Predictive maintenance in Electric Steel Arc Plants is crucial due to the high thermal, electrical, and mechanical stress involved in EAF operations. While my published research has not been specifically centered on EAFs, the core machine learning architectures and methods we've employed—such as deep learning, anomaly detection, and signal pattern recognition—are highly transferable to industrial environments like steel production.

Key ML approaches that are promising for EAF predictive maintenance include:

• Time-Series Forecasting Models (like LSTMs) for sensor data from electrode wear or transformer health.
• Anomaly Detection using Autoencoders to identify patterns in vibration, arc stability, or power consumption.
• Transfer Learning, especially when labeled failure data is scarce—leveraging pre-trained models from similar heavy industrial setups.
• Hybrid Techniques combining image-based inspections (thermal, visual) with telemetry logs using multimodal architectures.

From my own work—particularly in integrating real-time data ingestion pipelines, and fault classification models—I see direct applicability in modeling failure risks in components like electrode arms, hydraulic actuators, and refractory linings.

If you’re interested, I’d be happy to discuss how some of our deep learning implementations for anomaly prediction and maintenance scoring (especially from the work on intelligent alert systems and deep fake pattern detection) can be contextualized to electric arc operations.

Dear Joseph Ozigis Akomodi ,

Predictive Maintenance (PdM) emerged as one of the pillars of Industry 4.0, and becamecrucial for enhancing operational efficiency, allowing to minimize downtime, extend lifespanof equipment, and prevent failures. A wide range of PdM tasks can be performed usingArtificial Intelligence (AI) methods, which often use data generated from industrial sensors.The steel industry, which is an important branch of the global economy, is one of the potentialbeneficiaries of this trend, given its large environmental footprint, the globalized nature ofthe market, and the demanding working conditions. (PDF) Artificial Intelligence Approaches for Predictive Maintenance in the Steel Industry: A Survey. Available from: https://www.researchgate.net/publication/380756803_Artificial_Intelligence_Approaches_for_Predictive_Maintenance_in_the_Steel_Industry_A_Survey [accessed Jul 25 2025].

Preprint Artificial Intelligence Approaches for Predictive Maintenanc...

Regards,

Shafagat

Using AI and ML for Predictive Maintenance in EAF Steel Plants The use of advanced, data-driven techniques for Predictive Maintenance can improve reliability, reduce downtime, and optimize operations of Critical Equipment in EAF Steel Plants. “Sensor Integration”: gather real-time data(temperature, Vibration, Electrical current/voltage etc.) in various forms. And using AI model that you need on prediction / find the problem issue.

Artificial Intelligence (AI) and Machine Learning (ML) are being increasingly used in predictive maintenance in electric arc steel plants to improve operational reliability and reduce unexpected downtime. AI algorithms can detect subtle patterns and anomalies in sensor data from electrodes, transformers, and cooling systems in the furnace to predict equipment failures (Liu et al., 2021). Machine learning models like neural networks and support vector machines use historical maintenance records and real-time operational data to estimate the expected usage of these specific equipment and make interventions in advance of failure (Zhang et al., 2019).

AI-enabled predictive maintenance systems allow steel plants to shift from traditional reactive or scheduled maintenance to condition-based strategies, optimizing resource allocation and scheduling. Read Notes AI enables the shift from reactive or scheduled to condition-based maintenance, optimizing resource allocation and scheduling. They reduce maintenance costs and increase furnace availability by reducing inspections and avoiding breakdowns (Wang et al., 2020). Furthermore, AI-based on the Internet of Things (IoT) enhances real-time monitoring and decision-making, improving maintenance responsiveness and precision. This allows plants to maximize efficiency, safety, and equipment life. Read Notes AI plus IoT enhances real-time monitoring and decision-making, improving maintenance (Chen et al., 2022). In addition, AI and ML in predictive maintenance facilitate data-driven process improvements by identifying process parameters that reduce the lifespan of equipment, like equipment wear.

Advanced analytics can detect correlations between process variables and maintenance tasks, which can lead to improvements in design and operations to avoid these types of failures. Read Notes Advanced analytics can identify correlations between operations and the need for maintenance, as well as suggestions for design and process improvements (Chen et al., 2022). This total view improves maintenance performance and process efficiency and industry in electric arc steel production. Consequently, they offer significant advantages for predictive maintenance and help make steel manufacturing smarter and more robust.

References

Chen, Y., Liu, X., & Wang, T. (2022). Data-driven predictive maintenance in steel manufacturing: Machine learning approaches and applications. Journal of Manufacturing Systems, 62, 121–134.

Liu, X., Wang, T., & Chen, J. (2021). Machine learning-enabled predictive maintenance for electric arc furnace steel plants. IEEE Transactions on Industrial Informatics, 17(4), 2895–2904.

Wang, Y., Chen, J., & Li, H. (2020). Real-time monitoring and predictive maintenance in electric arc furnace steelmaking. Control Engineering Practice, 102, 104557.

Zhang, L., Wang, T., & Li, X. (2019). AI-driven predictive maintenance in electric arc furnace operations: A review. Journal of Manufacturing Processes, 45, 340–352.

Predictive maintenance in Electric Arc Furnace (EAF) steel plants using AI/ML involves collecting sensor data from equipment (e.g., electrodes, transformers), applying anomaly detection and failure prediction algorithms to prevent downtime, and integrating models with plant control systems for real-time intervention and optimization.