A wind turbine blade deformation anomaly monitoring method, system and electronic device

Abstract

The application provides a wind turbine blade deformation anomaly monitoring method, system and electronic equipment, and belongs to the technical field of wind turbines. The application is used to solve the problems of existing blade deformation anomaly monitoring, which relies on special sensors, has high deployment cost and low blade deformation anomaly recognition accuracy. By obtaining the operation data of the wind turbine unit, the variable pitch load data of each blade, the equivalent phase data and the speed fluctuation data, the effective data section is filtered and the target working condition interval is determined; based on the equivalent phase data, the power, load and speed fluctuation in each blade preset phase window are extracted to form the phase window characteristics; the phase window characteristics of continuous multiple rotation periods are counted to obtain various fluctuation indexes, which are compared with the index correlation reference to obtain the coupling deviation characteristics; and the abnormal blade and the abnormal type are determined according to the coupling deviation characteristics. The application can be used for wind turbine unit operation state monitoring and blade abnormal operation and maintenance review, and can reduce the deployment cost and improve the recognition accuracy.

Description

Technical Field

[0001] This application belongs to the field of wind turbine technology, and in particular relates to a method, system and electronic equipment for monitoring abnormal deformation of wind turbine blades. Background Technology

[0002] Wind turbine blades are key components for capturing wind energy in wind turbine generators. As wind turbines develop towards larger megawatts and longer blades, the wind loads, gravity loads, centrifugal loads, and pitch loads they bear during operation become more complex. After long-term operation, these blades are prone to abnormal deformation, fatigue damage, or aerodynamic performance degradation. Abnormal blade conditions can affect the generator's power generation efficiency and operational safety; therefore, it is necessary to monitor wind turbine blades for abnormalities during operation.

[0003] In existing technologies, distance measurement data from multiple cross-sections of the blades is mainly collected by a two-dimensional microwave radar sensor array deployed on the wind turbine tower. This data is then combined with tower design parameters and blade rotation models to construct a dynamic deformation model, thereby identifying blade deformation anomalies. Alternatively, vibration acceleration signals are collected, and combined with wind turbine operating data to extract blade damage impact feature sets, which are then used to calculate damage factors and cumulative blade damage indices.

[0004] However, the aforementioned existing technologies still have certain limitations. Radar array-based deformation monitoring schemes require the additional deployment of dedicated sensors and involve multi-source ranging data synchronization and spatial coordinate transformation, resulting in high on-site modification and maintenance costs. Damage monitoring schemes based on vibration acceleration signals rely on vibration sensor data, reflecting more the vibration response of the unit or structure, and making it difficult to stably correlate fluctuations in common operating data such as overall power and speed to specific blades. Furthermore, unit operating parameters vary significantly under different wind speeds, speeds, and pitch angles. Relying solely on fixed thresholds, single vibration characteristics, or direct deformation measurements makes them susceptible to changes in normal operating conditions, thus requiring further improvement in the accuracy and deployment convenience of blade deformation anomaly identification. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, and electronic device for monitoring abnormal deformation of wind turbine blades, so as to solve the problems of poor accuracy and deployment convenience in the identification of abnormal blade deformation in the prior art.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for monitoring abnormal deformation of wind turbine blades, comprising: The system acquires the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data, and the speed fluctuation data. The operating data includes wind speed, speed, total power, and pitch angle. The pitch load data is the torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is the phase representation quantity obtained by integrating the speed. Valid data segments are filtered from the operational data and the target operating condition intervals are determined. The operating condition intervals are divided based on the wind speed, engine speed, and pitch angle of the historical valid data segments. Based on equivalent phase data, the overall power fluctuation, pitch load fluctuation and speed fluctuation are extracted from the effective data segment within the preset phase window corresponding to each blade to form the phase window characteristics of each blade. The power fluctuation index, load fluctuation index, and speed fluctuation index of each blade are obtained by statistically analyzing the phase window characteristics within multiple consecutive rotation cycles. The coupling deviation characteristics of each blade are obtained by comparing them with the index correlation benchmark. The index correlation benchmark includes the time delay coupling relationship and phase synchronization coupling relationship between the power fluctuation index, load fluctuation index, and speed fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval. Based on the time-delay deviation of load fluctuation index and power fluctuation index, and the phase synchronous deviation of load fluctuation index and speed fluctuation index in the coupling deviation characteristics, the abnormal blades and abnormal types are determined.

[0007] In one feasible implementation, filtering valid data segments from operational data and determining their corresponding target operating condition ranges includes: If the wind speed change, rotational speed change, total power change, and pitch angle change all meet the corresponding preset stability thresholds within the target time period, and there are no shutdown, power limitation, yaw adjustment, braking, or pitch system fault markers within the target time period, then the target time period is determined as a valid data segment. Based on the wind speed, engine speed, and pitch angle in the valid data segment, the wind speed range, engine speed range, and pitch angle range are determined respectively. The operating condition range that corresponds to the wind speed range, engine speed range, and pitch angle range is determined as the target operating condition range to which the valid data segment belongs.

[0008] In one feasible implementation, the equivalent phase data is obtained by determining the angular increment between adjacent sampling times based on the rotational speed and sampling period, and then accumulating the angular increment. The preset phase window for each blade is obtained by dividing a single rotation cycle according to the number of blades and the installation phase difference of each blade.

[0009] In one feasible implementation, based on equivalent phase data, the overall power fluctuation, pitch load fluctuation, and speed fluctuation are extracted from the effective data segment within a preset phase window corresponding to each blade to form the phase window characteristics of each blade, including: Determine the phase position corresponding to each sampling time based on the equivalent phase data; When any blade enters the corresponding preset phase window, extract the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window; For the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window, calculate at least one of the mean, peak-to-valley difference, rate of change, and dispersion to obtain the phase window characteristics of the corresponding blade.

[0010] In one feasible implementation, the speed fluctuation data is the deviation of the speed within the effective data segment from the reference speed corresponding to the target operating condition range; Statistical analysis of the phase window characteristics over multiple consecutive rotation cycles yields power fluctuation, load fluctuation, and speed fluctuation indices for each blade, including: Based on the historical valid data segments corresponding to the target operating condition range, the power reference window characteristics, load reference window characteristics, and speed reference window characteristics are determined respectively. The power fluctuation index is obtained by comparing the phase window characteristics corresponding to the power fluctuation of the whole machine within multiple consecutive rotation cycles with the power reference window characteristics. The load fluctuation index is obtained by comparing the phase window characteristics corresponding to the pitch load fluctuation over multiple consecutive rotation cycles with the load reference window characteristics. The speed fluctuation index is obtained by comparing the phase window features corresponding to the speed fluctuations within multiple consecutive rotational cycles with the speed reference window features. Power fluctuation index, load fluctuation index, and speed fluctuation index all include at least one of deviation amplitude, deviation change rate, deviation duration period, and deviation dispersion.

[0011] In one feasible implementation, the index correlation benchmark includes the benchmark time difference range between the load fluctuation index and the power fluctuation index, the benchmark change direction correspondence, and the benchmark synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window. The coupling deviation characteristics include the deviation of the actual time difference between the load fluctuation index and the power fluctuation index from the reference time difference range, the deviation of the actual change direction correspondence between the load fluctuation index and the power fluctuation index from the reference change direction correspondence, and the deviation of the actual synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window from the reference synchronous change amplitude.

[0012] In one feasible implementation, the method further includes: The peak phase position of the target blade load is determined based on the peak phase position of the phase window characteristics corresponding to the pitch load fluctuation within multiple consecutive rotation cycles. Calculate the phase offset between the peak load phase position and the center of the preset phase window corresponding to the target blade; If the phase offset exceeds the preset offset threshold, the load peak phase position is used as the center of the adjusted phase window, the preset phase window corresponding to the target blade is redefined, and the phase window characteristics of the target blade are determined based on the redefined preset phase window.

[0013] In one feasible implementation, based on the time-delay deviation of the load fluctuation index and the power fluctuation index, and the phase synchronization deviation of the load fluctuation index and the speed fluctuation index in the coupling deviation characteristics, the abnormal blades and abnormal types are determined, including: The confidence level of aerodynamic anomaly of the target blade is determined based on the time delay deviation and directional deviation between the load fluctuation index and the power fluctuation index. The structural anomaly confidence level of the target blade is determined based on the synchronous deviation between the load fluctuation index and the speed fluctuation index. If the confidence level of aerodynamic anomaly meets the preset anomaly conditions, or the confidence level of structural anomaly meets the preset anomaly conditions, the target blade is identified as an anomalous blade. Based on the relationship between the confidence levels of aerodynamic anomaly and structural anomaly, the anomaly type of the target blade is determined to be one of aerodynamic deformation anomaly, structural deformation anomaly, or combined deformation anomaly.

[0014] Secondly, this application provides a wind turbine blade deformation anomaly monitoring system, comprising: The acquisition module is used to acquire the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data and the speed fluctuation data. The operating data includes wind speed, speed, total power and pitch angle. The pitch load data is the torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is the phase representation quantity obtained by integrating the speed. The determination module is used to filter valid data segments from the operating data and determine the target operating condition interval to which they belong. The operating condition interval is obtained by dividing the wind speed, rotational speed and pitch angle of the historical valid data segments. The extraction module is used to extract the overall power fluctuation, pitch load fluctuation and speed fluctuation from the effective data segment within the preset phase window corresponding to each blade based on the equivalent phase data, so as to form the phase window features of each blade. The comparison module is used to statistically analyze the phase window characteristics within multiple consecutive rotation cycles to obtain the power fluctuation index, load fluctuation index, and speed fluctuation index of each blade, and compare them with the index correlation benchmark to obtain the coupling deviation characteristics of each blade. The index correlation benchmark includes the time delay coupling relationship and phase synchronization coupling relationship between the power fluctuation index, load fluctuation index, and speed fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval. The determination module is also used to determine abnormal blades and abnormal types based on the time-delay deviation of load fluctuation index and power fluctuation index and the phase synchronization deviation of load fluctuation index and speed fluctuation index in the coupling deviation characteristics.

[0015] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute computer programs to implement the steps of the wind turbine blade deformation anomaly monitoring method as described in the first aspect above.

[0016] The wind turbine blade deformation anomaly monitoring method provided in this application utilizes the wind turbine's operating data, the pitch load data of each blade, the equivalent phase data, and the speed fluctuation data to monitor blade deformation anomalies. It does not rely on dedicated monitoring equipment such as radar arrays or blade-embedded sensors, which helps reduce on-site deployment and maintenance costs. At the same time, by filtering effective data segments and determining the target operating condition range, the current data can be compared with historical effective data under the corresponding operating conditions, which can reduce the interference of wind speed, speed, and pitch angle changes on anomaly judgment and improve the stability of monitoring results.

[0017] This application extracts the overall power fluctuation, pitch load fluctuation, and speed fluctuation within a preset phase window corresponding to each blade based on equivalent phase data. It also determines the coupling deviation characteristics through the time delay coupling relationship and phase synchronization coupling relationship between the power fluctuation index, load fluctuation index, and speed fluctuation index. This allows for the establishment of a correspondence between abnormal fluctuations in the overall operating data and specific blades, and the identification of abnormal blades and abnormal types based on different deviation relationships, thereby improving the accuracy and usability of blade deformation anomaly monitoring. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1A flowchart illustrating a method for monitoring abnormal deformation of wind turbine blades provided in this application embodiment; Figure 2 A schematic diagram of a wind turbine blade phase window provided in an embodiment of this application; Figure 3 This application provides a schematic diagram of an abnormal blade and an abnormality type determination process. Figure 4 A schematic diagram of a wind turbine blade deformation anomaly monitoring system provided in this application embodiment; Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] To address the problems of existing technologies, this application provides a method, apparatus, device, computer storage medium, and computer program product for monitoring abnormal deformation of wind turbine blades. The method for monitoring abnormal deformation of wind turbine blades provided in this application will be described first.

[0022] Figure 1 A flowchart illustrating a method for monitoring abnormal deformation of wind turbine blades according to an embodiment of this application is shown. Figure 1 As shown, S110. Acquire the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data, and the speed fluctuation data. The operating data includes wind speed, speed, total power, and pitch angle. The pitch load data is the torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is the phase representation quantity obtained by integrating the speed.

[0023] The pitch load data corresponds to the load magnitude of the pitch mechanism for each blade, and can be calculated by collecting the pitch motor current from the pitch driver or pitch control cabinet corresponding to each blade. In the specific calculation, the pitch motor current can first be zero-point corrected and filtered, and then the motor output torque can be obtained according to the motor torque constant. This torque can then be converted into the pitch load torque of the corresponding blade by combining the reduction ratio and transmission efficiency of the pitch reduction mechanism.

[0024] Equivalent phase data refers to the angular position corresponding to each sampling time during the continuous rotation of the impeller. This data can be obtained by integrating the impeller speed data acquired by the main control system or speed encoder. Speed ​​fluctuation data represents the change in impeller speed relative to a stable operating level. After determining the target operating range, the speed fluctuation data is obtained by subtracting the sampled speed value from the reference speed corresponding to the target operating range. In the preprocessing stage before the target operating range is determined, a temporary speed deviation sequence can be generated using the smoothed speed within the corresponding time period. Once the target operating range is determined, the speed fluctuation data can be recalculated using the reference speed corresponding to the target operating range. Alternatively, the speed change rate or the smoothed speed deviation sequence can be further calculated.

[0025] In practice, operating data can be obtained from the main control system of the wind turbine generator set, the wind farm monitoring platform, or the data acquisition terminal. The wind speed can come from the nacelle anemometer or the wind speed channel corrected by the main control system. The rotational speed can come from the rotor speed encoder or the rotor speed channel converted by the generator speed encoder through the transmission ratio. The total power can come from the converter, the power metering device, or the power channel of the main control system. The pitch angle can come from the pitch angle channel fed back by the pitch controller of each blade.

[0026] Since the data may come from different controllers or different sampling channels, a unified timeline can be established using the main control system timestamp or the data acquisition terminal timestamp, and a target sampling time sequence can be generated according to a preset sampling period. For data with a sampling frequency higher than the target sampling frequency, the mean, median, or most recent valid sample value can be taken within the time window corresponding to the target sampling time. For data with a sampling frequency lower than the target sampling frequency, linear interpolation of adjacent valid sampling points or keeping the previous valid value can be used to obtain the data corresponding to the target sampling time. For channels with known controller communication delays, delay compensation can be performed on the corresponding data timestamps based on the channel delay calibration value or historical synchronization pulse records.

[0027] Simultaneously, the pitch motor current can be read from the pitch driver of each blade. For example, blades A, B, and C correspond to current sequences IA, IB, and IC, respectively. After filtering the current sequences, the load torque sequences LA, LB, and LC are calculated according to the motor torque constant, reduction ratio, and transmission efficiency, respectively. Next, the sampled impeller speed R can be converted into angular velocity, and accumulated point by point according to the method that the angular increment between adjacent sampling points is equal to the angular velocity multiplied by the sampling period, to obtain the phase sequence Φ. When the accumulated phase exceeds one rotation cycle, the remainder can be taken over one rotation cycle to make the phase fall within the range of 0 degrees to 360 degrees.

[0028] Then, the rotational speed sequence R can be compared with the smoothed rotational speed R0 within the same time period or the reference rotational speed R0 under the corresponding operating conditions to obtain the rotational speed fluctuation sequence ΔR. For example, during the continuous sampling process of a wind turbine generator, the wind speed V, rotational speed R, total power P, pitch angle B, pitch load torque LA to LC of blades A to C, phase Φ, and rotational speed fluctuation ΔR corresponding to time points T1 to Tn can be obtained.

[0029] By acquiring operational data, blade pitch load data, equivalent phase data, and speed fluctuation data using the methods described above, the existing control and operational monitoring data of the wind turbine generator can be used to form the data foundation required for blade anomaly monitoring without deploying additional radar arrays or built-in blade sensors. Simultaneously, all types of data are acquired or converted on the same sampling time axis, and resampling, interpolation, delay compensation, and missing value processing are performed on a unified time axis. When a target sampling moment lacks any key data in wind speed, speed, total power, pitch angle, or corresponding blade pitch load, and cannot be supplemented by adjacent valid sampling points, that target sampling moment is marked as an invalid sampling point and removed in subsequent valid data segment filtering and phase window feature calculation. This facilitates subsequent corresponding analysis of changes in total power, pitch load, and speed, reducing errors caused by inconsistencies in time data from different sources.

[0030] S120. Filter valid data segments from the operating data and determine the target operating condition interval to which they belong. The operating condition interval is obtained by dividing the wind speed, rotational speed and pitch angle of the historical valid data segments.

[0031] The valid data segment is a data fragment selected from the operational data that indicates relatively stable unit operation without significant control intervention or malfunction. The historical valid data segment is a normal operating data fragment obtained from historical operational data according to the same stability rules and after health status screening. Health status screening includes: removing data fragments containing blade abnormality alarms, pitch system alarms, speed abnormality alarms, power abnormality alarms, unit maintenance records, manually verified abnormal records, and transitional operation records before and after blade cleaning, maintenance, or replacement; in cases where historical data for the same unit is insufficient, historical data from units of the same model, with the same blade configuration and similar operating environment can be selected, and abnormal data fragments can be removed according to the same rules.

[0032] Operating condition intervals can be obtained by dividing the wind speed, engine speed, and pitch angle in the historical valid data segments. Specifically, the average wind speed, average engine speed, and average pitch angle of each historical valid data segment can be calculated first. Then, the wind speed interval, engine speed interval, and pitch angle interval can be divided according to the value distribution of the three types of parameters, and the three types of intervals can be combined to form multiple operating condition intervals.

[0033] In practice, the operational data can be segmented according to a fixed duration or a fixed number of sampling points to obtain multiple target time periods. Then, the changes in wind speed, engine speed, total power, and pitch angle within each target time period are calculated, and the unit's operating status, power control status, yaw action status, braking status, and pitch fault status are read. Target time periods that meet the requirements for stable operation can be identified as valid data segments. Then, based on the statistical values ​​of wind speed, engine speed, and pitch angle within these valid data segments, pre-divided wind speed, engine speed, and pitch angle intervals are matched. The combination of these three intervals determines the target operating condition interval to which the valid data segment belongs.

[0034] In one feasible implementation, step S120, which involves filtering valid data segments from the operational data and determining their corresponding target operating condition ranges, includes: If the wind speed change, rotational speed change, total power change, and pitch angle change all meet the corresponding preset stability thresholds within the target time period, and there are no shutdown, power limitation, yaw adjustment, braking, or pitch system fault markers within the target time period, then the target time period is determined as a valid data segment. The target time period is a candidate data segment obtained by dividing continuous operating data. It can be a sliding time window or adjacent and non-overlapping time windows. The wind speed change, engine speed change, total power change, and pitch angle change are the differences between the maximum and minimum values ​​of the corresponding parameters within the target time period. When the sampling noise is large, the difference between the high quantile and low quantile values ​​can also be used.

[0035] The preset stability threshold can be determined based on the unit's designed operating range, on-site commissioning records, or historical normal operating data. When using historical normal operating data, data segments containing shutdowns, power limitations, yaw adjustments, braking, alarms, and pitch faults can be removed first. Then, the distribution of the variation amplitude of each parameter in the remaining data segments can be statistically analyzed, and the corresponding quantile value or empirical upper limit can be selected as the stability threshold.

[0036] In practice, the variation amplitudes of the wind speed sequence V, engine speed sequence R, total turbine power sequence P, and pitch angle sequence B within the target time period Wj can be calculated separately. Taking the wind speed variation amplitude as an example, it can be calculated using the following formula: ;in, The wind speed variation within the target time period Wj. The wind speed value corresponding to sampling time t. The maximum wind speed within the target time period Wj. This represents the minimum wind speed within the target time period Wj. The variations in engine speed, total power output, and pitch angle can be calculated using the same method.

[0037] Next, each change amplitude is compared with the wind speed stability threshold θV, the engine speed stability threshold θR, the power stability threshold θP, and the pitch angle stability threshold θB, respectively; at the same time, each status marker within the target time period is checked. If each change amplitude is less than or equal to the corresponding stability threshold, and there are no shutdown, power limiting, yaw adjustment, braking, or pitch system fault markers within the target time period, then the target time period is determined as a valid data segment.

[0038] Based on the wind speed, engine speed, and pitch angle in the valid data segment, the wind speed range, engine speed range, and pitch angle range are determined respectively. The operating condition range that corresponds to the wind speed range, engine speed range, and pitch angle range is determined as the target operating condition range to which the valid data segment belongs.

[0039] In practice, operating condition intervals can be constructed first using historical valid data segments. For each historical valid data segment, the average wind speed, average engine speed, and average pitch angle can be calculated to obtain historical operating condition samples. Then, wind speed intervals, engine speed intervals, and pitch angle intervals are divided according to preset intervals, historical quantiles, or sample densities. Afterward, any wind speed interval, any engine speed interval, and any pitch angle interval are combined into an operating condition interval. If the number of historical valid data segments within a combination is insufficient, the combination can be merged with adjacent wind speed intervals, adjacent engine speed intervals, or adjacent pitch angle intervals.

[0040] For example, if the wind speed in the historical operating condition sample is divided into Vm1, Vm2, and Vm3, the rotational speed is divided into Rk1, Rk2, and Rk3, and the pitch angle is divided into Bl1, Bl2, and Bl3, then Vm2, Rk2, and Bl1 can be combined to form an operating condition interval. For the current valid data segment W1, if its average wind speed falls into Vm2, its average rotational speed falls into Rk2, and its average pitch angle falls into Bl1, then the operating condition interval corresponding to Vm2, Rk2, and Bl1 is determined as the target operating condition interval to which W1 belongs.

[0041] By filtering valid data segments and determining target operating condition ranges, operational fluctuations caused by non-blade deformation factors such as shutdowns, power limitations, yaw adjustments, and brake or pitch system failures can be eliminated. Furthermore, the current valid data segment can be compared with historical valid data segments under the same or similar operating conditions. This reduces the impact of normal changes in wind speed, engine speed, and pitch angle on anomaly detection, improving the stability and interpretability of subsequent comparisons of power fluctuations, load fluctuations, and engine speed fluctuations.

[0042] S130. Based on equivalent phase data, extract the overall power fluctuation, pitch load fluctuation and speed fluctuation from the effective data segment within the preset phase window corresponding to each blade to form the phase window characteristics of each blade.

[0043] Figure 2A schematic diagram of the phase window of a wind turbine blade is shown in one embodiment of this application.

[0044] like Figure 2 As shown, the impeller includes a hub and blades A, B, and C, which are spaced apart along the circumference of the impeller. The 0°, 120°, and 240° markings on the outer side of the impeller circumference represent the phase positions of the three blades, respectively, and the three fan-shaped shaded areas represent the preset phase windows corresponding to each blade. In the basic implementation, the three preset phase windows can use the same half-width to extract the fluctuation data corresponding to blades A, B, and C respectively within one rotation cycle; during subsequent phase window calibration, the center of the window corresponding to a single blade can also be adjusted according to the peak load phase position.

[0045] The preset phase window is the phase range corresponding to a single blade within one rotation cycle, which can be determined based on the number of blades, the blade installation phase difference, and the allowable phase width. Overall power fluctuation is the change in overall power relative to the corresponding reference power or smoothed power within the valid data segment; pitch load fluctuation is the change in pitch load data for each blade relative to the corresponding reference load or smoothed load. The phase window characteristic is a data set obtained by statistically analyzing overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window corresponding to the same blade, and may include one or more of the following: mean, peak-to-valley difference, rate of change, and dispersion.

[0046] In practice, the equivalent phase data can be used to determine the phase position corresponding to each sampling time within the effective data segment, and then combined with... Figure 2 The blade phase distribution shown is used to determine the preset phase window for each blade based on the installation phase difference between blades A, B, and C on the impeller. Then, when any blade enters its corresponding preset phase window, the overall power fluctuation, the blade's pitch load fluctuation, and the rotational speed fluctuation are extracted from the same time range. The extracted data are then statistically calculated to obtain the phase window characteristics of that blade within that rotation cycle. Repeating this process for multiple consecutive rotation cycles yields multiple sets of phase window characteristics for each blade.

[0047] In one feasible implementation, the equivalent phase data is obtained by determining the angular increment between adjacent sampling times based on the rotational speed and sampling period, and accumulating the angular increment; the preset phase window corresponding to each blade is obtained by dividing a single rotational period according to the number of blades and the installation phase difference of each blade.

[0048] The rotation angle increment is the angle the impeller rotates between two adjacent sampling times, which can be calculated from the impeller speed and the sampling period. The sampling period is the time interval between two adjacent samplings, which can be determined by the sampling configuration of the data acquisition terminal, the main control system, or the wind farm monitoring platform. The installation phase difference is the difference in installation angle of each blade in the circumferential direction of the impeller. For a three-bladed unit, the installation phase difference between adjacent blades can usually be obtained by dividing a single rotation period by the number of blades, or it can be corrected using the unit installation parameters or the blade installation angle recorded in the main control system.

[0049] In practice, the impeller speed can be converted into angular velocity, and then the angular velocity can be multiplied by the sampling period to obtain the angular increment between adjacent sampling times. For example, if the speed R is measured in revolutions per minute, the angular increment can be calculated using the following formula: ;in, This represents the rotation increment corresponding to the i-th sampling interval, in degrees; The impeller speed at the i-th sampling time is expressed in revolutions per minute. The sampling period is in seconds. Since one revolution is 360 degrees and one minute is 60 seconds, the unit conversion for coefficient 6 is 360 divided by 60. Then, the rotation angle increments are accumulated to obtain the phase data: ;in, This represents the phase position corresponding to the i-th sampling time, in degrees; This represents the phase position corresponding to the previous sampling time. This means taking the remainder of the phase accumulation result within the parentheses by 360 degrees, so that the phase position falls within the range of 0 to 360 degrees.

[0050] For a three-blade unit, if the installation phase of blade A is 0 degrees, and the installation phases of blade B and blade C relative to blade A are 120 degrees and 240 degrees respectively, then 0 degrees, 120 degrees and 240 degrees can be used as the window center, and the preset phase window corresponding to each blade can be formed according to the preset window width.

[0051] like Figure 2 As shown, the three sector-shaped shaded areas correspond to the preset phase windows of blades A, B, and C, respectively. Each sector-shaped shaded area can be centered on the phase position of the corresponding blade and expanded by half the window width h in the direction of phase increase and phase decrease, respectively. Figure 2 The three sector-shaped shaded areas are drawn with equal angle width to represent the case where each blade uses the same window width; in actual processing, different window widths can also be set for different blades or the window center can be adjusted according to the historical load peak phase distribution of different blades.

[0052] In one feasible implementation, based on equivalent phase data, the overall power fluctuation, pitch load fluctuation, and speed fluctuation are extracted from the effective data segment within a preset phase window corresponding to each blade to form the phase window characteristics of each blade, including: Determine the phase position corresponding to each sampling time based on the equivalent phase data; Phase position refers to the angular position of each sampling moment within a single rotation cycle in the valid data segment, which can be obtained by taking the cumulative phase modulo 360 degrees. Using phase position, the total power, pitch load, and speed fluctuations collected at different times can be mapped to the angular range within the same rotation cycle.

[0053] In practice, the starting sampling time of the effective data segment can be used as the starting point for phase calculation, or the impeller zero-position signal or encoder zero-position signal recorded by the main control system can be used as the starting point for phase calculation. If an impeller zero-position signal exists, the phase position can be calibrated to 0 degrees when the zero position is triggered; if no zero-position signal exists, the phase at the beginning of the effective data segment can be set as the initial phase, and the phase position at subsequent sampling times can be continuously updated using the speed integral.

[0054] For each sampling moment, the sampling time, phase position, overall power fluctuation, pitch load fluctuation of the corresponding blade, and speed fluctuation can be saved simultaneously. The sampling time is the target sampling moment on a unified time axis, and the overall power fluctuation, pitch load fluctuation of the corresponding blade, and speed fluctuation are data after resampling, interpolation, delay compensation, and missing value processing. If the data at the target sampling moment is marked as an invalid sampling point, that sampling moment will not participate in the feature statistics within the phase window. This ensures that the data at the same sampling moment remains consistent in both the time and phase dimensions.

[0055] When any blade enters the corresponding preset phase window, extract the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window; Overall power fluctuations can be obtained by subtracting the sampled overall power value from the smoothed power or the corresponding reference power within the effective data segment and the target operating condition range. Pitch load fluctuations can be obtained by subtracting the pitch load torque of the corresponding blade from the corresponding reference load or smoothed load. The data within the preset phase window are the data corresponding to the sampling points whose phase positions fall within the window range; if the phase window spans 0 degrees, the window can be split into a segment close to 360 degrees and a segment close to 0 degrees for extraction.

[0056] In practice, the center of the window corresponding to blade A can be determined first. and window half width Forming a phase range Then, iterate through each sampling point within the valid data segment. If the phase position of a sampling point falls within this range, extract the overall power fluctuation, blade A pitch load fluctuation, and speed fluctuation corresponding to that sampling point. The overall power fluctuation can be calculated according to... Calculate, where, This represents the overall power fluctuation at the i-th sampling time. This is the sampled value of the total power. To smooth power or reference power. Pitch load fluctuations can be adjusted according to... Calculate, where, Let be the pitch load fluctuation of blade A at the i-th sampling time. The pitch load torque of blade A is... This represents the smoothed load or reference load corresponding to blade A. Rotational speed fluctuations can be obtained using the steps described above. .

[0057] For the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window, calculate at least one of the mean, peak-to-valley difference, rate of change, and dispersion to obtain the phase window characteristics of the corresponding blade.

[0058] The mean is the average level of the same type of fluctuation data within a preset phase window. The peak-to-valley difference is the difference between the maximum and minimum values ​​of the same type of fluctuation data. The rate of change can be the ratio of the change in fluctuation data between adjacent sampling points within the window to the sampling period, or the ratio of the change in fluctuation data between the start and end sampling points of the window to the corresponding time length. The dispersion can be the standard deviation, variance, or mean absolute deviation. The phase window characteristics can be composed of power fluctuation characteristics, load fluctuation characteristics, and speed fluctuation characteristics. For example, the phase window characteristics of blade A within a certain rotation cycle can be expressed as... Where A represents the blade and q represents the rotation cycle number.

[0059] In practice, data extracted from the same blade, the same rotation period, and the same preset phase window can be statistically analyzed separately. Taking overall power fluctuation as an example, if there are n sampling points within the window, the average power fluctuation can be calculated using the following formula: ;in, This represents the average power fluctuation within the window. This represents the number of sampling points within the window. Let represent the power fluctuation corresponding to the i-th sampling point within the window. The peak-to-valley difference of the power fluctuation can be calculated using the following formula: ;in, The peak-to-valley difference of power fluctuation within the window is used. The mean, peak-to-valley difference, rate of change, and dispersion of pitch load fluctuation and speed fluctuation can be calculated using the same method. Finally, the calculated power fluctuation statistics, load fluctuation statistics, and speed fluctuation statistics can be combined to form the phase window characteristics of the corresponding blade.

[0060] For example, combined Figure 2 In a three-bladed wind turbine generator set at a wind farm, the effective data segment W1 obtained in the aforementioned steps can be used as the processing object. The installation phases of blades A, B, and C in the rotor circumference direction differ by 120 degrees sequentially. The preset phase window centers for blades A, B, and C can be set to 0 degrees, 120 degrees, and 240 degrees respectively, with the same half-width h for each window. The rotational speed sequence R within W1 is calculated and accumulated according to the sampling period Δt to obtain the phase position φ corresponding to each sampling moment.

[0061] When the phase position of the sampling point within a certain rotation cycle falls into Figure 2 When sampling the fan-shaped shaded area corresponding to blade A, the overall power fluctuation ΔP, the pitch load fluctuation ΔLA, and the speed fluctuation ΔR of blade A can be extracted from these sampling points. The mean, peak-to-valley difference, rate of change, and dispersion are then calculated to obtain the phase window feature F(A,q) of blade A within that rotation cycle. Subsequently, the same method is applied to the fan-shaped shaded areas corresponding to blades B and C, yielding the phase window features F(B,q) and F(C,q), respectively. Repeating this process for multiple consecutive rotation cycles yields the phase window feature sequence for each blade across multiple rotation cycles.

[0062] In one feasible implementation, the method further includes: Based on the peak phase position of the phase window characteristics corresponding to the pitch load fluctuation within multiple consecutive rotation cycles, the peak load phase position of the target blade is determined; the phase offset between the peak load phase position and the center of the preset phase window corresponding to the target blade is calculated; if the phase offset exceeds the preset offset threshold, the peak load phase position is used as the adjusted center of the phase window, the preset phase window corresponding to the target blade is redefined, and the phase window characteristics of the target blade are determined based on the redefined preset phase window.

[0063] In this embodiment, the load peak phase position is the statistical result of the phase position corresponding to the peak value of the pitch load fluctuation of the target blade within multiple consecutive rotation cycles. It can be determined by the median or circumferential average of the peak phase positions corresponding to multiple rotation cycles. The phase offset is the angular deviation between the load peak phase position and the center of the preset phase window corresponding to the target blade. The preset offset threshold can be determined based on the offset distribution of the load peak phase position of each blade relative to the center of the corresponding window in historical normal operation data, or it can be set in combination with the phase sampling resolution and the impeller speed fluctuation range.

[0064] In practice, the currently processed blade can be designated as target blade A, and any rotation cycle among multiple consecutive rotation cycles can be designated as the q-th rotation cycle. For the phase window of target blade A within the q-th rotation cycle, the sampling point with the largest absolute value of pitch load fluctuation can be found, and the phase position corresponding to this sampling point can be determined as the peak phase position within that rotation cycle. The calculation method can be as follows: ;in, The peak phase position of the target blade A in the qth rotation cycle is expressed in degrees. Let A be the phase window of the target blade A during the qth rotation cycle; This represents the phase position corresponding to the i-th sampling point, in degrees; Let represent the pitch load fluctuation of the target blade A at the i-th sampling point.

[0065] After obtaining the peak phase positions corresponding to multiple consecutive rotation cycles, obviously outlier peak phase positions can be eliminated, and the remaining peak phase positions can be statistically analyzed to obtain the load peak phase position of target blade A. Since the phases can be consecutive at 0 degrees and 360 degrees, the signed offset of each peak phase position relative to the center of the original preset phase window can be calculated first. ;in, Let be the signed offset of the target blade A in the qth rotation cycle, in degrees; The center of the original preset phase window corresponding to target blade A is given in degrees. Then, multiple... Take the median or average value to obtain the statistical value of the phase offset of the target blade A. and with As the phase offset magnitude, The positive or negative sign indicates the direction of adjustment of the phase window center.

[0066] like Greater than the preset offset threshold The adjusted phase window center is then determined according to the following formula: ;in, The center of the phase window after adjustment for target blade A, in degrees; This is a preset offset threshold, in degrees. This represents the phase shift amplitude of the target blade A. Then, using... Centered on the target blade A and maintaining the original window half-width h, or re-determining the window half-width h based on the historical load peak phase distribution, a new phase window is formed. Then, following the method in S130, the overall power fluctuation, pitch load fluctuation, and speed fluctuation of the target blade A are re-extracted within the new phase window, and the mean, peak-to-valley difference, rate of change, and dispersion are recalculated to obtain the updated phase window characteristics of the target blade A.

[0067] By extracting overall power fluctuations, pitch load fluctuations, and speed fluctuations within preset phase windows corresponding to each blade based on equivalent phase data, the power and speed changes at the overall system level can be allocated to specific blade windows according to the blade rotation phase, thereby establishing a correspondence between common operating data and specific blades. Furthermore, calibrating the preset phase windows by the load peak phase position can reduce the impact of installation deviations, phase calculation errors, or phase drift caused by long-term operation on the window extraction results, improving the consistency between phase window characteristics and actual blade stress changes.

[0068] S140. Statistically analyze the phase window characteristics within multiple consecutive rotation cycles to obtain the power fluctuation index, load fluctuation index, and speed fluctuation index of each blade, and compare them with the index correlation benchmark to obtain the coupling deviation characteristics of each blade. The index correlation benchmark includes the time delay coupling relationship and phase synchronization coupling relationship between the power fluctuation index, load fluctuation index, and speed fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval.

[0069] The power fluctuation index, load fluctuation index, and speed fluctuation index are the deviations of the overall power fluctuation, pitch load fluctuation, and speed fluctuation relative to the corresponding reference window characteristics over multiple consecutive rotational cycles. These can include one or more of the following: deviation amplitude, deviation rate of change, number of cycles of deviation duration, and deviation dispersion. The index correlation benchmark is the correlation between different indices obtained statistically from historical valid data segments within the same target operating condition range. This can include the reference time difference range between the load fluctuation index and the power fluctuation index, the correspondence of the reference change direction, and the reference synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window.

[0070] In practical implementation, the phase window characteristics of blade A within Q consecutive rotation cycles can be used as an example to decompose the phase window characteristics within each rotation cycle into power fluctuation window characteristics. Load fluctuation window characteristics and speed fluctuation window characteristics Where q represents the rotation cycle number. Then, the power reference window characteristics, load reference window characteristics, and speed reference window characteristics are determined using the historical valid data segment corresponding to the target operating condition interval C. The window characteristics within the current consecutive rotation cycles are compared with the corresponding reference window characteristics to obtain the power fluctuation index of blade A. Load fluctuation index and speed fluctuation index After that, and The actual time difference and actual direction of change between them, and and The actual synchronous change amplitudes between the two are compared with the index correlation benchmark to obtain the coupling deviation characteristics of blade A. Blades B and C can be processed in the same way.

[0071] In one feasible implementation, the speed fluctuation data is the deviation of the speed within the effective data segment from the reference speed corresponding to the target operating condition range; the power fluctuation index, load fluctuation index and speed fluctuation index of each blade are obtained by statistically analyzing the phase window characteristics within multiple consecutive rotation cycles, including: determining the power reference window characteristics, load reference window characteristics and speed reference window characteristics respectively based on the historical effective data segments corresponding to the target operating condition range.

[0072] The reference speed corresponding to the target operating condition range can be determined based on historical valid data segments within that range. Specifically, the average, median, or average of the speeds within the historical valid data segments after removing outliers can be calculated as the reference speed for the target operating condition range. For example, it can be determined using the following formula: ;in, The reference speed corresponding to operating condition range C can be expressed in revolutions per minute; M is the number of historical valid data segments within operating condition range C. Let be the average rotational speed within the m-th historical valid data segment. For the i-th sampling time within the current valid data segment, it can be calculated according to... Calculate the speed fluctuation data, where, The rotational speed fluctuation corresponding to the i-th sampling time. Let be the rotational speed at the i-th sampling time.

[0073] In practice, the historical power fluctuation window features, historical load fluctuation window features, and historical speed fluctuation window features of each blade can be extracted from the historical valid data segment corresponding to the target operating condition interval C, following the same phase window division method as S130. Taking blade A as an example, the median or mean of the historical window features can be taken to obtain the power reference window features. Load reference window features and speed reference window features If the window feature includes multiple components such as mean, peak-to-valley difference, rate of change, and dispersion, then the benchmark window feature also adopts the same component structure, so that the current window feature and the benchmark window feature are consistent in data dimension and unit.

[0074] The power fluctuation index is obtained by comparing the phase window characteristics corresponding to the overall power fluctuation over multiple consecutive rotational cycles with the power reference window characteristics; the load fluctuation index is obtained by comparing the phase window characteristics corresponding to the pitch load fluctuation over multiple consecutive rotational cycles with the load reference window characteristics; and the speed fluctuation index is obtained by comparing the phase window characteristics corresponding to the speed fluctuation over multiple consecutive rotational cycles with the speed reference window characteristics.

[0075] In practice, the difference between the current window features and the corresponding baseline window features for each rotation cycle can be calculated. Taking blade A as an example, the difference can be calculated as follows: , , ;in, , and These represent the power window characteristic deviation, load window characteristic deviation, and speed window characteristic deviation of blade A during the q-th rotational cycle. If the window characteristics are multi-component, the above calculations can be performed separately for each component. Then, based on the deviation results over Q consecutive rotational cycles, the deviation amplitude, deviation rate of change, deviation duration period, and deviation dispersion can be calculated to obtain the power fluctuation index. Load fluctuation index and speed fluctuation index .

[0076] Power fluctuation index, load fluctuation index, and speed fluctuation index all include at least one of deviation amplitude, deviation change rate, deviation duration period, and deviation dispersion.

[0077] Taking the power fluctuation index as an example, the deviation amplitude can be the average or maximum absolute value of the power window characteristic deviation over Q consecutive rotational cycles; the deviation change rate can be the ratio of the deviation change between adjacent rotational cycles to the rotational cycle time; the deviation duration period can be the number of consecutive cycles in which the power window characteristic deviation exceeds the corresponding deviation threshold; and the deviation dispersion can be the standard deviation of the power window characteristic deviation over Q consecutive rotational cycles. The load fluctuation index and speed fluctuation index can be calculated using the same method, except that the corresponding data are replaced with the load window characteristic deviation and the speed window characteristic deviation, respectively.

[0078] In one feasible implementation, the index association benchmark includes the benchmark time difference range between the load fluctuation index and the power fluctuation index, the benchmark change direction correspondence, and the benchmark synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window; the coupling deviation features include the deviation of the actual time difference between the load fluctuation index and the power fluctuation index from the benchmark time difference range, the deviation of the actual change direction correspondence between the load fluctuation index and the power fluctuation index from the benchmark change direction correspondence, and the deviation of the actual synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window from the benchmark synchronous change amplitude.

[0079] In practical implementation, the historical valid data segment corresponding to the target operating condition interval C can be used to construct the indicator correlation benchmark. The historical valid data segment used to construct the indicator correlation benchmark is the historical normal sample, which is first screened by operating status and health status. The operating status screening is used to remove control intervention segments such as shutdown, power limitation, yaw adjustment, braking, and pitch system failure; the health status screening is used to remove data segments containing blade abnormal alarms, pitch load abnormal alarms, power abnormal alarms, speed abnormal alarms, maintenance work orders, abnormal conclusions of manual review, and data within the preset transition time before and after blade maintenance or replacement. If the number of historical normal samples in the same target operating condition interval is lower than the preset sample number threshold, the target operating condition interval can be merged with adjacent wind speed intervals, adjacent speed intervals, or adjacent pitch angle intervals and re-counted; if it is still insufficient after merging, historical normal samples obtained from the same type of wind turbine generator set through the same screening rules can be used as supplementary samples.

[0080] First, following the same phase window extraction method as S130 and the same index calculation method as this step, calculate the historical power fluctuation index, historical load fluctuation index, and historical speed fluctuation index for each blade in the historical valid data segment. Then, determine the change judgment threshold based on the distribution of each index in the historical normal samples. For example, determine the high quantile of the absolute value of the historical load fluctuation index as the load change judgment threshold, the high quantile of the absolute value of the historical power fluctuation index as the power change judgment threshold, and the high quantile of the absolute value of the historical speed fluctuation index as the speed change judgment threshold. After the unit completes blade maintenance, replacement, main control parameter adjustment, or long-term operating environment changes, the above historical normal sample screening and index statistics process can be repeated, and the original index correlation benchmark can be replaced with the newly obtained index correlation benchmark.

[0081] Next, within each historical valid data segment, the rotation period or sampling time at which the load fluctuation index first exceeds the load change judgment threshold is recorded, and the rotation period or sampling time at which the power fluctuation index first exceeds the power change judgment threshold is also recorded. Based on the difference between these two corresponding times, the historical time difference between the load fluctuation index and the power fluctuation index within that historical valid data segment is obtained. By statistically analyzing the historical time differences obtained from multiple historical valid data segments, the lower and higher quantile values ​​can be used to form a baseline time difference range. ,in, and These represent the lower and upper limits of the reference time difference range between the load fluctuation index and the power fluctuation index under the target operating condition range C, respectively.

[0082] Simultaneously, a baseline correlation of change directions can be constructed based on the changing directions of load fluctuation indicators and power fluctuation indicators in historical valid data segments. Specifically, when a historical load fluctuation indicator exceeds the load change judgment threshold, its main deviation direction can be determined to be either increasing or decreasing; when a historical power fluctuation indicator exceeds the power change judgment threshold, its main deviation direction can be determined to be either increasing or decreasing. If both increase or both decrease, it is recorded as a same-direction relationship; if one increases and the other decreases, it is recorded as an opposite-direction relationship. By statistically analyzing the directional relationships of multiple historical valid data segments under the target operating condition interval C, the directional relationship that occurs most frequently or whose proportion exceeds a preset proportion threshold can be determined as the baseline correlation of change directions. If neither the same-direction nor the opposite-direction relationship reaches the preset ratio threshold, the corresponding relationship of the reference change direction can be set to an unrestricted direction relationship.

[0083] To determine the baseline synchronous change amplitude between load fluctuation and speed fluctuation indices, their normalized change amplitudes can be calculated within the same preset phase window. Specifically, historical load fluctuation and speed fluctuation indices can be normalized first, for example, by dividing the load fluctuation index by the historical standard deviation of the load fluctuation index under the target operating condition range C, and dividing the speed fluctuation index by the historical standard deviation of the speed fluctuation index under the target operating condition range C. Then, the absolute value of the difference or the correlation coefficient between the two normalized change amplitudes within the same preset phase window can be calculated to obtain the historical synchronous change amplitude of a single historical valid data segment. Statistical analysis of the historical synchronous change amplitudes of multiple historical valid data segments can be performed, and the mean, median, or normal range can be used as the baseline synchronous change amplitude. It can also form a reference synchronous change range. .

[0084] When using an index-based benchmark, we can take blade A as an example and use the load fluctuation index over the current Q consecutive rotation cycles. Power fluctuation index and speed fluctuation index The actual time difference, actual direction of change, and actual synchronous change amplitude of blade A are determined. Specifically, the moment when the load fluctuation index of blade A first exceeds the load change judgment threshold and the moment when the power fluctuation index of blade A first exceeds the power change judgment threshold can be recorded; the difference between the two is the actual time difference. .like fall into Then the time difference deviation =0; if Less than ,but ;like Greater than ,but .

[0085] For the directional deviation, the corresponding relationship between the actual change direction can be determined based on the main deviation directions of the load fluctuation index and power fluctuation index of blade A. .like If the directional relationship is not specified, then the directional deviation is... It can be 0; if and If they are consistent, then the directional deviation is... Set to 0; if the two are inconsistent, then the direction deviation is... Choose 1, or a preset directional deviation value. For the synchronization deviation, the load fluctuation index and speed fluctuation index of blade A can be normalized in the same way as historical samples to calculate the actual synchronization change amplitude. If the benchmark for the related indicator is adopted, the benchmark's synchronous change range will be used. Then the synchronization deviation can be If the benchmark associated with the indicator is the range of synchronous changes in the benchmark. Then when When the value falls within this range, the synchronization deviation is 0; when it is less than the lower limit, the synchronization deviation is... When it exceeds the upper limit, the synchronization deviation is .

[0086] Therefore, the coupling deviation characteristic of blade A can be expressed as: ;in, This represents the coupling deviation characteristic of blade A. This represents the time difference deviation between the load fluctuation index and the power fluctuation index. This represents the directional deviation between the load fluctuation index and the power fluctuation index. This represents the synchronous deviation between the load fluctuation index and the speed fluctuation index. The coupling deviation characteristics of blades B and C can be obtained using the same method. and .

[0087] For example, for the aforementioned valid data segment W1, the target operating condition interval C to which W1 belongs can be determined first, and multiple historical samples can be obtained from the historical valid data segments corresponding to operating condition interval C. For these historical samples, the historical power fluctuation index, historical load fluctuation index, and historical speed fluctuation index of blade A, blade B, and blade C can be calculated respectively, and the reference time difference range can be obtained accordingly. Correspondence of reference change direction and the magnitude of synchronous change of the benchmark Then, statistics are performed on Q consecutive rotation cycles within the current W1. If the load fluctuation index of blade A changes significantly before the power fluctuation index, then the result is obtained based on the corresponding times of the two. And compared with the reference time difference range to obtain Then compare the actual direction of change of blade A. and ,get Finally, the actual synchronous change amplitude of the load fluctuation index and the speed fluctuation index of blade A within the same preset phase window is calculated. and with Comparison Finally, the coupling deviation characteristics of blade A were obtained. Performing the same treatment on blades B and C yields the following results: and .

[0088] By constructing an index correlation benchmark in the historical valid data segment corresponding to the target operating condition range, and comparing the current power fluctuation index, load fluctuation index, and speed fluctuation index with this index correlation benchmark, it is possible to avoid relying solely on a single fixed threshold to judge blade anomalies. By simultaneously considering the time delay relationship and directional relationship between the load fluctuation index and the power fluctuation index, as well as the phase synchronization relationship between the load fluctuation index and the speed fluctuation index, abnormal deviations can be identified from multiple operational response dimensions, making the coupled deviation characteristics more reflective of changes in the blade's aerodynamic and structural states.

[0089] S150. Based on the time-delay deviation of the load fluctuation index and the power fluctuation index in the coupling deviation characteristics, and the phase synchronous deviation of the load fluctuation index and the speed fluctuation index, determine the abnormal blades and the abnormal type.

[0090] Figure 3 This illustration shows a schematic diagram of the abnormal blade and abnormality type determination process provided in one embodiment of this application.

[0091] like Figure 3As shown, the coupling deviation characteristics of the target blade can be obtained first, and then the aerodynamic anomaly confidence and structural anomaly confidence can be calculated based on the time delay deviation, directional deviation and synchronization deviation. Then it is determined whether any anomaly condition is met. If not, the target blade is determined to be normal. If it is met, the anomaly type is determined based on the condition met and the difference between the two types of confidence.

[0092] An abnormal blade is defined as a blade whose coupling deviation characteristics meet the anomaly judgment criteria within the target operating condition range corresponding to the current valid data segment. Anomaly types can include aerodynamic deformation anomalies, structural deformation anomalies, and combined deformation anomalies. Aerodynamic deformation anomalies mainly correspond to abnormal load-power response relationships caused by changes in blade aerodynamic shape, surface condition, or aerodynamic forces. When changes occur in blade aerodynamic shape, surface roughness, icing, contamination, or aerodynamic force distribution, the response sequence and direction of the target blade pitch load fluctuation relative to the overall engine power fluctuation will deviate from the historical normal relationship under the same operating conditions. Therefore, the time lag deviation and directional deviation between the load fluctuation index and the power fluctuation index are used to characterize aerodynamic deformation anomalies. Structural deformation anomalies mainly correspond to abnormal load-speed synchronization caused by changes in blade structural stiffness, bending-torsional state, or local deformation. When the blade structural stiffness, bending-torsional coupling state, or local deformation state changes, the synchronization amplitude between pitch load fluctuations and impeller speed fluctuations within the same phase window will deviate from the historical normal relationship under the same operating conditions. Therefore, the synchronization deviation between load fluctuation index and speed fluctuation index is used to characterize structural deformation anomalies. Composite deformation anomalies indicate significant deviations in both aerodynamic and structural relationships, corresponding to the situation where the same target blade simultaneously meets both aerodynamic and structural anomaly conditions, and the difference between the two confidence levels does not exceed the composite judgment difference.

[0093] In specific implementation, it can be done according to Figure 3The illustrated process handles blades A, B, and C respectively. Taking blade A as an example, in S301, the coupling deviation feature E(A) of blade A is obtained, where E(A) includes time difference deviation Eτ(A), directional deviation EG(A), and synchronization deviation ES(A). In S302, the aerodynamic anomaly confidence level of blade A is calculated based on the time difference deviation Eτ(A) and directional deviation EG(A), and the structural anomaly confidence level of blade A is calculated based on the synchronization deviation ES(A). In S303, it is determined whether the aerodynamic anomaly confidence level or the structural anomaly confidence level meets the corresponding anomaly conditions; if neither is met, S304 is executed to determine that blade A is normal; if at least one is met, S305 is executed to determine the anomaly type based on the condition met and the difference between the two confidence levels. If the aerodynamic anomaly condition is met alone, or both conditions are met and the aerodynamic anomaly confidence level dominates, then execute S306 to determine it as an aerodynamic deformation anomaly. If the structural anomaly condition is met alone, or both conditions are met and the structural anomaly confidence level dominates, then execute S307 to determine it as a structural deformation anomaly. If both conditions are met and the difference between the two confidence levels does not exceed the composite judgment difference, then execute S308 to determine it as a composite deformation anomaly. Blade B and blade C are processed in the same way.

[0094] In one feasible implementation, the abnormal blade and abnormal type are determined based on the time-delay deviation between the load fluctuation index and the power fluctuation index, and the phase synchronization deviation between the load fluctuation index and the speed fluctuation index in the coupling deviation characteristics. This includes determining the aerodynamic anomaly confidence level of the target blade based on the time-delay deviation and directional deviation between the load fluctuation index and the power fluctuation index.

[0095] The aerodynamic anomaly confidence level is the degree of certainty that the target blade has experienced aerodynamic deformation anomalies, and can be a dimensionless value between 0 and 1. Time delay deviation corresponds to the deviation of the response sequence between load fluctuation indicators and power fluctuation indicators from the reference time difference range, while directional deviation corresponds to the deviation of the increase / decrease direction relationship between load fluctuation indicators and power fluctuation indicators from the reference change direction relationship. To avoid directly adding different dimensions, the time difference deviation can be normalized first, and then weighted and calculated with the directional deviation.

[0096] In practical implementation, taking blade A as an example, the time difference deviation normalization scale can be determined first based on the historical valid data segment under the target operating condition interval C. The normalization scale can be taken as the high quantile of the time difference deviation in historical normal samples, or as the width of the baseline time difference range or a field experience value. Then, the normalized value of the time delay deviation is calculated using the following formula: ;in, Let be the normalized value of the time delay deviation of blade A. This represents the time difference deviation of blade A. The time difference deviation from the normalized scale corresponding to the target operating condition interval C. Indicates when When it is greater than 1, take 1; when Take when less than or equal to 1 Directional deviation It can be 0 or 1, or a preset directional deviation value between 0 and 1.

[0097] The confidence level for aerodynamic anomalies can be calculated using the following formula: ;in, Let be the confidence level of the aerodynamic anomaly of blade A. For time-delay deviation weights, For the direction deviation weight, and The weights can be determined based on historical verification samples; if there are no anomaly labeled samples, the two weights can be set to the same value, or the time lag deviation weight can be made slightly higher than the direction deviation weight.

[0098] The confidence level of structural anomaly of the target blade is determined based on the synchronous deviation between the load fluctuation index and the speed fluctuation index.

[0099] The structural anomaly confidence level is the degree of certainty that the target blade has undergone structural deformation anomalies, and can be a dimensionless value between 0 and 1. The synchronization deviation corresponds to the deviation of the load fluctuation index and the speed fluctuation index from the baseline synchronization fluctuation index within the same preset phase window. Since pitch load and speed are different physical quantities, and the synchronization deviation has already been calculated using a normalization method in the preceding steps, the structural anomaly confidence level can be directly obtained by normalizing the synchronization deviation.

[0100] In practice, the normalization scale of synchronization deviation can be determined based on the historical valid data segment under the target operating condition interval C. The normalization scale can be taken as the high quantile of the synchronous deviation in historical normal samples, or as the width of the baseline synchronous variation range. Taking blade A as an example, the structural anomaly confidence level can be calculated using the following formula: ;in, Let A be the confidence level for structural anomaly of leaf A. Let A be the synchronization deviation of blade A. The normalized scale for the synchronization deviation corresponding to the target operating condition interval C. Indicates when When it is greater than 1, take 1; when Take when less than or equal to 1 .

[0101] If it is necessary to improve the ability to suppress short-term fluctuations, it can also be modified in combination with the number of cycles of deviation duration. For example, the structural anomaly confidence level can be retained only when the number of cycles in which the synchronous deviation exceeds the synchronous deviation threshold reaches a preset number of cycles; otherwise, the structural anomaly confidence level can be reduced.

[0102] If the confidence level of aerodynamic anomaly meets the preset anomaly conditions, or the confidence level of structural anomaly meets the preset anomaly conditions, the target blade is identified as an anomalous blade. Based on the relationship between the confidence levels of aerodynamic anomaly and structural anomaly, the anomaly type of the target blade is determined to be one of aerodynamic deformation anomaly, structural deformation anomaly, or combined deformation anomaly.

[0103] Preset anomaly conditions can include confidence threshold conditions and persistence conditions. A confidence threshold condition can be that the confidence level of the aerodynamic anomaly is greater than or equal to the aerodynamic anomaly threshold. Or the confidence level of the structural anomaly is greater than or equal to the structural anomaly threshold. The aerodynamic and structural anomaly thresholds can be determined based on the confidence distribution of historical normal samples within the target operating condition range C. For example, the high quantile of the confidence level of historical normal samples can be used, or it can be corrected by combining manually reviewed samples or historical fault samples. The persistence condition can be that multiple consecutive judgment windows meet the confidence threshold condition, which is used to reduce misjudgments caused by instantaneous disturbances.

[0104] In specific implementation, taking blade A as an example, if and If so, then leaf A is not necessarily an abnormal leaf. or If so, then leaf A is identified as an abnormal leaf. The abnormality type can be determined according to the following rules: if only If the preset anomaly conditions are met, the anomaly type is aerodynamic deformation anomaly; if only... If the preset anomaly conditions are met, the anomaly type is structural deformation anomaly; if both conditions are met, their magnitudes are compared, and the anomaly type is determined when the difference between the two is less than or equal to the composite judgment difference. When the anomaly type is a complex deformation anomaly, when When the anomaly type is aerodynamic deformation anomaly, when At that time, the anomaly type was structural deformation anomaly. Among them, The preset difference used to distinguish between single dominant anomalies and compound anomalies can be set based on historical review samples or experience.

[0105] For example, for the aforementioned valid data segment W1, the coupling deviation characteristics of blade A have already been obtained in S140. You can first... Divide by the time difference deviation from the normalized scale corresponding to the target operating condition interval C ,get Combined with the directional deviation Calculate the confidence level of aerodynamic anomalies At the same time, the synchronization deviation will be... Divided by synchronous deviation normalization scale The structural anomaly confidence level was obtained. .

[0106] If the aerodynamic anomaly confidence level of blade A meets the aerodynamic anomaly threshold condition, but the structural anomaly confidence level does not meet the structural anomaly threshold condition, then blade A is identified as an anomalous blade, and the anomaly type is aerodynamic deformation anomaly. If the structural anomaly confidence level of blade B meets the structural anomaly threshold condition, but the aerodynamic anomaly confidence level does not meet the aerodynamic anomaly threshold condition, then blade B is identified as an anomalous blade, and the anomaly type is structural deformation anomaly. If both the aerodynamic and structural anomaly confidence levels of blade C meet their respective threshold conditions, and the difference between them does not exceed the composite judgment difference, then blade C is identified as an anomalous blade, and the anomaly type is composite deformation anomaly.

[0107] By calculating the confidence scores for aerodynamic and structural anomalies separately, and determining the anomalous blades and anomaly types based on the conditions for satisfying these two confidence scores and their relative magnitudes, it is possible to distinguish between aerodynamic deformation anomalies, structural deformation anomalies, and combined deformation anomalies while identifying anomalous blades. This allows maintenance personnel to take more targeted verification or maintenance measures based on the anomaly type, such as prioritizing checks on the aerodynamic condition of the blade surface, the blade structural condition, or a combination of both, thereby improving the application value of anomaly monitoring results.

[0108] In summary, the wind turbine blade deformation anomaly monitoring method provided in this application can utilize existing operating data, pitch motor current, speed data, and power data of the wind turbine generator set for blade deformation anomaly monitoring. It eliminates the need for additional radar arrays, built-in blade sensors, or dedicated vibration sensors, thereby reducing on-site modification and maintenance costs. By filtering valid data segments and determining the target operating condition range, current data can be compared with historical valid data under the same or similar operating conditions, reducing the impact of normal changes in wind speed, speed, and pitch angle on anomaly detection.

[0109] Furthermore, by using equivalent phase data and preset phase windows corresponding to each blade, fluctuations in overall power, pitch load, and speed can be allocated to the rotational phase range corresponding to specific blades, thereby establishing a correspondence between common operating data and specific blades. By constructing an index correlation benchmark and obtaining coupling deviation characteristics, abnormal deviations can be identified from the time delay relationship and directional relationship between load and power, as well as the phase synchronization relationship between load and speed, no longer relying solely on a single fixed threshold. This improves the accuracy and stability of abnormal blade identification and further distinguishes between aerodynamic deformation anomalies, structural deformation anomalies, and combined deformation anomalies, enhancing the reference value of monitoring results for operation and maintenance decisions.

[0110] Figure 4 This is a schematic diagram of a specific embodiment of a wind turbine blade deformation anomaly monitoring system provided in this application, with reference to... Figure 4 The system may include: The acquisition module 410 is used to acquire the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data and the speed fluctuation data. The operating data includes wind speed, speed, total power and pitch angle. The pitch load data is the torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is the phase representation quantity obtained by integrating the speed. The determination module 420 is used to filter valid data segments from the operating data and determine the target operating condition interval to which they belong. The operating condition interval is obtained by dividing the wind speed, rotational speed and pitch angle of the historical valid data segments. The extraction module 430 is used to extract the overall power fluctuation, pitch load fluctuation and speed fluctuation from the effective data segment within the preset phase window corresponding to each blade based on the equivalent phase data, so as to form the phase window features of each blade. The comparison module 440 is used to statistically analyze the phase window characteristics within multiple consecutive rotation cycles to obtain the power fluctuation index, load fluctuation index and speed fluctuation index of each blade, and compare them with the index association benchmark to obtain the coupling deviation characteristics of each blade. The index association benchmark includes the time delay coupling relationship and phase synchronization coupling relationship between the power fluctuation index, load fluctuation index and speed fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval. The determination module 420 is also used to determine the abnormal blades and abnormal types based on the time delay deviation of the load fluctuation index and the power fluctuation index and the phase synchronization deviation of the load fluctuation index and the speed fluctuation index in the coupling deviation characteristics.

[0111] The wind turbine blade deformation anomaly monitoring system of this application is used to implement the aforementioned wind turbine blade deformation anomaly monitoring method. Therefore, the specific implementation of the wind turbine blade deformation anomaly monitoring system can be found in the embodiment section of the wind turbine blade deformation anomaly monitoring method above. The specific implementation can be referred to the description of the corresponding embodiment, and will not be repeated here.

[0112] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application is shown.

[0113] The electronic device may include a processor 510 and a memory 520 storing computer program instructions.

[0114] Specifically, the processor 510 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0115] Memory 520 may include mass storage for data or instructions. For example, and not limitingly, memory 520 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where suitable, memory 520 may include removable or non-removable (or fixed) media. Where suitable, memory 520 may be internal or external to an electronic device. In a particular embodiment, memory 520 is a non-volatile solid-state memory.

[0116] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this disclosure.

[0117] The processor 510 reads and executes computer program instructions stored in the memory 520 to implement any of the wind turbine blade deformation anomaly monitoring methods in the above embodiments.

[0118] In one example, the electronic device may also include a communication interface 530 and a bus 540. Wherein, such as Figure 5 As shown, the processor 510, memory 520, and communication interface 530 are connected through bus 540 and complete communication with each other.

[0119] The communication interface 530 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0120] Bus 540 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 540 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0121] The electronic device can execute the wind turbine blade deformation anomaly monitoring method in the embodiments of this application, thereby realizing the wind turbine blade deformation anomaly monitoring method described in conjunction with the accompanying drawings.

[0122] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0123] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0124] The foregoing has provided a detailed description of the method, system, electronic equipment, and storage medium for monitoring abnormal deformation of wind turbine blades provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for monitoring abnormal deformation of wind turbine blades, characterized in that, include: The system acquires the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data, and the speed fluctuation data. The operating data includes wind speed, speed, total power, and pitch angle. The pitch load data is a torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is a phase representation quantity obtained by integrating the speed. Valid data segments are filtered from the operational data and their corresponding target operating conditions are determined. The operating conditions are divided based on the wind speed, rotational speed, and pitch angle of the historical valid data segments. Based on the equivalent phase data, the overall power fluctuation, pitch load fluctuation and speed fluctuation are extracted from the effective data segment within the preset phase window corresponding to each blade to form the phase window features of each blade. The power fluctuation index, load fluctuation index, and speed fluctuation index of each blade are statistically obtained by analyzing the phase window characteristics within multiple consecutive rotation cycles. The coupling deviation characteristics of each blade are obtained by comparing them with the index association benchmark. The index association benchmark includes the benchmark time difference range and benchmark change direction correspondence between the load fluctuation index and the power fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval, as well as the benchmark synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window. The coupling deviation features include the deviation of the actual time difference between the load fluctuation index and the power fluctuation index from the reference time difference range, the deviation of the actual change direction correspondence between the load fluctuation index and the power fluctuation index from the reference change direction correspondence, and the deviation of the actual synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window from the reference synchronous change amplitude. Based on the time delay and directional deviation between the load fluctuation index and the power fluctuation index, the aerodynamic anomaly confidence level of the target blade is determined. This aerodynamic anomaly confidence level is calculated using the formula... The calculation yielded that, Let be the confidence level of the aerodynamic anomaly of blade A. Let be the normalized value of the time delay deviation of blade A. This represents the directional deviation of blade A. For time-delay deviation weights, For the direction deviation weight, and ; The structural anomaly confidence level of the target blade is determined based on the synchronous deviation between the load fluctuation index and the speed fluctuation index. This structural anomaly confidence level is calculated using the formula... The calculation yielded that, Let A be the confidence level for structural anomaly of leaf A. Let A be the synchronization deviation of blade A. The normalized scale for the synchronization deviation corresponding to the target operating condition interval C. Indicates when When it is greater than 1, take 1; when Take when less than or equal to 1 ; If the confidence level of the aerodynamic anomaly meets the preset anomaly condition, or the confidence level of the structural anomaly meets the preset anomaly condition, then the target blade is identified as an anomalous blade, and based on the relationship between the confidence levels of the aerodynamic anomaly and the structural anomaly, the anomaly type of the target blade is determined to be one of aerodynamic deformation anomaly, structural deformation anomaly, or combined deformation anomaly.

2. The method according to claim 1, characterized in that, The step of filtering valid data segments from the operating data and determining their corresponding target operating condition ranges includes: If the wind speed change, rotational speed change, overall power change, and pitch angle change all meet the corresponding preset stability thresholds within the target time period, and there are no shutdown, power limitation, yaw adjustment, braking, or pitch system fault markers within the target time period, then the target time period is determined as the valid data segment. Based on the wind speed, rotational speed, and pitch angle in the effective data segment, the wind speed range to which the wind speed belongs, the rotational speed range to which the rotational speed belongs, and the pitch angle range to which the pitch angle belongs are determined respectively. The operating condition range that simultaneously corresponds to the wind speed range, the rotational speed range, and the pitch angle range is determined as the target operating condition range to which the effective data segment belongs.

3. The method according to claim 1, characterized in that, The equivalent phase data is obtained by determining the angle increment between adjacent sampling times based on the rotational speed and sampling period, and then accumulating the angle increment. The preset phase window for each blade is obtained by dividing a single rotation cycle according to the number of blades and the installation phase difference of each blade.

4. The method according to claim 3, characterized in that, Based on the equivalent phase data, the process of extracting overall power fluctuation, pitch load fluctuation, and speed fluctuation from the effective data segment within a preset phase window corresponding to each blade to form the phase window features of each blade includes: Determine the phase position corresponding to each sampling time according to the equivalent phase data; When any blade enters the corresponding preset phase window, the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window are extracted. For the overall power fluctuation, pitch load fluctuation, and speed fluctuation within the preset phase window, calculate at least one of the mean, peak-to-valley difference, rate of change, and dispersion to obtain the phase window characteristics of the corresponding blade.

5. The method according to claim 1, characterized in that, The speed fluctuation data is the deviation of the speed within the effective data segment from the reference speed corresponding to the target operating condition range; The statistical analysis of the phase window characteristics over multiple consecutive rotation cycles to obtain the power fluctuation index, load fluctuation index, and speed fluctuation index for each blade includes: Based on the historical valid data segments corresponding to the target operating condition range, the power reference window characteristics, load reference window characteristics, and speed reference window characteristics are determined respectively. The power fluctuation index is obtained by comparing the phase window features corresponding to the power fluctuation of the whole machine within multiple consecutive rotation cycles with the power reference window features. The load fluctuation index is obtained by comparing the phase window features corresponding to the pitch load fluctuation within multiple consecutive rotation cycles with the load reference window features. The speed fluctuation index is obtained by comparing the phase window features corresponding to the speed fluctuations within multiple consecutive rotation cycles with the speed reference window features. The power fluctuation index, the load fluctuation index, and the speed fluctuation index all include at least one of the following: deviation amplitude, deviation change rate, deviation duration period, and deviation dispersion.

6. The method according to claim 4, characterized in that, The method further includes: The peak phase position of the target blade load is determined based on the peak phase position of the phase window characteristics corresponding to the pitch load fluctuation within multiple consecutive rotation cycles. Calculate the phase offset between the peak load phase position and the center of the preset phase window corresponding to the target blade; If the phase offset exceeds a preset offset threshold, the preset phase window corresponding to the target blade is redefined using the load peak phase position as the center of the adjusted phase window, and the phase window characteristics of the target blade are determined based on the redefined preset phase window.

7. A wind turbine blade deformation anomaly monitoring system, characterized in that, include: The acquisition module is used to acquire the operating data of the wind turbine generator set, the pitch load data of each blade, the equivalent phase data and the speed fluctuation data. The operating data includes wind speed, speed, total power and pitch angle. The pitch load data is a torque representation quantity calculated from the pitch motor current corresponding to each blade. The equivalent phase data is a phase representation quantity obtained by integrating the speed. The determination module is used to filter valid data segments from the operating data and determine the target operating condition interval to which they belong. The operating condition interval is obtained by dividing the wind speed, rotational speed and pitch angle of the historical valid data segments. The extraction module is used to extract the overall power fluctuation, pitch load fluctuation and speed fluctuation from the effective data segment within the preset phase window corresponding to each blade based on the equivalent phase data, so as to form the phase window features of each blade. The comparison module is used to statistically analyze the phase window features within multiple consecutive rotation cycles to obtain the power fluctuation index, load fluctuation index, and speed fluctuation index of each blade, and compare them with the index association benchmark to obtain the coupling deviation characteristics of each blade. The index association benchmark includes the benchmark time difference range and benchmark change direction correspondence between the load fluctuation index and the power fluctuation index constructed based on the historical valid data segment corresponding to the target operating condition interval, as well as the benchmark synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window. The coupling deviation features include the deviation of the actual time difference between the load fluctuation index and the power fluctuation index from the reference time difference range, the deviation of the actual change direction correspondence between the load fluctuation index and the power fluctuation index from the reference change direction correspondence, and the deviation of the actual synchronous change amplitude of the load fluctuation index and the speed fluctuation index within the same preset phase window from the reference synchronous change amplitude. The determination module is also used to determine the aerodynamic anomaly confidence level of the target blade based on the time delay deviation and directional deviation between the load fluctuation index and the power fluctuation index. The aerodynamic anomaly confidence level is determined by the formula... The calculation yielded that, Let be the confidence level of the aerodynamic anomaly of blade A. Let be the normalized value of the time delay deviation of blade A. This represents the directional deviation of blade A. For time-delay deviation weights, For the direction deviation weight, and Based on the synchronous deviation between the load fluctuation index and the speed fluctuation index, the structural anomaly confidence level of the target blade is determined. This structural anomaly confidence level is calculated using the formula... The calculation yielded that, Let A be the confidence level for structural anomaly of leaf A. Let A be the synchronization deviation of blade A. The normalized scale for the synchronization deviation corresponding to the target operating condition interval C. Indicates when When it is greater than 1, take 1; when Take when less than or equal to 1 If the confidence level of the aerodynamic anomaly meets the preset anomaly condition, or the confidence level of the structural anomaly meets the preset anomaly condition, then the target blade is identified as an anomalous blade, and the anomaly type of the target blade is determined to be one of aerodynamic deformation anomaly, structural deformation anomaly, or combined deformation anomaly based on the relationship between the confidence levels of the aerodynamic anomaly and the structural anomaly.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the wind turbine blade deformation anomaly monitoring method as described in any one of claims 1 to 6.

Images