Corrosion characterization of steel wires based on persistent homology theory for magnetostrictive guided wave testing signal

Abstract

Corrosion of steel wires in cables is a frequent occurrence and is a major factor affecting the durability and safety of cables. Magnetostrictive guided waves have been employed to detect steel wire corrosion. However, due to the complexity of the geometry of the corrosion area, it is difficult to interpret the magnetostrictive guided wave testing signals. Hence, extracting features from testing signals to characterize the corrosion condition is a great challenge. This study proposes a topological feature for corrosion characterization of steel wire, which is extracted from the two-dimensional point cloud with optimal geometry structure of the testing signal based on persistent homology theory. Corrosion experiments were carried out on 30 galvanized steel wires. The experimental results indicate that the topological feature increases roughly linearly with the corrosion depth. Furthermore, compared to other features obtained from the time domain, frequency domain, and joint time–frequency domain, this topological feature can reflect the progression of corrosion more accurately.