Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{174914,
        author = {Sumant Girish Tare},
        title = {Optimization of Industrial Boilers Using Reinforcement Learning},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {11},
        pages = {1038-1044},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=174914},
        abstract = {Boiler efficiency is indispensable in industrial power systems in the reduction of operational cost as well as environmental degradation. Con- ventional control strategies based on fixed rules and reactive compensation are, however, not effective in dealing with the intricate nonlinear interdependencies within operational parameters. In order to overcome these shortcomings, we introduce an integrated scheme that employs ARIMA (Autoregressive Integrated Moving Average) forecasting as a precursor to Q-learning-based reinforcement learning for anticipatory boiler optimization. The historical sensor data is preprocessed and modeled by ARIMA to forecast operational trends and a Q-learning agent designs control actions by modeling boiler operation as an MDP. The agent learns optimal actions—like fuel flow adjustments and pressure settings—to maximize a reward function optimizing efficiency, fuel consumption, and emissions. An intuitive Gradio interface allows operators to set efficiency goals and see real-time adjustments in a transparent and manual manner, as necessary. Experimental tests show that our framework transforms boiler operation from reactive to proactive, attaining statistically significant efficiency gains over traditional methods. Although the method is data-quality sensitive and demands precise hyperparameter tuning, the outcomes confirm its potential for dynamic adaptation, cost reduction, and emission mitigation. This comprehensive solution provides a strong, enduring, and cost-effective solution for real-time optimization of boilers in complicated industrial environments.},
        keywords = {ARIMA, Reinforcement Learning, Boiler Efficiency},
        month = {April},
        }