Fault Detection and Classification for Robotic Test-bench

Abstract

Maintenance of industrial systems often cost as much as theirinitial investment. Implementing predictive maintenance viasystem health analysis is one of the strategies to reduce maintenancecosts. Health status and life estimation of the machineryare the most researched topics in this context. In this paper,we present our analysis for Sixth European Conference ofthe Prognostics and Health Management Society 2021 DataChallenge, which introduces a fuse test bench for qualitycontrolsystem, and asks fault detection and classification forthe test bench. We proposed classification workflows, whichdeploy gradient boosting, linear discriminant analysis, andGaussian process classifiers, and report their performance fordifferent window sizes. Our gradient boosting based solutionhas been ranked 4th in the data challenge.