A New Detection Method Based on Magnetic Leakage Theory and BP Neural Network for Broken Steel Strands in ACSR Conductor

Abstract

Aluminum conductor steel-reinforced (ACSR) conductor is characterized by inner tensile body of steel core and outer twisted aluminum wire strands, which is obviously different from common steel wire ropes (SWRs) or cables. However, the existing studies are mainly focused on the magnetic flux leakage (MFL) testing of the SWR defects, few or not related to broken steel strands in ACSR conductors. Thus, the leakage magnetic field distributions of SWR and ACSR defects are compared and analyzed by the theoretical and numerical calculations. By considering the nonconductivity characteristics of aluminum wire strands, the steel-air model is established to calculate the effects of fracture width and the number of broken steel strands on the MFL signal for ACSR conductors. Based on the spatial distribution characteristics of ACSR leakage field with low intensity, weak lifting-off effect, and weak circumferential correlation, a single-channel signal processing method with time–frequency domain coupling is proposed. Multiple sets of characteristic values of MFL signal both in time and time–frequency domains are extracted to construct a back propagation (BP) neural network for broken steel strand identification. The results have shown that the improved single-channel time–frequency analysis method has higher accuracy in identifying the ACSR fracture width and the number of broken steel strands compared to the image processing method with multichannel leakage data fusion. With the improved method, the detection error of the fracture width is less than 2 mm, and the accuracy of the number of broken strands can reach 91.67%.