A Fault Early Warning Method Based on Auto-Associative Kernel Regression and Auxiliary Classifier Generative Adversarial Network (AAKR-ACGAN) of Gas Turbine Compressor Blades

The compressor blades of the gas turbine continually operate under extreme conditions, including elevated temperature, increased pressure, rapid rotation speed, and high-load environments, and are also subjected to complex vibrations, which inevitably lead to performance degradation and failures. Early fault warning based on historical operation data and real-time working conditions can enhance the safety and economy of gas turbines, preventing severe accidents. However, previous studies often faced challenges, such as a lack of fault data, imbalanced datasets, and low data utilization, which limited the accuracy of the algorithms. This study proposes a fault warning technique for gas turbine compressor blades based on AAKR-ACGAN. First, a digital twin model of the gas turbine is constructed using long-term operation data and simulation data from the mechanism model. Then, an auto-associative kernel regression (AAKR) model is used for the fault warning, monitoring multiple parameters to provide effective early warnings of potential faults. Additionally, an auxiliary classifier generative adversarial network (ACGAN) is employed to fully extract hidden data features of the fault points, balance the dataset, and accurately simulate the process of fault occurrence and development. The proposed approach is utilized for the early detection of faults in the compressor blades of a high-capacity gas turbine, and its precision and applicability are confirmed. The multisource early warning indicator can provide an early warning of a failure up to one year in advance of its occurrence. It was also able to detect a severe surge that occurred six months before the failure, which is speculated to be one of the causes that led to the failure.