A mode separation and recognition method based on dual sparse representation for reference-free aliasing ultrasonic guided waves signal

Abstract

In the absence of undamaged signals as reference signals, the conventional mode separation and recognition methods for aliasing ultrasonic-guided waves signal often lead to issues, such as the inability to separate weak damage signals, inaccurate mode recognition and excessive interference signals. In this article, a mode separation and recognition method based on dual sparse representation (SR) is proposed, which can shield the influence of phase errors, for the accurate separation and recognition of multimodes signals and weak damage signals. First, a lock-in process is used to extract the co-frequency component amplitude signal, which has good detection capability for weak damage signals. Then, the co-frequency amplitude signal is separated into wave packets by adaptive SR, to enhance the separation effectiveness of aliasing signal with weak damage, and reduce interference signals. Finally, a primary–secondary principle-based SR is used to achieve precise mode recognition of the separated wave packets. The experimental results show that, compared with normal SR, the proposed dual SR method has higher damage signal separation rate, lower interference subsignal separation rate, and higher mode recognition rate. The interference subsignal separation rate can be improved from −900 to −2.5%, and the comprehensive mode recognition rate can be improved from 75 to 95%.