Machine Learning Algorithms for Quality Control Problem

Quality control is an essential component of modern manufacturing. In the era of Industry 4.0, methods are being developed to enhance its precision and efficiency, primarily through process automation. One facet of Industry 4.0 is Quality 4.0, for which, however there remains a scarcity of publications detailing practical use cases. Quality 4.0 represents a modern approach to quality management, integrating digital technologies, data-driven insights, and a customer-centric focus to enhance overall business performance and competitiveness. This paper presents a solution aimed at automating the quality control process within a chosen manufacturing company, employing machine learning algorithms such as k-Nearest Neighbors, decision trees, and neural networks. They allow computer systems to automatically learn and improve from experience by detecting patterns within data and enabling them to make predictions or decisions based on new inputs. Results obtained in the case study underscore the potential of machine learning algorithms in enhancing quality control processes. It has been demonstrated that the Neural Networks and Decision Tree Classifiers outperformed the k-Nearest Neighbors algorithm in accuracy and resilience to imbalanced classes. Class balancing techniques positively impacted the accuracy of identifying faulty products, particularly in scenarios with balanced class distributions. The study also emphasised the importance of individual analysis and multiple attempts in determining the most beneficial structure for Neural Networks due to their inherent complexity.