On-line monitoring and diagnosis of out-of-control signals in gear manufacturing processes based on a hybrid ensemble learning model
Various multivariate control charts have been demonstrated to be effective in detecting out-of-control signals based upon an overall statistic.The main problem of such charts is that they can only detect an out-of-control circumstance but cannot directly determine which quality characteristic or group of quality characteristics has caused the out-of-control signal.In addition,these charts cannot provide more detailed process information,such as unnatural control chart patterns(CCPs),which would be very useful for quality practitioners to locate the assignable causes that give rise to a out-of-control situation.This study proposes an effective hybrid ensemble learning model for recognition of CCPs in multivariate manufacturing processes.The experimental results indicate that the proposed model can effectively recognize not only single CCPs where only one of various quality characteristics changes to out-of-control state at a certain time but also mixed CCPs where all quality characteristics change to out-of-control state simultaneously.
Statistical process control Control charts Multivariate manufacturing processes Artificial neural networks Support vector machines
Yang Wen-An Huang Chao
School of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,China
国际会议
第十七届国际制造会议(The 17th International Manufacturing Conference in China)(IMCC 2017))
深圳
英文
1-6
2017-11-23（万方平台首次上网日期，不代表论文的发表时间）