A major milestone towards autonomous steel production
The s-X-AIPI project is reaching exciting new heights in its development, with key advancements that bring it closer to full implementation. One of the most significant achievements is the development of the entire data flow process, with only minor fine-tuning left. This data flow is critical to the project’s mission of integrating AI-based decision-making into steel production. Here’s a breakdown of the recent progress:
Complete Data Flow Development
The data flow architecture within the s-X-AIPI project is now operational, marking a significant milestone. The current process includes:
Daily Data Upload by Sidenor: Each day, raw production data is uploaded by Sidenor, providing a consistent stream of operational insights.
Data Processing and Storage by MSI: Once the raw data is uploaded, MSI processes and stores it in a structured format, ensuring its availability for downstream analysis.
Metadata Generation via AI (Developed by BFI): The real magic begins with the AI algorithms developed by BFI. These algorithms collect the raw production data, generating metadata that contains valuable contextual information about production events. This metadata is stored in the project’s database, forming a rich resource for further analysis.
Metadata Processing and Upload to Orion Context Broker (OCB): After metadata generation, it is processed and uploaded to the Orion Context Broker. This step ensures that the metadata is available for real-time monitoring and decision-making.
Analysis of the Autonomic Manager: The Autonomic Manager plays a critical role in detecting deviations or abnormal behavior based on the incoming metadata. It monitors AI functioning and issues alerts when there is a deviation from expected behaviors, allowing for proactive management of potential issues.
Alert Collection and Integration with Scrap Mix Optimizer: MSI collects these alerts and connects them to the Scrap Mix Optimizer. The optimizer then proposes an updated scrap composition based on real-time production data, ensuring optimal steel production quality.
Human-in-the-Loop (HITL) Acknowledgment: Finally, the human-in-the-loop (HITL) system receives the alert and reviews the solution proposed by the Scrap Mix Optimizer. This collaboration between AI and human expertise ensures that final decisions are made with a holistic view of both data-driven insights and human experience.
Dashboard Monitoring for Full Transparency
To enhance operational transparency, historical dashboards have been developed for every critical element of the data flow. These dashboards provide:
Real-time monitoring of each stage of the process.
Historical data visualization to track changes and improvements over time.
Control mechanisms that allow for constant oversight and immediate adjustments where necessary.
In addition to this, the project has developed Key Performance Indicator (KPI) dashboards to measure:
Cost Estimation: Tracking the financial implications of production decisions.
Oxygen and Energy Consumption: Monitoring resource use for optimization.
Ratio Between Analysis Heats and Production Heats: Ensuring consistency between experimental and actual production data.
Recipes by Steel Grade: Tracking the performance and outcomes of specific steel grade recipes.
Next Steps Toward Project Maturity
While the data flow and core functionalities are in place, the s-X-AIPI project continues to evolve. The integration of new developments and adjustments to AI algorithms and system components is ongoing. As the system grows in complexity and capability, there will be further opportunities to fine-tune both the technological and operational aspects to achieve total maturity.
In summary, the s-X-AIPI project is well on its way to becoming a fully autonomous, AI-driven production system. With its robust data flow, real-time monitoring, and integration of human expertise, it is set to revolutionize how steel production decisions are made—providing a powerful combination of data-driven insights and human intuition.
Stay tuned for more updates as we continue toward full implementation!
