Pipe Damage Detection Method Based on BSWI and SVR

A new method for pipe damage detection is presented based on the combination of finite element method (FEM) of B-spline wavelet on the interval (BSWI) and support vector regression (SVR). The pipe with various damage location and size is modeled by wavelet-based elements to gain precise frequencies for forward problem analysis. The SVR algorithm is employed to build the prediction and analysis model trained by the frequencies with respect to damage location and size. The trained SVR model is applied to detect the location and size of an unknown damage in the pipe according to the first three measured frequencies. Numerical simulation is presented to show the accuracy of the proposed method.