分割轮廓,将各段轮廓特征值视为多元变量,结合T2统计量设计变点识别方法。仿真性能比较研究的结果表明当线性轮廓的斜率或截距出现局部变化时,新方法的变点识别能力优于现有方法。因此本方法是一种有效识别线性轮廓局部变点的Phase Ⅰ分析工具。
To study the identification of changeable points when the local data of Phase Ⅰ linear profile changes. By segmenting the contour, the feature values of the contour segments are regarded as multivariate variables, and the method of variable point identification is designed by combining the T2 statistics. The simulation results show that when the slope or intercept of the linear profile changes locally, the new method has better change-point recognition ability than the existing methods. Therefore, this method is a Phase Ⅰ analysis tool that can effectively identify the local change points of linear profiles.