Two-Stage Cascaded High-Precision Early Warning of Wind Turbine Faults Based on Machine Learning and Data Graphization

被引：0

作者：

Fu, Yang ^{[1
,2
]}

Wang, Shuo ^{[2
]}

Jia, Feng ^{[1
,2
]}

Zhou, Quan ^{[3
]}

Ge, Xiaolin ^{[1
,2
]}

机构：

[1] Engineering Research Center of Offshore Wind Technology Ministry of Education (Shanghai University of Electric Power), Shanghai, China

[2] College of Electrical Engineering, Shanghai University of Electric Power, Shanghai, China

[3] State Grid Shanghai Shinan Electric Power Supply Company, Shanghai, China

来源：

Journal of Electrical Engineering and Technology | 2024年 / 19卷 / 03期

关键词：

Balancing - Convolutional neural networks - Data acquisition - Generative adversarial networks - Image enhancement;

D O I：

暂无

中图分类号：

学科分类号：

摘要：

Due to the limited accessibility of wind turbines (WTs) and the complexity of operation and maintenance (O&M), it is increasingly important to early warn the component faults of WTs, and the difficulties lie in balancing the comprehensiveness and delicacy of early warning. In this paper, a two-stage cascaded high-precision fault early warning method based on machine learning (ML) and data graphization is proposed. The first stage copes with the early warning of the main components, in which the supervisory control and data acquisition (SCADA) data are converted into Gramian Angular Field (GAF) images to establish the potential relationship of fault features at different time points, and the fault characteristics are extracted by convolutional neural network (CNN) to realize fault early warning for multiple main components simultaneously. The second stage focus on the fault subcomponents inside the main components further, in which the time generative adversarial network (TimeGAN) is adopted to enhance the fault code data samples, then the enhanced data in the form of grayscale images is input into the Vision Transformer (ViT) to train the subcomponent early warning model. The proposed method is validated with real SCADA data, the results show the effectiveness of the proposed method. © The Author(s) under exclusive licence to The Korean Institute of Electrical Engineers 2023.

引用

页码：1919 / 1931