CHEN, BINDI (2014) Automated On-line Fault Prognosis for Wind Turbine Monitoring using SCADA data. Doctoral thesis, Durham University.
|PDF (Bindi Chen PhD Thesis) - Accepted Version|
Current wind turbine (WT) studies focus on improving their reliability and reducing the cost of energy, particularly when WTs are operated offshore. A Supervisory Control and Data Acquisition (SCADA) system is a standard installation on larger WTs, monitoring all major WT sub-assemblies and providing important information. Ideally, a WT’s health condition or state of the components can be deduced through rigorous analysis of SCADA data. Several programmes have been made for that purpose; however, the resulting cost savings are limited because of the data complexity and relatively low number of failures that can be easily detected in early stages.
This thesis develops an automated on-line fault prognosis system for WT monitoring using SCADA data, concentrating particularly on WT pitch system, which is known to be fault significant. A number of preliminary activities were carried out in this research. They included building a dedicated server, developing a data visualisation tool, reviewing the existing WT monitoring techniques and investigating the possible AI techniques along with some examples detailing applications of how they can be utilised in this research.
The a-priori knowledge-based Adaptive Neuro-Fuzzy Inference System (APK-ANFIS) was selected to research in further because it has been shown to be interpretable and allows domain knowledge to be incorporated. A fault prognosis system using APK-ANFIS based on four critical WT pitch system features is proposed. The proposed approach has been applied to the pitch data of two different designs of 26 Alstom and 22 Mitsubishi WTs, with two different types of SCADA system, demonstrating the adaptability of APK-ANFIS for application to variety of technologies. After that, the Alstom results were compared to a prior general alarm approach to show the advantage of prognostic horizon. In addition, both results are evaluated using Confusion Matrix analysis and a comparison study of the two tests to draw conclusions, demonstrating that the proposed approach is effective.
|Item Type:||Thesis (Doctoral)|
|Award:||Doctor of Philosophy|
|Keywords:||Wind Power, SCADA Systems, Fault Prognosis, Artificial Intelligence, ANFIS, Pitch System|
|Faculty and Department:||Faculty of Science > Engineering and Computing Science, School of (2008-2017)|
|Copyright:||Copyright of this thesis is held by the author|
|Deposited On:||02 Oct 2014 12:29|