Machining Tool and Bearing Failure Analysis Using the Orthogonal Hilbert-Huang Transform on Vibration and Motor Current Datasets

Abstract

Early fault detection is the primary goal of condition-based maintenance (CBM); to identify an impending fault before failure occurs and provide the necessary maintenance in a timely manner. Machining tools, such as drill bits, require a significant amount of maintenance during industrial machining operations to ensure that workpiece tolerances are met. Faulty tools tend to machine outside expected tolerances and run the risk of causing permanent damage to the workpiece, resulting in added production costs and unwanted delays. Similarly, ball bearing maintenance is crucial for keeping a machine in acceptable working condition, and has been the primary focus of many CBM studies due to nonlinear bearing degradation. The orthogonal Hilbert-Huang transform (OHHT) is an improvement to the Hilbert-Huang transform (HHT) that is significantly more computationally efficient compared to other improved HHT algorithms. The adaptive nature of the HHT makes it suitable for analyzing nonlinear and nonstationary phenomena and it returns instantaneous energy and frequency results—an advantage over traditional Fourier analysis. This paper showcases the OHHT’s potential as a useful diagnostics tool for analyzing machine induced vibration and motor current signals from ball bearing and drilling datasets. Features from the OHHT were fed into a neural network classifier giving health results consistent with another literature study using the same data. Through transfer leaning, a trained neural network from the ball bearing dataset was used to classify drill bit health from the drilling dataset and gave expected health results.