Deposited dust detecting and cleaning method for fan blade

Abstract

The invention relates to the technical field of fan blade maintenance, in particular to an accumulated dust detecting and cleaning method for a fan blade. The method comprises the steps of dividing a blade into a plurality of dust deposition sensitive areas based on chord length distribution and historical dust deposition data of the fan blade; constructing a reference database of fan blade ash deposition detection, and generating a reference parameter set including regional basic ash deposition parameters, environment associated ash deposition parameters and working condition associated ash deposition parameters; based on the reference parameter set, performing multi-dimensional data acquisition on the plurality of ash deposition sensitive areas to generate a sensing data set; and performing multi-stage judgment on the sensing data set and the reference parameter set to generate a dust deposition abnormity judgment result, and formulating a cleaning strategy and executing cleaning operation according to the dust deposition abnormity judgment result. According to the method, the accuracy of fan blade dust deposition detection and the high efficiency of cleaning operation can be improved, and stable and efficient operation of the fan is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fan blade maintenance, in particular to a dust detection and cleaning method for fan blades. BACKGROUND

[0002] In recent years, fans are increasingly widely used in the field of energy production. The dust on the surface of the blades directly affects the operation efficiency and service life of the fan. Therefore, dust detection and cleaning of the blades have become a key link in fan maintenance. However, the existing dust treatment scheme for the blades lacks targeted division of the dust sensitivity of different areas of the blades in the detection stage, and does not construct a dynamic benchmark combining historical data, real-time environment and working condition parameters, which is prone to detection errors and omissions. In the cleaning stage, a unified cleaning strategy is often used, which cannot dynamically adapt to the scheme according to the dust area and properties, resulting in difficulty in meeting the actual maintenance requirements in terms of the accuracy and efficiency of dust treatment. SUMMARY

[0003] The purpose of the present application is to solve the above technical problems. The present application provides a dust detection and cleaning method for fan blades, which aims to divide three dust sensitive areas near the root, near the maximum chord length and near the tip, construct a dynamic benchmark parameter set combining historical data and real-time environment and working condition parameters, realize multi-dimensional data acquisition and three-level judgment to accurately identify dust abnormalities, adapt a directional cleaning strategy from a multi-mode cleaning component library based on the abnormal results, dynamically adjust the cleaning parameters, simultaneously integrate energy recovery and dynamic anti-fouling to prevent dust, verify the cleaning quality through the cleaning effect, and realize overall management of multiple fans with the help of wind farm level collaborative optimization, so as to improve the accuracy of fan blade dust detection, the efficiency of cleaning operation and the systematicness of maintenance management, and ensure stable and efficient operation of the fan.

[0004] In some embodiments of the present application, a dust detection and cleaning method for fan blades is provided, comprising:

[0005] Dividing the fan blade into multiple dust sensitive areas based on the chord length distribution and historical dust data of the fan blade;

[0006] Constructing a benchmark database for dust detection of the fan blade to generate a benchmark parameter set containing regional basic dust parameters, environment-related dust parameters and working condition-related dust parameters;

[0007] Based on the benchmark parameter set, multi-dimensional data acquisition is performed on the multiple dust sensitive areas to generate a perception data set;

[0008] Multi-level judgment is performed on the perception data set and the benchmark parameter set to generate a dust abnormality determination result, and a cleaning strategy is developed and a cleaning operation is performed according to the dust abnormality determination result.

[0009] In some embodiments of the present application, when the fan blades are divided into multiple dusting sensitive regions based on the chord length distribution and historical dusting data of the fan blades, the method comprises:

[0010] Based on the design drawings of the fan blades, chord length data of the blades at each spanwise position is obtained to form a chord length distribution curve;

[0011] The dusting detection and maintenance records of the fan within a preset historical period are retrieved to extract historical dusting data including the location, frequency and severity of dusting occurrence;

[0012] The chord length distribution curve and the historical dusting data are analyzed for correlation to determine the blade sections with high frequency of historical dusting as high sensitive regions;

[0013] According to the results of the correlation analysis and in combination with the structural characteristics and aerodynamic characteristics of the blades, the fan blades are divided into three dusting sensitive regions in the spanwise direction.

[0014] In some embodiments of the present application, when the reference parameter set containing the regional basic dusting parameters, the environment-related dusting parameters and the working condition-related dusting parameters is generated, the method comprises:

[0015] The historical regional basic dusting parameters are analyzed to determine the initial dusting characteristic threshold values of each dusting sensitive region;

[0016] The historical environment-related dusting parameters and working condition-related dusting parameters are trained through a machine learning model to generate environment correction coefficients and working condition correction coefficients;

[0017] The initial dusting characteristic threshold values are weighted and corrected based on the environment correction coefficients and working condition correction coefficients to generate initial dynamic threshold values for each region;

[0018] The initial dynamic threshold values are dynamically updated and calibrated based on real-time environment-related dusting parameters and working condition-related dusting parameters to output comprehensive dynamic threshold values for the dusting sensitive regions.

[0019] In some embodiments of the present application, when the initial dynamic threshold values are dynamically updated and calibrated based on real-time environment-related dusting parameters and working condition-related dusting parameters to output comprehensive dynamic threshold values for the dusting sensitive regions, the method comprises:

[0020] The comprehensive dynamic threshold values are obtained from the following calculation formula:

[0021]

[0022] wherein, represents the comprehensive dynamic threshold value of the zth dusting sensitive region; represents the typical initial dusting characteristic threshold value of the zth dusting sensitive region; Indicates the environmental correction factor; This indicates the working condition correction factor; This is the weighting factor for the environmental correction factor. Let be the weighting coefficient of the working condition correction factor, and satisfy . + =1.

[0023] In some embodiments of this application, when performing multi-dimensional data acquisition on the plurality of dust-sensitive areas based on the benchmark parameter set to generate a perception dataset, the following are included:

[0024] For each dust-sensitive area, a first sensor group for collecting dust accumulation structural characteristic data and a second sensor group for collecting blade state response data are deployed.

[0025] The first sensor group is used to collect the dust accumulation thickness and density in the corresponding area to generate a subset of structural characteristic data.

[0026] The second sensor group is used to collect images of the blade surface, vibration spectrum and noise spectrum of the corresponding area, and generate a subset of state response data.

[0027] The perception dataset is generated by fusing the structural characteristic data subset with the state response data subset.

[0028] In some embodiments of this application, the step of performing multi-level judgments on the sensing dataset and the benchmark parameter set to generate a dust accumulation anomaly judgment result includes:

[0029] First-level judgment: Compare the real-time data in the perception dataset with the dynamic threshold range in the benchmark parameter set. If the data exceeds the threshold, it is marked as a preliminary abnormal area.

[0030] Second-level judgment: For the preliminary abnormal area, analyze whether its abnormal trend is consistent with the ash generation model constructed based on environmental-related ash accumulation parameters and working condition-related ash accumulation parameters;

[0031] If they match, proceed to the next level of judgment;

[0032] If they do not match, then it is ruled out as an abnormality of dust accumulation;

[0033] Third-level judgment: For the region confirmed by the second-level judgment, the structural characteristic data subset and the state response data subset are cross-validated;

[0034] If the verification results indicate that there is a coupling correlation between the structural characteristic data and the state response data, then the final ash accumulation anomaly determination result is generated.

[0035] If the verification results indicate that there is no coupling relationship between the structural characteristic data and the state response data, then it is excluded as a dust accumulation anomaly.

[0036] In some embodiments of this application, the step of formulating a cleaning strategy and performing cleaning operations based on the dust accumulation anomaly determination result includes:

[0037] Based on the ash accumulation anomaly determination results, the location of the anomaly area and the ash accumulation attribute features are extracted;

[0038] Based on a predefined cleaning strategy mapping model, several cleaning components are adapted for each abnormal area from a multi-mode cleaning component library, wherein the cleaning strategy mapping model defines adaptation rules based on the area location and dust accumulation attribute characteristics.

[0039] The operating parameters of the appropriate cleaning components are dynamically set according to the severity of dust accumulation.

[0040] Control the corresponding cleaning components to perform targeted cleaning operations.

[0041] Some embodiments of this application also include:

[0042] Energy recovery and dynamic antifouling include:

[0043] The mechanical energy generated by the vibration of the blades during operation is converted into electrical energy and stored by an energy harvesting device integrated on the blade surface.

[0044] Real-time monitoring of environmental dust concentration and blade surface coating status;

[0045] When the concentration of dust in the environment exceeds a preset threshold, the airflow guiding structure on the blades is controlled to perform anti-fouling operation.

[0046] When the coating condition on the blade surface is below the preset standard, the coating maintenance device is controlled to perform a coating repair operation.

[0047] In some embodiments of this application, a cleaning effect verification step is also included, which includes:

[0048] Within a preset time after the cleaning operation is completed, data is re-collected from the cleaned dust-sensitive areas to generate a cleaning verification dataset.

[0049] Double-validate the clean validation dataset with the benchmark parameter set:

[0050] Determine whether the key physical parameters characterizing the amount of ash accumulation have recovered to the dynamic threshold range of the corresponding region;

[0051] Determine whether the ash accumulation feature parameters of different dimensions conform to the preset logical consistency relationship;

[0052] If both of the dual verifications pass, it is determined that the current cleaning operation is qualified;

[0053] If any one of the verifications fails, analyze the reasons based on the cleaning verification dataset, re-determine the cleaning strategy, adjust the parameters, and perform the cleaning operation again.

[0054] In some embodiments of the present application, it further includes a wind farm-level collaborative optimization step, and this step includes:

[0055] Upload the ash accumulation abnormality determination result, cleaning operation record, and operation data of a single wind turbine to the wind farm cloud platform;

[0056] The wind farm cloud platform updates the ash accumulation feature knowledge base and optimizes the ash accumulation prediction model based on the data of multiple wind turbines;

[0057] Predict the blade ash accumulation trend in the future period based on the ash accumulation prediction model and real-time environmental data;

[0058] According to the prediction result, combined with the wind speed range and the unit working condition, recommend the cleaning time window to the corresponding wind turbine.

[0059] Compared with the prior art, the ash accumulation detection and cleaning method for wind turbine blades in the embodiments of the present application has the following beneficial effects:

[0060] The beneficial effects of the present application are mainly reflected in the accuracy and dynamic adaptability of detection. By dividing three ash accumulation sensitive areas, namely near the blade root, near the maximum chord length, and near the blade tip, constructing a dynamic reference parameter set by combining historical data with real-time environment and working condition parameters, and then through three-level judgment and multi-dimensional data cross-verification, the ash accumulation abnormality of the blade can be accurately identified, avoiding misjudgment and missed judgment caused by fixed thresholds or single data, and ensuring that the ash accumulation detection result conforms to the actual operation scenario.

[0061] At the same time, the present application realizes the high efficiency and full-process guarantee of the cleaning operation. Based on the ash accumulation abnormality determination result, the targeted cleaning strategy is adapted, and the cleaning components and parameters can be dynamically adjusted according to different regions and different ash accumulation attributes, reducing ineffective cleaning; the energy recovery and dynamic anti-fouling are combined to prevent ash accumulation in advance, the cleaning quality is ensured by combining the cleaning effect verification, and then the multi-wind turbine overall management is realized through the wind farm-level collaborative optimization, forming a closed loop from detection, cleaning to prevention and optimization, effectively ensuring the stable and efficient operation of the wind turbine. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic flowchart of an ash accumulation detection and cleaning method for wind turbine blades in a preferred embodiment of the embodiments of the present application. DETAILED DESCRIPTION

[0063] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0064] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0065] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0066] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0067] In some embodiments of this application, a method for detecting and cleaning dust accumulation on wind turbine blades is provided, including:

[0068] Based on the chord length distribution of the wind turbine blades and historical dust accumulation data, the blades are divided into multiple dust accumulation sensitive areas.

[0069] A benchmark database for detecting dust accumulation on wind turbine blades was constructed, and a benchmark parameter set containing regional basic dust accumulation parameters, environmentally related dust accumulation parameters, and operating condition-related dust accumulation parameters was generated.

[0070] Based on the benchmark parameter set, multi-dimensional data acquisition is performed on the multiple dust-sensitive areas to generate a perception dataset.

[0071] The sensing dataset and the benchmark parameter set are compared at multiple levels to generate a dust accumulation anomaly determination result. Based on the dust accumulation anomaly determination result, a cleaning strategy is formulated and a cleaning operation is performed.

[0072] In this embodiment, the determination result includes abnormal areas and dust accumulation type abnormalities.

[0073] In some embodiments of this application, when dividing the wind turbine blades into multiple dust-sensitive areas based on the chord length distribution and historical dust accumulation data, the following methods are included:

[0074] Based on the design drawings of the wind turbine blades, the chord length data of the blades at each spanwise position is obtained to form a chord length distribution curve;

[0075] Retrieve the dust accumulation detection and maintenance records of the fan within a preset historical period, and extract historical dust accumulation data on the location, frequency, and severity of dust accumulation.

[0076] By performing a correlation analysis between the chord length distribution curve and the historical ash accumulation data, the blade sections where historical ash accumulation occurs most frequently are identified as highly sensitive areas.

[0077] Based on the results of the correlation analysis, and combined with the structural characteristics and aerodynamic properties of the blades, the wind turbine blades are divided into three dust-sensitive areas in the spanwise direction.

[0078] In this embodiment, the three dust-sensitive areas include the area near the blade root, the area near the maximum chord length, and the area near the blade tip. The area near the blade root easily accumulates pollutants washed away by rainwater. The area near the maximum chord length is a critical region for aerodynamic performance; even slight dust accumulation significantly impacts efficiency. The area near the blade tip has high linear velocity and is strongly affected by scouring, but dust accumulation there greatly affects dynamic balance.

[0079] In some embodiments of this application, generating a reference parameter set that includes regional basic ash accumulation parameters, environmentally related ash accumulation parameters, and operating condition related ash accumulation parameters includes:

[0080] Analyze historical regional baseline ash accumulation parameters to determine the initial ash accumulation characteristic thresholds for each ash accumulation sensitive area;

[0081] The historical environmental and working condition-related ash accumulation parameters are used to generate environmental correction coefficients and working condition correction coefficients through machine learning model training.

[0082] The initial ash accumulation characteristic threshold is weighted and corrected based on the environmental correction coefficient and the operating condition correction coefficient to generate the initial dynamic threshold for each region.

[0083] Based on real-time environmental and operational condition-related dust accumulation parameters, the initial dynamic threshold is dynamically updated and calibrated to output a comprehensive dynamic threshold for dust-sensitive areas.

[0084] In this embodiment, the initial dust accumulation characteristic threshold refers to the initial reference value of the dust accumulation characteristic parameters extracted from historical data for three dust accumulation sensitive areas: near the leaf root, near the maximum chord length, and near the leaf tip. These values ​​represent the normal range of dust accumulation under standard environmental and operating conditions.

[0085] In this embodiment, a machine learning model can automatically predict the most reasonable environmental and operating condition correction coefficients based on the input environmental and operating condition parameters. First, a historical dataset is prepared for training the model. This dataset comes from the historical operation records of the wind turbines. Each set of data contains two main parts: one is the input features, namely the historical environmental-related dust accumulation parameters and the historical operating condition-related dust accumulation parameters; the other is the target labels, which are the analyzed values ​​of the ideal environmental and operating condition correction coefficients under the corresponding conditions. The initial values ​​of these target labels are analyzed and labeled by domain experts based on the experience of numerous historical cases. The prepared historical dataset is input into the model, and the model will autonomously learn and discover the correlation patterns between the environmental and operating condition parameters and the two correction coefficients through algorithms. When new, real-time environmental and operating condition parameters are input, the model will immediately calculate the applicable environmental and operating condition correction coefficients under the current conditions based on the previously learned mapping relationships.

[0086] In some embodiments of this application, when dynamically updating and calibrating the initial dynamic threshold based on real-time environmental-related dust accumulation parameters and operating condition-related dust accumulation parameters, and outputting the comprehensive dynamic threshold of the dust accumulation sensitive area, the following methods are included:

[0087] The overall dynamic threshold is obtained by the following formula:

[0088]

[0089] in, This represents the comprehensive dynamic threshold of the z-th ash-sensitive region; This represents the typical initial ash accumulation characteristic threshold of the z-th ash accumulation sensitive region; Indicates the environmental correction factor; This indicates the working condition correction factor; This is the weighting factor for the environmental correction factor. Let be the weighting coefficient of the working condition correction factor, and satisfy . + =1.

[0090] In this embodiment, the weighting coefficients α and β are obtained through training with historical data, reflecting the relative importance of environmental and operating parameters on the impact of dust accumulation.

[0091] In some embodiments of this application, when performing multi-dimensional data acquisition on the plurality of dust-sensitive areas based on the benchmark parameter set to generate a perception dataset, the following are included:

[0092] For each dust-sensitive area, a first sensor group for collecting dust accumulation structural characteristic data and a second sensor group for collecting blade state response data are deployed.

[0093] The first sensor group is used to collect the dust accumulation thickness and density in the corresponding area to generate a subset of structural characteristic data.

[0094] The second sensor group is used to collect images of the blade surface, vibration spectrum and noise spectrum of the corresponding area, and generate a subset of state response data.

[0095] The perception dataset is generated by fusing the structural characteristic data subset with the state response data subset.

[0096] In this embodiment, the ash accumulation structural characteristic data refers to data that directly describes the physical properties of the dust and contaminants adhering to the blade surface. This data is collected using a first sensor array installed on the blade surface. The raw data on ash accumulation thickness and density collected in real time by the first sensor array are then processed through signal conditioning and analog-to-digital conversion before being aggregated in the local processing unit to generate a subset of structural characteristic data.

[0097] In this embodiment, blade state response data refers to various indirect characteristic data of changes in the operating state of wind turbine blades caused by dust accumulation. This data is collected through a second sensor group. A multi-sensor system is deployed on the wind turbine nacelle. A high-frequency industrial camera captures optical images of the blade surface, and image recognition algorithms analyze changes in image texture and color depth to infer the distribution and severity of dust accumulation. Vibration acceleration sensors installed at the blade root continuously record the vibration spectrum during blade rotation, as dust accumulation alters the blade's mass distribution and aerodynamic shape, causing slight but detectable shifts in its characteristic vibration frequencies and amplitudes. Acoustic sensors arranged around the nacelle collect the noise spectrum during wind turbine operation. Dust accumulation disrupts the aerodynamic balance of the blades, generating specific eddy shedding noise, which is reflected in changes in the frequency spectrum. All these image, vibration spectrum, and noise spectrum data collected by different sensors, after preprocessing and feature extraction, constitute a subset of state response data. This dataset reflects the dynamic impact of dust accumulation on the blade's operating state.

[0098] In this embodiment, generating the sensing dataset includes: the system assigning a unified timestamp and spatial location label to each data point to ensure that the thickness data from the first sensor group and the data from the camera and vibration sensor describe the state of the same blade region at the same time; associating data from different dimensions through a data fusion algorithm; and integrating all aligned and associated multi-dimensional data into a structured data object, which is the sensing dataset. It contains raw sensor readings and preliminary correlations between data points, providing a data foundation for subsequent multi-level judgments of dust accumulation anomalies.

[0099] In some embodiments of this application, the step of performing multi-level judgments on the sensing dataset and the benchmark parameter set to generate a dust accumulation anomaly judgment result includes:

[0100] First-level judgment: Compare the real-time data in the perception dataset with the dynamic threshold range in the benchmark parameter set. If the data exceeds the threshold, it is marked as a preliminary abnormal area.

[0101] Second-level judgment: For the preliminary abnormal area, analyze whether its abnormal trend is consistent with the ash generation model constructed based on environmental-related ash accumulation parameters and working condition-related ash accumulation parameters;

[0102] If they match, proceed to the next level of judgment;

[0103] If they do not match, then it is ruled out as an abnormality of dust accumulation;

[0104] Third-level judgment: For the region confirmed by the second-level judgment, the structural characteristic data subset and the state response data subset are cross-validated;

[0105] If the verification results indicate that there is a coupling correlation between the structural characteristic data and the state response data, then the final ash accumulation anomaly determination result is generated.

[0106] If the verification results indicate that there is no coupling relationship between the structural characteristic data and the state response data, then it is excluded as a dust accumulation anomaly.

[0107] In this embodiment, the ash accumulation generation model is a built-in knowledge base that describes typical patterns of ash formation and growth under different environmental and operating condition combinations. The system retrieves historical environmental and operating condition-related ash accumulation parameters corresponding to the initial abnormal area during the abnormal occurrence period, and analyzes whether these parameters constitute a combination of conditions that easily induces ash accumulation. If the data analysis shows that the abnormal trend happens to occur during a period of high humidity, low wind speed, and high dust concentration in the air, then the system will determine that the abnormal trend is consistent with the prediction of the ash accumulation generation model. Through this causal correlation analysis, the system can effectively eliminate abnormal data points caused by instantaneous sensor failure or brief strong wind impacts, thereby marking those abnormal areas that are truly caused by ash accumulation and conform to physical laws as candidate abnormal areas.

[0108] In this embodiment, coupling correlation refers to the causal relationship between the aforementioned changes in physical properties and dynamic response characteristics. This includes performing correlation calculations between the physical property change data and dynamic response characteristic data of candidate anomaly regions. The system can not only detect anomalies but also confirm whether the anomaly is caused by actual accumulated dust, thus providing a reliable decision-making basis for subsequently initiating a precision cleaning strategy.

[0109] In some embodiments of this application, the step of formulating a cleaning strategy and performing cleaning operations based on the dust accumulation anomaly determination result includes:

[0110] Based on the ash accumulation anomaly determination results, the location of the anomaly area and the ash accumulation attribute features are extracted;

[0111] Based on a predefined cleaning strategy mapping model, several cleaning components are adapted for each abnormal area from a multi-mode cleaning component library, wherein the cleaning strategy mapping model defines adaptation rules based on the area location and dust accumulation attribute characteristics.

[0112] The operating parameters of the appropriate cleaning components are dynamically set according to the severity of dust accumulation.

[0113] Control the corresponding cleaning components to perform targeted cleaning operations.

[0114] In this embodiment, the mapping model includes the following rules: the area near the leaf root is preferentially adapted to the pneumatic purging assembly; the area near the maximum chord length is preferentially adapted to the flexible mechanical wiping assembly; and the area near the leaf tip is preferentially adapted to the flexible vibration cleaning assembly.

[0115] In this embodiment, the cleaning component library includes a pneumatic blowing component for handling loose dust, a flexible mechanical wiping component for handling adhesive dust, a low-temperature phase change cleaning component for handling stubborn dust, and a flexible vibration cleaning component for handling thin layers of dust in the blade tip region.

[0116] In this embodiment, the plurality includes at least one.

[0117] In some embodiments of this application, the method for detecting and cleaning dust accumulation on wind turbine blades further includes:

[0118] Energy recovery and dynamic antifouling include:

[0119] The mechanical energy generated by the vibration of the blades during operation is converted into electrical energy and stored by an energy harvesting device integrated on the blade surface.

[0120] Real-time monitoring of environmental dust concentration and blade surface coating status;

[0121] When the concentration of dust in the environment exceeds a preset threshold, the airflow guiding structure on the blades is controlled to perform anti-fouling operation.

[0122] When the coating condition on the blade surface is below the preset standard, the coating maintenance device is controlled to perform a coating repair operation.

[0123] In this embodiment, the core idea behind the anti-fouling operation of the airflow guiding structure is to fundamentally reduce the adhesion of dust and pollutants to the blades by actively altering the airflow path on the blade surface under adverse weather conditions. This includes embedding a row of miniature airflow guide vanes in key areas such as the leading edge of the wind turbine blades; these components constitute the airflow guiding structure. Each miniature airflow guide vane contains a miniature actuator that receives electrical signals from the control system. When the system detects in real-time that the concentration of dust in the environment exceeds a preset threshold via environmental sensors, it indicates a high level of particulate matter in the air, which easily leads to severe dust accumulation on the blade surface. At this point, the control system immediately sends a command to these airflow guide vanes, causing them to extend precisely from inside the blade to a specific height, thereby creating a small, controllable aerodynamic turbulence at the leading edge of the blade. This turbulence effectively guides the main airflow, lifting most of the dust particles and blowing them away from the blade surface.

[0124] In this embodiment, controlling the coating maintenance device to perform coating repair operations is a maintenance technology for functional coatings on blade surfaces. It includes: the blade surface being pre-sprayed with a smart microcapsule coating containing a repair agent. Distributed coating condition monitoring sensors are integrated within the blade substrate. When the system determines, based on monitoring data, that the blade surface coating condition has fallen below a preset standard due to long-term wear or ultraviolet radiation, the control system activates the coating maintenance device. The control system applies specific external stimuli to the degraded coating area, which causes the internally sealed repair agent liquid to be released. The repair agent automatically levels and covers the damaged coating surface, restoring its original hydrophobic and smooth properties. This process ensures that the blade surface is always in a low-adhesion state, improving its resistance to dust accumulation at the material level.

[0125] In some embodiments of the present application, a cleaning effect verification step is further included, and this step includes:

[0126] Within a preset time after the cleaning operation is completed, data resampling is performed on the cleaned dust-sensitive area to generate a cleaning verification data set;

[0127] The cleaning verification data set is double-verified with the reference parameter set:

[0128] Judge whether the key physical parameters representing the dust accumulation amount return to the dynamic threshold range of the corresponding area;

[0129] Judge whether the dust accumulation characteristic parameters in different dimensions conform to the preset logical consistency relationship;

[0130] If both verifications pass, it is determined that the current cleaning operation is qualified;

[0131] If any verification fails, analyze the reasons based on the cleaning verification data set, re-determine the cleaning strategy and adjust the parameters, and then perform the cleaning operation again.

[0132] In this embodiment, the key physical parameters refer to: the dust accumulation thickness, which directly measures the thickness of the dust layer adhering to the blade surface through the first sensor group; the dust accumulation density, which estimates the compactness of the dust accumulation through sensor data analysis or in combination with other measurement means. The dust accumulation with a large thickness but a small density and the dust accumulation with a small thickness but a large density have different effects on the blade; the image characteristics of the blade surface, which takes pictures of the blade surface through the first sensor group and analyzes the changes in texture roughness, color depth, and reflectivity characteristics using image recognition algorithms. Dust accumulation will cause the surface texture to become rough, the color to become darker, and the reflectivity to decrease; the vibration spectrum characteristics, which collect data through the vibration acceleration sensor installed at the root of the blade. Dust accumulation will change the mass distribution and aerodynamic shape of the blade, resulting in a drift in its inherent vibration frequency, an increase in amplitude, or the appearance of new harmonic components in the spectrum; the noise spectrum characteristics, which collect the sound during the operation of the fan through the acoustic sensor. Dust accumulation destroys the aerodynamic balance and generates specific vortex shedding noise, and these abnormal sounds will be reflected in the increased energy in a specific frequency band of the noise spectrum.

[0133] In this embodiment, the operation of determining whether the key physical parameters characterizing the amount of dust accumulation have recovered to the dynamic threshold range of the corresponding area includes: after a preset time following the completion of the cleaning operation, the system will activate the sensors again on the dust-sensitive area that was just cleaned to collect a new set of data. This set of data is called the cleaning verification dataset, which includes key physical parameters such as new dust accumulation thickness and new vibration spectrum; the system retrieves the comprehensive dynamic threshold of the corresponding area from the latest benchmark parameter set. The system compares each key physical parameter in the cleaning verification dataset with the corresponding dynamic threshold range in the benchmark parameter set; and generates a verification result: if all monitored key physical parameters have recovered to their respective dynamic threshold ranges, the system determines that the cleaning operation is qualified; if any one or more key physical parameters still exceed their dynamic threshold ranges, the system determines that the cleaning effect is unsatisfactory and triggers a subsequent re-cleaning process.

[0134] In this embodiment, a preset logical consistency relationship is used for cross-validation between data from different dimensions to ensure the accuracy of the judgment and prevent misjudgments. This includes:

[0135] 1. Logical Relationship Between Thickness and Vibration: The default logic is that increased dust accumulation leads to a change in the blade's mass distribution, which should cause a decrease in a specific vibration frequency. The system checks the data after cleaning. If the thickness parameter has returned to normal, but the vibration frequency is still low, this may indicate incomplete cleaning or other mechanical malfunctions. Conversely, if both the thickness and vibration frequency return to normal, then the logic is consistent.

[0136] 2. Logical Relationship Between Image and Thickness: The default logic is that optical image analysis shows a region with dark color and rough texture, leading to a higher measured dust thickness in that region. After cleaning, image analysis shows the surface has become bright and smooth, while the thickness sensor reading is close to zero. This indicates that the image data and thickness data are logically consistent, jointly proving the cleaning was effective.

[0137] 3. Logical Relationship Between Location and Type: The default logic is that the area near the leaf root is more prone to accumulating viscous, wet, and heavy contaminants due to lower centrifugal force, resulting in a typically higher dust density. If, during cleaning effectiveness verification, the system detects that the dust density measurement in the leaf root area remains high, even if the thickness meets the standard, an alarm will be triggered because this is inconsistent with the typical dust accumulation characteristics of that area, potentially requiring an investigation into the appropriateness of the cleaning strategy.

[0138] The system not only checks whether each parameter meets the standard individually, but also checks whether the relationship between these parameters conforms to the physical laws of dust accumulation and removal. This greatly improves the reliability and intelligence of the entire system, avoiding misjudgments caused by single sensor failures or data interference.

[0139] Some embodiments of this application also include a wind farm-level collaborative optimization step, which includes:

[0140] The results of dust accumulation anomaly assessment, cleaning operation records, and operation data of a single wind turbine are uploaded to the wind farm cloud platform.

[0141] The wind farm cloud platform updates the ash accumulation feature knowledge base and optimizes the ash accumulation prediction model based on data from multiple wind turbines.

[0142] Based on the ash accumulation prediction model and real-time environmental data, the leaf ash accumulation trend is predicted in the future period.

[0143] Based on the forecast results, combined with the wind speed range and unit operating conditions, the recommended cleaning time window is for the corresponding wind turbine.

[0144] In this embodiment, the wind farm cloud platform serves as the intelligent hub of the wind farm cluster, possessing continuous self-learning and collaborative optimization capabilities. Its core operation involves dynamically updating the ash accumulation feature knowledge base and optimizing the ash accumulation prediction model based on real-time and historical data uploaded by multiple wind turbines. Specifically, the ash accumulation feature knowledge base is a case library storing complete features of past ash accumulation events. Each time a single wind turbine completes a cleaning cycle, its successful experience is stored as a new knowledge entry, thereby optimizing operational experience within the wind farm. Simultaneously, the cloud platform runs a machine learning-based ash accumulation prediction model. This model takes future weather forecasts, wind turbine operation plans, and historical patterns from the knowledge base as input, and outputs predictions of the probability and severity of ash accumulation risk for each wind turbine blade in the future. The model periodically self-optimizes by comparing the deviation between the prediction results and the actual situation, thereby continuously improving its prediction accuracy.

[0145] In this embodiment, based on the predictive model, the cloud platform recommends a cleaning time window for each wind turbine. This window is an optimal time period that integrates dust accumulation risk, wind speed range, power generation plan, and weather conditions. For example, the platform recommends performing cleaning at night during periods of low wind speed and low grid load, which can both utilize the best wind speed conditions to ensure cleaning effectiveness and minimize power generation losses caused by downtime for cleaning.

[0146] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. A method for detecting and cleaning dust accumulation on wind turbine blades, comprising: Based on the chord length distribution of the wind turbine blades and historical dust accumulation data, the blades are divided into multiple dust accumulation sensitive areas. A benchmark database for detecting dust accumulation on wind turbine blades was constructed, and a benchmark parameter set containing regional basic dust accumulation parameters, environmentally related dust accumulation parameters, and operating condition-related dust accumulation parameters was generated. Based on the benchmark parameter set, multi-dimensional data acquisition is performed on the multiple dust-sensitive areas to generate a perception dataset. The sensing dataset and the benchmark parameter set are compared at multiple levels to generate a dust accumulation anomaly determination result. Based on the dust accumulation anomaly determination result, a cleaning strategy is formulated and a cleaning operation is performed.

2. The method for detecting and cleaning dust accumulation on wind turbine blades according to claim 1, characterized in that, Based on the chord length distribution of wind turbine blades and historical dust accumulation data, when dividing the blades into multiple dust accumulation sensitive areas, these include: Based on the design drawings of the wind turbine blades, the chord length data of the blades at each spanwise position is obtained to form a chord length distribution curve; Retrieve the dust accumulation detection and maintenance records of the fan within a preset historical period, and extract historical dust accumulation data on the location, frequency, and severity of dust accumulation. By performing a correlation analysis between the chord length distribution curve and the historical ash accumulation data, the blade sections where historical ash accumulation occurs most frequently are identified as highly sensitive areas. Based on the results of the correlation analysis, and combined with the structural characteristics and aerodynamic properties of the blades, the wind turbine blades are divided into three dust-sensitive areas in the spanwise direction.

3. The method for detecting and cleaning dust accumulation on wind turbine blades according to claim 2, characterized in that, When generating the benchmark parameter set, which includes regional basic ash accumulation parameters, environmentally related ash accumulation parameters, and operating condition related ash accumulation parameters, the following steps are included: Analyze historical regional baseline ash accumulation parameters to determine the initial ash accumulation characteristic thresholds for each ash accumulation sensitive area; The historical environmental and working condition-related ash accumulation parameters are used to generate environmental correction coefficients and working condition correction coefficients through machine learning model training. The initial ash accumulation characteristic threshold is weighted and corrected based on the environmental correction coefficient and the operating condition correction coefficient to generate the initial dynamic threshold for each region. Based on real-time environmental and operational condition-related dust accumulation parameters, the initial dynamic threshold is dynamically updated and calibrated to output a comprehensive dynamic threshold for dust-sensitive areas.

4. The method according to claim 3, characterized in that, Based on real-time environmental and operational condition-related dust accumulation parameters, the initial dynamic threshold is dynamically updated and calibrated. When outputting the comprehensive dynamic threshold for the dust-sensitive area, the following steps are included: The overall dynamic threshold is obtained by the following formula: ； in, This represents the comprehensive dynamic threshold of the z-th ash-sensitive region; This represents the typical initial ash accumulation characteristic threshold of the z-th ash accumulation sensitive region; Indicates the environmental correction factor; This indicates the working condition correction factor; This is the weighting factor for the environmental correction factor. Let be the weighting coefficient of the working condition correction factor, and satisfy . + =1.

5. The method for detecting and cleaning dust accumulation on wind turbine blades as described in claim 4, characterized in that, Based on the aforementioned benchmark parameter set, multi-dimensional data acquisition is performed on the multiple dust-sensitive areas to generate a perception dataset, including: For each dust-sensitive area, a first sensor group for collecting dust accumulation structural characteristic data and a second sensor group for collecting blade state response data are deployed. The first sensor group is used to collect the dust accumulation thickness and density in the corresponding area to generate a subset of structural characteristic data. The second sensor group is used to collect images of the blade surface, vibration spectrum and noise spectrum of the corresponding area, and generate a subset of state response data. The perception dataset is generated by fusing the structural characteristic data subset with the state response data subset.

6. The method for detecting and cleaning dust accumulation on wind turbine blades as described in claim 5, characterized in that, When performing multi-level judgments on the sensing dataset and the benchmark parameter set to generate a dust accumulation anomaly judgment result, the following steps are included: First-level judgment: Compare the real-time data in the perception dataset with the dynamic threshold range in the reference parameter set. If the data exceeds the threshold, mark it as a preliminary abnormal area. Second-level judgment: For the preliminary abnormal area, analyze whether its abnormal trend conforms to the ash deposition generation model constructed based on the environment-related ash deposition parameters and the working condition-related ash deposition parameters. If it conforms, proceed to the next-level judgment. If it does not conform, exclude it as an ash deposition abnormality. Third-level judgment: For the area confirmed by the second-level judgment, cross-verify the structural characteristic data subset and the state response data subset. If the verification result shows that there is a coupling correlation between the structural characteristic data and the state response data, finally generate an ash deposition abnormality determination result. If the verification result shows that there is no coupling correlation between the structural characteristic data and the state response data, exclude it as an ash deposition abnormality.

7. The method for detecting and cleaning dust accumulation on wind turbine blades as described in claim 1, characterized in that, When formulating a cleaning strategy and performing a cleaning operation according to the ash deposition abnormality determination result, it includes: Based on the ash deposition abnormality determination result, extract the position of the abnormal area and the ash deposition attribute characteristics. According to the predefined cleaning strategy mapping model, adapt several cleaning components for each abnormal area from the multi-mode cleaning component library, where the cleaning strategy mapping model defines the adaptation rules based on the area position and the ash deposition attribute characteristics. Dynamically set the operating parameters of the adapted cleaning components according to the severity of the ash deposition. Control the corresponding cleaning components to perform directional cleaning operations.

8. The ash deposition detection and cleaning method for a fan blade according to claim 1, further comprising: Energy recovery and dynamic anti-fouling, including: ​ ​ ​ ​ 9. The method for detecting and cleaning dust accumulation on wind turbine blades as described in claim 1, characterized in that, ​ ​ ​ ​ ​ ​ ​ 10. The method for detecting and cleaning dust accumulation on wind turbine blades as described in claim 1, characterized in that, ​ ​ ​ ​ Based on the forecast results, combined with the wind speed range and unit operating conditions, the recommended cleaning time window is for the corresponding wind turbine.

Images