Monitoring method based on blade rotation sound and equipment for auxiliary monitoring

Abstract

The invention provides a monitoring method based on blade rotation sound and equipment for auxiliary monitoring, and the method comprises the steps: capturing weak airflow disturbance sound wave signals generated during the rotation of blades of a wind driven generator through a non-contact acoustic sensor array, and carrying out the standardization preprocessing of the captured sound wave signals; the method comprises the following steps: performing dynamic filtering processing on environmental noise of sound wave signals, integrating acoustic sensor data from different positions to form a comprehensive sound image, applying deep learning to model and predict the health condition of a blade, and integrating edge calculation, so that a part of data processing tasks are completed in a place close to an acoustic sensor; the Internet of Things platform is adopted to realize interconnection and intercommunication between equipment, implement a cloud edge collaborative optimization mode and automatically allocate tasks to the most appropriate processing node, sensors do not need to be installed on the blades, physical influences on the equipment are avoided, environmental noise is effectively filtered out, the signal-to-noise ratio of target sound wave signals is increased, and therefore the monitoring precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine blade detection, and in particular to a monitoring method based on blade rotation sound and equipment for auxiliary monitoring. Background Art

[0002] As an important part of renewable energy, the operating efficiency and safety of wind turbines are directly related to the stability and economy of energy production. Wind turbine blades are one of the key components. They are exposed to harsh environments for a long time and are easily damaged by wear, cracks, corrosion, etc., resulting in performance degradation or even failure. Traditional blade monitoring methods mainly rely on regular manual inspection and maintenance, which is not only time-consuming and labor-intensive, but also difficult to detect potential problems in a timely manner, which may lead to sudden failures, increase maintenance costs and downtime.

[0003] At present, commonly used blade monitoring technologies include vibration analysis, infrared thermal imaging, ultrasonic testing, etc. Although these methods can reflect the status of the blades to a certain extent, they have the following limitations: 1. It is necessary to install sensors on the blades or perform close-range detection, which is complicated to operate and has an impact on the equipment.

[0004] 2. Data processing delay: A large amount of data needs to be transmitted to the cloud for processing, resulting in a long response time and inability to achieve real-time monitoring.

[0005] 3. Environmental interference: External environmental noise (such as wind noise, mechanical noise) will seriously interfere with the monitoring signal and reduce the monitoring accuracy.

[0006] 4. Limited coverage: A single point or a small number of sensors cannot fully cover the entire blade surface, making it difficult to obtain global information. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides a monitoring method based on blade rotation sound. In order to achieve the above purpose, the present invention adopts the following technical solutions: A monitoring method based on blade rotation sound comprises the following steps: The weak airflow disturbance sound wave signals generated by the rotation of wind turbine blades are captured by a non-contact acoustic sensor array; Use machine learning algorithms to standardize and preprocess the captured sound wave signals; The minimum mean square error algorithm or Kalman filter is used to dynamically filter the ambient noise of the sound wave signal. The formula for updating the weight of the minimum mean square error algorithm is: ; in, It is The weight vector for the iteration, is the step size parameter, is the error signal, is the input signal; Use multi-channel signal fusion technology to integrate acoustic sensor data from different locations to form a comprehensive sound image; Apply deep learning to model and predict blade health; Integrate edge computing so that some data processing tasks can be completed close to the acoustic sensor; Use the Internet of Things platform to achieve interconnection between devices, as well as data exchange and service management with the cloud; Implement cloud-edge collaborative optimization to automatically assign tasks to the most suitable processing nodes based on network conditions and computing resources.

[0008] Furthermore, the method of the machine learning algorithm comprises the following steps: The captured sound wave signal is processed by removing the DC component and normalizing the amplitude; By using short-time Fourier transform, wavelet transform or Mel frequency cepstrum coefficient technology, the time-frequency characteristics that can reflect the state of the blade are extracted from the preprocessed sound wave signal. The formula of short-time Fourier transform is: ; in, is a time domain signal, is the window function, is the time offset, is the angular frequency; Use supervised learning to train one or more deep learning models using labeled historical datasets, regularization techniques, and early stopping. The use of adaptive mechanisms enables the model to automatically adjust internal parameters according to changes in the external environment such as wind speed and temperature; After the model training is completed, it will be deployed in the actual monitoring system to analyze the real-time collected sound wave signals to determine whether there are potential problems with the blades and predict possible failures in the future. If an abnormality is detected, the system will issue an alarm to notify maintenance personnel.

[0009] Furthermore, the method of standardization preprocessing comprises the following steps: Decompose the large-scale acoustic wave data analysis task into several subtasks, each of which corresponds to the data of a specific time period or a specific wind turbine generator set; Establish a distributed computing cluster, including multiple computing nodes, and use the load balancing algorithm to reasonably allocate each subtask to different computing nodes for execution; Use gRPC or ZeroMQ data transmission protocols to quickly and accurately transfer the data required by the subtask to the corresponding computing nodes; When all subtasks have completed the calculation, the master node is responsible for collecting the results returned by each computing node and merging them; Prepare backup nodes for key subtasks during the task allocation phase. Once a problem occurs with the original node, the backup node immediately takes over the unfinished task. By monitoring the working status of each computing node, the task allocation strategy is dynamically adjusted, and future computing needs are predicted in advance based on historical operation data.

[0010] Furthermore, a quantum random number generator is used to initialize the weight values ​​in the machine learning algorithm, and the high entropy random numbers provided by the quantum random number generator are used to improve the selection of initial weights.

[0011] Furthermore, it also includes equipping one or more drones, which are programmed to fly along a predetermined path to carry acoustic sensors close to wind turbines for close-range inspection, while transmitting the collected data back to a ground station in real time for analysis.

[0012] Furthermore, the method of using a drone to inspect blades includes the following steps: Use genetic algorithms or ant colony algorithms to design optimal flight paths for drones based on flight distance, energy consumption, safety, and coverage factors; A wireless communication link is used between the drone and the ground station to support real-time transmission of high-definition audio and video streams; Integrated GPS and inertial measurement unit to enable drones to fly autonomously in complex environments without human intervention; The data collected by the drone is encrypted and transmitted to the ground station, which is responsible for data storage, processing and analysis. All data is time-stamped for subsequent comparison and tracking.

[0013] Furthermore, the drone is equipped with a visual recognition module that can identify specific marks or defects on the blades and combine them with the acoustic monitoring results to provide a blade health diagnosis report, specifically: The drone is equipped with a high-definition camera to capture images of the blade surface from multiple angles; Use computer vision technology to pre-process the collected images, including denoising, contrast enhancement and edge detection, and apply feature extraction algorithms to identify key points and texture information in the images; Combined with pre-trained deep learning models, it can classify and quantify blade cracks, corrosion or other damage in images; The results of visual recognition are correlated with the acoustic monitoring data to comprehensively evaluate the overall health of the blades.

[0014] The present invention also provides a device for auxiliary monitoring, comprising: Microphone array: An integrated high-sensitivity microphone array is used to capture the weak airflow disturbance sound wave signals generated by the rotation of wind turbine blades; Embedded processor: used to perform preliminary feature extraction and anomaly detection data processing tasks locally; Data transmission module: equipped with wireless communication module, supports sending complete data to cloud server for in-depth analysis; Power management system: adopts solar charging system and intelligent power management system; Data compression module: For data that needs to be uploaded to the cloud, it is compressed first; Local storage: built-in large-capacity Flash memory to temporarily store large amounts of monitoring data; User interface: Provides a user interface that allows maintenance personnel to view real-time monitoring results, set parameters, and receive alarm notifications.

[0015] Furthermore, the embedded processor has a built-in artificial intelligence chip to accelerate the execution of machine learning algorithms.

[0016] Furthermore, it also includes an AR glasses system. By wearing AR glasses, maintenance personnel can directly see the historical monitoring data of the blades, current status indications, and virtual information of possible problem areas superimposed on the real world when checking the wind turbines on site, thereby guiding maintenance work.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. By using multi-channel signal fusion technology and adaptive noise elimination technology, the present invention can effectively filter out environmental noise and enhance the signal-to-noise ratio of the target sound wave signal, thereby improving monitoring accuracy, especially for small cracks or early damage, it can be discovered and warned in time, reducing the occurrence of sudden failures; 2. Integrated edge computing capabilities enable some data processing tasks to be completed close to the acoustic sensor, reducing data transmission delays and achieving true real-time monitoring. Maintenance personnel can receive abnormal alarms as soon as possible and take timely measures to avoid the expansion of faults. 3. Through the application of deep learning models, the present invention can model and predict the health status of blades, identify potential problems in advance, and optimize maintenance plans, which not only reduces unnecessary regular inspections, but also extends the service life of blades and reduces maintenance costs; 4. Distributed computing resources and fault-tolerant mechanisms are used to ensure that the system can still operate normally even when some nodes fail. At the same time, the cloud-edge collaborative optimization strategy automatically allocates tasks according to network conditions and computing resources, ensuring high efficiency and low power consumption of the system; 5. Combined with the UAV visual recognition module and acoustic monitoring results, the present invention can provide a more accurate blade health diagnosis report. Maintenance personnel can use AR glasses to view virtual information superimposed on the real world on site, and intuitively understand the blade's historical monitoring data, current status indications, and possible problem areas, thereby guiding maintenance work. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is further described below in conjunction with embodiments.

[0019] Embodiment 1 The present invention provides a monitoring method based on blade rotation sound, comprising the following steps: Perform standardized preprocessing on the captured sound wave signal, including removing the DC component and normalizing the amplitude; The minimum mean square error (LMS) algorithm or Kalman filter is used to dynamically filter the ambient noise of the sound wave signal. The formula for updating the weight of the minimum mean square error algorithm is: ; in, It is The weight vector for the iteration, is the step size parameter, is the error signal, is the input signal; Use multi-channel signal fusion technology to integrate acoustic sensor data from different locations to form a comprehensive sound image; Use deep learning to model and predict the health status of leaves. Specifically, convolutional neural networks (CNN) or recurrent neural networks (RNN), especially long short-term memory networks (LSTM), can be used. The state update formula of LSTM units is: ; ; ; ; ; ; in, , , are the activation values ​​of the forget gate, input gate, and output gate, respectively. is a candidate cell state, is the current cell state, is the current hidden state; Integrate edge computing capabilities so that some data processing tasks can be completed close to the acoustic sensor, reducing latency and alleviating the burden on cloud servers; Use the Internet of Things (IoT) platform to achieve interconnection between devices, as well as data exchange and service management with the cloud; Implement cloud-edge collaborative optimization to automatically assign tasks to the most suitable processing nodes based on network conditions and computing resources.

[0020] In this embodiment, a quantum random number generator is used to initialize weight values ​​in a machine learning algorithm, and the high entropy random numbers provided by the quantum random number generator are used to improve the selection of initial weights.

[0021] Embodiment 2 In the first embodiment, the specific steps of the machine learning algorithm are: The captured sound wave signal is processed by removing the DC component and normalizing the amplitude; Using short-time Fourier transform (STFT), wavelet transform (WT) or Mel-frequency cepstral coefficient (MFCC) technology, the time-frequency characteristics that can reflect the state of the blade are extracted from the preprocessed sound wave signal. The formula of short-time Fourier transform is: ; in, is a time domain signal, is the window function, is the time offset, is the angular frequency; Use supervised learning to train one or more deep learning models using labeled historical datasets, regularization techniques, and early stopping. The use of adaptive mechanisms enables the model to automatically adjust internal parameters according to changes in wind speed and temperature external environment. For example, through online learning, the model weights are continuously updated to make it more suitable for actual working conditions; After the model training is completed, it is deployed in the actual monitoring system to analyze the real-time collected sound wave signals to determine whether there are potential problems with the blades and predict possible failures in the future. If an abnormality is detected, the system will issue an alarm to notify maintenance personnel.

[0022] Embodiment 3 In the first embodiment, the method for using distributed computing resources is as follows: Decompose the large-scale acoustic wave data analysis task into several subtasks, each of which corresponds to the data of a specific time period or a specific wind turbine generator set; Establish a distributed computing cluster, including multiple computing nodes, and use the load balancing algorithm to reasonably allocate each subtask to different computing nodes for execution. The load balancing algorithm can use polling method or minimum connection number method; Use gRPC or ZeroMQ data transmission protocols to quickly and accurately transfer the data required by the subtask to the corresponding computing nodes; When all subtasks have completed the calculation, the master node is responsible for collecting the results returned by each computing node and merging them. The merging process can be achieved through the weighted average method; Prepare backup nodes for key subtasks during the task allocation phase. Once a problem occurs with the original node, the backup node can immediately take over the unfinished task. By monitoring the working status of each computing node, the task allocation strategy is dynamically adjusted, and future computing needs are predicted in advance based on historical operation data.

[0023] Embodiment 4 Based on the first embodiment, drones are used to assist monitoring: Use genetic algorithms or ant colony algorithms to design optimal flight paths for drones based on flight distance, energy consumption, safety, and coverage factors; A wireless communication link is used between the drone and the ground station to support real-time transmission of high-definition audio and video streams; Integrated global positioning system (GPS) and inertial measurement unit (IMU) to enable drones to fly autonomously in complex environments without human intervention; The data collected by the drone is encrypted and transmitted to the ground station, which is responsible for data storage, processing and analysis. All data is time-stamped for subsequent comparison and tracking.

[0024] Embodiment 5 Based on the fourth embodiment, the drone is equipped with a visual recognition module, which can identify specific marks or defects on the blades and provide a blade health diagnosis report in combination with the acoustic monitoring results, specifically: The drone is equipped with a high-definition camera to capture images of the blade surface from multiple angles; Use computer vision technology to pre-process the collected images, including denoising, contrast enhancement and edge detection, and apply feature extraction algorithms to identify key points and texture information in the images; Combined with pre-trained deep learning models, it can classify and quantify blade cracks, corrosion or other damage in images; The results of visual recognition are correlated with the acoustic monitoring data to comprehensively evaluate the overall health of the blades.

[0025] Embodiment 6 The present invention also provides a device for auxiliary monitoring, comprising: Microphone array: An integrated high-sensitivity microphone array is used to capture the weak airflow disturbance sound wave signals generated by the rotation of wind turbine blades; Embedded processor: Built-in high-performance embedded processor that can perform preliminary data processing tasks locally, such as feature extraction and anomaly detection; Data transmission module: equipped with wireless communication modules such as Wi-Fi, Bluetooth or LoRa, supporting sending complete data to cloud servers for in-depth analysis; Power management system: The solar charging system and intelligent power management system ensure that the device can work stably for a long time. At the same time, it has a low power consumption mode and automatically enters a sleep state when no activity is detected, further extending the battery life. Data compression module: For data that needs to be uploaded to the cloud, it is first compressed using lossless or lossy compression algorithms (such as FLAC or MP3) to reduce data volume and save transmission time and bandwidth costs; Local storage: Built-in large-capacity Flash memory to temporarily store large amounts of monitoring data to cope with network instability or interruption; User interface: Provides a user interface that allows maintenance personnel to view real-time monitoring results, set parameters, and receive alarm notifications.

[0026] In this embodiment, in order to achieve accelerated reasoning, the embedded processor has a built-in artificial intelligence chip to accelerate the execution of the machine learning algorithm.

[0027] In this embodiment, maintenance personnel wearing AR glasses can directly see the historical monitoring data of the blades, current status indications, and virtual information of possible problem areas superimposed on the real world when checking the wind turbines on site, thereby guiding maintenance work.

[0028] The present invention can realize real-time, accurate and comprehensive monitoring of wind turbine blades, and improve the reliability and maintenance efficiency of the system.

[0029] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A monitoring method based on blade rotation sound, characterized in that: The following steps are involved: The weak airflow disturbance sound wave signals generated by the rotation of wind turbine blades are captured by a non-contact acoustic sensor array; Use machine learning algorithms to standardize and preprocess the captured sound wave signals; The minimum mean square error algorithm or Kalman filter is used to dynamically filter the ambient noise of the sound wave signal. The formula for updating the weight of the minimum mean square error algorithm is: ； in, It is The weight vector for the iteration, is the step size parameter, is the error signal, is the input signal; Use multi-channel signal fusion technology to integrate acoustic sensor data from different locations to form a comprehensive sound image; Apply deep learning to model and predict blade health; Integrate edge computing so that some data processing tasks can be completed close to the acoustic sensor; Use the Internet of Things platform to achieve interconnection between devices, as well as data exchange and service management with the cloud; Implement cloud-edge collaborative optimization to automatically assign tasks to the most suitable processing nodes based on network conditions and computing resources.

2. The monitoring method based on blade rotation sound according to claim 1, characterized in that: The method of the machine learning algorithm comprises the following steps: The captured sound wave signal is processed by removing the DC component and normalizing the amplitude; By using short-time Fourier transform, wavelet transform or Mel frequency cepstrum coefficient technology, the time-frequency characteristics that can reflect the state of the blade are extracted from the preprocessed sound wave signal. The formula of short-time Fourier transform is: ； in, is a time domain signal, is the window function, is the time offset, is the angular frequency; Use supervised learning to train one or more deep learning models using labeled historical datasets, regularization techniques, and early stopping. The use of adaptive mechanisms enables the model to automatically adjust internal parameters according to changes in the external environment such as wind speed and temperature; After the model training is completed, it will be deployed in the actual monitoring system to analyze the real-time collected sound wave signals to determine whether there are potential problems with the blades and predict possible failures in the future. If an abnormal situation is detected, the system will issue an alarm to notify maintenance personnel.

3. The monitoring method based on blade rotation sound according to claim 1, characterized in that: The method of the standardized pretreatment comprises the following steps: Decompose the large-scale acoustic wave data analysis task into several subtasks, each of which corresponds to the data of a specific time period or a specific wind turbine generator set; Establish a distributed computing cluster, including multiple computing nodes, and use the load balancing algorithm to reasonably allocate each subtask to different computing nodes for execution; Use gRPC or ZeroMQ data transmission protocols to quickly and accurately transfer the data required by the subtask to the corresponding computing nodes; When all subtasks have completed the calculation, the master node is responsible for collecting the results returned by each computing node and merging them; Prepare backup nodes for key subtasks during the task allocation phase. Once a problem occurs with the original node, the backup node immediately takes over the unfinished task. By monitoring the working status of each computing node, the task allocation strategy is dynamically adjusted, and future computing needs are predicted in advance based on historical operation data.

4. The monitoring method based on blade rotation sound according to claim 1, characterized in that: A quantum random number generator is used to initialize the weight values ​​in the machine learning algorithm, and the high entropy random numbers provided by the quantum random number generator are used to improve the selection of initial weights.

5. The monitoring method based on blade rotation sound according to claim 1, characterized in that: It also includes equipping one or more drones that are programmed to fly along a predetermined path to carry acoustic sensors close to wind turbines for close-range inspections, while transmitting the collected data back to a ground station in real time for analysis.

6. The monitoring method based on blade rotation sound according to claim 5, characterized in that: The method of using drones to inspect blades includes the following steps: Use genetic algorithms or ant colony algorithms to design optimal flight paths for drones based on flight distance, energy consumption, safety, and coverage factors; A wireless communication link is used between the drone and the ground station to support real-time transmission of high-definition audio and video streams; Integrated GPS and inertial measurement unit to enable drones to fly autonomously in complex environments without human intervention; The data collected by the drone is encrypted and transmitted to the ground station, which is responsible for data storage, processing and analysis. All data is time-stamped for subsequent comparison and tracking.

7. The monitoring method based on blade rotation sound according to claim 5, characterized in that: The drone is equipped with a visual recognition module that can identify specific marks or defects on the blades and combine it with the acoustic monitoring results to provide a blade health diagnosis report, specifically: The drone is equipped with a high-definition camera to capture images of the blade surface from multiple angles; Use computer vision technology to pre-process the collected images, including denoising, contrast enhancement and edge detection, and apply feature extraction algorithms to identify key points and texture information in the images; Combined with pre-trained deep learning models, it can classify and quantify blade cracks, corrosion or other damage in images; The results of visual recognition are correlated with the acoustic monitoring data to comprehensively evaluate the overall health of the blades.

8. A device for auxiliary monitoring, characterized in that: Used to provide hardware support for the monitoring method of any one of claims 1 to 7, comprising: Microphone array: An integrated high-sensitivity microphone array is used to capture the weak airflow disturbance sound wave signals generated by the rotation of wind turbine blades; Embedded processor: used to perform preliminary feature extraction and anomaly detection data processing tasks locally; Data transmission module: equipped with wireless communication module, supports sending complete data to cloud server for in-depth analysis; Power management system: adopts solar charging system and intelligent power management system; Data compression module: For data that needs to be uploaded to the cloud, it is compressed first; Local storage: built-in large-capacity Flash memory to temporarily store large amounts of monitoring data; User interface: Provides a user interface that allows maintenance personnel to view real-time monitoring results, set parameters, and receive alarm notifications.

9. The device for auxiliary monitoring according to claim 8, characterized in that: The embedded processor has a built-in artificial intelligence chip to accelerate the execution of machine learning algorithms.

10. The device for auxiliary monitoring according to claim 8, characterized in that: It also includes an AR glasses system. Maintenance personnel can wear AR glasses and directly see the historical monitoring data of the blades, current status indications, and virtual information of possible problem areas superimposed on the real world when viewing wind turbines on site, thereby guiding maintenance work.

Images