Optimization Design of Internal Cooling Structures for Turbine Blades Based on Machine Learning: A Comprehensive Review

With the steady growth of performance requirements of gas turbines, the optimization design of turbine blade cooling structures has emerged as a critical technical challenge. This paper provides a comprehensive review of the research progress on machine learning methods in the optimization design of turbine blade cooling structures. First, common internal cooling methods for turbine blades are reviewed, including rib turbulated cooling, pin-fin cooling, and impingement cooling. Second, the application and limitations of traditional optimization methods in the optimization design of turbine blade internal cooling structures are analyzed in detail. On this basis, the application of machine learning methods in the optimization design of turbine blade internal cooling structures is comprehensively reviewed, covering mainstream machine learning models such as artificial neural networks, deep learning, generative adversarial networks, and physics-informed neural networks, as well as mainstream optimization algorithms including genetic algorithms, non-dominated sorting genetic algorithms, particle swarm optimization, and simulated annealing. Furthermore, emerging research directions such as data augmentation techniques, temperature field reconstruction, model interpretability, and optimization structure evolution patterns are discussed. The reviewed literature demonstrates that machine learning methods can substantially enhance the efficiency and accuracy of cooling structure optimization, offering robust support for the cooling design of next-generation gas turbines.