Wind turbine blade state detection method fusing blade noise and clearance video monitoring

Abstract

The invention relates to a wind turbine blade state detection method fusing blade noise and clearance video monitoring, and the method comprises the steps: extracting blade sound features and blade dynamic features through obtaining blade sound sensing data and clearance video sensing data at the same time; the health state of the wind turbine blade structure is judged according to the blade rotating state and the image information, the implicit damage state of the blade is found from the voiceprint map of the blade noise, the actual damage condition of the blade is accurately judged through combination of the two states, and interference of other damage to judgment of the health state of the blade is avoided. And early warning of blade damage is realized.

Description

Technical Field

[0001] This application relates to the field of blade condition detection technology, and in particular to a method for detecting the condition of wind turbine blades that integrates blade noise and overhead video monitoring. Background Technology

[0002] As the core aerodynamic component of a wind power generation system, the structural integrity of wind turbine blades directly affects the safe operation and economy of the entire wind turbine unit. With the continuous increase in the capacity of individual wind turbines and the increasing complexity of the operating environment, blade structural health monitoring has become a key research direction in the field of wind power operation and maintenance.

[0003] Currently, commonly used methods for monitoring the acoustic signature of wind turbine blades typically only focus on the blade noise itself, neglecting the operational health status of the wind turbine blades and lacking monitoring of the blades' operating attitude. This can easily lead to misclassification of abnormal blade noise under poor operating attitude as blade damage, resulting in misjudgment.

[0004] To address this issue, this method incorporates monitoring of the wind turbine blade's operating attitude. False alarms are avoided by fusing blade noise with overhead video monitoring. This provides a novel solution to this problem. Summary of the Invention

[0005] The following is an overview of the subject matter described in detail herein. This overview is not intended to limit the scope of the claims.

[0006] The main objective of this invention is to propose a wind turbine blade condition detection method that integrates blade noise and airspace video monitoring, thereby improving monitoring accuracy, reducing misjudgments, and achieving accurate early warning of wind turbine blade damage.

[0007] To achieve the above objectives, embodiments of the present invention provide a method for detecting the condition of wind turbine blades by integrating blade noise and overhead video monitoring, the method comprising: 1) A multimodal data acquisition system consisting of an acoustic sensor array evenly distributed below the wind turbine tower to collect blade operating noise signals; and a high-definition camera deployed on the top of the wind turbine tower to acquire real-time video streams of the blade clearance area.

[0008] 2) Synchronous data acquisition and time alignment mechanism: Establish a precise time synchronization mechanism between the acoustic sensor array and the high-definition camera to ensure that the noise signal and video frame data are precisely aligned in the time dimension, achieving millisecond-level synchronization accuracy.

[0009] 3) Noise feature extraction and analysis module, including a noise feature extraction module, is used to extract time-domain features, frequency-domain features and wavelet features from the collected blade noise signals, construct a multi-dimensional noise feature vector, and characterize the blade operating status.

[0010] 4) Video image processing and analysis system, including a video image processing system, used to preprocess the acquired overhead video, perform background modeling, target detection and motion trajectory analysis, and identify the position, shape and motion characteristics of the blades in the video.

[0011] 5) Integrate feature extraction and feature-level fusion algorithms. Design a feature-level fusion algorithm to normalize the extracted noise feature vector and video feature vector, and construct a comprehensive feature vector through a weighted fusion strategy to achieve effective fusion of multimodal information.

[0012] 6) Deep learning state recognition model: Construct a state recognition model based on deep learning, using convolutional neural networks (CNN) to process video features, recurrent neural networks (RNN) to process temporal noise features, and achieving deep fusion of multimodal features through an attention mechanism.

[0013] 7) Anomaly detection and fault diagnosis module: This module integrates anomaly detection and fault diagnosis, establishes a baseline for normal operating conditions based on historical data, and detects abnormal patterns that deviate from the normal state through statistical analysis and machine learning algorithms, thereby identifying early faults such as blade cracks and damage.

[0014] 8) Real-time monitoring and early warning system: Establish a real-time monitoring and early warning system to evaluate the results of the fusion analysis in real time. When an abnormal state is detected, trigger a multi-level early warning mechanism, including local alarms, remote notifications, and maintenance suggestions.

[0015] 9) Adaptive learning and model update mechanism: The adaptive learning mechanism can continuously optimize and update the state recognition model based on new monitoring data, improve detection accuracy and robustness, and adapt to different environmental conditions and working states.

[0016] 10) Based on the storage and management architecture, construct a distributed data storage and management architecture to achieve efficient storage, retrieval and analysis of massive monitoring data, and support historical data backtracking, trend analysis and long-term performance evaluation.

[0017] 11) Human-computer interaction and visualization interface, which presents detection results, early warning information, historical trends and other information in intuitive chart form, and supports multi-dimensional data display and interactive analysis.

[0018] 12) The system integration and remote monitoring platform supports centralized monitoring of multiple wind turbine clusters, enabling data sharing, collaborative analysis, and remote maintenance, thereby improving the overall operation and maintenance efficiency of wind farms.

[0019] The wind turbine blade condition detection method that integrates blade noise and overhead video monitoring provided in this application has at least the following beneficial effects: By acquiring overhead video data and calculating the blade's operating attitude based on attitude image changes and the distance between the blade tip and the tower during blade movement, a preliminary assessment of the blade's structural health is made. Building upon this, the acoustic signature features and dynamic graphic features of the blade are fused to evaluate blade damage. This avoids misclassifying abnormal blade noise under unfavorable operating attitudes as blade damage, thus improving the accuracy of the assessment. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a wind turbine blade condition detection method that integrates blade noise and overhead video monitoring, provided in one embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0023] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0025] like Figure 1 One embodiment of this application provides a method for detecting the condition of wind turbine blades by integrating blade noise and overhead video monitoring. The method includes: Step A11: Acquire the sound signal of the blades using an acoustic sensor array; Step A12: Convert the acoustic sensor array data into temporally continuous sound data, and convert it into acoustic print image morphological features through time-frequency analysis; Step A13: Extract and analyze noise features based on the morphological characteristics of the leaf acoustic print image; Step A21: Acquire blade running posture image signals using the clearance video sensor; Step A22: Extract the time-continuous image data from the blade clearance video sensor data, and extract the distance from the blade tip to the tower. Step A23: Analyze the leaf morphology and motion characteristics based on the dynamic graphic features of the leaf. Step A31: Establish a precise time synchronization mechanism between the acoustic sensor array and the high-definition camera to achieve millisecond-level synchronization accuracy.

[0026] Step A32 involves normalizing the extracted noise feature vector and the video feature vector, and constructing a comprehensive feature vector through a weighted fusion strategy to achieve effective fusion of multimodal information.

[0027] Step A33: A convolutional neural network (CNN) is used to process video features, a recurrent neural network (RNN) is used to process temporal noise features, and an attention mechanism is used to achieve deep fusion of multimodal features.

[0028] Step A34: Establish a baseline for normal operating conditions based on historical data, and detect abnormal patterns that deviate from the normal state through statistical analysis and machine learning algorithms to identify early faults such as blade cracks and damage.

[0029] Compared with existing technologies, the wind turbine blade status detection method of this embodiment, which integrates blade noise and airspace video monitoring, monitors airspace video while collecting blade noise data and performs fusion analysis on the two signals, can effectively determine the health status of the wind turbine blade attitude. This avoids the misjudgment problem caused by noise changes resulting from changes in wind turbine blade attitude.

[0030] The above is a detailed description of the preferred embodiments of this application. However, the embodiments of this application are not limited to the above-described implementation methods. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the embodiments of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of the embodiments of this application.

Claims

1. A method for detecting the condition of wind turbine blades by integrating blade noise and overhead video monitoring, characterized in that, The method includes: a multimodal data acquisition system; an acoustic sensor array uniformly arranged below the wind turbine tower for collecting blade operating noise signals; and a high-definition camera deployed on the top of the wind turbine tower for acquiring real-time video streams of the blade clearance area.

2. The method according to claim 1, characterized in that: It has a synchronous data acquisition and time alignment mechanism, and establishes a precise time synchronization mechanism between the acoustic sensor array and the high-definition camera to ensure that the noise signal and video frame data are precisely aligned in the time dimension, achieving millisecond-level synchronization accuracy.

3. The method according to claim 1, characterized in that: It has a noise feature extraction and analysis module, including a noise feature extraction module, which is used to extract time-domain features, frequency-domain features and wavelet features from the collected blade noise signals, construct a multi-dimensional noise feature vector, and characterize the blade operating status.

4. The method according to claim 1, characterized in that: It is equipped with a video image processing and analysis system, including a video image processing system, which is used to preprocess the acquired overhead video, perform background modeling, target detection and motion trajectory analysis, and identify the position, shape and motion characteristics of the blades in the video.

5. The method according to claim 1, characterized in that: It possesses fusion feature extraction and feature-level fusion algorithms. The feature-level fusion algorithm is designed to normalize the extracted noise feature vector and video feature vector, and construct a comprehensive feature vector through a weighted fusion strategy to achieve effective fusion of multimodal information.

6. The method according to claim 5, characterized in that: It possesses a deep learning state recognition model, constructs a state recognition model based on deep learning, uses convolutional neural networks (CNN) to process video features, recurrent neural networks (RNN) to process temporal noise features, and achieves deep fusion of multimodal features through an attention mechanism.

7. The method according to claim 6, characterized in that: It has an anomaly detection and fault diagnosis module, integrates anomaly detection and fault diagnosis module, establishes a normal operating status baseline based on historical data, and detects abnormal patterns that deviate from the normal state through statistical analysis and machine learning algorithms, so as to identify early faults such as blade cracks and damage.

8. The method according to claim 1, characterized in that: It has a real-time monitoring and early warning system, which evaluates the results of the fusion analysis in real time. When an abnormal state is detected, a multi-level early warning mechanism is triggered, including local alarms, remote notifications, and maintenance suggestions.

9. The method according to claim 6, characterized in that: It has an adaptive learning and model update mechanism. The adaptive learning mechanism can continuously optimize and update the status recognition model based on new monitoring data, improve detection accuracy and robustness, and adapt to different environmental conditions and working states.

10. The method according to claim 1, characterized in that: It features a data storage and management architecture, constructing a distributed data storage and management architecture to achieve efficient storage, retrieval, and analysis of massive monitoring data, supporting historical data backtracking, trend analysis, and long-term performance evaluation.

11. The method according to claim 8, characterized in that: It features a human-computer interaction and visualization interface, presenting detection results, early warning information, historical trends, etc. in intuitive charts and graphs, and supports multi-dimensional data display and interactive analysis.

12. The method according to claim 8, characterized in that: It has a system integration and remote monitoring platform that supports centralized monitoring of multiple wind turbine clusters, enabling data sharing, collaborative analysis and remote maintenance, and improving the overall operation and maintenance efficiency of wind farms.

Images