AI-Driven Predictive Quality Engineering for Zero-Defect Manufacturing

Modern manufacturing quality control methods are becoming more and more insufficient because of their inability to prevent defects before their occurrence. The present paper introduces an AI-driven predictive quality engineering model, which is aimed at facilitating zero-defect manufacturing in a real-time predictive quality engineering framework with closed-loop feedback control. The proposed system will be tested with the help of a large-scale dataset in the sphere of automotive manufacturing that comprises approximately 2.5 million records gathered during the period of 12 months. To detect the variations in structured and temporal processes, advanced machine learning models, such as XGBoost and LSTM are used. The accuracy of prediction and recall of 94.2 and 92.8 respectively are demonstrated in the experimental results and would guarantee accurate detection of defects with few false negatives. Implementation of the framework leads to 41 percent reduction in defect rates and 28 percent reduction in scrap, which has a significant positive impact on production efficiency. Moreover, the system can be inferred in real-time with less than 120 ms of latency, indicating that the system is suitable to use in industry. The article confirms the usefulness of AI-guided predictive quality engineering as a scalable method to attain near zero-defect manufacturing.