Search

Wind turbines located in high humidity and high altitude areas are often accompanied by blade icing, which adverses the operating efficiency and even causes safety accidents. Early identification of blade icing will help improve the operating efficiency of the wind turbine. This paper proposes an icing diagnosis method for wind turbine blades based on feature optimization and the one-dimensional convolutional neural network (1D-CNN). First, feature optimization is achieved by feature selection and feature reconstruction. The XGBoost algorithm is used to calculate the importance of each feature and select the features comprehensively that reflect blade icing. Second, the important features related to blade icing are reconstructed by using the deviation principle to extract the deviation information of features accurately when blades ice. Finally, the features screened by XGBoost and the reconstructed features are combined into the final feature set as the input of the 1D-CNN, which takes the temporal and spatial characteristics of data into account, to diagnose the icing state of blades. The method is validated on the data set collected from a real wind farm. The experimental results show that the proposed icing diagnosis method for wind turbine blades is superior to the traditional deep learning methods. It is favorable to improve the efficiency of wind turbine operation and maintenance. [ABSTRACT FROM AUTHOR]

Blade icing of a wind turbine will affect the startup performance of the blades, resulting in the loss of power generation of the wind turbine, and even affect the safety of production and operation. In order to reflect the blade icing state of wind turbines as truthfully and objectively as possible, this paper proposes a wind turbine blade icing state recognition model based on the combination of vine-Copula network model and Long Short-Term Memory (LSTM)-Autoencoder algorithm. First, the vine-Copula model is used to analyze the correlation between the various parameters in supervisory control and data acquisition (SCADA) system and the blade icing state, thereby constructing a high-dimensional vine-Copula structure. Then, removing the features that are not directly related to the blade icing state, the final vine-Copula model and related features are obtained. The filtered features are input into the LSTM-Autoencoder algorithm, then the "memory" function and non-linear feature extraction capabilities of the LSTM-Autoencoder algorithm are used to obtain the evaluation results of the blade icing state of wind turbines. The experimental results show that the indicators of the wind turbine blade icing state recognition based on this method are overall better than the indicators of the Recurrent Neural Network-Autoencoder algorithm without feature reduction and the LSTM-Autoencoder algorithm without feature reduction and traditional classification algorithms. [ABSTRACT FROM AUTHOR]

An image recognition model based on a deep learning network is proposed for the automatic extraction of image features and the accurate and efficient detection of wind turbine blade damage. The Otsu threshold segmentation method is used to segment the blade image to eliminate the influence of the image background on the detection task. In order to improve the recognition performance of the proposed deep learning model, transfer learning and an ensemble learning classifier are used in a convolutional neural network model. Transfer learning is used to enhance the ability of the proposed model to extract abstract features and accelerate the convergence efficiency, whereas the random forest-based ensemble learning classifier is used to improve the accuracy of detecting the blade defects. The performance of the proposed model is verified by using unmanned aerial vehicle (UAV) images of the wind turbine blades. The proposed model provided better performance than the support vector machine (SVM) method, the basic deep learning model and the deep learning model combined with the ensemble learning approach. • A new method of blade damage detection based on deep learning is proposed. • Transfer learning and ensemble learning classifier are used in the proposed model. • The proposed model can detect damage automatically and nondestructively by images. [ABSTRACT FROM AUTHOR]

The icing problem of wind turbine blades in northern China has a serious impact on the normal and safe operation of the unit. In order to effectively predict the icing conditions of wind turbine blades, a deep fully connected neural network optimized by machine learning (ML) algorithms based on big data from the wind farm is proposed to diagnose the icing conditions of wind turbine blades. This study first uses the random forest model to reduce the features of the supervisory control and data acquisition (SCADA) data that affect blade icing, and then uses the K-nearest neighbor (KNN) algorithm to enhance the active power feature. The features after the random forest reduction and the active power mean square error (MSE) feature enhanced by the KNN algorithm are combined and used as the input of the fully connected neural network (FCNN) to perform and an empirical analysis for the diagnosis of blade icing. The simulation results show that the proposed model has better diagnostic accuracy than the ordinary back propagation (BP) neural network and other methods. [ABSTRACT FROM AUTHOR]

Damaged wind turbine (WT) blades have an imbalanced load and abnormal vibration, which affects their safe and stable operation or even results in blade rupture. To solve this problem, this study proposes a new method to detect damage in WT blades using wavelet packet energy spectrum analysis and operational modal analysis. First, a wavelet packet transform is used to analyze the tip displacement of the blades to obtain the energy spectrum. The damage is detected preliminarily based on the energy change in different frequency bands. Subsequently, an operational modal analysis method is used to obtain the modal parameters of the blade sections and the damage is located based on the modal strain energy change ratio (MSECR). Finally, the professional WT simulation software GH (Garrad Hassan) Bladed is used to simulate the blade damage and the results are verified by developing an online fault diagnosis platform integrated with MATLAB. The results show that the proposed method is able to diagnose and locate the damage accurately and provide a basis for further research of online damage diagnosis for WT blades. [ABSTRACT FROM AUTHOR]