English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Reliable fault detection is an essential requirement for safe and efficientoperation of complex mechanical systems in various industrial applications.Despite the abundance of existing approaches and the maturity of the faultdetection research field, the interdependencies between condition monitoringdata have often been overlooked. Recently, graph neural networks have beenproposed as a solution for learning the interdependencies among data, and thegraph autoencoder (GAE) architecture, similar to standard autoencoders, hasgained widespread use in fault detection. However, both the GAE and the graphvariational autoencoder (GVAE) have fixed receptive fields, limiting theirability to extract multiscale features and model performance. To overcome theselimitations, we propose two graph neural network models: the graph waveletautoencoder (GWAE), and the graph wavelet variational autoencoder (GWVAE). GWAEconsists mainly of the spectral graph wavelet convolutional (SGWConv) encoderand a feature decoder, while GWVAE is the variational form of GWAE. Thedeveloped SGWConv is built upon the spectral graph wavelet transform which canrealize multiscale feature extraction by decomposing the graph signal into onescaling function coefficient and several spectral graph wavelet coefficients.To achieve unsupervised mechanical system fault detection, we transform thecollected system signals into PathGraph by considering the neighboringrelationships of each data sample. Fault detection is then achieved byevaluating the reconstruction errors of normal and abnormal samples. We carriedout experiments on two condition monitoring datasets collected from fuelcontrol systems and one acoustic monitoring dataset from a valve. The resultsshow that the proposed methods improve the performance by around 3%~4% comparedto the comparison methods.

A Novel Unsupervised Graph Wavelet Autoencoder for Mechanical System Fault Detection