A method and device for monitoring a wind turbine variable pitch device

Abstract

The application specifically discloses a kind of wind turbine variable pitch device monitoring method and device, it discards the monitoring mode based on static historical baseline of tradition, instead constructs a kind of adaptive dynamic cross reference model.In any variable pitch event, the real-time dynamic response of any one blade to be measured is compared with the average dynamic response of the remaining blades of the same unit at the same time, thereby generating a deviation signal that can effectively filter out common operating condition interference and purely reflect the relative performance difference.Further, the deviation signal is decomposed into dynamic action error and dynamic consumption error two dimensions, by the information fusion analysis to these two dimensions, not only can the inconsistency degree of variable pitch action be accurately quantified, but also can the physical root cause behind inconsistency be preliminarily diagnosed according to the combination characteristics of different error components, thereby solving the technical problems of low monitoring accuracy, susceptible to operating condition interference and lack of diagnostic capability.

Description

Technical Field

[0001] This application relates to the field of wind turbine pitch system monitoring, and more specifically, to a monitoring method and apparatus for a wind turbine pitch device. Background Technology

[0002] The pitch control system of a wind turbine generator is one of its core control and safety protection systems. By adjusting the pitch angle of the three blades in real time, it optimizes wind energy capture efficiency and provides aerodynamic braking under extreme wind conditions. The reliability of the pitch control system directly affects the generator's power generation performance and operational safety. Under ideal operating conditions, the three blades should achieve perfectly synchronized and consistent pitch control under a unified command. However, during long-term service, the pitch drive chains of each blade (including motors, drivers, backup power supplies, and mechanical transmission components) inevitably experience performance degradation, and the rate of degradation often differs. This leads to deviations in the actual dynamic response of the three blades to the same pitch command, a problem known as pitch inconsistency. This inconsistency causes aerodynamic load imbalances between the blades, exacerbating generator vibration and structural fatigue. This not only reduces power generation efficiency but, in severe cases, can induce permanent damage to critical components, posing a significant safety hazard. Therefore, developing a solution capable of accurately and in real-time monitoring the consistency of the wind turbine pitch control system is crucial for ensuring the long-term stable operation of wind turbine generators.

[0003] Existing technologies typically employ the following methods: One common approach is to rely on fault codes built into the wind turbine's main control or pitch controller for post-event diagnosis. However, this method can only respond to severe faults that have already occurred and cannot effectively warn of gradual performance degradation and early inconsistencies. Another method involves analyzing low-frequency (minute-level) data collected by SCADA (Supervisory Control and Data Acquisition) systems. However, pitch control is a rapid dynamic process lasting several seconds, and low-frequency average data cannot capture the instantaneous deviation details during the process, thus missing diagnostic opportunities. Some solutions attempt to establish a health baseline model for individual blades based on historical data, identifying anomalies by comparing current actions with historical baselines. However, this method is extremely sensitive to changing external conditions (such as wind speed, turbulence, and ambient temperature) and cannot adapt to the normal performance aging of the system itself. It is prone to generating numerous false alarms due to outdated reference benchmarks, resulting in poor practicality. The fundamental flaw in these existing solutions is that they either lack the necessary time resolution or rely on a static reference frame that cannot adapt to changes in operating conditions, making it difficult to accurately quantify and diagnose the minute differences in the relative performance between multiple blades.

[0004] Therefore, an optimized monitoring scheme for wind turbine pitch control devices is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a monitoring method and apparatus for a wind turbine pitch control device.

[0006] According to one aspect of this application, a monitoring method for a wind turbine pitch control device is provided, comprising: In response to the detection of a change in the pitch angle command, a pitch event window is collected. The pitch event window includes multiple synchronous data frames, each of which includes a timestamp, the actual pitch angle of each blade, and the motor current of each blade. Input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade; For the first The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade Dynamic wear error of each blade; For the first The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

[0007] According to another aspect of this application, a monitoring device for a wind turbine pitch control system is provided, comprising: The pitch change practice acquisition module is used to collect pitch change event windows in response to the detection of pitch angle command changes. The pitch change event window includes multiple synchronous data frames, each of which includes a timestamp, the actual pitch angle of each blade, and the motor current of each blade. The dynamic cross-reference module is used to input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade; The deviation calculation module is used to calculate the deviation of the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade Dynamic wear error of each blade; The diagnostic alarm module is used to monitor the first... The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

[0008] Compared with existing technologies, the monitoring method and apparatus for wind turbine pitch control provided in this application abandon the traditional monitoring mode based on static historical baselines and instead construct an adaptive dynamic cross-reference model. In any pitch control event, the real-time dynamic response (including actual angle and motor current) of any blade under test is compared with the average dynamic response of the other blades in the same unit at the same moment, thereby generating a deviation signal that effectively filters out interference from common operating conditions and purely reflects relative performance differences. Furthermore, this deviation signal is decomposed into two dimensions: dynamic action error and dynamic consumption error. By fusing and analyzing the information from these two dimensions, not only can the degree of inconsistency in pitch control be accurately quantified, but also, based on the combined characteristics of different error components, a preliminary diagnosis can be made of the physical root causes behind the inconsistency (such as increased mechanical resistance or abnormal control system), thus solving the technical problems of low monitoring accuracy, susceptibility to operating condition interference, and lack of diagnostic capabilities. Attached Figure Description

[0009] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a flowchart of a monitoring method for a wind turbine pitch control device according to an embodiment of this application; Figure 2 This is a schematic diagram of data flow in a monitoring method for a wind turbine pitch control device according to an embodiment of this application; Figure 3 For the monitoring method of a wind turbine pitch device according to an embodiment of this application, the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade A flowchart of the dynamic wear error of each blade; Figure 4 For the monitoring method of a wind turbine pitch device according to an embodiment of this application, the first... The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... A flowchart of the pitch consistency health indicators and alarm information for each blade. Figure 5 This is a block diagram of a monitoring device for a wind turbine pitch control system according to an embodiment of this application. Detailed Implementation

[0011] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0012] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0013] While this application makes various references to certain modules in the apparatus according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules are merely illustrative, and different aspects of the apparatus and methods may use different modules.

[0014] Flowcharts are used in this application to illustrate the operations performed by the apparatus according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0015] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0016] To address the technical problem of existing wind turbine pitch monitoring methods being susceptible to interference from changes in operating conditions due to their reliance on static baselines, resulting in low monitoring accuracy and an inability to effectively warn of early inconsistencies, this application proposes a monitoring method for wind turbine pitch devices. This method first captures the complete pitch event window triggered by pitch angle command changes in real time, acquiring high-frequency synchronous data containing the actual angles of all blades and the motor current. Further, a dynamic cross-reference model is constructed: for any blade under test, this model uses the average performance of the other blades in the same unit at the same moment as its dynamic and ideal reference benchmark, thereby generating a dynamic reference angle sequence and a dynamic reference current sequence for the blade under test. Subsequently, by comparing and accumulating the actual sequence of the blade under test with these two dynamic reference sequences point by point, this method quantifies the dynamic action error and dynamic consumption error from two dimensions: action response and energy consumption. Finally, the errors from these two dimensions are fused with a consistency health index to generate a health score that comprehensively reflects the severity of inconsistency. Based on this score, diagnostic alerts are issued, thereby achieving high-precision, adaptive monitoring and early warning of inconsistency in the pitch system under complex and variable operating conditions.

[0017] Figure 1 This is a flowchart of a monitoring method for a wind turbine pitch device according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow in a monitoring method for a wind turbine pitch control device according to an embodiment of this application. Figure 1 and Figure 2 As shown, a monitoring method for a wind turbine pitch control device according to an embodiment of this application includes the following steps: S100, in response to detecting a change in pitch angle command, acquiring a pitch event window, wherein the pitch event window includes multiple synchronous data frames, each synchronous data frame including a timestamp, the actual pitch angle of each blade, and the motor current of each blade; S200, inputting the pitch event window into a dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence for the first blade; S300, for the first blade... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of the first blade; S400, for the first blade The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

[0018] Specifically, in step S100, in response to detecting a change in the pitch angle command, a pitch event window is acquired. This pitch event window includes multiple synchronous data frames, each containing a timestamp, the actual pitch angle of each blade, and the motor current of each blade. It should be understood that pitch inconsistency is a transient characteristic that only manifests during the dynamic change of blade angles. Continuously acquiring high-frequency data would generate massive amounts of redundant information, while low-frequency data would lose crucial dynamic process details. Therefore, in response to detecting a change in the pitch angle command, a pitch event window is acquired, comprising multiple synchronous data frames, each containing a timestamp, the actual pitch angle of each blade, and the motor current of each blade. This accurately captures a complete, high-resolution data snapshot from the start to the end of the pitch maneuver. This provides a data foundation with complete dynamic process information and a high signal-to-noise ratio for subsequent dynamic cross-reference modeling and inconsistency quantification analysis, ensuring the accuracy of the diagnosis and the efficiency of the calculation.

[0019] More specifically, in a concrete example of this application, the process begins with a monitoring module continuously listening to the pitch angle command issued by the wind turbine main controller. Once the monitoring module detects that the change in the pitch angle command value exceeds a preset trigger threshold, such as 0.1 degrees, the system determines that a pitch event has begun and records the current moment as the start time of the event. Subsequently, the system initiates a data acquisition program, which, within a preset time window, such as 5 seconds, or until the actual pitch angle feedback values ​​of all blades stabilize near the new command value, frequently acquires raw data streams of actual pitch angle and motor current from the angle encoders and motor drivers of each blade. During the acquisition process, a time synchronization unit uses a high-precision clock protocol to assign a unified timestamp to all data points acquired at the same physical moment and combines these data into a synchronization data frame. This process is executed continuously throughout the event window, ultimately merging all generated synchronization data frames to form a structured pitch event window data block, which is then output to subsequent processing steps.

[0020] Specifically, in step S200, the pitch event window is input into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade. It should be understood that using fixed historical data as a health benchmark cannot adapt to the real-time changing operating conditions of the wind turbine, leading to high susceptibility to misjudgments under external disturbances such as gusts and turbulence, thus reducing the reliability of the monitoring system. Therefore, in the technical solution of this application, the pitch event window is further input into the dynamic cross-reference model to obtain the dynamic reference current sequence of the first blade. The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade is used to construct a performance benchmark for each blade that dynamically adapts to the current actual operating conditions and is defined by the real-time behavior of its companion blades. This effectively isolates the influence of common operating modes, allowing subsequent deviation calculations to accurately focus on the true inconsistencies caused by the performance degradation of individual blades, greatly improving the sensitivity and accuracy of monitoring.

[0021] More specifically, in the embodiments of this application, the pitch event window is input into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence for the blade includes: the dynamic cross-reference model processes the pitch event window using the following formula to obtain the first blade. The dynamic reference angle sequence of each blade is given by the following formula:

[0022] in, For the first The actual pitch angle of each blade For the first The actual pitch angle of each blade For the first The dynamic reference angle of the first blade. Furthermore, the dynamic cross-reference model processes the pitch event window using the following formula to obtain the first blade's dynamic reference angle. The dynamic reference current sequence for each blade is given by the following formula:

[0023] in, For the first The motor current for each blade, For the first The motor current for each blade, For the first Dynamic reference current for each blade.

[0024] In a specific example of this application, the process begins with receiving the pitch event window data block generated in the previous step. First, the target blade to be analyzed is determined, for example, blade number 1. Then, each synchronous data frame in the event window is traversed. When processing the data frame corresponding to the first timestamp, the actual pitch angles of blades number 2 and 3 are extracted, their arithmetic mean is calculated, and this result is used as the dynamic reference angle of blade number 1 at that moment. Simultaneously, the motor currents of blades number 2 and 3 are extracted, their arithmetic mean is calculated, and this is used as the dynamic reference current of blade number 1 at that moment. This calculation process is repeated sequentially for all subsequent timestamps within the event window until the last data frame is processed. Finally, all calculated dynamic reference angle values ​​are arranged in chronological order, forming a complete dynamic reference angle sequence for blade number 1, and all dynamic reference current values ​​constitute the dynamic reference current sequence for blade number 1. After that, the system switches the target blade to blade number 2, and uses the data of blades number 1 and 3 as a reference to repeat the above process to generate a unique reference sequence for blade number 2, and finally completes the operation for blade number 3.

[0025] Specifically, in step S300, for the first The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade The dynamic consumption error of each blade. It is understandable that different fault modes in a pitch system exhibit different characteristics in terms of dynamic response and energy consumption. For example, increased mechanical friction can simultaneously lead to sluggish action and increased current, while control parameter drift may only manifest as sluggish action. Simply merging the deviation information would lose crucial information for distinguishing fault types. Therefore, in the technical solution of this application, a two-dimensional inconsistency deviation calculation is further performed on the dynamic reference angle sequence and the dynamic reference current sequence of the i-th blade to obtain the dynamic action error and dynamic consumption error of the i-th blade. This transforms the original time-series deviation data into two independent scalar indicators with clear physical meaning. This provides a two-dimensional error vector containing rich feature information for subsequent health status assessment and fault diagnosis, thereby significantly improving the depth and accuracy of diagnosis.

[0026] Figure 3 For the monitoring method of a wind turbine pitch device according to an embodiment of this application, the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade A flowchart illustrating the dynamic wear error of each blade. (See attached flowchart.) Figure 3 As shown, step S300 includes: S310, extracting the first... The actual pitch angle sequence of the first blade and the... The motor current sequence of the first blade; S320, based on the first blade. The actual pitch angle sequence of the first blade and the... The dynamic reference angle sequence of the nth blade is used to calculate the nth blade. The instantaneous angle deviation sequence of the first blade; S330, based on the first blade The dynamic reference current sequence of the first blade and the first blade Calculate the motor current sequence of the nth blade. The instantaneous current deviation sequence of the blade; S340, respectively for the blade... The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade.

[0027] Accordingly, in steps S310, S320, and S330, the first [event] is extracted from the pitch event window. The actual pitch angle sequence of the first blade and the... The motor current sequence of the nth blade, based on the nth blade. The actual pitch angle sequence of the first blade and the... The dynamic reference angle sequence of the nth blade is used to calculate the nth blade. The instantaneous angle deviation sequence of the blade, and based on the first blade... The dynamic reference current sequence of the first blade and the first blade Calculate the motor current sequence of the nth blade. The instantaneous current deviation sequence of each blade. It should be understood that since the original actual sequence and reference sequence can only macroscopically reflect the dynamic process and cannot directly quantify the degree of difference between them at each moment, the subsequent calculation of cumulative error lacks a direct data basis. Therefore, in the technical solution of this application, the actual pitch angle sequence and the motor current sequence of the i-th blade are further extracted from the pitch event window. Based on the actual pitch angle sequence and the dynamic reference angle sequence of the i-th blade, the instantaneous angle deviation sequence of the i-th blade is calculated. Furthermore, based on the dynamic reference current sequence and the motor current sequence of the i-th blade, the instantaneous current deviation sequence of the i-th blade is calculated. This makes the inconsistency information implicit in the original data explicit, generating two instantaneous error streams representing motion deviation and energy consumption deviation, respectively. This provides accurate and unbiased data input for the subsequent calculation of cumulative deviation indicators, ensuring the accuracy of the final error assessment.

[0028] In a specific example of this application, the process is based on receiving a pitch event window and a dynamic reference angle sequence and a dynamic reference current sequence generated for the i-th blade. First, the actual pitch angle sequence and actual motor current sequence of the i-th blade are located and extracted from the pitch event window. Then, a vector operation is performed on the actual pitch angle sequence and dynamic reference angle sequence of the i-th blade, subtracting each element and taking the absolute value. The result constitutes a new time series, namely the instantaneous angle deviation sequence of the i-th blade. In parallel, using the exact same processing logic, the actual motor current sequence and dynamic reference current sequence of the i-th blade are processed to generate the instantaneous current deviation sequence of the i-th blade. Finally, these two newly generated instantaneous deviation sequences are output together to the next calculation stage.

[0029] Accordingly, in step S340, the first... The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade. It should be understood that, since the instantaneous deviation sequence is a time-series vector containing multiple data points, it cannot be directly used for the final assessment of health status, nor is it easy to effectively fuse multi-dimensional information. Therefore, in the technical solution of this application, the cumulative deviation index is further calculated for the instantaneous current deviation sequence and the instantaneous angle deviation sequence of the i-th blade to obtain the dynamic motion error and dynamic wear error of the i-th blade. This integrates the time-series deviation information throughout the pitching event process, condensing it into two single scalar values ​​that represent the overall inconsistency of the event. This provides stable, quantifiable, and dimensionally unified input features for subsequent health indicator fusion and alarm decision-making, greatly simplifying the analysis model and enhancing the robustness of the assessment results.

[0030] Specifically, in the embodiments of this application, the first... The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of the blade includes: calculated using the following formula for the blade... The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of each blade is given by the following formula:

[0031] in, For the first The instantaneous angle deviations in the instantaneous angle deviation sequence of each blade, The sampling time interval, For the first The number of instantaneous angle deviations in the instantaneous angle deviation sequence of each blade. For the first The dynamic motion error of each blade.

[0032] Furthermore, the following formula is used for the first The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic wear error of each blade is given by the following formula:

[0033] in, For the first The instantaneous current deviation of each blade in the instantaneous current deviation sequence of the blades. The sampling time interval, For the first The dynamic wear error of each blade.

[0034] More specifically, this process is executed by a cumulative calculation module that receives the instantaneous angle deviation sequence and instantaneous current deviation sequence of the i-th blade generated in the previous stage. First, the module sums all values ​​in the instantaneous angle deviation sequence to obtain a cumulative sum. Then, this cumulative sum is multiplied by a preset data sampling time interval parameter in the system; the result of this product is defined as the dynamic motion error of the i-th blade. In parallel, the module performs the same operation on the instantaneous current deviation sequence: first, it sums all values ​​in the sequence, then it multiplies the sum by the data sampling time interval to obtain the dynamic consumption error of the i-th blade. Finally, these two calculated scalar error values ​​are output together for subsequent health status assessment.

[0035] Specifically, in step S400, for the first The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The method involves monitoring the pitch consistency health indicators and alarm information for each blade. It's understandable that existing pitch system inconsistency monitoring methods use a linear weighted summation approach when fusing health indicators. This approach has a fundamental technical flaw: it forcibly compresses a richly informative two-dimensional error vector (i.e., a vector composed of dynamic motion error and dynamic attrition error) into a single one-dimensional scalar health indicator. This dimensionality reduction directly leads to the permanent loss of crucial diagnostic details. Specifically, different physical fault sources, such as increased mechanical friction and control loop instability, exhibit drastically different characteristic patterns in the two-dimensional error space. The former typically shows a simultaneous increase in both motion error and attrition error, while the latter may only show a sharp increase in motion error. However, through linear weighting, these two distinct fault modes may be incorrectly mapped to the same health indicator value, causing the system to lose its ability to distinguish and identify specific fault types. Furthermore, this method fails to effectively model the co-occurrence of motion error and attrition error. In a healthy state, both errors should ideally remain at extremely low levels. When both deviate significantly simultaneously, this is an anomalous signal much stronger than a deviation from a single indicator. The linear summation model treats the two errors as independent contributions, failing to impose the appropriate and more severe penalty weights on this high-risk co-occurrence phenomenon, essentially ignoring the joint probability distribution describing the system state. Furthermore, relying on static weights set based on expert experience makes the model lack adaptability to different wind conditions and operating conditions, thus greatly limiting its robustness and accuracy. To overcome the above technical deficiencies, in a preferred embodiment of this application, a preferred solution is proposed. This preferred solution maps two-dimensional error information to the complex domain and combines it with a probabilistic statistical model for analysis, thereby achieving accurate assessment and in-depth diagnosis of inconsistencies in the pitch system. In other words, the technical solution of this application further integrates and diagnoses the dynamic motion error and dynamic consumption error of the i-th blade using consistency health indicators to obtain the pitch consistency health indicator and alarm information of the i-th blade. This allows the construction of a decision model that can comprehensively assess the severity of the two-dimensional error vector and interpret its inherent patterns. This enables a leap from simple anomaly detection to in-depth fault diagnosis, generating not only a comprehensive indicator that accurately reflects the system's health status but also outputting alarm information with clear physical directionality, providing precise guidance for subsequent maintenance work.

[0036] Figure 4 For the monitoring method of a wind turbine pitch device according to an embodiment of this application, the first... The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... A flowchart illustrating the pitch consistency health indicators and alarm information for each propeller blade. (Example:) Figure 4 As shown, step S400 includes: S410, based on the first The dynamic motion error of the first blade and the first blade The dynamic consumption error of the first blade is used to construct the second blade. The complex anomaly vector of the first blade; S420, calculate the first blade based on binary Gaussian likelihood. The complex anomaly vector of the nth blade The pitch consistency health index of each blade; S430, based on the first blade. The complex anomaly vector of the nth blade and the nth blade The alarm information is generated based on the pitch consistency health index of each blade.

[0037] Accordingly, in step S410, based on the first The dynamic motion error of the first blade and the first blade The dynamic consumption error of the first blade is used to construct the second blade. The complex anomaly vector of the i-th blade. It should be understood that, due to the use of linear weighting and other methods to fuse dynamic motion errors and dynamic consumption errors, a two-dimensional error vector containing rich diagnostic information is inevitably compressed into a single one-dimensional scalar. This information dimensionality reduction operation directly leads to the permanent loss of key diagnostic details, causing fault modes with vastly different properties to be incorrectly mapped to the same health indicator value. Therefore, in the technical solution of this application, a complex anomaly vector of the i-th blade is further constructed based on the dynamic motion error and dynamic consumption error of the i-th blade. This employs a mathematical carrier that can completely preserve two-dimensional error information, avoiding information loss during the fusion process. In this way, the abstract abnormal state can be visualized as a vector point on the complex plane. The distance of this vector point to the origin (i.e., the modulus of the complex number) intuitively reflects the overall severity of the anomaly, while the angle between this point and the positive direction of the real axis (i.e., the argument of the complex number) contains the inherent properties or type information of the anomaly, laying a data foundation for subsequent leaps from anomaly detection to fault root cause diagnosis.

[0038] Specifically, in a specific example of this application, the implementation of this process begins with receiving the dynamic motion error and dynamic consumption error of the i-th blade calculated in the previous step. First, these two error values ​​are normalized, for example, by dividing their respective values ​​by the maximum value or a specific quantile value obtained statistically from historical healthy operation data, thereby obtaining the dimensionless normalized dynamic motion error (…). ) and normalized dynamic consumption error ( Subsequently, the normalized dynamic motion error is used as the real part of the complex number, and the normalized dynamic consumption error is used as the imaginary part of the complex number. Based on the definition of a complex number, a complex anomaly vector for the i-th blade is constructed. Finally, this complex anomaly vector, containing complete information about the two-dimensional error, is output for subsequent probabilistic health indicator calculations and diagnostic analysis.

[0039] Accordingly, in step S420, the first... The complex anomaly vector of the nth blade The pitch consistency health index for each blade. It should be understood that the simple linear weighted summation method fails to effectively model the co-occurrence of dynamic motion errors and dynamic consumption errors, and relies on static, expert-experience-based weights, resulting in a lack of adaptability to different operating conditions, thus greatly limiting its robustness and accuracy. Therefore, a health measurement method with greater statistical significance and sensitivity than linear weighted summation is needed. In the technical solution of this application, the pitch consistency health index of the i-th blade is further calculated based on the complex anomaly vector of the i-th blade using binary Gaussian likelihood. This replaces the rigid, manually set weights with a dynamic, data-driven probabilistic model, thereby quantifying the degree to which the current abnormal state deviates from the joint probability distribution of the healthy state. This allows for the utilization of the natural property of Gaussian functions to impose heavier penalties on error co-occurrence, more sensitively capturing the early signs of failure, and generating a pitch consistency health index that accurately reflects the probability of anomaly occurrence.

[0040] Specifically, in a concrete example of this application, the process is implemented based on a pre-built probabilistic health status model. This model models the health status as a zero-mean bivariate Gaussian distribution at the origin of the complex plane by analyzing a large number of complex anomaly vectors in historical normal operation data, and learns the standard deviation parameters of this distribution along the real and imaginary axes. When the complex anomaly vector of the i-th blade generated in the previous step is received... First, the real part representing the normalized dynamic motion error and the imaginary part representing the normalized dynamic consumption error are extracted. Then, these two component values, along with the pre-learned standard deviation parameter, are substituted into the probability density function of a bivariate Gaussian distribution to calculate the probability density value of the current complex anomaly vector under the health model. Finally, the system subtracts this probability density value from 1, and the difference is defined as the pitch consistency health index of the i-th blade. A very low probability event corresponds to a very high health index value, and this output is used for the final alarm decision.

[0041] Accordingly, in step S430, based on the first The complex anomaly vector of the nth blade and the nth blade The alarm information is generated based on the pitch consistency health index of each blade. It should be understood that since a single pitch consistency health index can only reflect the severity of the anomaly but cannot reveal its underlying physical cause, maintenance personnel still need to perform complex troubleshooting work after receiving the alarm. Therefore, in the technical solution of this application, the alarm information is further generated based on the complex anomaly vector of the i-th blade and the pitch consistency health index of the i-th blade. This utilizes the phase angle information of the complex anomaly vector to classify the identified anomaly state into fault types, because different angular regions of the complex plane are given clear physical meanings. This enables a leap from general anomaly detection to refined fault diagnosis. The final output alarm information not only includes the anomaly level but also a clear fault type indication, thus providing accurate data support for maintenance decisions.

[0042] Specifically, in a concrete example of this application, the process first compares the pitch consistency health index of the i-th blade calculated in the previous step with a preset alarm threshold. If the health index exceeds the alarm threshold, the system determines that the i-th blade is abnormal and immediately initiates a fault diagnosis procedure. In this procedure, the system extracts the complex anomaly vector of the corresponding i-th blade and calculates its phase angle. Subsequently, the system generates specific alarm information based on the predefined angle range into which the phase angle falls. For example, if the phase angle is between the mechanical fault pre-set threshold (e.g., 30 degrees) and the control fault threshold (e.g., 60 degrees), the system generates an alarm indicating a high probability of increased mechanical friction in blade i; if the phase angle is less than the mechanical fault pre-set threshold, an alarm is generated indicating a suspected control or sensor problem; if the phase angle is greater than the control fault threshold, an alarm is generated indicating a suspected drive or motor efficiency problem. Finally, complete alarm information, including severity level and diagnostic type, is output to the monitoring system.

[0043] More specifically, the system based on this phase angle The predefined angular range into which the fault falls is used to classify the root cause of the fault. It should be understood that different angular regions of the complex plane are given clear physical meanings; for example, smaller positive angles (close to the real axis) may correspond to control or sensor problems, while an angle of approximately 45 degrees may point to mechanical friction problems. This makes the phase angle a powerful fault classification feature.

[0044] For example, according to The angle range of the fall will generate corresponding alarm information: if The warning indicates a high probability of increased mechanical friction in blade i. Both motion error and wear error will increase. To anticipate mechanical failures, for example, 30°. To control the fault threshold, for example, 60°. Otherwise... If so, a warning will be issued: there is a suspected control or sensor problem with propeller i. The motion error is high, while the consumption error is normal. Otherwise, if... If the warning message is received, it indicates a possible driver or motor efficiency problem with propeller i. Consumption error is high, while the motion error is normal.

[0045] Furthermore, a monitoring device for a wind turbine pitch control system is also provided.

[0046] Figure 5 This is a block diagram of a monitoring device for a wind turbine pitch control system according to an embodiment of this application. Figure 5 As shown, a monitoring device 100 for a wind turbine pitch control device according to an embodiment of this application includes: a pitch control practice acquisition module 110, used to acquire a pitch control event window in response to a detected change in pitch angle command, the pitch control event window including multiple synchronous data frames, each synchronous data frame including a timestamp, the actual pitch angle of each blade, and the motor current of each blade; and a dynamic cross-reference module 120, used to input the pitch control event window into a dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the first blade; deviation calculation module 130, used for the first blade... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade The dynamic consumption error of the first blade; the diagnostic alarm module 140, used to monitor the dynamic consumption error of the first blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

[0047] As described above, the wind turbine pitch control device monitoring device 100 according to the embodiments of this application can be implemented in various wireless terminals, such as servers with monitoring algorithms for wind turbine pitch control devices. In one possible implementation, the wind turbine pitch control device monitoring device 100 according to the embodiments of this application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the wind turbine pitch control device monitoring device 100 can be a software module in the operating device of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the wind turbine pitch control device monitoring device 100 can also be one of many hardware modules of the wireless terminal.

[0048] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A monitoring method for a wind turbine pitch control system, characterized in that, include: In response to the detection of a change in the pitch angle command, a pitch event window is collected. The pitch event window includes multiple synchronous data frames, each of which includes a timestamp, the actual pitch angle of each blade, and the motor current of each blade. Input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade; For the first The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade Dynamic wear error of each blade; For the first The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

2. The monitoring method for a wind turbine pitch system according to claim 1, characterized in that, Input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence for the blade includes: the dynamic cross-reference model processes the pitch event window using the following formula to obtain the first blade. The dynamic reference angle sequence of each blade is given by the following formula: in, For the first The actual pitch angle of each blade For the first The actual pitch angle of each blade For the first The dynamic reference angle of each blade.

3. The monitoring method for a wind turbine pitch system according to claim 1, characterized in that, Input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence for the blade includes: the dynamic cross-reference model processes the pitch event window using the following formula to obtain the first blade. The dynamic reference current sequence for each blade is given by the following formula: in, For the first The motor current for each blade, For the first The motor current for each blade, For the first Dynamic reference current for each blade.

4. The monitoring method for a wind turbine pitch system according to claim 1, characterized in that, For the first The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade includes: Extract the first from the pitch event window The actual pitch angle sequence of the first blade and the... The motor current sequence for each blade; Based on the The actual pitch angle sequence of the first blade and the... The dynamic reference angle sequence of the nth blade is used to calculate the nth blade. The instantaneous angle deviation sequence of each blade; Based on the The dynamic reference current sequence of the first blade and the first blade Calculate the motor current sequence of the nth blade. The instantaneous current deviation sequence of each blade; For the first The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade.

5. The monitoring method for a wind turbine pitch system according to claim 4, characterized in that, For the first The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of the blade includes: calculated using the following formula for the blade... The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of each blade is given by the following formula: in, For the first The instantaneous angle deviations in the instantaneous angle deviation sequence of each blade, The sampling time interval, For the first The number of instantaneous angle deviations in the instantaneous angle deviation sequence of each blade. For the first The dynamic motion error of each blade.

6. The monitoring method for a wind turbine pitch system according to claim 4, characterized in that, For the first The instantaneous current deviation sequence of the first blade and the first blade The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic motion error of the first blade and the first blade The dynamic wear error of the blade includes: calculated using the following formula for the blade... The cumulative deviation index is calculated by performing a sequence of instantaneous angle deviations of the first blade to obtain the value of the second blade. The dynamic wear error of each blade is given by the following formula: in, For the first The instantaneous current deviation of each blade in the instantaneous current deviation sequence of the blades. The sampling time interval, For the first The dynamic wear error of each blade.

7. The monitoring method for a wind turbine pitch system according to claim 1, characterized in that, For the first The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade include: Based on the The dynamic motion error of the first blade and the first blade The dynamic consumption error of the first blade is used to construct the second blade. Complex anomaly vectors of each blade; The first calculation is based on the binary Gaussian likelihood. The complex anomaly vector of the nth blade Health indicators for pitch consistency of each blade. Based on the The complex anomaly vector of the nth blade and the nth blade The alarm information is generated based on the pitch consistency health index of each blade.

8. A monitoring device for a wind turbine pitch control system, characterized in that, include: The pitch change practice acquisition module is used to collect pitch change event windows in response to the detection of pitch angle command changes. The pitch change event window includes multiple synchronous data frames, each of which includes a timestamp, the actual pitch angle of each blade, and the motor current of each blade. The dynamic cross-reference module is used to input the pitch event window into the dynamic cross-reference model to obtain the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of each blade; The deviation calculation module is used to calculate the deviation of the first... The dynamic reference angle sequence of the first blade and the first... The dynamic reference current sequence of the blade is used to calculate the two-dimensional inconsistency deviation to obtain the first blade. The dynamic motion error of the first blade and the first blade Dynamic wear error of each blade; The diagnostic alarm module is used to monitor the first... The dynamic motion error of the first blade and the first blade The dynamic wear error of each blade is used to fuse consistent health indicators and generate diagnostic alarms to obtain the first... The pitch consistency health indicators and alarm information for each blade.

Images