Fan blade diagnosis method and device, electronic equipment and storage medium

Abstract

The invention discloses a fan blade diagnosis method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a voiceprint signal generated in the operation process of a fan blade; filtering the voiceprint signal to obtain a target voiceprint signal; decomposing the target voiceprint signal to obtain a decomposed voiceprint signal; the decomposed voiceprint signal is reconstructed, and a reconstructed voiceprint signal is obtained; calculating each sub-band energy feature of the reconstructed voiceprint signal to obtain a target energy feature; inputting the target energy characteristics into a fan blade diagnosis model for prediction to obtain a diagnosis result; wherein the fan blade diagnosis model is determined by training historical voiceprint signals. According to the technical scheme, potential faults of the fan blade can be effectively identified, early warning is performed in advance, the maintenance cost is reduced, and the operation efficiency of a wind power plant is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power equipment monitoring, and in particular to a wind turbine blade diagnosis method, device, electronic equipment and storage medium. Background Art

[0002] With the rapid development of the new energy industry, wind power installed capacity has hit new highs. However, wind turbines generally have problems such as high failure rate, high cost of repairing large components and long downtime, followed by very high operation and maintenance costs. As one of the most important components of wind turbines, wind turbine blades are mostly made of fiberglass composite materials. Due to the particularity of the manufacturing process, internal defects such as delamination, wrinkles, bubbles, etc. may occur during the manufacturing process. Wind turbine blades are exposed to the external environment for a long time, and almost all external stimuli act on the wind turbine blades. These diffusely distributed defects gradually expand and penetrate into macro cracks under the action of random alternating stress and then form serious damage. If they are not discovered and handled in time, they will cause blade breakage, seriously affecting the safe operation of the wind turbine unit and causing great energy and economic losses to the entire wind farm.

[0003] At present, the judgment of internal defects of blades is still mainly based on the appearance inspection or knocking of inspectors. Not only does it rely on subjective experience and technical level, but it is also difficult to identify internal defects and initial damage of blades. Once damage is found on the surface of the blade, it often means that the damage is already very serious. At this time, the blade not only has high repair costs and long downtime, but also has the risk of serious accidents. Summary of the invention

[0004] The present invention provides a fan blade diagnosis method, device, electronic equipment and storage medium, which can effectively identify potential faults of the fan blade.

[0005] According to one aspect of the present invention, a method for diagnosing a fan blade is provided, the method comprising:

[0006] Acquire the soundprint signal generated during the operation of the fan blades;

[0007] Filtering the voiceprint signal to obtain a target voiceprint signal;

[0008] Decomposing the target voiceprint signal to obtain a decomposed voiceprint signal; reconstructing the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculating the energy characteristics of each sub-band of the reconstructed voiceprint signal to obtain a target energy characteristic;

[0009] The target energy feature is input into a fan blade diagnostic model for prediction to obtain a diagnostic result; wherein the fan blade diagnostic model is determined by training historical voiceprint signals.

[0010] According to another aspect of the present invention, a fan blade diagnostic device is provided, the device comprising:

[0011] A voiceprint signal acquisition module is used to acquire the voiceprint signal generated during the operation of the fan blades;

[0012] A target voiceprint signal obtaining module is used to filter the voiceprint signal to obtain a target voiceprint signal;

[0013] The target energy feature acquisition module is used to decompose the target voiceprint signal to obtain a decomposed voiceprint signal; reconstruct the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculate the energy features of each sub-band of the reconstructed voiceprint signal to obtain a target energy feature;

[0014] The diagnosis result obtaining module is used to input the target energy characteristics into the fan blade diagnosis model for prediction to obtain the diagnosis result; wherein the fan blade diagnosis model is determined by training the historical voiceprint signal.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a wind blade diagnostic method described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a wind turbine blade diagnosis method described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention obtains the soundprint signal generated during the operation of the wind turbine blade, filters the soundprint signal to obtain the target soundprint signal, then decomposes the target soundprint signal to obtain the decomposed soundprint signal, reconstructs the decomposed soundprint signal to obtain the reconstructed soundprint signal, calculates the energy characteristics of each sub-band of the reconstructed soundprint signal to obtain the target energy characteristics, and then inputs the target energy characteristics into the wind turbine blade diagnosis model for prediction to obtain the diagnosis result. This technical solution can effectively identify the potential faults of the wind turbine blade, give early warning, reduce maintenance costs, and improve the operation efficiency of the wind farm.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a fan blade diagnosis method provided according to Embodiment 1 of the present invention;

[0022] Figure 2 A schematic diagram of a fan blade diagnosis process provided by Embodiment 2 of the present invention;

[0023] Figure 3 This is a flow chart of a fan blade detection system based on voiceprint signals provided in the second embodiment of the present application;

[0024] Figure 4 A specific location diagram of the fan sensor arrangement provided in the embodiment of the present application;

[0025] Figure 5 A schematic diagram of the structure of a fan blade diagnostic device provided in Embodiment 3 of the present invention;

[0026] Figure 6 It is a structural schematic diagram of an electronic device for implementing a wind turbine blade diagnosis method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 This is a flow chart of a fan blade diagnosis method provided according to the first embodiment of the present invention. This embodiment is applicable to the case of diagnosing a fan blade fault. The method can be executed by a fan blade diagnosis device. The fan blade diagnosis device can be implemented in the form of hardware and / or software. The fan blade diagnosis device can be configured in a device. For example, the device can be a background server or other device with communication and computing capabilities. Figure 1 As shown, the method includes:

[0031] S110. Acquire a soundprint signal generated during the operation of the fan blades.

[0032] In this solution, the voiceprint is a sound wave spectrum that carries speech information and is displayed by an electroacoustic instrument. The voiceprint signal is a direct reflection of the status of the fan blades and contains a lot of information about the health status of the fan blades.

[0033] In this embodiment, the soundprint signals generated by the fan blades during operation under different working conditions can be collected in real time based on sensors. The sensors may include soundprint sensors, audio sensors, etc.; different working conditions may include different weather, different wind speeds, etc., that is, the soundprint signals generated by the fan blades during operation under different weather and different wind speeds can be collected based on sensors.

[0034] Optionally, obtaining a soundprint signal generated during the operation of the fan blades includes:

[0035] Acquire the sound print signal generated during the operation of the wind turbine blade based on the pre-installed audio sensor; wherein, the audio sensor comprises a first audio sensor and a second audio sensor; the first audio sensor is arranged at the hub of the wind turbine blade root; the second audio sensor is arranged at a preset height of the tower outer wall, the height of the blade wind sweeping lower edge, and the outer wall of the hub.

[0036] Among them, audio sensors are devices that can sense the sound in the surrounding environment, usually composed of one or more microphones. They can convert sound signals into electrical signals for processing by computers, embedded systems and other devices.

[0037] Specifically, the audio sensors are arranged inside and outside the wind turbine respectively. The internal audio sensor, i.e., the first audio sensor, is arranged at the blade root hub by a fixing fixture to ensure stability and accuracy. The external audio sensor, i.e., the second audio sensor, is arranged at a preset height on the outer wall of the tower, at the height of the lower edge of the blade sweeping the wind, and at the outer wall of the hub, which can effectively cover various key parts of the wind turbine and fully capture the sound print signals generated during operation. Among them, the preset height can be set according to the installation situation of the wind turbine. For example, the preset height can be 2 meters, that is, the second audio sensor is arranged at a height of 2 meters on the outer wall of the tower, at the height of the lower edge of the blade sweeping the wind, and at the outer wall of the hub.

[0038] By arranging audio sensors, the sound print signals generated during the operation of the fan blades can be obtained in real time, and the fault diagnosis of the fan blades can be realized, providing the key theoretical basis and key technical support for the healthy operation of the fan equipment.

[0039] S120: Filter the voiceprint signal to obtain a target voiceprint signal.

[0040] In this solution, various filters can be used to filter the voiceprint signal to obtain the target voiceprint signal.

[0041] S130, decomposing the target voiceprint signal to obtain a decomposed voiceprint signal; reconstructing the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculating the energy characteristics of each sub-band of the reconstructed voiceprint signal to obtain a target energy characteristic.

[0042] In this scheme, the energy characteristics of the target voiceprint signal can be extracted based on wavelet packet analysis. Specifically, it includes: 1. Wavelet packet decomposition: the target voiceprint signal is decomposed into multiple layers through a series of filter groups, and each layer of decomposition includes high-frequency and low-frequency components; 2. Wavelet packet reconstruction: the decomposed voiceprint signal is reconstructed to combine the signals of each sub-band, restore the approximate original signal, and obtain the reconstructed voiceprint signal; 3. Energy feature extraction: calculate the energy characteristics of each sub-band to obtain the target energy characteristics. The specific formulas include:

[0043] Initial conditions:

[0044] W 0,0 (t) = x(t);

[0045] Among them, W 0,0 , x(t) is the target voiceprint signal.

[0046] Breakdown steps:

[0047] W j+1,2k (t)=∑ n h(n)W j,k (t-2 j n);

[0048] W j+1,2k+1 (t)=∑ n g(n)W j,k (t-2 j n);

[0049] Among them, W j,k is the signal of the kth node in the jth layer; h(n) is the low-pass filter coefficient; g(n) is the high-pass filter coefficient.

[0050] Refactoring steps:

[0051] W j,k (t)=∑ n [h(n)W j+1,2k (t+2 j n)+g(n)W j+1,2k+1 (t+2 j n)];

[0052] Energy characteristics:

[0053]

[0054] Among them, E j,k (n) is the energy of the signal of the kth node in the jth layer.

[0055] S140, inputting the target energy feature into a fan blade diagnosis model for prediction to obtain a diagnosis result; wherein the fan blade diagnosis model is determined by training historical voiceprint signals.

[0056] Among them, the fan blade diagnostic model is used to detect the fault conditions of the fan blades in the target energy characteristics, which can accurately obtain the operating status of the fan blades and provide key theoretical basis and key technical support for the healthy operation of the fan equipment.

[0057] In this solution, the wind turbine blade diagnosis model includes the ECAPA-TDNN neural network model. Among them, ECAPA-TDNN is an advanced neural network structure in the field of voiceprint recognition, and is still recognized as one of the most effective convolutional neural network (CNN) architectures. The architecture of the ECAPA-TDNN neural network model includes input feature processing: preprocessing the input voiceprint features, including standardization and feature selection. Short-term semantic feature extraction: extracting the semantic features of the signal in a short period of time. Long-term information fusion: integrating the long-term feature information of the signal to improve the model's recognition ability for long-term dependencies. Fully connected layer and classification: using the fully connected layer to classify the extracted features. Loss function and optimization: using the cross entropy loss function and the Adam optimization algorithm for training. Model training: training the model through training data, optimizing the network weights, and obtaining a fault detection model.

[0058] In this embodiment, the diagnosis results include normal, slight damage, severe damage, and failure. The wind turbine blade diagnosis model outputs the diagnosis result of the wind turbine blade according to the input target energy characteristics.

[0059] Among them, the fan blade diagnosis model is suitable for fan blades of different types and specifications. It can maintain efficient fault detection performance under various complex working conditions and environments, showing strong adaptability and broad application prospects.

[0060] Optionally, the training process of the fan blade diagnosis model includes:

[0061] Obtain the historical voiceprint signal of the wind turbine blade diagnostic model to be trained;

[0062] Controlling the to-be-trained fan blade diagnosis model to perform a fan blade diagnosis task based on the historical voiceprint signal;

[0063] The fan blade diagnosis model to be trained is adjusted according to the fan blade diagnosis task to obtain a trained and updated fan blade diagnosis model.

[0064] Among them, a comprehensive data set containing data from multiple wind farms can be constructed based on historical voiceprint signals. This data set covers the noise signals radiated by wind turbine blades from multiple wind farms. Specifically, the training and test data of the data set are 30 seconds long, with a sampling rate of 20kHz, and training set packages and test set packages are established respectively. By processing the training data with augmentation technology, it is expected to cover more scenarios of wind turbine blade failures, enhance the diversity of the data set, and improve the generalization ability of the model. Data augmentation effectively avoids overfitting of the model and enhances the reliability and accuracy of the model in practical applications.

[0065] In this solution, the wind turbine blade diagnosis task is used to detect the fault conditions of wind turbine blades in historical voiceprint signals, and can accurately obtain the operating status of wind turbine blades. By controlling the training wind turbine blade diagnosis model through historical voiceprint signals to perform wind turbine blade diagnosis tasks, it can effectively identify potential faults of wind turbine blades, give early warnings, reduce maintenance costs, and improve the operating efficiency of wind farms.

[0066] In this embodiment, the parameters in the fan blade diagnosis model to be trained can be adjusted according to the fan blade diagnosis task, so as to obtain a fan blade diagnosis model after training and updating. The parameters may include a learning rate, a loss function, and the like.

[0067] Furthermore, the wind turbine blade fault detection problem can be equivalent to a one- or two-class classification problem. Classification problems often use accuracy (A), recall (R), precision (P), and F-1 score (F1) to measure the performance of classification algorithms. TP (True Positive) indicates the number of samples whose labels are positive and whose prediction results are also positive, FN (False Negative) indicates the number of samples whose labels are positive but whose prediction results are negative, FP (False Positive) indicates the number of samples whose labels are negative but whose prediction results are positive, and TN (True Negative) indicates the number of samples whose labels are negative and whose prediction results are also negative. For all the above indicators, the higher the value, the better the corresponding algorithm effect.

[0068] Optionally, adjusting the fan blade diagnosis model to be trained according to the fan blade diagnosis task includes:

[0069] Determining a loss function value corresponding to the wind turbine blade diagnosis task;

[0070] According to the loss function value, the network parameters of the wind turbine blade diagnostic model to be trained are adjusted.

[0071] In this embodiment, the loss function may be mean square error loss (mean square error loss function) or l1 loss (regression loss function) or the like.

[0072] Specifically, the historical voiceprint signal can be used as input, and the fan blade diagnosis task can be performed through the fan blade diagnosis model to be trained, so as to determine the loss function value corresponding to the fan blade diagnosis task. Based on the loss function value, the network parameters of the fan blade diagnosis model to be trained are adjusted to achieve the optimal loss function value.

[0073] By adjusting the network parameters of the wind turbine blade diagnostic model to be trained and building a wind turbine blade diagnostic model, the potential faults of the wind turbine blades can be effectively identified based on the wind turbine blade diagnostic model, and early warning can be given to reduce maintenance costs and improve the operating efficiency of the wind farm.

[0074] Optionally, after obtaining the diagnosis result, the method further includes:

[0075] A fan blade fault warning is performed according to the diagnosis result, and the fan blade diagnosis model is optimized according to the diagnosis result.

[0076] In this solution, when the diagnosis result is a fault, the system triggers an alarm mechanism. The alarm mechanism may include sound and light alarms and notification of relevant personnel. The notification of relevant personnel may be by phone, email, etc.

[0077] Furthermore, the system continuously evaluates the diagnostic results and calculates performance indicators such as accuracy, recall, and F1 score to ensure the reliability of the model. The wind turbine health index integrates diagnostic results, maintenance records, operating hours and other factors to update the health status of the wind turbine in real time and provide performance optimization suggestions. All diagnostic results, health status assessments and fault information are uploaded to the cloud platform in real time for storage and management. The system uses big data technology to perform trend analysis on the stored data to help the operation team conduct comprehensive health management of wind turbine equipment.

[0078] Through automated fault detection, real-time warning and intelligent decision support, the frequency of manual inspection and maintenance has been greatly reduced. The system can detect and warn of potential faults in advance, thus avoiding large-scale equipment damage, reducing wind turbine downtime and maintenance costs, and improving the overall economic benefits of the wind farm.

[0079] The technical solution of the embodiment of the present invention obtains the soundprint signal generated during the operation of the wind turbine blade, filters the soundprint signal, obtains the target soundprint signal, then decomposes the target soundprint signal to obtain the decomposed soundprint signal, reconstructs the decomposed soundprint signal to obtain the reconstructed soundprint signal, calculates the energy characteristics of each sub-band of the reconstructed soundprint signal to obtain the target energy characteristics, and then inputs the target energy characteristics into the wind turbine blade diagnosis model for prediction to obtain the diagnosis result. By executing this technical solution, through the combination of wavelet packet analysis and neural network model, the normal operation audio and fault audio of the wind turbine blade can be accurately identified, thereby significantly improving the accuracy of fault detection; supporting real-time online monitoring of wind turbine blades, timely discovering the damage of the blades and issuing warnings immediately, ensuring the safe and stable operation of the wind turbine; the diagnosis process from sound collection, signal processing to fault diagnosis and warning, without manual intervention, fully realizes automation, reduces the cost and labor intensity of manual inspection; can achieve low-cost, efficient and rapid system deployment of wind farms, and significantly reduce the maintenance and overhaul costs of wind turbines.

[0080] Embodiment 2

[0081] Figure 2 This is a schematic diagram of a fan blade diagnosis process provided by the second embodiment of the present invention. The relationship between this embodiment and the above embodiment is a detailed description of the voiceprint signal processing process. Figure 2 As shown, the method includes:

[0082] S210: Acquire a soundprint signal generated during the operation of the fan blades.

[0083] S220, processing the voiceprint signal based on a preset sampling interval and a preset sampling period to obtain a sampled voiceprint signal, and transmitting the sampled voiceprint signal to a storage device through a wireless network and an optical fiber network.

[0084] The sampling interval and sampling period can be set according to the voiceprint signal sampling requirements. For example, the sampling interval can be set to 30 minutes and the sampling period can be set to 30 seconds.

[0085] In this solution, the voiceprint signal contains a large amount of data. In order to effectively reduce the data volume, transmission and storage pressure caused by the voiceprint data collection throughout the day, and at the same time ensure the integrity and timeliness of key data, edge computing and dynamic interval sampling strategies can be adopted. Specifically, when the wind turbine is operating normally, the sampling interval is 30 minutes and the sampling period is 30 seconds to save bandwidth and storage. Under specific working conditions, the sampling frequency is increased to capture more details. At the same time, the sensor collects data and transmits it to the edge gateway to monitor the data in real time. When the voiceprint signal frequency exceeds the preset threshold, the log is stored and the data is uploaded to the server for archiving and further analysis.

[0086] Furthermore, the signal transmission adopts a hybrid network architecture of optical fiber deployment and wireless network. The sampled voiceprint signal data is transmitted to the edge gateway in the control cabinet through the internal wireless network, converted into a limited network protocol (Ethernet) after preliminary processing, and then transmitted to the wind farm control room through the optical fiber interface, and finally connected to the optical fiber backbone network and transmitted to the server for further storage and analysis. This architecture combines the flexibility of wireless communication and the high-speed stability of optical fiber networks, ensuring the efficiency and security of data transmission while reducing wiring costs.

[0087] Through dynamic sampling strategies, a transmission architecture combining optical fiber and wireless networks is constructed to ensure efficient and secure data transmission. At the same time, edge computing technology is used to improve data processing capabilities and reduce network load. Real-time, low-latency fault monitoring is achieved, significantly improving data collection efficiency.

[0088] S230: Obtain a sampled voiceprint signal.

[0089] In this solution, the sampled voiceprint signal can be obtained from the memory.

[0090] S240: Filter the sampled voiceprint signal through a Butterworth bandpass filter to obtain a filtered voiceprint signal.

[0091] Among them, the Butterworth bandpass filter is a filter that allows signals to pass within a specific frequency range and blocks signals outside the range from passing through.

[0092] In this solution, the acquired sampled voiceprint signal is initially filtered through a Butterworth bandpass filter to obtain a filtered voiceprint signal. The transfer function H(s) of the Butterworth bandpass filter is defined as: Where G is the filter cutoff frequency; s is the complex frequency unit, ω c is the filter cutoff frequency and n is the filter order.

[0093] S250: Filter the filtered voiceprint signal using empirical mode decomposition to obtain a target voiceprint signal.

[0094] Furthermore, the filtered voiceprint signal is filtered again by empirical mode decomposition (EMD) to obtain the target voiceprint signal. The steps are as follows: 1. Extreme point detection: find all the maximum and minimum values ​​in the signal; 2. Envelope construction: construct the upper envelope and the lower envelope respectively by the maximum and minimum values; 3. Average value calculation: calculate the average value of the upper envelope and the lower envelope to obtain the trend component of the signal; 4. IMF extraction: subtract the trend component from the voiceprint signal to obtain a preliminary IMF. Repeat this process until the remaining signal has no maximum or minimum values. Among them, the high-frequency and low-frequency noise components are removed by EMD, and the main characteristic frequency bands of the signal are retained to obtain the filtered and denoised signal. The EMD decomposition formula is as follows: Among them, IMF i (t) is the i-th intrinsic mode function; r(t) is the final residual component.

[0095] By combining the Butterworth bandpass filter with the empirical mode decomposition technique, noise can be effectively removed, important features can be retained, and the accuracy of the signal can be improved.

[0096] S260, decomposing the target voiceprint signal to obtain a decomposed voiceprint signal; reconstructing the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculating the energy characteristics of each sub-band of the reconstructed voiceprint signal to obtain a target energy characteristic.

[0097] S270, inputting the target energy feature into a fan blade diagnostic model for prediction to obtain a diagnostic result; wherein the fan blade diagnostic model is determined by training historical voiceprint signals.

[0098] In this program, Figure 3 is a flow chart of a fan blade detection system based on voiceprint signals provided in the second embodiment of the present application, such as Figure 3As shown in the figure, sensor layout: audio sensors are arranged inside and outside the wind turbine respectively. Acquisition of voiceprint signals: The voiceprint signals generated by the wind turbine blades under different working conditions are collected in real time through the arranged voiceprint sensors; edge computing and dynamic interval sampling are used to collect all-weather voiceprint data, that is, sampled voiceprint signals are obtained. At the same time, the sensor collected data is transmitted to the edge gateway for real-time monitoring of the data; the signal transmission adopts a hybrid architecture combining optical fiber and wireless network. The voiceprint sensor data is transmitted to the edge gateway through the wireless network, converted into Ethernet protocol after preliminary processing, and then transmitted to the wind farm control room through optical fiber and connected to the backbone network, and finally transmitted to the server. Signal filtering and noise reduction: The acquired sampled voiceprint signal is initially filtered through a Butterworth bandpass filter to obtain a filtered voiceprint signal; the filtered voiceprint signal is filtered again through empirical mode decomposition to obtain the target voiceprint signal. Fault diagnosis method: Extract target capability features based on wavelet packet analysis; build a comprehensive data set containing data from multiple wind farms; build a neural network for wind turbine blade fault detection, namely, a wind turbine blade diagnosis model; use accuracy, recall, precision, and F1 value to measure the performance of the wind turbine blade diagnosis model. Real-time diagnostic response warning and performance evaluation: The system continuously evaluates the diagnostic results, calculates performance indicators such as accuracy, recall, and F1 score to ensure the reliability of the model. The wind turbine health index integrates diagnostic results, maintenance records, operating hours, and other factors to update the health status of the wind turbine in real time and provide performance optimization suggestions. All diagnostic results, health status assessments, and fault information are uploaded to the cloud platform in real time for storage and management. The system uses big data technology to perform trend analysis on the stored data to help the operation team conduct comprehensive health management of wind turbine equipment.

[0099] The technical solution of the embodiment of the present invention obtains the soundprint signal generated during the operation of the fan blade, processes the soundprint signal based on a preset sampling interval and a preset sampling period to obtain a sampled soundprint signal, and transmits the sampled soundprint signal to a memory through a wireless network and an optical fiber network. Then, the sampled soundprint signal is obtained from the memory, the sampled soundprint signal is filtered by a Butterworth bandpass filter to obtain a filtered soundprint signal, and the filtered soundprint signal is filtered by empirical mode decomposition to obtain a target soundprint signal. Then, the target soundprint signal is decomposed to obtain a decomposed soundprint signal, the decomposed soundprint signal is reconstructed to obtain a reconstructed soundprint signal, the energy characteristics of each sub-band of the reconstructed soundprint signal are calculated to obtain a target energy characteristic, and then the target energy characteristic is input into the fan blade diagnosis model for prediction to obtain a diagnosis result. By implementing this technical solution, combined with the Butterworth bandpass filter and empirical mode decomposition technology, it is possible to effectively remove noise, retain important features, and improve the accuracy of the signal; the algorithm combining wavelet packet analysis with the ECAPA-TDNN neural network is used to improve the accuracy and robustness of wind turbine blade fault diagnosis, and to identify potential wind turbine blade faults at an early stage. It realizes the automated monitoring and intelligent decision support of wind turbine equipment, which not only greatly reduces the cost of manual inspection and maintenance, but also significantly reduces wind turbine downtime and improves the overall economic benefits of wind farms. It has strong practical application value and provides an innovative technical solution for the health management and operation and maintenance of wind power equipment.

[0100] In this solution, taking the fan blade soundprint detection as an example, a fan blade soundprint detection method based on characteristic values ​​is provided, and the steps are as follows:

[0101] Step 1: Sensor Placement

[0102] In order to comprehensively monitor the operating status of the fan blades, it is first necessary to arrange audio sensors at key locations of the fan to capture all the soundprint signals of the fan operation. Figure 4 The specific location diagram of the fan sensor arrangement provided in the embodiment of the present application is as follows: Figure 4 As shown, 1. Blade root hub sensor 1: An audio sensor is installed at the blade root hub of the fan to capture the soundprint signal of the blade root. 2. Tower outer wall sensor 3: An audio sensor is installed at a height of 2 meters on the tower outer wall to capture the soundprint signal of the fan blade sweeping the wind. 3. Blade sweeping lower edge sensor 2: An audio sensor is installed at the blade sweeping lower edge to capture the sound characteristics of the blade under different working conditions.

[0103] Step 2: Acquire voiceprint signal

[0104] The sensor collects the soundprint signal of the fan blade in real time and goes through the following processing steps:

[0105] 1. Signal collection: During normal operation of the fan, the soundprint signal is collected every 30 minutes, and each collection lasts 30 seconds. If the fan has an abnormal situation (such as drastic changes in wind speed, equipment vibration, etc.), the sampling frequency will be automatically increased.

[0106] 2. Data transmission: The voiceprint signal is transmitted to the wind farm control system via wireless network and optical fiber, and finally uploaded to the cloud platform for storage and subsequent analysis.

[0107] Step 3: Signal filtering and noise reduction

[0108] Step 3.1: In order to remove environmental noise, the collected voiceprint signal is filtered through a Butterworth bandpass filter.

[0109] Step 3.2: Use empirical mode decomposition to remove low-frequency noise, retain high-frequency voiceprint features, and obtain the target voiceprint signal.

[0110] Step 4: Fault Diagnosis Algorithm

[0111] Step 4.1: Extracting acoustic eigenvalue parameters of target voiceprint signal. The acoustic eigenvalue of the voiceprint signal is extracted from the target voiceprint signal using wavelet packet analysis.

[0112] Step 4.2: Dataset construction. Based on the extracted voiceprint features, a comprehensive dataset containing multiple wind farm data is constructed. The dataset includes wind turbine blade radiation noise signals from wind farms in different geographical locations.

[0113] Step 4.3: Neural network construction for wind turbine blade fault detection. Neural network construction for wind turbine blade fault detection. A wind turbine blade fault detection method based on the ECAPA-TDNN network is used.

[0114] Step 4.4: Determine the evaluation index. The fan blade fault detection problem can be equivalent to a one- or two-class classification problem. Classification problems often use accuracy A, recall R, precision P, and F1 value to measure the performance of the classification algorithm.

[0115] Step 5: Wind turbine real-time diagnosis, response, warning and performance evaluation

[0116] Step 5.1: Real-time diagnosis and reasoning: In the fan blade fault diagnosis system, the health status of the fan is evaluated through the real-time collected soundprint signal. The collected soundprint signal is collected by the audio sensor installed on the fan and transmitted to the data processing unit for analysis. The data processing unit inputs the real-time audio signal into the trained ECAPA-TDNN neural network model for reasoning. The network model outputs the health status of the fan blade based on the feature parameters of the previous training. The classification of health status includes: normal, slightly damaged, severely damaged and faulty.

[0117] Step 5.2: Alarm: When the system detects that the health status of the fan blade is faulty, the system will trigger the alarm mechanism. The alarm mechanism includes: Sound and light alarm: The sound and light alarm device will sound an alarm at the fan site to remind the on-site staff to deal with it in time. Remote notification: The system will notify the relevant maintenance personnel and management personnel through email, SMS, APP, etc. to ensure rapid response and troubleshooting.

[0118] Step 5.3: Real-time data upload and big data analysis: All diagnostic results, health status assessments, fault information, and wind turbine operating data are uploaded to the cloud platform in real time for storage and management. The cloud platform uses big data technology to perform trend analysis on the stored historical data and generate data reports. Through trend analysis, the operation team can more clearly understand the operating status of the equipment, the frequency of faults, and the changing trends of health status.

[0119] Embodiment 3

[0120] Figure 5 A schematic diagram of the structure of a fan blade diagnostic device provided in Embodiment 3 of the present invention.

[0121] like Figure 5 As shown, the device comprises:

[0122] The voiceprint signal acquisition module 510 is used to acquire the voiceprint signal generated during the operation of the fan blade;

[0123] The target voiceprint signal obtaining module 520 is used to filter the voiceprint signal to obtain the target voiceprint signal;

[0124] The target energy feature obtaining module 530 is used to decompose the target voiceprint signal to obtain a decomposed voiceprint signal; reconstruct the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculate the energy features of each sub-band of the reconstructed voiceprint signal to obtain a target energy feature;

[0125] The diagnosis result obtaining module 540 is used to input the target energy feature into the fan blade diagnosis model for prediction to obtain a diagnosis result; wherein the fan blade diagnosis model is determined by training the historical voiceprint signal.

[0126] Optionally, the voiceprint signal acquisition module 510 is specifically used to:

[0127] Acquire the sound print signal generated during the operation of the wind turbine blade based on the pre-installed audio sensor; wherein, the audio sensor comprises a first audio sensor and a second audio sensor; the first audio sensor is arranged at the hub of the wind turbine blade root; the second audio sensor is arranged at a preset height of the tower outer wall, the height of the blade wind sweeping lower edge, and the outer wall of the hub.

[0128] Optionally, the device further comprises:

[0129] The sampled voiceprint signal acquisition module is used to process the voiceprint signal based on a preset sampling interval and a preset sampling period to obtain a sampled voiceprint signal, and transmit the sampled voiceprint signal to a storage device via a wireless network and an optical fiber network.

[0130] Optionally, the target voiceprint signal obtaining module 520 is specifically used to:

[0131] Obtaining a sampled voiceprint signal;

[0132] Filtering the sampled voiceprint signal through a Butterworth bandpass filter to obtain a filtered voiceprint signal;

[0133] The filtered voiceprint signal is filtered using empirical mode decomposition to obtain a target voiceprint signal.

[0134] Optionally, the diagnosis result obtaining module 540 includes:

[0135] A historical voiceprint signal acquisition unit, used to acquire the historical voiceprint signal of the wind turbine blade diagnostic model to be trained;

[0136] A fan blade diagnosis task execution unit, used for controlling the fan blade diagnosis model to be trained to execute the fan blade diagnosis task based on the historical voiceprint signal;

[0137] The fan blade diagnosis model training unit is used to adjust the fan blade diagnosis model to be trained according to the fan blade diagnosis task to obtain a trained and updated fan blade diagnosis model.

[0138] Optional, wind turbine blade diagnostic model training unit, specifically used for:

[0139] Determining a loss function value corresponding to the wind turbine blade diagnosis task;

[0140] According to the loss function value, the network parameters of the wind turbine blade diagnostic model to be trained are adjusted.

[0141] Optionally, the device further comprises:

[0142] The diagnosis result module is used to provide a fan blade fault warning according to the diagnosis result, and to optimize the fan blade diagnosis model according to the diagnosis result.

[0143] A fan blade diagnostic device provided by an embodiment of the present invention can execute a fan blade diagnostic method provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0144] Embodiment 4

[0145] Figure 6 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0146] like Figure 6 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0147] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0148] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a wind turbine blade diagnosis method.

[0149] In some embodiments, a wind blade diagnostic method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of a wind blade diagnostic method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform a wind blade diagnostic method in any other appropriate manner (e.g., by means of firmware).

[0150] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0151] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0152] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0153] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0154] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0155] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0156] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0157] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A fan blade diagnosis method, characterized in that: include: Acquire the soundprint signal generated during the operation of the fan blades; Filtering the voiceprint signal to obtain a target voiceprint signal; Decomposing the target voiceprint signal to obtain a decomposed voiceprint signal; reconstructing the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculating the energy characteristics of each sub-band of the reconstructed voiceprint signal to obtain a target energy characteristic; The target energy feature is input into a fan blade diagnostic model for prediction to obtain a diagnostic result; wherein the fan blade diagnostic model is determined by training historical voiceprint signals.

2. The method according to claim 1, characterized in that Acquire the soundprint signal generated during the operation of the fan blades, including: Acquire the sound print signal generated during the operation of the wind turbine blade based on the pre-installed audio sensor; wherein, the audio sensor comprises a first audio sensor and a second audio sensor; the first audio sensor is arranged at the hub of the wind turbine blade root; the second audio sensor is arranged at a preset height of the tower outer wall, the height of the blade wind sweeping lower edge, and the outer wall of the hub.

3. The method according to claim 1, characterized in that After acquiring the soundprint signal generated during the operation of the fan blade, the method further includes: The voiceprint signal is processed based on a preset sampling interval and a preset sampling period to obtain a sampled voiceprint signal, and the sampled voiceprint signal is transmitted to a storage device via a wireless network and an optical fiber network.

4. The method according to claim 3, characterized in that: Filtering the voiceprint signal to obtain a target voiceprint signal includes: Obtaining a sampled voiceprint signal; Filtering the sampled voiceprint signal through a Butterworth bandpass filter to obtain a filtered voiceprint signal; The filtered voiceprint signal is filtered using empirical mode decomposition to obtain a target voiceprint signal.

5. The method according to claim 1, characterized in that The training process of the fan blade diagnosis model includes: Obtain the historical voiceprint signal of the wind turbine blade diagnostic model to be trained; Controlling the fan blade diagnosis model to be trained to perform a fan blade diagnosis task based on the historical voiceprint signal; The fan blade diagnosis model to be trained is adjusted according to the fan blade diagnosis task to obtain a trained and updated fan blade diagnosis model.

6. The method according to claim 5, characterized in that Adjusting the fan blade diagnosis model to be trained according to the fan blade diagnosis task includes: Determining a loss function value corresponding to the wind turbine blade diagnosis task; According to the loss function value, the network parameters of the wind turbine blade diagnostic model to be trained are adjusted.

7. The method according to claim 1, characterized in that After obtaining the diagnosis result, the method further includes: A fan blade fault warning is performed according to the diagnosis result, and the fan blade diagnosis model is optimized according to the diagnosis result.

8. A fan blade diagnostic device, characterized in that: include: A voiceprint signal acquisition module is used to acquire the voiceprint signal generated during the operation of the fan blades; A target voiceprint signal obtaining module is used to filter the voiceprint signal to obtain a target voiceprint signal; The target energy feature acquisition module is used to decompose the target voiceprint signal to obtain a decomposed voiceprint signal; reconstruct the decomposed voiceprint signal to obtain a reconstructed voiceprint signal; calculate the energy features of each sub-band of the reconstructed voiceprint signal to obtain a target energy feature; The diagnosis result obtaining module is used to input the target energy characteristics into the fan blade diagnosis model for prediction to obtain the diagnosis result; wherein the fan blade diagnosis model is determined by training the historical voiceprint signal.

9. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a wind blade diagnostic method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a wind turbine blade diagnosis method according to any one of claims 1 to 7 when executed.

Images