Real Time Quality Assessment in a Production Line: Machine Learning Approach

In a modern manufacturing unit, there are some demanding requirements like minimization of system faults, boosting productivity, real-time monitoring, planned maintenance, and several others. Lack of data quality, consistency, and insufficient computational power resulted in limited artificial intelligence (AI) integration to enable a decision support system. However, machine learning (ML) models improved the uniformity of manufactured products, effectively lowering production costs while reducing manufacturing lead time and downtime in Industry 4.0. Thus, this paper addresses three main areas namely failure type prediction (along with chance of it), product segmentation, and quality assurance. It addresses failure type prediction with the implementation of linear SVC, KNN, and Random Forest Classifier (performed best) on the existing data set. On the other hand, product segmentation and quality assurance are successfully implemented with K means model and MobileNet V2 respectively in real time. Additionally, quality assurance is also implemented on Raspberry Pi as a computing platform.