(61a) Understanding Human-Artificial Intelligence Integration to Ensure Process System Safety

The Industry 4.0 revolution suggested transforming the process and manufacturing industries by shifting from automated to autonomous operations. Traditionally, process plants relied on automated controls governed by predefined parameters set by human operators. For instance, a pressure control system would function within established limits, only signaling alarms when an abnormal deviation occurred from a safety perspective. However, in such conventional automated systems, predicting the early abnormal situation and real-time diagnosis of the root cause during the abnormal condition remains a significant challenge. Moreover, delays in response to these abnormal events pose potential safety risks in process industries.

While traditional systems lack these predictive capabilities, recent advancements in Industry 4.0, such as Artificial Intelligence (AI) and machine learning, enable autonomous systems to detect, diagnose, and respond to process system abnormal behavior in real-time. These AI-driven systems significantly improve process safety by using data-driven insights. Therefore, AI gains knowledge based on the provided data and continuously refines its models to predict and prevent potential failures, enhance operational efficiency, and support proactive maintenance strategies. This iterative learning process allows AI to adapt to new patterns in system behavior, thereby improving accuracy in identifying abnormal conditions and contributing to a more resilient, reliable process control environment. However, with insufficient data or novel scenarios, AI may struggle to make accurate predictions or diagnoses, demanding human intervention to provide contextual insights and guide decision-making, ensuring safety and adaptability in unexplored conditions.

In response, Industry 5.0 introduces a human-centered approach known as Intelligent Augmentation (IA), which enhances human intelligence with AI support rather than replacing humans and working autonomously. This approach fosters a trustworthy AI framework for process industries by emphasizing collaborative decision-making between human operators and AI.

This paper addresses the following critical questions:

1. How can AI be integrated into process control in a trustworthy way?
2. What methods ensure that AI integration aligns with safety standards and supports unified operations?
3. What are the challenges of autonomous AI decision-making?
4. Is Intelligent Augmentation (IA) feasible in process systems?
5. How does IA blend human adaptability and situational awareness with AI’s data-processing capabilities to enhance decision-making while maintaining essential human involvement?
6. What are the risks associated with IA in process control?

By exploring these questions, this research aims to advance safe, effective human-AI collaboration in process automation, promoting a balanced, human-centered approach to autonomous control from the safety perspective.