Integrating SPC 4.0 and Machine Learning for Predictive Quality Management in Smart Manufacturing

Abstract

The evolution of quality management under Industry 4.0 has led to the emergence of Statistical Process Control 4.0 (SPC 4.0), an integrated framework combining real-time sensor analytics, machine learning (ML), and advanced statistical methods to predict and prevent manufacturing defects. This study presents a synthetic case comparing key performance indicators before and after SPC 4.0 deployment in an automotive assembly context. A simulated production line was configured to capture real-time data from vibration, temperature, and image-based sensors. These inputs fed into a dual-layer quality system comprising Hotelling’s T² control charts and ML classifiers (Gradient Boosting and CNN) for predictive defect detection. An alarm system triggered responses based on either statistical out-of-control signals or ML-derived defect probabilities exceeding a predefined threshold. Results show a 32% reduction in defect rate, a 33% decrease in customer complaints, an 85% improvement in mean time to detect (MTTD), and a 60% decline in manual inspection load. Gradient Boosting achieved an 88% accuracy (F1-score 0.82), while CNN reached 94% accuracy on vision-based tasks. The findings demonstrate that SPC 4.0 not only enhances quality control efficiency but also supports broader operational metrics such as equipment utilization and customer satisfaction. In conclusion, SPC 4.0 offers a replicable, high-impact strategy for proactive quality assurance, positioning it as a cornerstone of smart manufacturing initiatives.