Methods and electronic equipment for assessing electrical charge loss due to corrosion at the leading edge of wind turbine blades

Abstract

This application provides a method and electronic device for assessing power loss due to leading-edge corrosion of wind turbine blades. The method involves acquiring a pre-defined wind turbine blade feature-operating condition association library, collecting at least two target wind turbine blade features based on these features, determining the target operating condition corresponding to each feature in the library, and if the target operating condition indicates effective leading-edge corrosion, determining a target power correction coefficient in a pre-defined wind turbine blade feature-power correction coefficient mapping table based on the target blade features. Finally, the method calculates the cumulative power generation loss over a pre-defined period based on the target power correction coefficient. This application can automatically determine the cumulative power generation loss corresponding to the current target wind turbine feature, providing an intuitive reference for subsequent operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of wind turbine technology, and in particular to a method and electronic equipment for assessing electrical power loss due to corrosion at the leading edge of wind turbine blades. Background Technology

[0002] Calculating the power loss caused by leading-edge corrosion of wind turbine blades can quantify various anomalies into actual economic losses, providing a basis for decision-making to balance maintenance costs and power generation revenue. It can also objectively verify the actual effectiveness of different anti-corrosion technologies. To obtain accurate power loss assessment results, it is necessary to ensure accurate detection of leading-edge corrosion of wind turbine blades.

[0003] Currently, the industry's detection and assessment methods for wind turbine blade leading edge corrosion mainly rely on a single physical quantity and usually require manual intervention, resulting in problems such as a single information source and weak anti-interference ability. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and electronic equipment for assessing the electrical loss due to corrosion at the leading edge of wind turbine blades, which can improve the accuracy of assessing the electrical loss due to corrosion at the leading edge of wind turbine blades.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for assessing electrical charge loss due to corrosion at the leading edge of wind turbine blades, comprising: Obtain a preset wind turbine blade feature-operating condition association library, and collect at least two target wind turbine blade features based on the wind turbine blade features in the wind turbine blade feature-operating condition association library; Determine the target operating condition corresponding to the target wind turbine blade feature in the wind turbine blade feature-operating condition association library; If the target operating condition is effective for leading edge corrosion, then the target power correction coefficient is determined in the preset wind turbine blade characteristic-power correction coefficient mapping table according to the target wind turbine blade characteristics. The cumulative power generation loss value within the preset period is calculated based on the target power correction coefficient.

[0006] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows: A terminal for assessing electrical loss due to leading-edge corrosion of wind turbine blades includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor, when executing the computer program, implements each step of the aforementioned method for assessing electrical loss due to leading-edge corrosion of wind turbine blades.

[0007] The beneficial effects of this invention are as follows: It establishes a pre-defined association library of wind turbine blade features and operating conditions. Based on the wind turbine blade features relied upon for judging the operating condition, it collects at least two corresponding target wind turbine blade features, avoiding the problem of relying on a single feature for judgment and failing to obtain accurate operating condition judgment results when a single feature collection is abnormal. The operating conditions identified by multiple wind turbine blade features can corroborate each other. Only when the target operating condition corresponding to the target wind turbine blade feature is effective due to leading-edge corrosion is the cumulative power generation loss further calculated, ensuring the accuracy of the final calculated cumulative power generation loss and avoiding invalid calculation processes. The cumulative power generation loss value is calculated by matching the target power correction coefficient corresponding to the collected target wind turbine blade features with the pre-calibrated mapping relationship between wind turbine blade features and power correction coefficients. Through actual calibration, in subsequent calculations, only the corresponding target power correction coefficient needs to be directly obtained, without having to determine the target power correction coefficient separately for each calculation, improving calculation efficiency. Furthermore, it can automatically determine the cumulative power generation loss value corresponding to the current target wind turbine feature, providing an intuitive reference for subsequent operation and maintenance. Attached Figure Description

[0008] Figure 1 This is a flowchart of a method for assessing electrical charge loss due to corrosion at the leading edge of a wind turbine blade, according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the implementation of a wind turbine blade leading-edge corrosion electrical loss assessment method in a specific scenario, according to an embodiment of the present invention. Figure 3 This is a timing diagram illustrating the implementation of a wind turbine blade leading-edge corrosion electrical loss assessment method in a specific scenario, according to an embodiment of the present invention. Figure 4 This is a schematic diagram of a terminal for assessing electrical loss due to corrosion at the leading edge of a wind turbine blade, according to an embodiment of the present invention. Figure 5 This is an architectural diagram of a terminal for assessing electrical power loss due to corrosion at the leading edge of a wind turbine blade, according to an embodiment of the present invention. Detailed Implementation

[0009] Definitions:

[0010] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0011] In existing technologies, the following methods are commonly used to detect the degree of leading-edge corrosion of wind turbine blades: 1. Manual interpretation of visible light images: Photos of the blade leading edge are taken using aerial work platforms or drones, and professionals classify the corrosion into slight, moderate, and severe levels based on experience. 2. Single infrared thermal imaging detection: The temperature distribution difference on the blade surface during shutdown or operation is used to identify the local temperature rise caused by changes in the turbulent boundary layer due to corrosion. 3. Single vibration signal monitoring: Accelerometers installed in the nacelle or tower are used to analyze the changes in the characteristic frequency of the overall turbine vibration caused by aerodynamic imbalance of the blades. 4. Power loss estimation based on theoretical models: The theoretical power generation reduction is estimated by combining the relationship curve between blade surface roughness and lift / drag coefficient measured in the laboratory with wind speed data. All of these methods judge the state of the wind turbine through a single data source and fail to establish a closed-loop logic from "multi-source physical field anomalies" to "specific economic losses in power generation" and then to "triggering of farm-level operation and maintenance actions," affecting the accuracy of estimating power generation losses due to leading-edge corrosion of wind turbines.

[0012] It is evident that the relevant technologies primarily rely on qualitative judgments and lack quantitative correlation. While they can detect increased roughness at the leading edge of wind turbine blades, they cannot directly translate this change in physical state into power generation loss (kWh), the economic indicator most important to operation and maintenance managers.

[0013] Insufficient anti-interference capability and high false alarm rate. A single visible light image can easily misjudge mud, sand, and insect stains attached to the leading edge as substrate corrosion; a single infrared thermal image is easily affected by ambient temperature and sunlight angle; a single vibration signal is difficult to distinguish between corrosion and transient response caused by gusts of wind.

[0014] Data utilization is fragmented. Even when image and vibration sensors are installed simultaneously, the data is analyzed independently, failing to achieve collaborative verification across physical fields and time scales.

[0015] To address at least the aforementioned issues, this application collects features from at least two target wind turbine blades to determine the current operating condition of the turbine. Only when the leading-edge corrosion is effective is the cumulative power generation loss calculated. Furthermore, during the calculation of cumulative power generation loss, the correspondence between wind turbine blade features and power correction coefficients is pre-calibrated. In actual calculations, the corresponding target power correction coefficients are directly matched based on the target wind turbine blade features. This approach reduces computational resource consumption during the calculation of cumulative power generation loss and improves the accuracy of operating condition determination through mutual verification of operating conditions among multiple target wind turbine blade features, thereby enhancing the accuracy of the cumulative power generation loss calculation.

[0016] The following details a method for assessing electrical charge loss due to corrosion at the leading edge of wind turbine blades, as described in this invention. Please refer to [link / reference]. Figure 1The method 100 includes steps 110 to 140.

[0017] Step 110: Obtain a preset wind turbine blade feature-operating condition association library. Based on the wind turbine blade features in the library, collect at least two target wind turbine blade features. For example, Table 1 provides a simplified illustration of the wind turbine blade feature-operating condition association library, which provides a condition → conclusion mapping knowledge base. The leading edge texture complexity, heat diffusion anomaly area, and side lobe energy fluctuations are wind turbine blade features, and these three features are also collected when collecting the target wind turbine blade features. Wind turbine blade features include at least leading edge texture complexity, heat diffusion anomaly area, and side lobe energy fluctuations, and may also include point defect coverage, temperature difference between the local heat diffusion anomaly area and the main body temperature of the wind turbine blade, etc.

[0018] Table 1

[0019] Referring to Table 1 above, when only dry mud and sand adhere to the leading edge of the blade, the visible light texture complexity increases, but because the aerodynamic shape of the blade surface is not damaged, the infrared temperature difference characteristics and vibration sidelobe energy characteristics do not change significantly. The wind turbine blade is determined to be in an effective leading edge corrosion state only when at least two of the following parameters—texture complexity, abnormal area of ​​thermal diffusion, and sidelobe energy fluctuations—exceed the preset co-triggered threshold. If only a single sensor's monitoring value is abnormal, it is determined to be due to environmental interference, surface contamination, or operating condition fluctuations, and the original monitoring state is maintained.

[0020] Step 120: Determine the target operating condition corresponding to the target wind turbine blade feature in the wind turbine blade feature-operating condition association library. For example, referring to the illustration in Table 1 above, if the target wind turbine blade feature is texture complexity, thermal diffusion anomaly area, and side lobe energy fluctuation, compare the acquired values ​​of the target wind turbine blade feature with the set thresholds respectively. If all are greater than the threshold, the corresponding target operating condition is effective leading edge corrosion.

[0021] Step 130: If the target operating condition is effective for leading-edge corrosion, then determine the target power correction coefficient based on the target wind turbine blade characteristics in a preset wind turbine blade characteristic-power correction coefficient mapping table. For example, the power correction coefficient is obtained by pre-calibrating the difference between the power under different wind turbine blade characteristics with corrosion and the power of the wind turbine without corrosion through experiments, and power correction coefficients corresponding to different wind turbine blade characteristics are established.

[0022] Step 140: Calculate the cumulative power generation loss value within a preset period based on the target power correction coefficient. For example, the cumulative power generation loss value of the target wind turbine blade characteristics within a preset period can be calculated based on the power correction coefficient corresponding to the target wind turbine blade characteristics over a future period, providing an intuitive data reference for subsequent operation and maintenance.

[0023] As described above, a pre-defined database linking wind turbine blade features and operating conditions is used. At least two target wind turbine blade features are collected based on the blade features used to determine the operating condition. This avoids relying on a single feature for judgment, which can lead to inaccurate operating condition assessments when a single feature fails to capture. Operating conditions identified by multiple wind turbine blade features can corroborate each other. Only when the target operating condition corresponding to the target wind turbine blade feature is confirmed to have effective leading-edge corrosion is the cumulative power generation loss calculated. This ensures the accuracy of the final calculated cumulative power generation loss, effectively distinguishing between leading-edge fouling and actual corrosion, avoiding unnecessary shutdowns for inspection or maintenance due to misjudgment, and preventing ineffective calculations. The cumulative power generation loss value is calculated by matching the target power correction coefficient corresponding to the collected target wind turbine blade features with a pre-calibrated mapping relationship between wind turbine blade features and power correction coefficients. Through actual calibration, subsequent calculations only require directly obtaining the corresponding target power correction coefficient, eliminating the need to determine the target power correction coefficient separately for each calculation. This improves calculation efficiency and automatically determines the cumulative power generation loss value corresponding to the current target wind turbine feature, providing an intuitive reference for subsequent operation and maintenance.

[0024] In one optional implementation, step 110 includes collecting at least two target wind turbine blade features based on the wind turbine blade features in the wind turbine blade feature-operating condition association library.

[0025] Step 1110: Collect the characteristics of the target wind turbine blades using at least two of the visible light imaging unit, infrared thermal imaging unit, and vibration monitoring unit. For example, the visible light imaging unit can be a separate optical camera, mounted on a drone, a wall-climbing robot, a ground-based telephoto gimbal, or deployed on top of an offshore substation. The gimbal controls the camera to align with each turbine within the wind farm, capturing high-definition images of the wind turbine blades. The infrared thermal imaging unit can be a handheld infrared thermal imager mounted on a drone, or an infrared thermal imager installed on the wind turbine. It can also be coaxially deployed with the visible light imaging unit on the same gimbal at the substation. When the visible light imaging unit and the infrared thermal imaging unit are simultaneously mounted on the same drone or coaxially deployed on the same gimbal, it facilitates calibration of the same area on the wind turbine blade in the images captured by the visible light imaging unit and the infrared thermal imaging unit. The vibration monitoring unit can be a piezoelectric accelerometer installed on the wind turbine or a fiber optic strain sensor mounted on the surface of the wind turbine blades, or other devices capable of vibration monitoring. Visible light imaging unit, infrared thermal imaging unit, and vibration monitoring unit (collectively referred to as sensor unit) can also carry a unified timestamp tag to record the data acquisition time, enabling the acquisition of data collected by various sensor units at the same time, thereby comprehensively judging the working condition at the corresponding time.

[0026] As described above, by acquiring the characteristics of the target wind turbine blades through various detection methods, a foundation for multi-source feature fusion to judge the operating conditions of the wind turbine blades is established. This allows the system to eliminate single environmental interferences. For example, changes in light can increase the error of the visible light imaging unit, dirt can affect the imaging effect of the visible light imaging unit, and gusts of wind can affect the detection data of the vibration monitoring unit. The characteristics of the target wind turbine blades obtained from multiple data sources can be mutually verified to improve the accuracy of the operating condition confirmation results.

[0027] In one optional implementation, step 110, which involves collecting at least two target wind turbine blade features based on the wind turbine blade features in the wind turbine blade feature-operating condition association library, includes steps 1121 to 1123.

[0028] Step 1121: Acquire the leading edge texture complexity or point defect coverage of the wind turbine blade using a visible light imaging unit. The visible light imaging unit acquires high-resolution digital images of the leading edge region of the wind turbine blade and extracts the leading edge texture complexity or point defect coverage reflecting the microscopic geometric unevenness of the surface. Point defects refer to surface damage on the leading edge surface caused by corrosion, which are discretely distributed micro-pits, pinholes, or pits.

[0029] For example, the steps for obtaining the leading edge texture complexity are as follows: (1) Convert the visible light image of the leading edge region of the wind turbine blade acquired by the visible light imaging unit into a grayscale image, and divide the grayscale image into several image sub-blocks of equal size. (2) For each image sub-block, according to the preset pixel spacing and direction, count the frequency of grayscale value combinations of pixel pairs in the image sub-block, and construct a grayscale co-occurrence matrix. (3) Based on the grayscale co-occurrence matrix of each image sub-block, calculate the statistical feature values ​​reflecting the texture roughness, including contrast, correlation and entropy. Among them, contrast reflects the total amount of local grayscale changes in the image, correlation reflects the similarity of grayscale in the row or column direction, and entropy reflects the randomness of grayscale distribution in the image. The rougher the surface of the leading edge of the wind turbine blade, the greater the grayscale difference between adjacent pixels in the image, the higher the calculated contrast and entropy values, the lower the correlation, and the greater the leading edge texture complexity parameter. (4) Take a weighted average of the statistical feature values ​​of all sub-blocks to obtain the leading edge texture complexity of the entire image. The rougher the leading edge surface of the blade, the greater the grayscale difference between adjacent pixels in the image, the higher the calculated contrast and entropy values, the lower the correlation, and the greater the complexity of the leading edge texture.

[0030] For example, the steps to obtain the point defect coverage rate are as follows: perform binarization segmentation on the visible light image, identify the pixel connected regions in the leading edge region of the wind turbine blade whose gray value is lower than the gray value threshold of the surrounding normal surface, and calculate the percentage of the total area of ​​all connected regions to the total area of ​​the monitoring area at the leading edge of the wind turbine blade as the point defect coverage rate.

[0031] Step 1122: Acquire the temperature difference between the localized abnormal heat diffusion area of ​​the wind turbine blade and the main body temperature of the wind turbine blade using an infrared thermal imaging unit. For example, obtain the temperature field distribution along the leading edge span of the wind turbine blade during operation, and extract the area of ​​the localized abnormal heat diffusion area caused by premature airflow separation and the amplitude of the temperature difference between the area of ​​the localized abnormal heat diffusion area and the main body temperature of the blade.

[0032] Step 1123: Collect the sidelobe energy fluctuation characteristics of the high-order harmonics related to the rotor frequency of the wind turbine blades using the vibration monitoring unit. For example, install the vibration monitoring unit on the top of the wind turbine nacelle or tower, extract the low-frequency acceleration signal from the top of the nacelle or tower, and extract the sidelobe energy fluctuation characteristics of the high-order harmonics related to the rotor frequency from this acceleration signal. The order range of the high-order harmonics can be set from 1P (the rotor rotation frequency, i.e., the fundamental vibration frequency corresponding to one revolution of the rotor, numerically obtained by dividing the rotor speed by 60) to 6P (six times the rotor rotation frequency, i.e., the 6th harmonic of the 1P frequency). The sidelobe energy fluctuations are based on the historical reference sideband energy under intact wind turbine blade conditions; the sidelobe energy fluctuations can be calculated based on the spectral energy integration within a specific sidelobe search window.

[0033] The wind turbine blade features extracted in step 110 are divided into two categories: judgment features and descriptive features. The leading-edge texture complexity, the area of ​​local thermal diffusion anomalies, and sidelobe energy fluctuations are judgment features, used in step 120 for operating condition matching in the wind turbine blade feature-operating condition association library to determine whether the wind turbine blade is currently in an effective leading-edge corrosion state. Point defect coverage and temperature difference amplitude are descriptive features. Although descriptive features do not directly participate in the operating condition matching judgment logic, they can be combined with judgment features in step 130 to form a multi-dimensional feature combination value, serving as input conditions for the mapping map retrieval. The introduction of descriptive features increases the dimensionality of samples in the mapping map, improves the ability to distinguish different corrosion severity levels, and thus improves the matching accuracy of the power correction coefficient.

[0034] As described above, different devices are used to collect data on different characteristics affecting the performance of wind turbine blades, overcoming the interference and misjudgment caused by single sensors. The target wind turbine blade features collected by the visible light imaging unit can quantify the degree of geometric damage at the blade leading edge; the target wind turbine blade features collected by the infrared thermal imaging unit can capture the boundary layer transition and turbulent heating effects caused by corrosion, with high sensitivity and no interference from visible light changes or dirt; the target wind turbine blade features obtained by the vibration monitoring unit reflect the modulation intensity of the corrosion-induced imbalance on the wind turbine from the perspective of dynamic response, providing a data basis for subsequent confirmation of power generation loss.

[0035] In one alternative implementation, step 120 includes steps 121 to 123.

[0036] Step 121: Obtain the feature threshold corresponding to each wind turbine blade feature from the wind turbine blade feature-operating condition association library. For example, in addition to setting a fixed value in advance, the feature threshold can also be updated according to changes in different conditions.

[0037] Step 122: Compare each of the target wind turbine blade features with the corresponding feature threshold to obtain the comparison result corresponding to each of the target wind turbine blade features.

[0038] Step 123: Determine the target operating condition based on the comparison results and the wind turbine blade feature-operating condition association library.

[0039] As described above, the wind turbine blade feature-operating condition association library also stores the feature thresholds corresponding to each wind turbine blade feature. That is, by comparing the collected values ​​of the target wind turbine blade features with the corresponding feature thresholds, it is determined whether the wind turbine blade has entered a corrosion state. Referring to Table 1 above, the judgment results of various wind turbine blade features can corroborate each other, thus ensuring the accuracy of the final target operating condition judgment. Because some wind turbine blade features may require exceeding the corresponding feature threshold to indicate corrosion, while others may require falling below the corresponding feature threshold, the wind turbine blade feature-operating condition association library is matched based on the comparison results to determine the corresponding target operating condition, adapting to the different relationships between the target wind turbine blade features and feature thresholds.

[0040] In one alternative implementation, step 123 includes step 1231.

[0041] Step 1231: If the comparison result shows that there is only one target wind turbine blade with effective leading edge corrosion as identified by the comparison result of the characteristic threshold, then the target operating condition is determined to be that there is no effective leading edge corrosion.

[0042] As described above, if only one target wind turbine blade feature comparison result indicates effective leading-edge corrosion, while the comparison results of other target wind turbine blade features indicate no effective leading-edge corrosion, it indicates that a single detection unit (visible light imaging unit, infrared thermal imaging unit, or vibration monitoring unit) may malfunction, leading to misjudgment. In this case, the target operating condition is determined to be without effective leading-edge corrosion, avoiding misjudgment of the result of a single target wind turbine blade feature and improving the accuracy of the determined target operating condition.

[0043] In an alternative implementation, step 120 further includes steps 210 to 220.

[0044] Step 210: Determine the confidence level corresponding to the target operating condition in the wind turbine blade feature-operating condition association library.

[0045] Step 220: If the confidence level is lower than the preset value, a prompt will be made indicating that manual confirmation is required and the process will not proceed to steps 130 and 140, that is, the process of calculating the cumulative power generation loss value within the preset period will not proceed.

[0046] For example, if the target operating condition is effective leading-edge corrosion with high confidence, the system can confirm that there is indeed corrosion at the leading edge of the wind turbine blade, automatically enter the power loss calculation and operation and maintenance decision engine, and start calculating the cumulative power generation loss value within the preset period without manual confirmation. If the target operating condition is effective leading-edge corrosion with medium confidence, a semi-automatic process is triggered, including (1) automatically calculating the cumulative power generation loss value within the preset period, and (2) pushing the evaluation result with medium confidence to the operation and maintenance personnel, suggesting manual review. For example, the operation and maintenance personnel can retrieve the original data (visible light image, infrared image, etc.) corresponding to the wind turbine blade features to verify the judgment result of the target operating condition, but this will not block the system from automatically calculating the cumulative power generation loss value within the preset period. The system can also increase the acquisition frequency of wind turbine blade features to shorten the feature acquisition cycle, improve the confidence of the data, and eliminate interference factors from single acquisition. If the target operating condition is effective leading-edge corrosion with low confidence, a manual confirmation prompt will be triggered. The cumulative power generation loss value within the preset period will not be automatically calculated. After the manual confirmation that the target operating condition is indeed effective leading-edge corrosion, the calculation of the cumulative power generation loss value will be forcibly started. At the same time, the system records the judgment of "suspected interference" for reference in subsequent statistical analysis.

[0047] As described above, after obtaining the target operating condition, the confidence level corresponding to the final target operating condition is determined based on the judgment results of each wind turbine blade feature during the judgment process. If the confidence level is lower than the preset value, a prompt is made that manual confirmation is required. The process of introducing manual judgment when the confidence level of the automatically obtained target operating condition is low is introduced through the output method to ensure the accuracy of the judgment result of the final target operating condition.

[0048] In one optional implementation, step 130 includes steps 310 to 330 before determining the target power correction coefficient in a preset wind turbine blade characteristic-power correction coefficient mapping table based on the target wind turbine blade characteristics.

[0049] Step 310: Obtain wind tunnel test data of the same type of wind turbine blades as the wind turbine blades corresponding to the wind turbine blade feature-operating condition association library, and obtain the first wind turbine blade feature-first power correction coefficient mapping table.

[0050] For example, wind tunnel test data is obtained using a scaled-down model of the wind turbine. The acquired characteristic parameters include the leading-edge texture complexity corresponding to the artificial roughness level, and the sideband energy of the infrared temperature difference and torque fluctuation spectrum on the surface of the scaled-down blade model measured during the experiment. Sideband energy essentially corresponds to the same physical phenomenon as sidelobe energy, only the monitoring object and signal source are different; it can be used equivalently to sidelobe energy. The output power correction coefficient is calculated from the ratio of the measured torque of the rough blade to that of the smooth blade at the same wind speed. For example, first, the first torque of the smooth blade at a wind speed of 11 m / s is obtained, then the second torque of the rough blade at 11 m / s is obtained under a specific wind turbine blade characteristic. The power correction coefficient is obtained based on the ratio of the first torque to the second torque. By artificially changing the artificial roughness of the rough blade in the scaled-down wind turbine model, the characteristics of the wind turbine blade are changed, thus obtaining the power correction coefficients corresponding to different wind turbine blade characteristics.

[0051] Step 320: Obtain the first operating data of the test wind turbine blade under non-corrosion state and the second operating data under corrosion state, and obtain the second wind turbine blade characteristic-second power correction coefficient mapping table based on the first operating data and the second operating data.

[0052] For example, first and second operational data are acquired through SCADA (Supervisory Control and Data Acquisition) systems. The input characteristics of the SCADA data source include the visible light leading-edge texture complexity, the area of ​​abnormal infrared thermal diffusion regions, and vibration sidelobe energy fluctuations obtained from actual turbine inspections. The output power correction coefficient is obtained by weighting the power ratio of corroded turbines and uncorroded turbines in the same wind farm across all wind speed ranges (separate statistics are performed for different wind speeds). For instance, uncorroded wind turbines in the same wind farm are selected as a benchmark. The output power of uncorroded wind turbines is statistically analyzed at multiple preset wind speed ranges (e.g., 3 m / s, 10 m / s, 20 m / s), and then weighted and averaged according to the weight of each wind speed range in the annual wind frequency distribution to obtain the first power of the uncorroded wind turbines. Simultaneously, the output power of corroded wind turbines in the same wind farm is statistically analyzed at the same wind speed range, and then weighted and averaged with the same weight as the uncorroded wind turbines to obtain the second power of the corroded wind turbines. The ratio of the second power to the first power is calculated to obtain the power correction coefficient of the corroded wind turbine under the current turbine blade characteristics. As the degree of corrosion gradually worsens over time, monitoring the corroded wind turbine at different time periods can yield the second power and its power correction coefficient corresponding to different turbine blade characteristics.

[0053] In steps 310 to 320, the specific descriptions of the input items are shown in Table 2. The input for the wind tunnel test is the wind tunnel sample set obtained during the wind tunnel test using a scaled-down blade model. The input features include the texture complexity corresponding to the artificial roughness, the infrared temperature difference on the blade surface, and the sideband energy of the torque fluctuation spectrum. The output label includes the power correction coefficient calculated from the ratio of rough / smooth blades to wind speed torque. The first operating data under the non-corrosion state and the second operating data under the corrosion state are obtained through on-site inspection and operating data to form a SCADA sample set as input items. The specific input features are the texture complexity, infrared thermal diffusion anomaly area, and vibration sidelobe energy fluctuations obtained from unit inspection. The output label is the power correction coefficient weighted by the power ratio of the corroded / non-corroded units across all wind speed compartments. For a detailed description of the output items, please refer to Table 3. The output feature vector includes wind turbine blade features (leading edge texture complexity, local thermal diffusion anomaly region, side lobe energy fluctuation), which serve as the index key for map retrieval; it also includes the power correction coefficient corresponding to the wind turbine blade features, which is the final value after joint optimization by wind tunnel test and SCADA sample set, and is used to calculate the cumulative power generation loss value in step 140.

[0054] Table 2

[0055] Table 3

[0056] Step 330: Merge the first wind turbine blade feature-first power correction coefficient mapping table with the second wind turbine blade feature-second power correction coefficient mapping table to obtain a preset wind turbine blade feature-power correction coefficient mapping table.

[0057] For example, a bidirectional training mechanism using least-squares residual fitting or nonlinear regression can be employed to jointly optimize wind tunnel calibration coefficients and SCADA inversion coefficients, forming the final wind turbine blade characteristic-power correction coefficient mapping table. In the forward operation of the bidirectional training mechanism, the physical law of wind turbine blade surface roughness-power attenuation established by wind tunnel experiments serves as the initial framework, constraining the reasonable boundaries of SCADA data inversion, and filling sparse feature intervals of the SCADA samples with wind tunnel-derived values. In the reverse operation of the bidirectional training mechanism, the power ratio obtained from the statistical analysis of field SCADA operating data across all wind speed ranges is used as the true reference. The deviation between the wind tunnel-derived values ​​and the field-derived values ​​in each feature interval is calculated, and this deviation is used to correct the wind tunnel baseline curve, ensuring that the corrected curve passes the SCADA inversion value.

[0058] The fusion of physical and statistical constraints is an asymmetric bidirectional constraint: wind tunnel data provides the physical mechanism boundary, and SCADA data provides statistical calibration. The two are fused through a weighted adaptive mechanism, mathematically equivalent to a weighted fusion of the two data sources. The roles of wind tunnel data and SCADA data are complementary and asymmetric: within any feature interval, the data-rich side is dominant, and the data-scarce side is secondary, jointly constraining the final power correction coefficient. This includes the following steps (1) to (3).

[0059] Step (1) Deviation elimination: Calculate the difference distribution between the two (the mapping table of the first wind turbine blade characteristics-first power correction coefficient calculated by the wind tunnel and the mapping table of the second wind turbine blade characteristics-second power correction coefficient obtained by SCADA measurement) in different feature dimension intervals.

[0060] The differential distribution calculation process includes the following steps a to e.

[0061] Step a: Feature Dimension Interval Unit Division. The multidimensional feature space is divided into grids according to preset intervals. For example, texture complexity intervals of 0.3-0.4, 0.4-0.5, 0.5-0.6... are combined with the area intervals of local thermal diffusion anomalies and the sidelobe energy fluctuation intervals to form several feature dimension interval units. The leading edge texture complexity, the area of ​​local thermal diffusion anomalies, and the sidelobe energy fluctuations are each divided into several intervals according to preset step sizes or preset quantiles, and combined through Cartesian products to form multidimensional feature dimension interval units. Each feature dimension interval unit corresponds to a grid region in the feature space, used to collect wind tunnel samples and SCADA samples falling into that region for subsequent difference analysis and weighted fusion.

[0062] Step b: Sample aggregation within the interval. For each feature dimension interval unit, all wind tunnel samples whose multi-dimensional feature combination values ​​fall within that interval are aggregated into the wind tunnel sample set, and all SCADA samples whose multi-dimensional feature combination values ​​fall within that interval are aggregated into the SCADA sample set. Each sample carries its corresponding power correction (coefficient) value.

[0063] Step c: Determining the representative value for the interval. For the wind tunnel sample set, since different power correction coefficient values ​​will be obtained at different wind speeds, the average value of the power correction coefficient values ​​of all wind tunnel samples within the interval is taken as the representative value of the wind tunnel for that interval. For the SCADA sample set, the average value of the power correction coefficient values ​​of all SCADA samples within the interval is taken as the representative value of the SCADA for that interval.

[0064] Step d: Difference calculation. For each feature dimension interval cell, calculate: Interval difference = Wind tunnel representative value - SCADA representative value.

[0065] Step e: Summarize the difference distribution. Summarize the differences of all feature dimension intervals to form the complete difference distribution of the mapping map. For feature dimension intervals with large differences, if the sample size is small, the wind tunnel representative value is still relied upon entirely. The difference information is not used to change the weight allocation. If the difference is large and the SCADA sample is sufficient, it indicates that there is a true bias in this interval. The SCADA representative value is accepted, and the wind tunnel curve is corrected using the difference. If the difference is large but the SCADA sample is insufficient, although a bias is found, the SCADA data is insufficient to support reliable correction. The wind tunnel representative value is still accepted, and the physical inference results are maintained. If the difference is small and the SCADA sample is sufficient, the model is accurate, but sufficient real data is still accepted. If the difference is small but the SCADA sample is insufficient, the model is accurate, and there is insufficient data to refute it. The wind tunnel representative value is accepted. It can be seen that the difference reveals "whether attention is needed", and the sample size determines "whether SCADA can be adopted". The two are judged independently.

[0066] Step (2) Weighted optimization. For feature dimension intervals with abundant on-site SCADA data (such as feature dimension intervals in the light to moderate corrosion range), the inversion data is assigned a higher confidence weight; for extreme working conditions such as severe corrosion that are difficult to continue operating, the deduction of the wind tunnel mechanism model is mainly relied upon.

[0067] The weight calculation formula is as follows: ; In the formula, there exists n <n min n min ≤n≤n max and n>n max Three scenarios; In the formula, w SCADA This represents the weight of the SCADA representative value, where n is the number of valid SCADA samples within a certain feature dimension interval. min This is the minimum number of samples required for SCADA representative values ​​to be adopted; below this value, the system relies entirely on wind tunnel representative values. max This is the saturation point for the number of samples at which the SCADA representative value can independently support decision-making; values ​​above this point are considered fully trustworthy. When n is between n... min With n max When the confidence weight increases linearly with the increase of the sample size, the confidence weight increases linearly.

[0068] Weight of wind tunnel representation value .

[0069] The power correction coefficient value of the feature dimension interval cell is finally written into the mapping spectrum and obtained by weighted fusion:

[0070] In the formula, P finalThe final power correction value, i.e., the power correction coefficient value, is P. SCADA This represents the SCADA value for that range. This represents the wind tunnel value for this interval.

[0071] The determination of valid SCADA samples in step (2) is carried out as follows: steps a to e. This is to screen the validity of the operational data corresponding to the obtained SCADA samples. The execution order of steps a to e is not limited.

[0072] Step a, Operating condition screening: Select only the operating data segment of the wind turbine under normal power generation conditions without faults, power curtailment, or abnormal shutdown, and within the data segment, there should be no obvious abnormal jumps in parameters such as wind speed and power.

[0073] Step b, Corrosion Feature Matching: The corrosion status of the wind turbine blades corresponding to this SCADA operating data segment must be matched with a valid corrosion feature record that is confirmed by multi-source feature acquisition (visible light, infrared, vibration) and judgment by the operating condition association library within the same time window; the allowable deviation of the time window is preset according to the inspection frequency (e.g., not exceeding 24 hours).

[0074] Step c, Benchmark Unit Verification: The non-corroded benchmark unit used for comparison must also pass the normal power generation state screening in step a above.

[0075] Step d: Data integrity check: The sample should contain complete SCADA record fields, with no missing or outliers.

[0076] Step e, External interference elimination: Eliminate periods of abnormal power fluctuations caused by abnormal power grid frequency, extreme weather (such as icing, typhoons, etc.).

[0077] Step (3) Curve Smoothing. Using Bézier curves or spline interpolation logic, discrete calibration sample points are fitted into a continuous mapped surface, eliminating step jumps. The calibration process for obtaining calibration sample points involves uniformly applying the same weighted fusion formula to all feature dimension interval units. The magnitude of the difference is only used to identify the degree of deviation between the wind tunnel model and the real data, and does not participate in determining the calculation path of the final power correction coefficient.

[0078] The curve smoothing process employs radial basis function interpolation to fit discrete calibration sample points into a continuous mapped surface. Specifically, for any input wind turbine blade feature vector x, its corresponding power correction coefficient f(x) is as follows.

[0079]

[0080] in Let x be a radial basis function. iLet p be the vector of the wind turbine blade features at the i-th calibration sample point. This surface guarantees first-order or even second-order continuity, avoiding step jumps between adjacent intervals in the feature space. j (x) represents the j-th polynomial basis function, M represents the total number of terms in the polynomial basis function, N represents the total number of discrete calibration sample points, and w i The weight coefficients β of the i-th radial basis function are represented by... j Let w represent the coefficients of the j-th polynomial basis function. i and β j Solve using the difference condition and the orthogonality condition.

[0081] As described above, the final wind turbine blade characteristic-power correction coefficient mapping table is obtained through mutual verification of wind tunnel test data and actual monitoring data. By combining data obtained from the test environment and test data obtained from the actual operating environment, the accuracy of the correspondence between the power correction coefficient and the wind turbine blade characteristics is ensured. Furthermore, before performing actual power loss calculation, the mapping relationship between wind turbine blade characteristics and power correction coefficient is determined in advance. When calculating power loss, the corresponding target power correction coefficient can be directly matched according to the target wind turbine blade characteristics, thereby improving calculation efficiency.

[0082] In an alternative implementation, step 130 further includes steps 131 to 132.

[0083] Step 131: Calculate the candidate wind turbine blade feature with the highest similarity to the target wind turbine blade feature in the preset wind turbine blade feature-power correction coefficient mapping table using a similarity matching algorithm.

[0084] Step 132: Use the power correction coefficient corresponding to the candidate wind turbine blade characteristics as the target power correction coefficient.

[0085] As described above, in the process of determining the target power correction coefficient, the candidate wind turbine blade features with the highest similarity are obtained from the wind turbine blade feature-power correction coefficient mapping table based on the target wind turbine blade features. That is, in the process of constructing the wind turbine blade feature-power correction coefficient mapping table, the power loss under each wind turbine blade feature has been pre-calculated. Therefore, in the process of calculating the power loss, directly matching the candidate wind turbine blade features corresponding to the target wind turbine blade features can obtain the corresponding power correction coefficient representing the power loss, thereby improving the calculation efficiency.

[0086] In one alternative implementation, step 140 includes steps 141 to 143.

[0087] Step 141: Obtain the real-time wind speed of the environment in which the wind turbine blades are currently located. For example, this can be achieved through the real-time wind speed versus theoretical power curve provided by the SCADA system.

[0088] Step 142: Calculate the moment-to-moment power generation loss for each moment within the preset period based on the real-time wind speed and the power correction coefficient. The moment-to-moment power generation loss is calculated by combining the real-time wind speed, theoretical power curve, and power correction coefficient.

[0089] Step 143: Summing up the power generation loss at all the times mentioned to obtain the cumulative power generation loss value.

[0090] As described above, in the process of calculating the cumulative power generation loss, by combining the real-time wind speed of the environment in which the wind turbine blades are currently located, the power generation loss is calculated and accumulated moment by moment. This can accurately capture the real power generation attenuation caused by corrosion under wind speed fluctuations, and can directly obtain the power loss caused by the leading edge corrosion of the wind turbine blades, providing an objective quantification of the loss caused by corrosion.

[0091] In an alternative implementation, step 140 is followed by step 150: generation of hierarchical operation and maintenance decisions based on economic thresholds.

[0092] For example, step 150 includes steps 151 to 155.

[0093] Step 151: Set economic thresholds, including routine maintenance thresholds, small window maintenance thresholds, and forced shutdown repair thresholds.

[0094] Step 152: When the cumulative power generation loss value is less than the routine maintenance threshold, the system maintains normal monitoring status and only notes a slight attenuation trend in the routine monthly report; Step 153: When the cumulative power generation loss value is between the routine maintenance threshold and the small wind window maintenance threshold, the system automatically shortens the monitoring cycle of the wind turbine blades, collects wind turbine blade characteristics at a higher frequency, continuously tracks the loss growth trend, and temporarily does not trigger maintenance recommendations. Step 154: When the cumulative power generation loss is greater than or equal to the small wind window maintenance threshold, but has not reached the mandatory shutdown repair threshold, the system combines the future wind speed forecast data from the meteorological forecast data to calculate the opportunity cost of power generation for performing maintenance actions in different future time windows, and automatically recommends the low wind speed maintenance period with the lowest opportunity cost. The system determines that the current corrosion state has caused significant economic impact, meeting the economic triggering condition for maintenance operations. At this time, the system automatically calls the meteorological forecast interface to obtain the wind speed forecast time series data for the future preset time period, and combines it with the standard shutdown duration required for maintenance operations to calculate the opportunity cost of power generation for performing maintenance actions in different candidate time windows. The system selects the window with the lowest opportunity cost as the recommended maintenance period and pushes maintenance suggestions to the operation and maintenance task management terminal, for example: "Increased aerodynamic performance degradation of the leading edge has been detected. It is recommended to perform leading edge protective film repair operations in the low wind speed window within the next N hours."

[0095] The formula for calculating opportunity cost is: ; In the formula, C opp (T s ,T e ) indicates from the start time T s At the end time T e Within the candidate window, the total opportunity cost of performing maintenance. forecast (t) represents the theoretical power output of the wind turbine at the predicted wind speed at time t. r represents the target power correction factor. Δt represents the calculation time step. C price This indicates the on-grid electricity price for wind farms.

[0096] Step 155: When the forced shutdown threshold is less than the cumulative power generation loss value, a Level 1 alarm is pushed, and it is recommended to immediately shut down the system for leading-edge repair; for example, a Level 1 alarm can be pushed to the field-level monitoring system.

[0097] In one optional implementation, the operation and maintenance decision thresholds support dynamic updates based on changes in external parameters. Specifically, the routine maintenance threshold and the small wind window maintenance threshold are determined based on the ratio of the comprehensive cost of a single inspection to the grid-connected electricity price, and the ratio of the total cost of a single maintenance operation to the grid-connected electricity price, respectively. When the above cost parameters or the grid-connected electricity price are adjusted, the system automatically recalculates and updates the corresponding thresholds. The forced shutdown repair threshold is set based on the structural damage tolerance standard provided by the wind turbine blade manufacturer. When the manufacturer updates the tolerance standard, or the system detects that the growth rate of the cumulative power loss per unit time exceeds the preset accelerated deterioration threshold, or the vibration monitoring characteristic value shows a step-like abnormal change, the forced shutdown repair threshold is re-evaluated and updated. The above threshold update mechanism is independent of the operation and maintenance decision action triggered in step 140 based on the comparison of the current cumulative power generation loss value with the economic threshold. The former determines "where the warning line is set", and the latter determines "whether the warning line has been reached".

[0098] Please refer to Table 4. This scheme includes two types of thresholds. The first is the feature threshold (feature-coordinated triggering threshold) in step 120, which is used to determine whether the wind turbine blades have leading-edge corrosion. Each judgment feature corresponds to a feature threshold, which is obtained based on the statistical distribution of the features of each wind turbine blade under normal conditions. The second is the economic threshold (operation and maintenance decision threshold) in step 140, which is used to determine whether maintenance actions need to be performed. It is determined by the cumulative power generation loss value and is divided into three levels: routine maintenance threshold, small wind window maintenance threshold, and forced shutdown repair threshold. In order to balance the economic loss of the cumulative power generation loss value caused by leading-edge corrosion with the maintenance cost, its configuration can be dynamically adjusted according to the operating parameters.

[0099] Table 4

[0100] This paper applies the aforementioned method for assessing power loss due to leading-edge corrosion of wind turbine blades to a specific scenario. A certain offshore wind farm is located far from the coastline, has a large total installed capacity, and houses multiple high-power direct-drive wind turbines. Due to the marine salt spray environment, the leading-edge corrosion rate of the blades is significantly higher than that of onshore turbines. Because the sea conditions severely limit the sea opening window, traditional manual visual inspections can only be performed 1-2 times per year, and these inspections cannot directly quantify the actual impact of corrosion on power generation. (See attached document) Figures 2-3 Specifically, this includes the following five points.

[0101] I. Hardware Deployment and Data Acquisition Configuration.

[0102] 1. Visible light imaging unit: A high-definition industrial camera is installed on the top of the nacelle of each unit, equipped with a telephoto lens and an automatic gimbal. It automatically takes panoramic images of the leading edge span of the three blades at sunrise every day. The image resolution meets the requirements for identifying details on the leading edge surface.

[0103] 2. Infrared thermal imaging unit: The same pan-tilt unit is equipped with a medium-wave cooled infrared thermal imager, which scans and measures the temperature of the leading edge of the blades during the rated power operation of the unit. The spatial resolution meets the requirements for fine analysis of the temperature field on the blade surface.

[0104] 3. Vibration monitoring unit: A low-frequency acceleration sensor is installed in the main bearing housing of the engine room to continuously collect vibration time-domain signals and calculate the spectral characteristics in real time.

[0105] 4. Data Synchronization: All data is aggregated through tower-based switches and transmitted to the land-based central control server via the submarine fiber optic ring network.

[0106] II. Feature Extraction Examples

[0107] Visible light processing: The image algorithm extracted a texture complexity (leading edge texture complexity) in a specific region along the leading edge span that was significantly higher than the baseline value, and the coverage of point defects was significantly higher than the normal range.

[0108] Infrared processing: The area of ​​the localized abnormal heat diffusion region extracted in the same region and the temperature difference amplitude between the region and the main body temperature of the fan blades are significantly higher than those under normal conditions.

[0109] Vibration treatment: Spectrum analysis shows that the energy fluctuations of the side lobes near the rotor rotation frequency have increased compared to historical baseline values.

[0110] III. Example of multi-source collaborative verification.

[0111] The system inputs the above three features (leading edge texture complexity, local thermal diffusion anomaly region, and side lobe energy fluctuation) into the wind turbine blade feature-operating condition association library.

[0112] The leading edge texture complexity, abnormal thermal diffusion region, and side lobe energy fluctuation all exceed their respective preset feature thresholds. If the "≥2 items" judgment condition is met, the system marks the blade as a valid leading edge corrosion condition.

[0113] IV. Example of power loss assessment.

[0114] The system retrieves the target power correction coefficient corresponding to the current target wind turbine blade characteristics from the wind turbine blade characteristic-power correction coefficient mapping table. This coefficient indicates that the actual power is significantly lower than the theoretical power. Combined with the theoretical power generation of the wind turbine over a past period recorded by SCADA, the cumulative power generation loss is calculated. Based on the grid connection price, the corrosion of this single blade has caused a certain amount of economic loss.

[0115] V. Examples of Operation and Maintenance Decision Generation

[0116] The current cumulative power generation loss has exceeded the system's preset small wind window maintenance threshold, but has not reached the mandatory shutdown repair threshold. The system automatically calls the meteorological forecast interface to obtain wind speed forecast data for the next few days: there is a significant period of low wind speed in the near future, preceded and followed by full power generation or high wind speed conditions. After calculating the opportunity cost of each candidate maintenance window, the system pushes a suggestion to the maintenance fleet dispatch terminal, recommending that the leading edge protective film repair work be carried out at sea within the low wind speed window, and providing an estimated power loss that can be avoided by executing the recommended window.

[0117] It is evident that this scenario achieves a complete closed loop of unmanned remote sensing, automatic loss quantification, and optimal sea departure time recommendation based on sea and wind conditions, significantly improving the economic efficiency of offshore wind power operation and maintenance.

[0118] Please refer to Figure 4 The present invention also provides a wind turbine blade leading edge corrosion power loss assessment terminal 400, including a memory 401, a processor 402, and a computer program stored in the memory and executable on the processor. When the processor 402 executes the computer program, it implements the various steps of a wind turbine blade leading edge corrosion power loss assessment method as described above.

[0119] The technical effect achieved by the wind turbine blade leading edge corrosion power loss assessment terminal of this application is the same as the above method, and will not be repeated here.

[0120] Please refer to Figure 5This paper provides an architecture for implementing the aforementioned method for assessing power loss due to leading-edge corrosion of wind turbine blades. It includes a physical sensing layer, a data aggregation and edge computing layer, a platform fusion analysis layer, and an application display and interaction layer. The physical sensing layer uses a visible light camera and an infrared thermal imager at the leading edge of the blade, and a vibration acceleration sensor at the nacelle or tower of the wind turbine to collect data. The data aggregation and edge computing layer includes a time synchronization module, an edge data acquisition gateway, and a data preprocessing and feature extraction unit. The edge data acquisition gateway receives data sent by the physical sensing layer. The platform fusion analysis layer includes a multi-source spatiotemporal alignment module, a wind turbine blade feature-operating condition correlation verification rule library, a multi-dimensional feature-power attenuation mapping library (including a wind turbine blade feature-power correction coefficient mapping table), a power loss calculation engine, and an operation and maintenance decision logic controller. The power loss calculation engine receives real-time wind speed data from SCADA, and the operation and maintenance decision logic controller connects to a meteorological forecast data interface to receive meteorological forecast data. The operation and maintenance decision logic controller sends operation and maintenance decision data to the application display and interaction layer. The application display and interaction layer includes a large screen in the field-level monitoring center, an operation and maintenance mobile terminal APP, and a work order interface for the asset management system.

[0121] In summary, this application discloses a method and terminal for assessing the electrical energy loss due to leading-edge corrosion of wind turbine blades. Based on a multi-dimensional physical field collaborative verification and engineering mapping architecture for assessing electrical energy loss due to leading-edge corrosion, a two-level mapping and a single-level verification mechanism are established. During the acquisition of wind turbine blade features, multi-dimensional feature collaborative verification is implemented, cross-validating the feature parameters extracted by visible light, infrared, and vibration sensors after temporal and spatial alignment. Only when multiple source features simultaneously point to "abnormal fluid state in the leading-edge boundary layer" is the current corrosion condition confirmed, thereby eliminating interference from single-sensor noise. Feature-aerodynamic parameter correlation mapping utilizes a pre-constructed wind turbine blade feature-power correction coefficient mapping table calibrated through historical operational big data and physical simulation to obtain the aerodynamic performance degradation characterization quantity corresponding to the current corrosion state. Electrical energy loss and economic decision mapping involves inputting the aerodynamic performance degradation characterization quantity into a power loss estimation model based on the actual power curve of the wind farm to obtain the cumulative electrical energy loss, which is then used to trigger different levels of maintenance work orders.

[0122] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for assessing electrical charge loss due to corrosion at the leading edge of wind turbine blades, characterized in that, include: Obtain a preset wind turbine blade feature-operating condition association library, and collect at least two target wind turbine blade features based on the wind turbine blade features in the wind turbine blade feature-operating condition association library; Determine the target operating condition corresponding to the target wind turbine blade feature in the wind turbine blade feature-operating condition association library; If the target operating condition is effective for leading edge corrosion, then the target power correction coefficient is determined in the preset wind turbine blade characteristic-power correction coefficient mapping table according to the target wind turbine blade characteristics. The cumulative power generation loss value within the preset period is calculated based on the target power correction coefficient.

2. The method according to claim 1, characterized in that, The step of collecting at least two target wind turbine blade features based on the wind turbine blade features-operating condition association library includes: The characteristics of the target wind turbine blades are collected by at least two of the visible light imaging unit, infrared thermal imaging unit, and vibration monitoring unit.

3. The method according to claim 1, characterized in that, The step of collecting at least two target wind turbine blade features based on the wind turbine blade features-operating condition association library includes: The leading edge texture complexity or point defect coverage of wind turbine blades is acquired using a visible light imaging unit. The temperature difference between the local heat diffusion abnormal area of ​​the wind turbine blade and the main body temperature of the wind turbine blade is collected by the infrared thermal imaging unit. The sidelobe energy fluctuation characteristics of the high-order harmonics related to the rotor frequency of the wind turbine blades are collected by the vibration monitoring unit.

4. The method according to claim 1, characterized in that, Determining the target operating condition corresponding to the target wind turbine blade feature in the wind turbine blade feature-operating condition association library includes: Obtain the feature threshold corresponding to each wind turbine blade feature from the wind turbine blade feature-operating condition association library; Each target wind turbine blade feature is compared with the corresponding feature threshold to obtain a comparison result for each target wind turbine blade feature. The target operating condition is determined based on the comparison results and the wind turbine blade feature-operating condition association library.

5. The method according to claim 4, characterized in that, The step of determining the target operating condition based on the comparison results and the wind turbine blade feature-operating condition association library includes: If the comparison result indicates that only one target wind turbine blade has effective leading-edge corrosion as identified by the comparison result of the characteristic threshold, then the target operating condition is determined to be that there is no effective leading-edge corrosion.

6. The method according to claim 1, 4, or 5, characterized in that, The step of determining the target operating condition corresponding to the target wind turbine blade feature in the wind turbine blade feature-operating condition association library further includes: Determine the confidence level corresponding to the target operating condition in the wind turbine blade feature-operating condition association library; If the confidence level is lower than the preset value, a prompt will appear indicating that manual confirmation is required and the process will not proceed to the step of calculating the cumulative power generation loss value within the preset period.

7. The method according to claim 1, characterized in that, Before determining the target power correction coefficient based on the target wind turbine blade characteristics in a preset wind turbine blade characteristic-power correction coefficient mapping table, the process includes: Obtain wind tunnel test data of the same type of wind turbine blades as the wind turbine blades corresponding to the wind turbine blade feature-operating condition association library, and obtain the first wind turbine blade feature-first power correction coefficient mapping table; Obtain the first operating data of the test wind turbine blade under non-corrosion state and the second operating data under corrosion state, and obtain the second wind turbine blade characteristic-second power correction coefficient mapping table based on the first operating data and the second operating data; By integrating the first wind turbine blade feature-first power correction coefficient mapping table and the second wind turbine blade feature-second power correction coefficient mapping table, a preset wind turbine blade feature-power correction coefficient mapping table is obtained.

8. The method according to claim 7, characterized in that, The step of determining the target power correction coefficient based on the target wind turbine blade characteristics in a preset wind turbine blade characteristic-power correction coefficient mapping table includes: The similarity matching algorithm is used to calculate the candidate wind turbine blade feature with the highest similarity to the target wind turbine blade feature in the preset wind turbine blade feature-power correction coefficient mapping table; The power correction coefficient corresponding to the characteristics of the candidate wind turbine blades is used as the target power correction coefficient.

9. The method according to claim 1, characterized in that, The calculation of the cumulative power generation loss value within a preset period based on the target power correction coefficient includes: Obtain the real-time wind speed of the environment in which the wind turbine blades are currently located; Calculate the moment-by-moment power generation loss at each moment within the preset period based on the real-time wind speed and the power correction coefficient; The cumulative power generation loss value is obtained by summing up the power generation losses at all the times mentioned.

10. A terminal for assessing electrical loss due to corrosion at the leading edge of a wind turbine blade, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the method for assessing electrical energy loss due to leading-edge corrosion of wind turbine blades as described in any one of claims 1 to 9.

Images