Paper Contents
Abstract
Industrial waste shredders are often subject to clogging with bulky or entangled materials that induce reverse motor events, which can lead to an increase in wear and loss of production. This work introduces a supervised artificial neural network (ANN)-based system for detecting clog-prone waste in real time based on a relationship between the visual content of a camera mounted on a shredder and motor reversal behavior. The model was trained with supervised data on 400 manually annotated images and achieved 92% accuracy with a 0.5 second alert time in real-time validation tests. As a proof-of-concept system, it is well suited for long-term data collection and is a strong candidate for implementation in predictive maintenance of the equipment. The flexible architecture is also relevant in other industrial settings that can pair visual and mechanical signals with negative events.
Copyright
Copyright © 2025 wasantha samarathunga. This is an open access article distributed under the Creative Commons Attribution License.
