Deep learning-based fault identification in condition monitoring
Hariom Dhungana, Suresh Kumar Mukhiya, Pragya Dhungana, Benjamin Karic
TL;DR
The paper tackles real-time fault identification in rolling element bearings using vibration signals by converting time-series data into 2D images with multiple encodings and classifying fault types/sizes via a CNN. It systematically compares encoding methods (Pixel Strength, GAF, MTF, Recurrence, and GAF-MTF) and demonstrates that Pixel Strength offers the best trade-off, achieving end-to-end inference in around 6.5 ms with high accuracy on the CWRU dataset. Dual-channel encodings (GAF-MTF) and multi-speed training provide notable accuracy gains, while single-encoding Pixel Strength enables rapid deployment for time-sensitive maintenance decisions. The work highlights the practical potential of image-based DL approaches for fast, interpretable fault identification in industrial condition monitoring, including remote settings where latency is critical.
Abstract
Vibration-based condition monitoring techniques are commonly used to identify faults in rolling element bearings. Accuracy and speed of fault detection procedures are critical performance measures in condition monitoring. Delay is especially important in remote condition monitoring and time-sensitive industrial applications. While most existing methods focus on accuracy, little attention has been given to the inference time in the fault identification process. In this paper, we address this gap by presenting a Convolutional Neural Network (CNN) based approach for real-time fault identification in rolling element bearings. We encode raw vibration signals into two-dimensional images using various encoding methods and use these with a CNN to classify several categories of bearing fault types and sizes. We analyse the interplay between fault identification accuracy and processing time. For training and evaluation we use a bearing failure CWRU dataset.