Soundness Evaluation of Wire Ropes using Magnetic Sensor Array System by Kernel-SVM

Abstract

The damage detection of the wire ropes is using magnetic sensors and search coils based on magnetic flux leakage testing method. In recent years, a method of detecting damage in the circumferential direction of a wire rope using installing multiple magnetic sensors in the wire rope's circumferential direction has been implemented. However, it is difficult to evaluate the damage of multiple sensor data based on their correlation. Even if a damage in front of the sensor is the same as a damage between the sensors, it cannot be correctly evaluated as the same damage. Therefore, in this study, in order to focus on correlation of sensor data obtained from a magnetic sensor array system containing multiple magnetic sensors, the continuous wavelet transform (CWT) and principal component analysis (PCA) from the detected values at the wire rope flaw detector according to experiments using a multi sensor the feature amount extracted, by performing classification by kernel support vector machine (Kernel-SVM), and consider the possibility of this method as a damage identification method for wire ropes.