Comparison of Hilbert Transform and Complex Demodulation for Defect Identification in Cutting Discs using Vibration-Based Feature Extraction

This paper presents a novel concept for vibration-based feature extraction to identify damages in cutting discs of Tunnel Boring Machines (TBM). Defect frequencies resulting from repeated interaction of rock and disc defects are analysed. The data set is represented by the normal force acting on the edge of a cutting disc and the rock. Two different methods, the Hilbert transform and the complex demodulation, are used to generate the envelope of the time series, which was used to analyse whether the signal shows a feature representing an existing defect in the frequency domain. For the first proof of concept two numerical models were used - a multi-body system and a peridynamics 3D model simulating time series of normal forces. With both models, the linear motion of the disc on a rock sample with constant velocity was simulated. An experimental setup, mechanically similar to the simulations, was used in two experiments for further comparison. All implemented defects could be detected using vibration data of forces and one of the proposed data analysis techniques.