| Summary: | 碩士 === 國立交通大學 === 資訊科學與工程研究所 === 108 === Automated material handling system (AMHS) has replaced human handling in the operation of the factory. Not only can it speed up the cycle time of the factory, but it also reduces the vibration caused by human transportation. Hence, the quality of the product becomes more stable. When the automated material handling system fails and cannot function properly, the operation of the entire factory will be interrupted, causing huge losses for the company. Therefore, if the abnormal situation can be found before the equipment is broken, the related departments will have more time to respond, and the maintenance can be performed at the most appropriate time, which can greatly reduce the cost. In the field of anomaly detection, there are many types of anomalies. In the thesis, we utilize different techniques to detect two types of anomaly: point anomalies and collective anomalies. Finally, we will compare the performance of each method in the case of various anomalies, and analyze the pros and cons of each method.
|
|---|
