Proceedings of the
9th International Conference of Asian Society for Precision Engineering and Nanotechnology (ASPEN2022)
15 – 18 November 2022, Singapore
doi:10.3850/978-981-18-6021-8_OR-08-0273

Predictive analytics for downtime prevention in Plastic Injection Molding Line

Junt Kong Chan, Shiming Ang, Hongtao Wang, and Yan Shen

Agency for Science, Technology and Research 3 Cleantech Loop, #05-06, CleanTech Two, Singapore 637143

ABSTRACT

This paper discusses the implementation of predictive analytics a plastic injection molding production line. The production line was analyzed to understand the process and identify the machines for their function, parameters available and communication capability / protocols. Sensors for environmental parameters, water, and compressed air supply were added. An IIoT architecture was designed for communication interoperability between the edge computer and the line machines. This allowed each machine data to be captured, synchronized and stored in a database. A dashboard was designed with user friendly interface (UI), allows operators to easily comprehend production status; and process experts can analyse historical data via charts. For the analytics, the line data was studied with the process experts to identify line issues and key parameters which drive these issues. It was found that the line would shut down a few times a day for an unknown reason, at that time. Using analytic tools and studying data captured from the production line, key parameters causing the unplanned shutdown were identified. The event was able to be predicted ahead of time. Based on the findings regarding the issue root cause, the predictive analytics model was developed and fine-tuned. The model provided alerts with 10 minutes lead time at 75% confidence.

Keywords: Industrial Internet of Things, Predictive Analytics, Visualization



PDF Download