Data-Driven Damage Classification Using Guided Waves in Pipe Structures

Damage types are important for structural condition assessment, however, conventionally guided wave-based inspections, the characteristics extracted from wave packets usually used to detect, locate and quantify damages, but not classify them. In this work, data-driven method is proposed common damages in pipe utilizing signals obtained numerous damage detection tests. The fundamental torsional mode T(0,1) selected conduct reduce complexity of signal processing its almost non-dispersive property. A total 520 groups experimental data under different degrees were verify method. Finally, with help a deep neural network (DNN) algorithm, all response pipes clearly classified quite high probability.