Search

Load identification is the premise of wind turbine blade design and control. In wind farms, due to the lack of a thorough understanding of the blade loads, catastrophic accidents occur occasionally, and the causes of some accidents cannot be reasonably explained. To further promote the depth of blade load research and application, the development of the blade industry, load theory, load measurement, and load control of large-scale Horizontal Axis Wind Turbines (HAWTs) are comprehensively reviewed in this paper. The key issues and the corresponding progress involved in the blade research and application are discussed in detail, covering aerodynamic load calculation and analysis, load measurement in both laboratory and field environments, and load control of wind turbine blades. Additionally, the main matters needing attention in the blade research and application in the future are outlined, including more accurate and faster load calculation, strain perception and load mapping in the field, reconstruction of historical service load, and time-history optimization of the load control algorithm. Some effective countermeasures, e.g., reconstructing the historical service loads on wind turbine blades with the help of both blade load sensors and the Supervisory Control and Data Acquisition (SCADA) system, are also proposed. • A comprehensive assessment of blade load research of wind turbines is presented. • The key issues and their progress involved in blade research and application are discussed. • Future developments and challenges in blade load research are proposed. [ABSTRACT FROM AUTHOR]

Summary: Condition monitoring and fault diagnosis of main bearings of large‐scale wind turbines is critical for improving its reliability and reducing operating and maintenance costs, especially in the early stages. To achieve the goal, this paper proposes a novel deep learning approach named stacked sparse autoencoder multi‐layer perceptron (SSAE‐MLP) with a new framework by utilizing supervisory control and data acquisition (SCADA) data for wind turbine main bearing temperature prediction. After the SCADA parameter variables related to the temperature change of the main bearing are extracted, the input characteristic vector is constructed. Then, the multiple sparse autoencoders are stacked to learn the deep features inside the input data by applying the greedy layerwise unsupervised learning algorithm. Finally, a regression predictor is added to the top layer of the stacked sparse autoencoder model for supervised learning to fine‐tune the overall network. Comparative experiments show that the proposed approach has superior performance for wind turbine main bearing temperature prediction. [ABSTRACT FROM AUTHOR]

Due to the complex and variable conditions under which wind turbines operate, existing working condition classification methods are inadequate for condition assessment of the main transmission system. Because working conditions are too few after classification, it cannot effectively describe the complex and variable working conditions of wind turbine. This can lead to high false-alarm rates in the condition monitoring, which affect normal operations. This paper proposes a working condition classification method for the main transmission system of wind turbines based on supervisory control and data acquisition (SCADA) data. Firstly, correlation analysis of SCADA data acquired by wind farm is used to select the parameters relevant to the main transmission system. Secondly, according to the wind turbine control principle, the working conditions are initially divided into four phases: shutdown, start-up, maximum wind energy tracking, and constant speed. The k-means clustering algorithm is used to subdivide the maximum wind energy-tracking phase and constant speed phase, which account for a larger proportion of the working conditions, to achieve better classification. Finally, a case study is used to demonstrate the calculation of alarm thresholds and alarm rates for each working condition. The results are compared with the direct use of k-means clustering for working condition classification. It is concluded that the proposed method can significantly reduce the false-alarm rate of the vibration detection process. [ABSTRACT FROM AUTHOR]