Wind turbine frequent principal fault detection based on a self-attentive LSTM encoder-decoder model

With the development of intelligent monitoring technology, internet information technology, and data storage technology, data-driven fault detection and diagnosis method in the wind turbine system has become a focus of research in recent years. In this paper, we design a data-driven architecture for...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Chinese Control Conference S. 4171 - 4176
Hauptverfasser: JIANG, Na, LI, Ning
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: Technical Committee on Control Theory, Chinese Association of Automation 01.07.2020
Schlagworte:
Data models
Fault detection
Feature extraction
Feature sequence extraction
Frequent principal fault detection
Logic gates
SALSTMED
Time series
Time series analysis
Wind power generation
Wind turbine system
Wind turbines
ISSN:1934-1768
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:With the development of intelligent monitoring technology, internet information technology, and data storage technology, data-driven fault detection and diagnosis method in the wind turbine system has become a focus of research in recent years. In this paper, we design a data-driven architecture for the wind turbine frequent principal fault detection. Considering the sequential relationship in the wind power data, we introduce the long short-term memory (LSTM) model. Additionally, to retain more necessary information hidden in the wind power time series, which is too long to make the performance of the LSTM model poor, we propose a novel self-attentive LSTM encoder-decoder(SALSTMED) model to learn the high-level feature sequence other than the feature vector. Further, the dataset collected from a real wind farm is employed to verify the performance of the proposed approach. The results indicate that the proposed approach is effective for the wind turbine frequent principal fault detection.
ISSN:1934-1768
DOI:10.23919/CCC50068.2020.9188717