| Summary: | 碩士 === 元智大學 === 資訊工程學系 === 105 === With the smart manufacturing and industrial 4.0 concepts sweeping across the globe, the thing that how to through analyzing data to be more efficient on using sensor data has become important. In the traditional production line, usually for a single production line to evaluate the impact of product and production line’s performance. General when products are produced in the production line, the production line is independent operation. However, this production experience is hard to use in other similar production line. In a production line, it usually covers dozens or hundreds of sensors and parameter settings. After a long period of testing, engineers can know the production status of product and production line parameter settings and the relationship between the sensors. If we are able to apply related settings of mature production line to other new line, we can effectively improve the speed of production line deployment and the ability to debug.
In this thesis, we provide user a visualization system. According to analyzing the multiple sensor data of the production line and the information of the parameter settings, we can evaluate the relation of the sensor in the same production line and in different production line. By hierarchical clustering, we can visualize the correlation between the sensors. As a result, we can get from the corresponding sensors to learn how to read the parameter of the sensor and quickly understand the relationship between the sensor and the production status on the new production line. Furthermore, in our study, we further compare the results of sensor clustering with different sets of backgrounds and help user compare multiple different time period, products and setting in different production line in order to optimize the setting parameter and find unusual sensor.
|
|---|
