Summary: | 碩士 === 國立成功大學 === 製造資訊與系統研究所 === 106 === Nowadays, Automatic Virtual Metrology (AVM) has been widely used in the high-tech industry and the machine tool industry to achieve online real-time inspection. However, there are still many issues to be improved in the AVM implementation of the machine tool industry. One of the biggest challenges is how to build an optimal model automatically. As the modeling processes are complicated with higher standards, and algorithm and statistical expertise are required, it takes a lot of manpower and time for model creation in the machine tool industry. In order to resolve the problems mentioned above to automatically build an optimal model, three major topics starting from the pre-modeling data pre-processing to the whole modeling process are discussed in depth in this research: 1) optimal processing signal acquisition , 2) Automated Model Creation (AMC), and 3) model management. Among them, the optimal processing signal acquisition mainly deals with the issue on how to find out the effective machining signals in data pre-processing to enhance the AVM accuracy for more accurate model creation. Then Automated Model Creation (AMC) is integrated with GED-plus-AVM System (GAVM) to simplify the model creation processes, and algorithms are utilized to find out the most appropriate model for each measurement item. Moreover, factory-wide model management is adopted by developing a model changing mechanism and setting the indicators and light signals of model freshness. Through the webpage interface, not only that users can see the model status at a glance, but model replacement can also be performed at any time to achieve the ultimate goal of managing the factory-wide models.
|