Description of ActivityThis research sits under the Connected Everything Network to address Digital Manufacturing Industrial Opportunities of Flexible Manufacturing. The project aimed to assess the feasibility of using a general purpose Advanced Abnormal Perception algorithm (AAP) for SME factories with automated production. It succeeded in demonstrating that such an approach can be used to quickly customize plug-and-play anomaly detection systems for SME. This achievement is a move towards providing a low-cost means of improving an SME factory’s production line efficiency, quality control, and maintenance. In this research, we developed a self-supervised learning AAP algorithm, which is a general anomaly algorithm that can be used for production line health monitoring or product quality control. It can significantly reduce the involvement of data engineers compared to other traditional AP algorithms. Test results show that the accuracy is as high as 93% for the defects detection of the product. To visualise the process performance and predict product throughput based on the information provide by AAP, we also used Discrete Event Simulation (DES) to model the production line of the KCC Ltd. The DES that created through this research shows capabilities to work as a digital twin to real-time monitoring the physical production through the connection between DES and cloud-based MySQL database.
|Period||25 Jun 2019|
|Held at||University of Nottingham, United Kingdom|
|Degree of Recognition||International|