Enhanced Creep-Fatigue Life Prediction for Geothermal Turbine Blades via Multi-Modal Bayesian Neural Network Integration

**Abstract:** Geothermal turbine blades operate under extreme temperature and strain conditions, experiencing complex creep and fatigue interactions that significantly limit operational lifespan. Existing life prediction models often struggle to accurately capture these coupled phenomena, leading to conservative designs and reduced efficiency. This paper proposes a novel Bayesian Neural Network (BNN) framework integrating multi-modal data (microstructural imaging, finite element analysis (FEA) outputs, and stress-time history) to achieve significantly improved creep-fatigue life prediction for advanced Inconel 718 alloy turbine blades. The BNN accounts for inherent uncertainties in material properties and operating conditions, providing probabilistic lifetime estimates and enabling condition-based maintenance strategies. Extensive numerical simulations and validation against accelerated testing data demonstrate a 25% improvement in life prediction accuracy compared to conventional empirical methods. This framework directly translates to optimized blade designs, reduced downtime, and increased power generation efficiency in geothermal power plants.