Improved Fault Diagnosis in Hydraulic Systems with Gated Convolutional Autoencoder and Partially Simulated Data

This paper examines the effectiveness of neural network algorithms for hydraulic system fault detection and a novel neural network architecture is suggested. The proposed gated convolutional autoencoder was trained on a simulated training set augmented with just 0.2% data from the real test bench, dramatically reducing the time needed to spend with the actual hardware to build a high-quality fault detection model. Our fault detection model was validated on a test bench and showed accuracy of more than 99% of correctly recognized hydraulic system states with a 10-s sampling window. This model can be also leveraged to examine the decision boundaries of the classifier in the two-dimensional embedding space.

Automated summary

This paper explores the effectiveness of neural network algorithms in hydraulic system fault detection and introduces a novel neural network architecture. The proposed gated convolutional autoencoder, trained on a simulated training set augmented with just 0.2% data from the real test bench, significantly reduces the time needed with the actual hardware. The fault detection model demonstrated over 99% accuracy on a test bench, with the ability to examine decision boundaries of the classifier in a two-dimensional embedding space.