小批量零件加工质量数据的动态建模  

Part Quality Modelling of Smallbatch in the Flexible Manufasturing Environment

作　　者：毛宁[1] 陈庆新[1] 刘霞[1] 

机构地区：[1]汕头大学

出　　处：《机械科学与技术》1996年第4期579-584,共6页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家863计划CIMS主题基金资助项目

摘　　要：分析了在小批量的产品生产方式下传统零件加工过程的统计过程控制方法的局限性；说明了基于自回归模型的简单最小二乘算法处理数据自相关性的可行性与不足之处；针对单批次小批量数据中随机噪声水平相对较高、而噪声统计特性没有充分展开的主要矛盾，借鉴自适应卡尔曼平滑算法对于观测噪声的一定抑制机制，提出将该类方法应用到小批量零件的生产加工的质量建模中来；最后，利用实验数据对以上提出的各种方法进行了比较研究，充分说明了简单最小二乘算法的可行性。In the way of smallbatch production, this paper first analyses the confines of traditional Statistical Process Control (SPC) algorithm. And then feasibility and defect of the simple leastsquares algorithm, which is used to establish an autoregressive model, are thoroughly expounded. When the power level of random noise is high and the statistical features of noise have not been demonstrated sufficiently yet, the most effective approach to improve the precision of estimate is to decrease the total level of the noise. Considering the restraining mechanism of adaptive Kalman smoothering algorithm, this algorithm is efficiently applied to the quality modelling of smallbatch production. At last, a group of experiment data is used to compare the effects of three algorithms above stated, with these experimental results, the feasibility of simple leastsquares algorithm and effort of adaptive Kalman smoothering algorithm are well explained.

关 键 词：质量控制 小子样建模 ARMA模型 

分 类 号：TB114[理学—概率论与数理统计]