A Quantitative Diagnosis Method for Aero-Engine Blade Cracks Based on 3D-BTC

The aero-engine is known as the heart of the aircraft, and the turbine is particularly important as one of the key parts. The blades on the turbine discs are often dangerous due to crack failure. Therefore, a quantitative diagnosis method of blade crack parameters based on Transformer-ResNet-GRU-Channel Attention using three dimensional blade tip clearance(3D-BTC) signal data is proposed in this paper. A quantitative diagnosis experiment was carried out after 3D-BTC data was obtained by dynamic test bench. The results show that the identification accuracy of the crack depth, position, pitch angle and roll angle of the blade reaches 99.58%, 100.00%, 100.00% and 99.17%, respectively. At the same time, the ablation experiment proved that the proposed method combined the advantages of each module to improve the diagnostic accuracy, and further proved the superiority of the proposed method by comparing with other methods.