Designing and Developing Quality Control of Processes Using the Failure Mode and Effects Analysis Method and Machine Learning

Abstract

Medium-density fiberboard (MDF) production involves multiple intricate stages. Uneven thickness formation, large volumes of dynamic data from automated systems, and rapidly evolving technologies create substantial challenges for maintaining consistent quality control. Conventional approaches rely heavily on expert judgment and lack predictive capability, leaving a critical gap in timely and accurate risk assessment. This study addresses these challenges by integrating Failure Mode and Effects Analysis (FMEA) with machine learning techniques to evaluate and predict risks throughout the MDF production process. Real production data from an industrial facility were used to ensure practical relevance. Domain experts first assessed the Severity (S), Occurrence (O), and Detection (D) parameters using the PFMEA method. Predictive models—including K-Nearest Neighbors, Support Vector Machine, Neural Network, and an Ensemble Method—were then developed to estimate risk scores. The findings show that the Neural Network and Ensemble Method achieved the highest overall accuracy. This integrated approach reduces subjective bias, enhances predictive precision, and supports informed decision-making for quality control and risk mitigation in industrial MDF production.