Structural Response Analysis of Runner Blade Adjustment Mechanism of Kaplan Turbine by Modal Decomposition and Function-Driven

This study focuses on runner blades of hydraulic turbines, aiming to achieve accurate prediction of blade torque and structural stress analysis under eight operating conditions through fluid-structure interaction analysis.A combined method of Variational Mode Decomposition (VMD) and Long Short-Term Memory (LSTM) network is adopted to process blade torque signals: VMD decomposes complex torque time-series signals into intrinsic mode functions with different frequency characteristics, while LSTM network, leveraging its advantage in capturing long-term dependencies in sequence data, reconstructs precise torque variation patterns, with the root mean square error (RMSE) ranging from 0.02% to 0.51% across all operating conditions, fully verifying its effectiveness.Stress analysis shows that there are three key highstress areas in Kaplan turbines, namely the corner of the tumbler arm, the middle part of the side surface of the connecting plane, and the diameter-changing section of the thin shaft on the ear lug, among which the diameter-changing section of the thin shaft on the ear lug has the highest stress level and should be the key target for structural optimization.Correlation analysis indicates that when the runner blade angle is 0 and -5, stress fluctuations are the most drastic, while an increase in the angle (e.g., exceeding -5) can significantly improve operational stability, and combining stress analysis with the full characteristic diagram of the hydraulic turbine can effectively delineate safe and stable operating regions.The results demonstrate that the VMD-LSTM framework can efficiently complete torque prediction with limited computational resources, and the stress analysis results provide targeted guidance for structural reinforcement, offering important references for the design optimization, durability improvement, and safe operation of hydraulic turbines under different hydrological conditions.