Wind turbine gearbox fault prediction algorithm based on feature subspace

Yubin Pan, Rongjing Hong, Jie Chen, Hua Wang

Research output: Contribution to conferencePaperpeer-review

Abstract

As to wind turbine gearbox, vibration signals are usually non-stationary and non-Gaussion feature, which makes it very difficult to extract useful information. In this paper, we propose a wind turbine gearbox fault prediction algorithms using the feature subspace based kernel principal component analysis (FS-KPCA) and exponentially weighted moving average (EWMA). After that, the vibration signal in normal operation is usedto build the feature subspace where the current monitoring data are mapped to. Then,the detection is achieved viacomparing the hyper-plane defined by the orthogonal basis between the reference and a current state of the system through the concept of subspace angle.Subspace angle is used as an indicator to detect the fault and its threshold is defined by SPC(Statistical Process Control) principle. EWMA is employedto improve the sensitivity of the detection method and reduce false alarms. Finally, the proposed method is applied to the actual wind turbine, which further demonstrate the effectiveness.

Original languageEnglish
StatePublished - 2017
Event1st World Congress on Condition Monitoring 2017, WCCM 2017 - London, United Kingdom
Duration: 13 Jun 201716 Jun 2017

Conference

Conference1st World Congress on Condition Monitoring 2017, WCCM 2017
Country/TerritoryUnited Kingdom
CityLondon
Period13/06/1716/06/17

Fingerprint

Dive into the research topics of 'Wind turbine gearbox fault prediction algorithm based on feature subspace'. Together they form a unique fingerprint.

Cite this