Helical Guided Wave Imaging Through Multi-task Deep Learning Optimization

Deep learning has been widely used in ultrasonic guided wave nondestructive testing (NDT). However, the limited view caused by the limited transducer arrangement on the pipe will greatly affect the inspection quality. A method of using helical guided waves to supplement the view is proposed. The acoustic field characteristics of axial guided waves on the pipe are analyzed, and the conversion relationship between the spiral angle and the circumferential order of the guided wave is obtained. The guided wave mode used for imaging is determined based on the view supplement model. Using dispersion curves as mode interpreters, a multi-task optimized hybrid guided wave mode imaging method is proposed, and the feasibility of the method is verified by a simulation model based on real signals.