AI-Driven Predictive Maintenance in Steam Power Plants: A Hybrid LSTM-XGBoost Model for Cost Optimization and Efficiency Enhancement

Steam power plants face critical challenges in operational efficiency due to unpredictable equipment failures and suboptimal fuel consumption. This study proposes a novel hybrid AI framework combining Long Short-Term Memory (LSTM) networks and XGBoost algorithms to predict turbine failures and optimize fuel allocation. Leveraging real-time sensor data from a ۵۰۰ MW plant (۲۰۱۸–۲۰۲۳), the model reduced unplanned downtime by ۲۳%, lowered maintenance costs by ۱۸%, and improved fuel efficiency by ۱۲% compared to traditional methods. Statistical validation (p < ۰.۰۱) confirms the model’s superiority over standalone AI approaches, offering a scalable solution for the energy sector.

کلیدواژه ها:

Predictive maintenance ، Artificial intelligence ، Steam turbines ، Energy optimization ، Hybrid models