Title: Local wavenumber estimation for small damages based on artificial neural network
Abstract: The local wavenumber feature of ultrasonic guided waves can be used to detect wall-thinning because the thickness at each spatial location determines the wavenumber at that point. However, this paper shows that the accuracy of the local wavenumber estimation results is closely related to the size of the damage. When the size is small to a certain extent, the estimated local wavenumber can be significantly lower than the true value. In order to obtain the local wavenumber accurately, an artificial neural network (ANN) method is introduced in this paper to relate the estimated local wavenumber to the true wavenumber. A two-dimensional finite element guided wave model is developed to generate the rich inputs (estimated local wavenumber), while the outputs (true local wavenumber) are obtained by a semi-analytical finite element (SAFE) method. The ANN method is experimentally validated with a laser-generated ultrasonic system, and the error between the predicted local wavenumber and the true wavenumber is within 5% for a wide range of damage sizes, thus ensuring the accuracy of assessing wall-thinning.
Publication Year: 2023
Publication Date: 2023-05-11
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
Cited By Count: 1
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot