A Fault Early Warning Method for Centrifugal Compressors Based on Multi-Task Gaussian Processes
Centrifugal compressors play a vital role in critical industrial sectors such as oil & gas and petrochemical factories, where they operate under challenging conditions like high-temperature, high-speed, high-pressure, and high-corrosion environments. However, these conditions also make them pron...
Gespeichert in:
Hauptverfasser: | , , |
---|---|
Format: | Buchkapitel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Centrifugal compressors play a vital role in critical industrial sectors such as oil & gas and petrochemical factories, where they operate under challenging conditions like high-temperature, high-speed, high-pressure, and high-corrosion environments. However, these conditions also make them prone to failures, leading to significant losses from unplanned downtime. Therefore, accurate monitoring of their operational status is essential to ensure safety and reliability. In response to this need, we propose a fault prediction method based on Multi-Task Gaussian Process (MTGP). This method combines vibration characteristic data and aerodynamic performance data from centrifugal compressors as inputs. Through empirical examples, we have demonstrated that the prediction performance of our MTGP model surpasses that of commonly used models such as long short-term memory neural networks and Gaussian process regression. By leveraging both vibration characteristics and aerodynamic performance data, this holistic method enables early detection of potential issues, allowing for proactive maintenance and minimizing the risk of unplanned downtime. |
---|---|
ISSN: | 2211-0984 2211-0992 |
DOI: | 10.1007/978-3-031-73407-6_24 |