Tour inspection method and device for fan blade of wind driven generator and medium

Abstract

The invention provides a fan blade inspection method and device for a wind driven generator and a medium. The method comprises the steps that blade image information, real-time environment information of blades and real-time aerodynamic acoustic signals of the blades of the wind driven generator in an alpine region collected by an unmanned aerial vehicle in the flight inspection process are obtained; determining blade defect parameters according to the blade image information; determining a first risk index of the blade according to the blade defect parameter and the real-time environment information; determining a second risk index of the blade according to the real-time aerodynamic acoustic signal; determining a current risk state index of the blade according to the first risk index and / or the second risk index; and generating and sending an action control instruction of the wind driven generator according to the current risk state index of the blade. According to the invention, the real-time closed-loop response of the fan blade inspection in the alpine region is realized.

Description

Technical Field

[0001] This invention relates to the field of wind turbine inspection technology, and in particular to a method, device and medium for inspecting the blades of a wind turbine generator. Background Technology

[0002] Currently, during wind turbine inspections, visual data acquired by drones is typically stored first. After the drone returns, ground personnel or a central server perform offline analysis to make final maintenance decisions. Alternatively, high-speed data links can be used to transmit data back to the ground in real time. However, there is a response delay caused by the analysis and decision-making process between acquiring risk information and the control system executing a response. In high-altitude, cold regions, the icing on wind turbine blades can change rapidly within a short period, making this response delay extremely detrimental. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method, device and medium for inspecting wind turbine blades of a wind turbine generator, so as to solve the problem of response delay during the inspection of wind turbine blades of a wind turbine generator.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: In a first aspect, embodiments of the present invention provide a method for inspecting the blades of a wind turbine generator, comprising: Acquire image information of wind turbine blades in high-altitude and cold regions, real-time environmental information of the blades, and real-time aeroacoustic signals of the blades collected by UAVs during flight inspections. Based on the blade image information, determine the blade defect parameters of the wind turbine. Based on the blade defect parameters and the real-time environmental information of the blade, the first risk index of the blade is determined; The second risk index of the blade is determined based on the real-time aeroacoustic signals of the blade. The current risk status index of the blade is determined based on the first risk index and / or the second risk index. Based on the current risk status index of the blade, a motion control command for the wind turbine is generated and sent to the control system of the wind turbine, so that the control system executes the corresponding action according to the motion control command.

[0005] Optionally, based on the blade image information, the blade defect parameters of the wind turbine are determined, including: The blade image information is then analyzed to identify icing, cracks, and three-dimensional geometric dimensions, resulting in the identification results for icing, cracks, and three-dimensional geometric dimensions. Based on the identification results of icing, cracks, and three-dimensional geometry, the blade defect parameters of the wind turbine were determined.

[0006] Optionally, based on the blade defect parameters and the real-time environmental information of the blade, a first risk index for the blade is determined, including: Based on the blade defect parameters, the real-time environmental information of the blade, and multiple preset interpolation center points, multiple distances are determined; The weight of each distance is determined based on a preset kernel function and the multiple distances; The first risk index is determined based on the weights and the distance.

[0007] Optionally, a second risk index for the blade is determined based on the real-time aeroacoustic signals of the blade, including: The wind speed and blade pitch angle of the current operating condition of the blades are obtained from the control system of the wind turbine. Based on the wind speed and blade pitch angle under the current operating conditions, an acoustic feature summary is obtained by matching from a preset acoustic baseline library and used as a reference acoustic summary. The real-time start-up acoustic signal of the blades is converted to the frequency domain, and a real-time acoustic feature summary is extracted. Obtain the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band; The second risk index of the blade is determined based on the difference and the preset threshold.

[0008] Optionally, an acoustic feature summary is obtained from a preset acoustic baseline library based on the wind speed and blade pitch angle under the current operating conditions, serving as a baseline acoustic summary, including: The wind speed and blade pitch angle under the current operating conditions are quantified into query conditions. Based on the query criteria, an acoustic feature summary is obtained by matching the preset acoustic baseline library, which serves as the benchmark acoustic summary.

[0009] Optionally, a second risk index for the blade is determined based on the difference and a preset threshold, including: If the difference is greater than the first preset threshold, the difference is determined to be an acoustic feature variation; The second risk index of the blade is determined based on the changes in the acoustic characteristics.

[0010] Optionally, the current risk status index of the blade is determined based on the first risk index and / or the second risk index, including: If the acoustic feature change is greater than the second preset threshold, then the current risk status index of the blade is determined to be the second risk index. If the acoustic feature variation is less than or equal to the second preset threshold, the current risk status index of the blade is determined based on the first risk index and the second risk index.

[0011] Optionally, the wind turbine's operation control command is generated based on the current risk state index of the blade, including: If the current risk status index is greater than the first preset control threshold, a shutdown control command of the safety level is generated; If the current risk status index is less than the second preset control threshold, a safety control command for load reduction level is generated. Otherwise, generate motion control instructions for reducing the load level.

[0012] Secondly, embodiments of the present invention also provide a wind turbine blade inspection device for a wind turbine generator, comprising: The edge computing module is used to acquire image information of wind turbine blades in high-altitude and cold regions, real-time environmental information of the blades, and real-time aeroacoustic signals of the blades collected by the UAV during flight inspection; determine the blade defect parameters of the wind turbine based on the blade image information; determine the first risk index of the blade based on the blade defect parameters and the real-time environmental information of the blade; determine the second risk index of the blade based on the real-time aeroacoustic signals of the blade; determine the current risk state index of the blade based on the first risk index and / or the second risk index; and generate action control commands for the wind turbine based on the current risk state index of the blade. The communication module sends control commands to the control system of the wind turbine, causing the control system to execute corresponding actions according to the action control commands.

[0013] Thirdly, embodiments of the present invention also provide a computer-readable storage medium storing a program that, when executed by a processor, implements the method described in the first aspect.

[0014] The above-described solution of the present invention has at least the following beneficial effects: The above-described solution of the present invention acquires image information of wind turbine blades in high-altitude and cold regions, real-time environmental information of the blades, and real-time aeroacoustic signals of the blades collected by a UAV during flight inspection. Based on the blade image information, it determines the blade defect parameters; based on the blade defect parameters and the real-time environmental information of the blades, it determines a first risk index of the blades; based on the real-time aeroacoustic signals of the blades, it determines a second risk index of the blades; based on the first risk index and / or the second risk index, it determines a current risk state index of the blades; based on the current risk state index of the blades, it generates motion control commands for the wind turbines and sends them to the wind turbine control system, causing the control system to execute the corresponding actions according to the motion control commands. This invention, by combining environmental information, aeroacoustic signals, and blade image information, efficiently and accurately achieves real-time closed-loop response to dynamic risks in high-altitude and cold regions. Attached Figure Description

[0015] Figure 1 This is a flowchart of an embodiment of the wind turbine blade inspection method of the present invention; Figure 2 This is a schematic diagram of an embodiment of the wind turbine blade inspection device of the present invention; Figure 3 This is a flowchart illustrating an application scenario of the wind turbine blade inspection method of the present invention. Figure 4 This is a schematic diagram illustrating an application scenario of the wind turbine blade inspection method of the present invention. Detailed Implementation

[0016] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0017] like Figure 1 As shown, an embodiment of the present invention proposes a method for inspecting the blades of a wind turbine generator, comprising: Step 11: Acquire the image information of the wind turbine blades in high-altitude and cold regions, the real-time environmental information of the blades, and the real-time aeroacoustic signals of the blades collected by the UAV during the flight inspection process. Step 12: Determine the blade defect parameters of the wind turbine based on the blade image information; Step 13: Determine the first risk index of the blade based on the blade defect parameters and the real-time environmental information of the blade. Step 14: Determine the second risk index of the blade based on the real-time aeroacoustic signal of the blade; Step 15: Determine the current risk status index of the blade based on the first risk index and / or the second risk index; Step 16: Generate a motion control command for the wind turbine based on the current risk status index of the blade, and send it to the control system of the wind turbine, so that the control system executes the corresponding action according to the motion control command.

[0018] For step 11, in this embodiment, the process of acquiring blade image information may include: a high-definition visible light camera is carried by an inspection drone, and the camera is used as an imaging sensor to collect image data streams of the blade surface in real time and continuously along a predetermined route during the drone's flight inspection mission.

[0019] In this embodiment, the real-time environmental information of the blade includes the current real-time ambient temperature of the blade. As an example, this can be obtained in real-time through environmental sensors such as a digital thermometer mounted on a drone, for example, by acquiring a... The real-time ambient temperature value.

[0020] In this embodiment, a microphone with a high signal-to-noise ratio can be installed on the drone. When the drone flies over the blade surface, the mixed acoustic signals can be continuously recorded through this microphone to obtain the real-time aeroacoustic signal of the blade.

[0021] In an optional embodiment of the present invention, step 12, determining the blade defect parameters of the wind turbine based on the blade image information, includes: Step 121: Identify the icing, cracks and three-dimensional geometric dimensions of the blade image information to obtain the identification results of icing, cracks and three-dimensional geometric dimensions; Step 122: Determine the blade defect parameters of the wind turbine based on the identification results of icing, cracks, and three-dimensional geometry.

[0022] For step 121, the identification results of icing include: areas on the blade surface that conform to the visual characteristics of icing (such as abnormal color, texture, and outline), and the type of these areas is "icing"; the identification results of cracks include: areas on the blade surface that conform to the visual characteristics of cracks, such as linear or mesh-like patterns, and the type of these areas is "cracks"; the identification results of three-dimensional geometric dimensions include: the size data of the defects on the blade surface (such as crack length of 0.4m) and the location information of the defects on the blade surface.

[0023] As an example, the process of identifying icing and cracks in the blade image information can be found in the following: 1. Acquire image I and perform grayscale processing on image I to obtain a grayscale image. : ; Among them, R G B Let x, y, and y represent the red, green, and blue channel intensity values ​​of each pixel at coordinates (x, y) in image I, respectively. This represents the image after grayscale processing.

[0024] 2. For grayscale images For Gaussian filtering, please refer to: ; ; in, This represents an image processed by Gaussian filtering, where k represents the threshold of the filter kernel radius (e.g., k=2). This represents the preset standard deviation, controlling the smoothness (e.g., σ=1.5), where x represents the x-coordinate of a pixel and y represents the y-coordinate of a pixel. This represents the offset of x. This represents the offset in the y-axis. This represents the weighting function, used to implement Gaussian smoothing filtering for noise reduction.

[0025] 3. Based on the grayscale image Extract grayscale image The ice-covered features, crack features, and texture features in the text.

[0026] Extracting grayscale images For icing characteristics, please refer to: ; ; in, Indicates icing characteristics, This indicates the total number of pixels within the area of ​​interest. This represents the arithmetic mean of the grayscale values ​​of all pixels within the region of interest, and is used to calculate the dispersion of grayscale values ​​within that region.

[0027] Extracting grayscale images For crack characteristics, please refer to: ; ; in, This represents a binary image at the edge (1 at the edge, 0 otherwise). This indicates the high and low thresholds of the Canny operator. This represents the edge density value.

[0028] Extracting grayscale images For texture features, please refer to: ; in, This represents an 8-neighborhood local binary pattern operator with a radius of 1. Indicates calculation Variance of the encoding Representing texture features, cracks disrupt the consistency of surface texture; therefore, the larger the crack, the worse the texture. The larger.

[0029] 4. The ice-covering features, crack features, and texture features are fused to obtain the ice-covering and crack recognition results of the blade image information.

[0030] ; ; ; in, This represents a feature matrix composed of icing features, crack features, and texture features. , These represent the preset weights. , This indicates the preset bias. , These represent the scores for icing features and crack features, respectively.

[0031] Features scoring above a preset value are used to identify icing and cracks. For example, category "1" indicates the presence of icing features, category "2" indicates the presence of crack features, category "12" indicates the presence of both icing and crack features, and category "0" indicates the absence of both icing and crack features. For example, if... , If all values ​​are greater than the preset value, the recognition result is "12". If only If the value is greater than the preset value, the recognition result is "1".

[0032] As an example, the process of identifying the three-dimensional geometric dimensions of the blade image information can be referred to as follows: 1. Determine the value of each pixel in step 121. , , If the value is greater than a preset threshold, then the pixel is determined to be the pixel corresponding to ice and / or cracks.

[0033] 2. According to Map the two-dimensional coordinates of the pixels corresponding to icing and / or cracks to three-dimensional coordinates. Where i = 1, 2, ..., n represents the total number of pixels corresponding to icing and / or cracks. () represents the pixel coordinates of icing and / or cracks in the image. This represents the pose matrix of the UAV at the current moment, including the UAV's current position coordinates and attitude information (pitch, yaw, roll angle). This represents a three-dimensional digital surface model of the pre-defined inspection blade. This represents the back projection function, a standard computer vision geometric calculation process. Its principle is to combine camera intrinsic parameters and UAV pose to construct a path from the camera lens through the pixel (…). The line of sight. Calculate the relationship between this line of sight and the 3D model of the blade. The intersection point is the three-dimensional spatial location of the defect on the blade. .

[0034] 3. According to Determine the three-dimensional geometric dimensions, where, This represents the pixel-to-physical-size conversion matrix factor.

[0035] For step 122, the blade defect parameters are obtained by constructing a matrix based on the identification results of icing, cracks, and three-dimensional geometry. Among them, ,in, This represents a matrix constructed based on the identification results of icing, cracks, and three-dimensional geometric dimensions. The identification result for icing or cracks is indicated by, for example, "0", "1", "12" or "2". Represents three-dimensional geometric dimensions.

[0036] In an optional embodiment of the present invention, step 13, determining the first risk index of the blade based on the blade defect parameters and the real-time environmental information of the blade, includes: Step 131: Determine multiple distances based on the blade defect parameters, the real-time environmental information of the blade, and multiple preset interpolation center points; Step 132: Determine the weight of each distance based on the preset kernel function and the multiple distances; Step 133: Determine the first risk index based on the weight and the distance.

[0037] For step 131, the blade defect parameters and the real-time environmental information of the blade are first combined into a multi-dimensional query point. ,in, ,in, Represents real-time ambient temperature (which can be measured by a thermometer on the drone or by a temperature measurement received by the drone from the fan under test), in units of... , Represents a 3-dimensional vector. This represents a matrix constructed based on the identification results of icing, cracks, and three-dimensional geometric dimensions. This indicates the identification results for icing and cracks. Represents three-dimensional geometric dimensions.

[0038] Then, N interpolation center points are preset. ,according to Calculate multidimensional query points To each center point The Euclidean distance, of which , Indicates the query point. Indicates the center point , Indicates query point The Euclidean distance to the i-th center point.

[0039] For step 132, the preset kernel function is a Gaussian kernel, according to... European distance Convert to weights, where, Represents the Gaussian kernel function. The bandwidth parameter (or shape parameter) of the kernel function is a constant greater than 0, determined during model training. It controls the rate at which the weights decay with distance. The larger the value, the slower the weight decay, and the greater the influence of distant points. Indicates the first Each center point is relevant to the current query point. The influence weight is determined by the distance. The closer the distance, the greater the weight; the farther the distance, the closer the weight approaches 0.

[0040] For step 133, according to The first risk index was determined, among which, Indicates the first Interpolation center point The corresponding known risk index, This represents the first risk index, with a value range of 0-100.

[0041] In an optional embodiment of the present invention, step 14, determining the second risk index of the blade based on the real-time aeroacoustic signal of the blade, includes: Step 141: Obtain the wind speed and blade pitch angle of the current operating condition of the blades from the control system of the wind turbine. Step 142: Based on the wind speed and blade pitch angle under the current operating conditions, obtain an acoustic feature summary from the preset acoustic baseline library and use it as a reference acoustic summary. Step 143: Convert the real-time start-up acoustic signal of the blade to the frequency domain and extract the real-time acoustic feature summary; Step 144: Obtain the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band; Step 145: Determine the second risk index of the blade based on the difference and the preset threshold.

[0042] For step 141, the UAV sends a request to the wind turbine to be tested via a pre-set communication link to obtain the current operating conditions of the blades, including wind speed v (in m / s) and blade pitch angle. The unit is degree.

[0043] In an optional embodiment of the present invention, step 142, which involves matching an acoustic feature summary from a preset acoustic baseline library based on the wind speed and blade pitch angle under the current operating conditions, and using this summary as a reference acoustic summary, may include: Step 1421: Quantify the wind speed and blade pitch angle of the current operating condition into query conditions; Step 1422: Based on the query conditions, obtain an acoustic feature summary from the preset acoustic baseline library and use it as a benchmark acoustic summary.

[0044] For step 1421, since the acoustic baseline library is stored discretely (e.g., wind speed at 2 m / s intervals, pitch angle at 5...), (Interval storage), therefore, it is necessary to store continuous real-time operating conditions. Quantify to the nearest warehouse operating point. For example, it can be based on... Continuous real-time operating conditions Quantified to the closest warehouse operating point, among which, Indicates the quantized query key. This represents the quantization function, for example, , .

[0045] For step 1422, use the query key. Search the acoustic baseline library to obtain the baseline acoustic summary ( ),in, The mean vector representing the characteristic vectors of healthy leaves. The covariance matrix represents the characteristic vector of a healthy leaf.

[0046] As an example, the process of setting up a preset acoustic baseline library can be found here: First, after confirming that the wind turbine is in good condition, the drone is controlled to fly close to the wind turbine under different controllable operating conditions (including combinations of different wind speeds and different blade pitch angles) and collect aeroacoustic signals. Next, for each set of raw acoustic signals acquired under each working condition, standardized signal processing (which may include Fast Fourier Transform, i.e., FFT) is performed to convert them to the frequency domain, and their key acoustic feature summaries (which may include the energy distribution of a specific frequency band or the amplitude of the main harmonic component) are extracted. Finally, these operating parameters (wind speed, pitch angle) are associated with the corresponding acoustic feature summaries and stored in the database to obtain the preset acoustic baseline library.

[0047] For step 143, the real-time startup acoustic signal of the blade is converted to the frequency domain, and a real-time acoustic feature summary is extracted, including: The real-time startup acoustic signal s(t) of the wind turbine under test is acquired using the drone's built-in microphone. The original time-domain signal s(t) is then subjected to bandpass filtering (to remove drone noise) and frame-by-frame windowing to obtain frames of the signal. For each frame The spectrum is obtained by performing a Fast Fourier Transform (FFT), passed through a Mel filter bank, and the logarithmic energy is taken. A Discrete Cosine Transform (DCT) is then performed on the log-Mel spectrum, and the first 13 coefficients are used to form the feature vector of the frame. The feature vectors of the current moment or the most recent few frames are used as real-time acoustic feature summaries. .

[0048] For step 144, obtaining the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band includes: according to Determine the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band, where, This represents a summary of real-time acoustic features. The mean vector representing the characteristic vectors of healthy leaves. The covariance matrix representing the feature vectors of healthy leaves. This represents the inverse of the covariance matrix.

[0049] In an optional embodiment of the present invention, step 145, determining the second risk index of the blade based on the difference and a preset threshold, includes: Step 1451: If the difference is greater than the first preset threshold, determine the difference. The acoustic characteristics have changed; Step 1452, based on the acoustic feature variation Determine the second risk index for the blade.

[0050] For step 1451, if the difference is less than or equal to the first preset threshold, it is determined that there is no acoustic feature change (i.e., the acoustic feature change becomes 0); otherwise, the difference is determined. This is due to changes in acoustic characteristics.

[0051] For step 1452, according to

[0052] Determine the second risk index. Among them, This indicates the second risk index. This represents the amplification factor (a constant > 1, such as 1.5). It is used to enhance sensitivity in high-risk areas; once a threshold is exceeded, the risk index increases rapidly. Indicates the threshold for acoustic anomaly criteria. This represents the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band.

[0053] In an optional embodiment of the present invention, step 15, determining the current risk status index of the blade based on the first risk index and / or the second risk index, includes: Step 151: If the acoustic feature change is greater than the second preset threshold, then the current risk status index of the blade is determined as the second risk index. Step 152: If the acoustic feature variation is less than or equal to the second preset threshold, determine the current risk status index of the blade based on the first risk index and the second risk index.

[0054] If the second preset threshold in step 151 is greater than the first preset threshold in step 1451, and the acoustic feature variation is greater than the second preset threshold, it indicates that the acoustic feature variation is the main cause, and the current risk state index of the blade is determined as the second risk index; if the acoustic feature variation is less than or equal to the second preset threshold, the average of the first risk index and the second risk index can be considered as the current risk state index.

[0055] In an optional embodiment of the present invention, step 16, generating a wind turbine operation control command based on the current risk state index of the blade, includes: Step 161: If the current risk status index is greater than the first preset control threshold, generate a shutdown level action control command. Step 162: If the current risk status index is less than the second preset control threshold, generate a safety level action control command. Step 163, otherwise, generate the action control command for the load reduction level.

[0056] In this embodiment, the current risk state index of the current blade is set to be... Its value range is [0, 100]. Let the first preset control threshold be... The second preset control threshold is And satisfy ,For example, , .

[0057] For step 161: when This indicates that the blades are in a high-risk state, and the risk is approaching or has reached a critical level. At this time, the most stringent safety measures must be implemented immediately, generating a shutdown-level action control command to stop the wind turbine from rotating immediately and to send alarm information back to the ground control center.

[0058] For step 162: when When the time is right, it indicates that the blades are in a low-risk state, and the impact of defects or mutations is minor. To balance power generation efficiency and equipment safety, preventative protection measures are taken, generating safety-level action control commands to maintain the normal operation of the wind turbine under test.

[0059] For step 163: when At this point, it indicates that the blade is in a medium-risk state, with a clear defect or functional abnormality, but not yet reaching the level of emergency high risk. To prevent the risk from escalating during continuous operation, the current operation needs to be interrupted for inspection or protection. The matching value is obtained by rounding up, and then matched against a preset table (see the table below for reference). The final load reduction scheme is determined based on the matching result. For example, Round up to get , satisfy Therefore, by iterating through the preset table to find the matching values, the UAV sends control information to the wind turbine to be tested, indicating that the target power is reduced by 60%.

[0060] Table 1. Preset Tables

[0061] In one application scenario, refer to Figure 3 , Figure 4 Before the drone takes off to perform its mission, the preset acoustic baseline library and the relevant algorithms for collecting blade defects are loaded and deployed together in the memory of the edge computing unit carried by the drone. The edge computing unit is an affordable processor module with basic floating-point operation capabilities, which can be an embedded system board based on the ARM architecture.

[0062] During the drone's inspection flight along a predetermined route, the edge computing unit simultaneously and in parallel executes multiple data acquisition tasks using onboard sensors: Firstly, it acquires blade defect parameters. The imaging sensor onboard the UAV can acquire real-time image data of the blade surface using a high-definition visible light camera to identify and quantify visible defect parameters, including the type, three-dimensional spatial location, and geometric dimensions of icing, cracks, or structural damage. Secondly, it involves acquiring real-time environmental information about the blades. This can be obtained through real-time sampling by an environmental sensor, such as a digital thermometer mounted on a drone. Real-time ambient temperature value; Thirdly, it captures real-time aeroacoustic signals during blade inspection. This can be achieved by continuously recording the mixed acoustic signals generated as the drone flies over the blade surface using a high signal-to-noise ratio microphone onboard the drone. The real-time aeroacoustic signals are then used to determine whether the acquired sound is normal, thus identifying non-optically visible defects.

[0063] Finally, based on the blade defect parameters and the real-time environmental information of the blade, a first risk index of the blade is determined; based on the real-time aeroacoustic signal of the blade, a second risk index of the blade is determined; and based on the first risk index and / or the second risk index, the current risk state index of the blade is determined, thus overcoming the limitations of a single sensor.

[0064] When the current risk status index is found to have triggered a preset control threshold ( The edge computing unit immediately determines that control intervention at the wind turbine level is required; At this time, the edge computing unit does not transmit the massive raw image or acoustic data stream. Instead, it sends a simple, high-priority alarm message to the wind turbine's control system (SCADA) in real time through its communication module (which can be 4G / 5G or LoRa data link). The wind turbine then controls the wind turbine according to the action control command. Upon receiving the instruction, the wind turbine's control system will immediately and automatically execute the predetermined control action corresponding to the alarm level without waiting for manual confirmation. This action may include executing an emergency shutdown procedure or protective load reduction, thereby completing the control loop the instant the risk is identified. In addition, the present invention may also include optimizations for perception robustness; When the drone identifies its geometric dimensions as being below a preset geometric threshold, such as... When dealing with defect parameters, the output can be constrained to a risk index that is lower than a preset control threshold. This mechanism leverages an engineering understanding in the field of wind turbine operation and maintenance: the risk index of minute defects is extremely low. This allows the mechanism to automatically filter out false defects or misidentifications caused by CV algorithms in severe weather conditions such as snowflakes and water droplets, ensuring that the system does not generate false alarms for these harmless perceived noises.

[0065] In another application scenario, a wind turbine is encountering a sudden, strong cold front accompanied by freezing rain, causing the ambient temperature around the blades to drop rapidly from [previous temperature range]. sudden drop This operating condition poses two concurrent threats to the safety of the blade structure: first, the ductile-brittle transformation of the blade material due to low temperature significantly reduces its tolerance to existing minor defects; second, freezing rain can easily form a layer of high-density transparent ice on the blade surface that is difficult to distinguish optically.

[0066] The process of an unmanned aerial vehicle (UAV) performing an inspection task according to the method of this invention is as follows: When the drone flew to a certain point on the blade, its onboard imaging sensor identified a defect parameter. This parameter is characterized as a geometric dimension of only Tiny surface cracks, while real-time environmental parameters acquired by airborne environmental sensors. for However, the airborne microphone did not detect any changes in acoustic characteristics; according to and Determine the first risk index of the first blade of the wind turbine. ,For example .

[0067] Subsequently, the drone continued flying to blade B, where it failed to identify any visible defect parameters on the icy surface. If it is an empty value, then... and Determine the first risk index of the second blade of the wind turbine. ,For example ; However, based on the acquired real-time startup acoustic signal, a real-time acoustic feature summary is extracted, and the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band is used to determine the second risk index of the blade based on the difference and a preset threshold. =91.

[0068] Finally, take the first risk index. Compared with the first risk index The median maximum value is used as the final first risk index. Then, based on the first risk index Second risk index =90 determines the control command, due to the second risk index. Since the current risk status index is 91, which is greater than the second preset threshold of 90, the current risk status index is 91. , If a shutdown-level action control command is generated, the edge computing unit of the UAV immediately sends a shutdown-level control command to the control system of the wind turbine through the communication module. The main control system of the wind turbine then executes the emergency shutdown procedure, thereby completing active intervention before unbalanced loads or low-temperature brittle fracture occur, avoiding serious failures that may be caused by environmental factors or optical factors.

[0069] This invention deploys the state assessment logic of wind turbine blades from the ground workstation to the edge computing unit of the inspection drone, so that the defect parameters obtained during the inspection process do not need to be transmitted back to the ground for analysis. Instead, they are calculated instantly by the drone. This transforms a passive decision-making process that relies on offline analysis of data transmission into a closed-loop response process that discovers, evaluates and actively triggers control commands on the orbit, thus solving the inherent contradiction between the timeliness of inspection information and the instantaneity of control decisions. This invention enables reliable operation on UAVs by extracting blade defect parameters and real-time acoustic feature summaries, thus avoiding dependence on high-performance, high-cost airborne processors. By using blade defect parameters and real-time environmental parameters such as temperature as the basis for judging the risk index, the calculation of the risk index is no longer a static lookup table, but a dynamic reflection of the actual impact of changes in the material physical properties of the blade, such as the ductile-brittle transition, on the current defect risk level under specific environments such as high and low temperatures. This makes the instructions received by the wind turbine control system based on physical principles and situational awareness diagnostic results, avoiding risk misjudgment caused by ignoring environmental factors.

[0070] Figure 2 This is a schematic diagram of an embodiment 200 of a wind turbine blade inspection device for a wind turbine generator according to the present invention. The device includes: The edge computing module 201 is used to acquire image information of wind turbine blades in high-altitude and cold regions, real-time environmental information of the blades, and real-time aeroacoustic signals of the blades collected by the UAV during flight inspection; determine the blade defect parameters of the wind turbine based on the blade image information; determine a first risk index of the blade based on the blade defect parameters and the real-time environmental information of the blades; determine a second risk index of the blade based on the real-time aeroacoustic signals of the blades; determine the current risk state index of the blade based on the first risk index and / or the second risk index; and generate action control commands for the wind turbine based on the current risk state index of the blades. The communication module 202 sends control commands to the control system of the wind turbine, causing the control system to execute corresponding actions according to the action control commands.

[0071] Optionally, based on the blade image information, the blade defect parameters of the wind turbine are determined, including: The blade image information is then analyzed to identify icing, cracks, and three-dimensional geometric dimensions, resulting in the identification results for icing, cracks, and three-dimensional geometric dimensions. Based on the identification results of icing, cracks, and three-dimensional geometry, the blade defect parameters of the wind turbine were determined.

[0072] Optionally, based on the blade defect parameters and the real-time environmental information of the blade, a first risk index for the blade is determined, including: Based on the blade defect parameters, the real-time environmental information of the blade, and multiple preset interpolation center points, multiple distances are determined; The weight of each distance is determined based on a preset kernel function and the multiple distances; The first risk index is determined based on the weights and the distance.

[0073] Optionally, a second risk index for the blade is determined based on the real-time aeroacoustic signals of the blade, including: The wind speed and blade pitch angle of the current operating condition of the blades are obtained from the control system of the wind turbine. Based on the wind speed and blade pitch angle under the current operating conditions, an acoustic feature summary is obtained by matching from a preset acoustic baseline library and used as a reference acoustic summary. The real-time start-up acoustic signal of the blades is converted to the frequency domain, and a real-time acoustic feature summary is extracted. Obtain the difference between the real-time acoustic feature summary and the reference acoustic summary in the target frequency band; The second risk index of the blade is determined based on the difference and the preset threshold.

[0074] Optionally, an acoustic feature summary is obtained from a preset acoustic baseline library based on the wind speed and blade pitch angle under the current operating conditions, serving as a baseline acoustic summary, including: The wind speed and blade pitch angle under the current operating conditions are quantified into query conditions. Based on the query criteria, an acoustic feature summary is obtained by matching the preset acoustic baseline library, which serves as the benchmark acoustic summary.

[0075] Optionally, a second risk index for the blade is determined based on the difference and a preset threshold, including: If the difference is greater than the first preset threshold, the difference is determined to be an acoustic feature variation; The second risk index of the blade is determined based on the changes in the acoustic characteristics.

[0076] Optionally, the current risk status index of the blade is determined based on the first risk index and / or the second risk index, including: If the acoustic feature change is greater than the second preset threshold, then the current risk status index of the blade is determined to be the second risk index. If the acoustic feature variation is less than or equal to the second preset threshold, the current risk status index of the blade is determined based on the first risk index and the second risk index.

[0077] Optionally, the wind turbine's operation control command is generated based on the current risk state index of the blade, including: If the current risk status index is greater than the first preset control threshold, a shutdown control command of the safety level is generated; If the current risk status index is less than the second preset control threshold, a safety control command for load reduction level is generated. Otherwise, generate motion control instructions for reducing the load level.

[0078] It should be noted that this device is the same as the method described above. All implementations in the above method embodiments are applicable to the embodiments of this device and can achieve the same technical effect.

[0079] In another aspect, the present invention also provides a computer-readable storage medium storing a program that, when executed by a processor, implements the method described in the present invention. All implementations in the above method embodiments are applicable to the embodiments of this computer-readable storage medium and can achieve the same technical effects.

[0080] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0081] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0082] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0083] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0084] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0085] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0086] Furthermore, it should be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof, which can be achieved by those skilled in the art using their basic programming skills after reading the description of the present invention. Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a known general-purpose device. The object of the present invention can also be achieved simply by providing a program product containing program code for implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that, in the apparatus and method of the present invention, it is obvious that the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered equivalent solutions of the present invention. Furthermore, the steps of performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0087] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method of inspecting a wind turbine blade of a wind power generator, characterized in that, The method comprises the following steps: acquiring image information of blades of a wind turbine in an alpine region collected by a UAV during flight inspection, real-time environmental information of the blades, and real-time aerodynamic acoustic signals of the blades; determining blade defect parameters of the wind turbine according to the image information of the blades; determining a first risk index of the blades according to the blade defect parameters and the real-time environmental information of the blades; determining a second risk index of the blades according to the real-time aerodynamic acoustic signals of the blades; determining a current risk state index of the blades according to the first risk index and / or the second risk index; generating an action control instruction of the wind turbine according to the current risk state index of the blades, and sending the action control instruction to a control system of the wind turbine, so that the control system performs corresponding actions according to the action control instruction.

2. The windmill blade inspection method of wind power generator according to claim 1, characterized in that, The method comprises the following steps: identifying icing, cracks, and three-dimensional geometric dimensions of the image information of the blades to obtain identification results of the icing, the cracks, and the three-dimensional geometric dimensions; determining the blade defect parameters of the wind turbine according to the identification results of the icing, the cracks, and the three-dimensional geometric dimensions.

3. The method of claim 1, wherein the method further comprises: The method comprises the following steps: determining a plurality of distances according to the blade defect parameters, the real-time environmental information of the blades, and a plurality of preset interpolation center points; determining a weight of each distance according to a preset kernel function and the plurality of distances; determining the first risk index according to the weight and the distance.

4. The method of claim 1, wherein the method further comprises: The method comprises the following steps: obtaining a wind speed and a blade pitch angle of a current working condition of the blades from a control system of the wind turbine; matching an acoustic feature abstract from a preset acoustic baseline library according to the wind speed and the blade pitch angle of the current working condition, as a reference acoustic abstract; converting the real-time starting acoustic signal of the blades to a frequency domain to extract a real-time acoustic feature abstract; obtaining a difference between the real-time acoustic feature abstract and the reference acoustic abstract in a target frequency band; determining the second risk index of the blades according to the difference and a preset threshold.

5. The method of claim 4, wherein, The method comprises the following steps: quantifying the wind speed and the blade pitch angle of the current working condition as a query condition; matching the acoustic feature abstract from the preset acoustic baseline library according to the query condition, as the reference acoustic abstract.

6. The method of claim 4, wherein the method further comprises: The method comprises the following steps: if the difference is greater than the first preset threshold, determining that the difference is an acoustic feature abnormality; determining the second risk index of the blades according to the acoustic feature abnormality.

7. The method of claim 6, wherein the method further comprises: The method comprises the following steps: if the acoustic feature abnormality is greater than a second preset threshold, determining that the current risk state index of the blades is the second risk index; if the acoustic feature abnormality is less than or equal to the second preset threshold, determining the current risk state index of the blades according to the first risk index and the second risk index.

8. The method of claim 1, wherein, According to the current risk state index of the blade, an action control instruction of the wind turbine is generated, including: If the current risk state index is greater than a first preset control threshold, a shutdown control instruction of a safety level is generated; If the current risk state index is less than a second preset control threshold, a safety control instruction of a load reduction level is generated; Otherwise, an action control instruction of a load reduction level is generated.

9. A wind turbine blade inspection device for a wind power generator, characterized in that, It comprises: An edge computing module is configured to acquire blade image information of a wind turbine in an alpine region collected by a UAV during flight inspection, real-time environmental information of the blade, and real-time aerodynamic acoustic signals of the blade; According to the blade image information, a blade defect parameter of the wind turbine is determined; According to the blade defect parameter and the real-time environmental information of the blade, a first risk index of the blade is determined; According to the real-time aerodynamic acoustic signals of the blade, a second risk index of the blade is determined; According to the first risk index and / or the second risk index, a current risk state index of the blade is determined; and according to the current risk state index of the blade, an action control instruction of the wind turbine is generated; A communication module is configured to send the control instruction to a control system of the wind turbine, so that the control system performs corresponding actions according to the action control instruction.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program, and the program is executed by the processor to implement the method in any one of claims 1 to 8.

Images