A holistic zero-defect manufacturing framework enabled by AI and digital twins for smart production systems

This paper presents an original research contribution: a holistic Zero Defect Manufacturing (ZDM) framework integrating Artificial Intelligence (AI) and Digital Twin (DT) technologies to enable a validated predict-and-prevent quality management strategy. Artificial neural network-based defect prediction models are embedded within a real-time synchronised digital twin, forming a closed-loop system that simultaneously optimises quality, productivity, and sustainability through AI-enhanced scenario evaluation and automated PLC parameter adjustments. Process data from a real automotive injection-moulding production line were used to train the AI model and calibrate the digital twin; framework performance was measured on the physical system over six months (three-month baseline followed by three-month AI-enabled operation), with simulation serving as the scenario evaluation engine only. Results demonstrate: OEE from 62.5% to 80.3% (+28.5%), downtime from 13.8 to 5.0 h/month (−64%), energy from 3.62 to 2.43 kWh/unit (−33%), and defect rate from 5.37% to 1.31% (−76%); all p < 0.001 with very large effect sizes (Cohen’s d > 2.3). These synergistic gains substantially exceed single-objective implementations, confirming a scalable, data-driven pathway for sustainable ZDM in Industry 4.0.