This paper introduces a next-generation hybrid system for industrial process monitoring and control, integrating advanced statistical process control (SPC) charts with state-of-the-art artificial intelligence (AI) models. By combining robust adaptive charts such as Max-mixed EWMA and Bayesian SPC with deep learning architectures including Transformers, Graph Neural Networks (GNNs), and reinforcement/meta-learning agents, the framework achieves real-time detection, precise diagnosis, and autonomous recovery from process anomalies. Evaluation on a real-world manufacturing dataset demonstrates that the hybrid approach consistently outperforms traditional SPC and standalone neural models across key metrics, including detection delay, false alarm rate, recovery time, and interpretability. The modular architecture allows for flexible extension, human-in-the-loop transparency, and scalable deployment in dynamic, sensor-rich industrial environments. This work sets a new benchmark for smart manufacturing, highlighting the synergistic value of statistical-AI fusion for trustworthy and adaptive quality control.
Alwan et al. (Wed,) studied this question.