Proposed Cloud-assisted Machine Learning Classification Process implemented on Industrial Systems: Application to Critical Events Detection and Industrial Maintenance

Machine learning methods are in great demand nowadays, as more and more smart embedded logic, included in Industrial equipment, are considered while building industrial systems. Such control equipment is assigned with data mining and machine learning tasks. This work proposes a framework for the next generation Industry 5.0 machinery with logic. This framework furthers the Industry 4.0 concept of cloud interconnected industrial machines, offering real-time and ondemand responses to personnel queries. The authors implemented an embedded multi-layer perceptron classifier to strengthen their approach. The proof-of-concept classifier has been trained using sensory data from HEL.PE. oil refinery compressors. It can be used in such a way as to be capable of automatic cloud-driven self-train, operating on existing industrial equipment. The proposed self-trained model accuracy has also been examined, as well as its ability to fit IIoT devices and model self-training. As the authors propose, such framework-friendly machine learning algorithms can be highly chosen to solve industrial maintenance problems and detect critical events. This effort can be further enriched, especially when the underneath infrastructure allows us to use a deeper model, benefiting from the deep learning theory.