A method for identifying cross-domain prestress in in-service structures based on ultrasonic guided waves

Abstract

This invention provides a method for identifying cross-domain prestress in in-service structures based on ultrasonic guided waves. The method includes: acquiring ultrasonic guided signals of steel strands under in-service conditions; inputting the ultrasonic guided signals of the steel strands under in-service conditions into a target machine learning model to obtain predicted prestress values ​​for the steel strands under in-service conditions. The target machine learning model is trained using the mapped feature matrix of the source domain ultrasonic guided signals and the corresponding prestress as samples. The source domain ultrasonic guided wave signals are the ultrasonic guided signals of the steel strands under different prestresses in the experimental state. The mapping process aims to minimize the difference in feature parameter distribution between the source domain ultrasonic guided signals and the ultrasonic guided signals of the steel strands under in-service conditions. By implementing this invention, the problem of feature distribution differences between the source and target domains can be effectively solved, the requirement for target domain samples can be reduced, the data acquisition difficulty can be lowered, and the training efficiency of the model can be improved.

Images