摘要
多元质量控制过程中,为了准确诊断引起失控状态的变量或变量组合,需要对样本数据进行降维操作以消除数据之间的相关性。针对现有多质量特性降维方法忽略各采样批次之间相关性的问题,利用自回归滑动平均模型能够从数据当中提取出基于时间独立关系的特性,尝试将自回归滑动平均模型与传统PCA算法相结合,设计出一种改进的PCA算法,用于处理存在相关性的数据降维问题,并利用汽车曲轴生产实例验证其有效性。
In multivariate quality control,data dimension reduction is required to eliminate data’dependencies and diagnose failure factors when the system is out of control.As interaction between sample batches is often overlooked in existing methodologies,in this paper,the autoregressive and moving average model which is able to extract individualities from data based on time integrates the traditional PCA algorithm to construct an improved PCA algorithm for dealing with the problem of data dimension reduction.Ultimately,a case study in automotive crankshaft production process is presented and shows its effectiveness.
出处
《工业工程与管理》
CSSCI
北大核心
2014年第3期66-71,共6页
Industrial Engineering and Management
基金
国家自然科学基金资助项目(60903124)
高等学校博士学科点专项科研基金项目(20096118120003)
西安市科技计划项目(CX1255-4)
关键词
多元质量控制
自回归滑动平均模型
主成分分析
数据降维
multivariate quality control
autoregressive and moving average model
principle component analysis
data dimension reduction
