Forecast pipe defect size based on modified grey system and guided wave signal recognition with matching pursuit

This work aims to develop a method to predict the pipe defect size by modified grey system and guided wave signal recognition with matching pursuit (MP). Firstly, the evolutionary programming using mutations based on the t probability distribution (tEP) is applied to the matching pursuit method. Then, a steel pipe with a notch of different depth is tested by guided wave testing system and the measured signals are decomposed by MP with chirplet atoms and the amplitudes of defect echo are extracted. At last, the signals amplitudes are made to be the original sequence to predict the defect grow by modified grey system. The forecast error of the defect size is analyzed and the error ratio is restricted within 2%-5%. Therefore, this method can effectively estimate the defect size and the steel pipe defect can be quantitatively identified.