Pipe damage detection method by combination of wavelet-based element and support vector regression

This paper proposes a new method based on wavelet-based element and support vector regression (SVR) for pipe crack detection. The cracked pipe is modeled using B-spline wavelet on the interval (BSWI) element to obtain the precise frequencies database associated with different crack location and depth. Subsequently, the database is employed as training samples to construct the crack prediction model by means of SVR algorithm. The first three frequencies measured are inputted to the model to predict the location and severity of unknown crack. Both the numerical simulation and experimental study have verified the validity of the proposed method