Reliability Modeling and Predictive Maintenance Integration for an Induced Draft Fan System Using Semi-Markov Process and Machine Learning-Based Predictive Models

Abstract

The proposed research enhances the dependability and efficiency levels of induced draft fan systems in thermal power plants due to the combination of semi-Markov process reliability modeling and machine learning-based predictive maintenance (PdM)methods. The system in question is a three-fan system that has two working units and one standby unit that runs cold but is not operational to give redundancy.The most important reliability metrics,such as mean time to system failure,system availability,repairman busy period, and predicted downtime,are analyzed analytically with the semi-Markov and regenerative point methods.In order to supplement the analytical reliability model,a PdM framework through machine learning with the help of a Random Forest classifier is created to predict possible failure learning conditions based on historical data on its operation.The model examines the parameters including the failure rate,the repair rate,the downtimes,and the operational capacity to detect the initial signs of system degradation.The proposed structure can greatly ensure the minimization of unplanned breakdowns and maximization of overall system performance due to the possible proactive scheduling of maintenance activities.Hybrid probabilistic reliability analysis and data-driven predictive modeling offer a solution to increasing operational decision-making in thermal power plants’maintenance systems.The suggested methodology shows that there is a prospect of relying on both reliability engineering and machine learning to realize effective real-time monitoring, enhanced availability,and cost-effective maintenance planning of vital industrial infrastructure.

Received: 2 November 2025 | Revised: 27 March 2026 | Accepted: 16 April 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data are available from the corresponding author (Kaushal Kumar) via email upon reasonable request.

Author Contribution Statement

Prawar Chaudhary: Conceptualization, Methodology, Software,Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization. Kaushal Kumar: Validation, Formal analysis, Investigation, Resources, Writing – original draft, Writing – review & editing, Supervision, Project administration.