Advanced Fault Detection in Steam Turbines Using Welch’s Method and Probabilistic Modeling

Rotodynamic systems using steam turbines are essential for power generation, requiring reliable operation under continuous use. Mechanical vibrations are the major cause of failures, often linked to component degradation and structural resonances that reduce lifespan. Existing fault detection methods use signal processing, probabilistic models, and finite element analysis but struggle with scalability, real-time applicability, and integration of structural and operational data. This paper introduces a data-driven approach combining accelerometers, signal processing, and probabilistic modeling to detect anomalies. Validated on a real power system, the methodology effectively identifies resonance-related faults in rotodynamic vibrations over long-term monitoring.