# Wind Turbine Gearbox Fault Detection Based on Sparse Filtering and Graph Neural Networks

arXiv: https://arxiv.org/abs/2303.03496 · HTML: https://mlsanity.com/paper/2303.03496/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 6 of its 38 references that are in the map.
- First posted on arXiv 6 Mar 2023.

## Builds on (most cited first, 6 of 6)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation](https://mlsanity.com/paper/1406.1078/) (2014)
- [Gated Graph Sequence Neural Networks](https://mlsanity.com/paper/1511.05493/) (2015)
- [Representation Learning on Graphs: Methods and Applications](https://mlsanity.com/paper/1709.05584/) (2017)
- [OntoWind: An Improved and Extended Wind Energy Ontology](https://mlsanity.com/paper/1803.02808/) (2018)
- [Towards Understanding Sparse Filtering: A Theoretical Perspective](https://mlsanity.com/paper/1603.08831/) (2016)

## Abstract (arXiv)

The wind energy industry has been experiencing tremendous growth and confronting the failures of wind turbine components. Wind turbine gearbox malfunctions are particularly prevalent and lead to the most prolonged downtime and highest cost. This paper presents a data-driven gearbox fault detection algorithm base on high frequency vibration data using graph neural network (GNN) models and sparse filtering (SF). The approach can take advantage of the comprehensive data sources and the complicated sensing networks. The GNN models, including basic graph neural networks, gated graph neural networks, and gated graph sequential neural networks, are used to detect gearbox condition from knowledge-based graphs formed using wind turbine information. Sparse filtering is used as an unsupervised feature learning method to accelerate the training of the GNN models. The effectiveness of the proposed method was verified on practical experimental data.