Monitoring / Diagnosing the On-line Measuring Capability of Machine Vision System

• Main Authors: Ko, Te-Cheng, 柯德政
• Other Authors: Shiau Yau-Ren
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/14054711413215575715

碩士 === 逢甲大學 === 工業工程研究所 === 86 === Machine Vision System (MVS) is one of many new technologies in adding to theefficiency of manufacturing processes and metrology today. The measurementcapability of MVS must be determined before applying it. Several methods havebeen developed for the off-line determination usage. MVS can be applied foron-line applications after the measurement capability is determined and it meets the requirement defined by users. However, the measurement capabilitycould get worse gradually during the on- line applications. Furthermore, it isnecessary to develop an on- line monitoring system for MVS to reduce the qualitycost. The cause of changing the performance of MVS also should be diagnosedquickly when the malfunction happens. The measurement capability of MVS then can be adjusted or reset. Therefore, an on-line diagnosis system for MVS is to reduce the production efficient.This research builds a control chart in older to monitor the measurement capability of MVS. The quality loss is used as the quality characteristic inthe control char. The advantage of using quality loss value is that user canknow the loss clearly when controlling the MVS. This research also analysis the situation of out of control occurred in the control chart. The purpose of analysis is the variance of measurementcapability may be adjusted or non-adjusted. If the variance of measurementcapability is adjusted , MVS is not necessary to reset. Reversibly, user hasto reset the MVS. If the variance of measurement capability is adjusted, this research can adjust the variance of measurement value, and let the varianceof measurement capability restore to original value. If non-adjusted, we use two methods to diagnosis variance of measurement capability. In first method,the contribution of percentage of each control factor is used to find out possible factors that make the variance of measurement capability. The other method is checking between several combinations of variance of measurementcapability and out-control point. By the ways, user can avoid the waste time toblindly diagnosis the system for repairing MVS.