Wind power variable pitch fault pre-diagnosis system and method based on digital twinning

Abstract

This invention discloses a wind turbine pitch fault pre-diagnosis system and method based on digital twins, belonging to the field of wind turbine pitch fault pre-diagnosis technology. Specifically, it includes the following steps: when the pitch equipment enters a fault state, continuously monitor the fault recovery process of the pitch equipment and construct a digital twin virtual model to determine whether the pitch equipment exhibits frequent fault state switching; when frequent fault state switching occurs, acquire and analyze the behavior switching information of the pitch equipment during the frequent fault state switching process in real time to determine the switching amplitude of the pitch equipment when frequent fault switching occurs; based on the determination result, dynamically adjust the fault diagnosis process of the virtual model to adapt to the fault switching mode of the pitch equipment. This invention solves the problem of delayed diagnostic response when the pitch equipment frequently switches faults, realizing dynamic control and accurate prediction of the fault diagnosis process.

Description

Technical Field

[0001] This invention relates to the field of wind turbine pitch fault pre-diagnosis technology, specifically to a wind turbine pitch fault pre-diagnosis system and method based on digital twins. Background Technology

[0002] Pitch control failure in wind turbines refers to a problem with the pitch control equipment used to adjust the blade angle of a wind turbine. This typically manifests as the blade angle being unable to adjust properly or inaccurately, preventing the turbine from generating electricity efficiently at different wind speeds, and potentially causing equipment damage or disrupting the normal operation of the wind turbine. Pitch control failures severely impact the power generation efficiency and safety of wind turbines, making timely detection and diagnosis crucial. Traditional fault diagnosis relies on periodic maintenance and manual inspection, which struggles to detect potential faults early, easily missing optimal maintenance opportunities and causing unnecessary downtime losses. Pre-diagnosis of wind turbine pitch control failures based on digital twin technology can map the actual equipment's performance onto a virtual model by monitoring the wind turbine's operating status in real time, enabling continuous monitoring and precise analysis. By comparing the differences between real-time data and the virtual model, digital twins can identify potential problems before a fault occurs and issue early warnings, avoiding the delayed response problems of traditional methods. This significantly improves the timeliness and accuracy of fault diagnosis, thereby reducing wind turbine maintenance costs and improving operational reliability and economic efficiency.

[0003] Existing digital twin-based wind turbine pitch system fault prediction technology combines the actual pitch system of a wind turbine with its virtual model to achieve real-time monitoring and prediction of faults. First, digital twin technology uses sensors to collect equipment operating data, reflecting the real-time status of the pitch equipment and converting it into a virtual model. Next, the virtual model, through data fusion and simulation, accurately simulates the operation of the wind turbine pitch equipment and compares it with actual operating data to analyze potential anomalies or deviations. Based on the deviation between the model and the data, the prediction algorithm uses various data analysis methods, such as fault diagnosis algorithms, pattern recognition, and machine learning, to identify possible fault types and their timing, and provides early warnings. Specific steps include data acquisition, data preprocessing, virtual model building and simulation, fault mode recognition, and early warning notification. Through these steps, digital twin-based wind turbine pitch system fault prediction technology can achieve early detection of faults in wind turbine pitch equipment, accurately predict the probability and location of faults, thus providing sufficient early warning for equipment maintenance and repair, and avoiding significant losses caused by the lag of traditional manual inspections and periodic checks.

[0004] The existing technology has the following shortcomings:

[0005] In the pitch control systems of wind turbines, frequent fault state switching may occur during the fault recovery process. For example, after a fault occurs, the pitch control system attempts to resume operation and adjusts the blade angle based on factors such as load and wind speed. During this process, the equipment frequently enters different fault states and recovers rapidly. Because existing digital twin-based wind turbine pitch control fault pre-diagnosis technologies cannot dynamically adjust the fault diagnosis process of the virtual model according to the switching amplitude of the pitch control system during frequent fault switching, the virtual model fails to adapt to the changes in equipment state in a timely manner, resulting in fault diagnosis that does not accurately reflect the actual operating state of the equipment. Specifically, the virtual model fails to identify and adjust the diagnostic strategy in a timely manner to cope with the rapid changes in equipment state. Due to this delay, the wind turbine fails to identify potential problems in a timely manner when a fault occurs, which may lead to the equipment continuing to operate under suboptimal conditions. Frequent misdiagnosis and missed diagnosis increase equipment maintenance costs, prolong downtime, and reduce the overall power generation efficiency and economic benefits of the wind farm.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a wind turbine pitch fault pre-diagnosis system and method based on digital twins to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a wind turbine pitch fault pre-diagnosis method based on digital twins, specifically including the following steps:

[0009] Real-time monitoring of the operation of the pitch control equipment of the wind turbine to determine whether the pitch control equipment has entered a fault state;

[0010] When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and a digital twin virtual model is constructed to determine whether the pitch control equipment exhibits frequent fault state switching.

[0011] When the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time, analyzed, and the switching amplitude of the pitch control equipment when frequent fault switching occurs is determined.

[0012] Based on the determined results, the fault diagnosis process of the virtual model is dynamically adjusted to adapt to the fault switching mode of the pitch equipment;

[0013] Based on the fault diagnosis results, fault prediction is performed and early warnings are issued to operators. At the same time, historical fault data and diagnosis results of the pitch equipment are recorded to optimize the diagnosis strategy of the virtual model.

[0014] Preferably, when the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and a digital twin virtual model is constructed to determine whether the pitch control equipment exhibits frequent fault state switching. Specifically:

[0015] When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and the recovery status information of the pitch control equipment during the fault recovery process is continuously acquired during the monitoring process.

[0016] Based on the acquired recovery status information, a digital twin virtual model is constructed to simulate the dynamic behavior of the pitch equipment during the fault recovery process, and the pitch equipment status of the virtual model is updated in real time.

[0017] Based on the constructed digital twin virtual model, the system updates and analyzes the equipment status in real time to determine whether the pitch equipment experiences frequent fault state switching during the fault recovery process. The specific judgment logic is as follows: within a preset time period, if the frequency of the transition between the fault and non-fault states of the pitch equipment exceeds a preset frequency threshold, and after each entry into the non-fault state, the pitch equipment fails to maintain the non-fault state for more than a set stable time period, then it is determined that the pitch equipment is experiencing frequent fault state switching.

[0018] Preferably, when the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and analyzed to determine the switching amplitude of the pitch control equipment when frequent fault state switching occurs. Specifically, this includes the following steps:

[0019] When the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and preprocessed after acquisition.

[0020] Multidimensional operational fluctuation information and execution response offset information are extracted from the preprocessed behavior switching information, and analyzed after extraction to generate fluctuation expansion coefficient and state alienation index, respectively.

[0021] A switching amplitude assessment model is constructed based on the generated fluctuation expansion coefficient and state alienation index to generate switching assessment coefficients;

[0022] Determine the pre-set threshold range of the switching evaluation coefficient, and compare it with the generated switching evaluation coefficient. Based on the comparison results, determine the switching amplitude of the pitch equipment when frequent fault switching occurs.

[0023] Preferably, the logic for obtaining the fluctuation expansion coefficient is as follows:

[0024] Multidimensional operational fluctuation information is extracted from the preprocessed behavior switching information. Specifically, this includes the blade angle change value, current load value, and pitch motor speed value at different times over a period of time during frequent fault state switching of the pitch control equipment, and these values ​​are denoted as θ. m I m and ω m θ m I represents the blade angle change at time m over a period of time during frequent fault state switching of the pitch control equipment. m ω represents the current load value at time m during a period of time when the pitch control equipment is undergoing frequent fault state switching. m This represents the rotational speed of the pitch motor at time m during a period of time when the pitch equipment is undergoing frequent fault state switching, where m = 1, 2, 3, ..., g, and g is a positive integer;

[0025] The average value calculation formula is used to determine the blade angle change θ at all times over a period of time during frequent fault state switching of the pitch control equipment. m The average value θ between _ Current load value I at all times m The average value between I - and the speed ω of the pitch motor at all times. m The average value ω between - ;

[0026] The fluctuation spread factor (FEC) is calculated by taking the blade angle change value θ at each moment during a period of time when the pitch control system is undergoing frequent fault state switching. m Its average θ - The absolute value of the difference between them is used to calculate the exponential function; the current load value I at each moment during a period of time in the pitch control equipment during frequent fault state switching is calculated. m Its average value I _ The square root function is calculated by taking the absolute value of the difference between them; the speed ω of the pitch motor at each moment during a period of time during frequent fault state switching of the pitch equipment is calculated. m Its average value ω _ The absolute value of the difference between the values ​​is taken, and then the logarithmic function is calculated by adding one. The three results obtained are summed, and the arithmetic mean of the summation results at all times is taken. Then, the natural logarithm is taken after adding one to the mean. The result is the fluctuation expansion coefficient FEC.

[0027] Preferably, the logic for obtaining the state alienation index is as follows:

[0028] Execution response offset information is extracted from the preprocessed behavior switching information. Specifically, this includes the deviation of the actual blade angle from the preset target angle at different times during a period of time when the pitch control equipment is undergoing frequent fault state switching, the blade angle adjustment range, and the impedance load torque borne by the pitch motor, and is denoted as Δθ. m α m and T m , Δθ m α represents the deviation of the actual blade angle from the preset target angle at time m within a certain period of time during frequent fault state switching of the pitch control equipment. m T represents the blade angle adjustment range at time m within a certain period of time during frequent fault state switching of the pitch control equipment. m This represents the impedance load torque borne by the pitch motor at time m during a period of time when the pitch equipment is undergoing frequent fault state switching, where m = 1, 2, 3, ..., g, and g is a positive integer;

[0029] The State Dissimilarity Index (SDI) is calculated by taking the blade angle adjustment magnitude α at each moment. m As the exponent of the exponential function, it represents the deviation Δθ between the actual blade angle and the preset target angle at each moment. m Adding 1 and using it as the independent variable of the logarithmic function, the square root of the product of the two is used as the first part; the impedance load torque T borne by the pitch motor at each moment is... m The square of the result plus 1 is used as the independent variable of the logarithmic function, and its natural logarithm is calculated as the second part. The results of the above two parts are summed in the range of m=1 to m=g, and divided by the total number of time points g. The average value obtained is the state alienation index SDI.

[0030] Preferably, a switching magnitude assessment model is constructed based on the generated fluctuation expansion coefficient FEC and state alienation index SDI, and a switching assessment coefficient SEC is generated by weighted summation.

[0031] Preferably, a pre-defined threshold range for the switching evaluation coefficient [SEc] is determined. min SEC max After determination, it is compared with the generated switching evaluation coefficient SEC. Based on the comparison results, the switching amplitude of the pitch equipment when frequent fault switching occurs is determined. The specific comparison analysis is as follows:

[0032] If SEC <SEC min When the pitch control equipment experiences frequent fault switching, the switching amplitude is low.

[0033] If SEC min ≤SEC≤SEC maxThe switching amplitude of the pitch control equipment when frequent fault switching occurs is of medium magnitude.

[0034] If SEC > SEC max When the pitch control equipment experiences frequent fault switching, the switching amplitude is high.

[0035] Preferably, based on the determined results, the fault diagnosis process of the virtual model is dynamically adjusted, specifically as follows:

[0036] When the result is determined to be low, keep the fault diagnosis threshold parameters of the current virtual model unchanged and continuously monitor the operating status of the pitch control equipment;

[0037] When the result is determined to be of moderate magnitude, the increase of the fault diagnosis sensitive parameters in the virtual model is adjusted, and the fault identification frequency is increased.

[0038] When the result is determined to be high, the diagnostic sensitive parameters in the virtual model are enhanced by a high amplitude, and the fault mode recognition weight configuration and diagnostic judgment range are adjusted simultaneously to improve the adaptability to abnormal states and the diagnostic accuracy.

[0039] Preferably, the wind turbine pitch fault pre-diagnosis system based on digital twin includes an operation status perception module, a recovery process modeling module, a behavior switching evaluation module, a diagnostic logic control module, and a prediction and early warning optimization module.

[0040] The operation status sensing module monitors the operation of the pitch control equipment of the wind turbine in real time and determines whether the pitch control equipment has entered a fault state.

[0041] The recovery process modeling module continuously monitors the fault recovery process of the pitch equipment when it enters a fault state, and builds a digital twin virtual model to determine whether the pitch equipment is experiencing frequent fault state switching.

[0042] The behavior switching assessment module acquires and analyzes the behavior switching information of the pitch equipment during the frequent fault state switching when the pitch equipment experiences frequent fault state switching, and determines the switching amplitude of the pitch equipment when frequent fault state switching occurs.

[0043] The diagnostic logic control module dynamically adjusts the fault diagnosis process of the virtual model based on the determined results to adapt to the fault switching mode of the pitch equipment.

[0044] The prediction and early warning optimization module predicts faults based on fault diagnosis results and issues early warnings to operators. It also records historical fault data and diagnosis results of the pitch equipment and optimizes the diagnosis strategy of the virtual model.

[0045] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0046] 1. This invention enables dynamic perception and diagnostic response throughout the entire operation of pitch control equipment, eliminating reliance on traditional fault identification mechanisms based on static rules or single parameters. By continuously monitoring the fault recovery process and constructing a digital twin virtual model based on real-time recovery status information, the system can dynamically reconstruct the operating behavior trajectory of the pitch control equipment and identify whether the equipment experiences frequent fault state switching. This overcomes the limitations of existing methods, such as slow response to fault mode changes and delayed identification, significantly enhancing the adaptability and response speed of the diagnostic system.

[0047] 2. This invention establishes a precise quantification mechanism for fault switching amplitude by constructing two high-dimensional comprehensive indicators, the "fluctuation expansion coefficient" and the "state alienation index," and further forming a switching evaluation coefficient. This overcomes the problem of existing technologies being unable to adjust diagnostic strategies based on specific switching amplitudes. In particular, by classifying responses to different switching amplitude levels (low, medium, and high), the system can automatically adjust the diagnostic parameters, judgment intervals, and fault channel weights of the virtual model based on the evaluation results. This enables the diagnostic process to have hierarchical adaptability, effectively reducing misjudgments and omissions, and improving the accuracy and stability of identifying complex dynamic fault states.

[0048] 3. The diagnostic strategy constructed in this invention not only possesses a feedforward response mechanism, but also drives the self-learning of the virtual model and optimization of the diagnostic strategy through the continuous accumulation and archiving of historical fault data and prediction results, thereby achieving intelligent evolution of the diagnostic system. By adjusting fault-sensitive parameters, judgment logic, and evaluation strategies in real time through software, this solution not only improves the operational reliability and maintenance efficiency throughout the equipment's lifecycle, but also possesses high software implementability and platform scalability, providing important technical support and broad engineering application prospects for intelligent operation and maintenance of wind farms. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0050] Figure 1 This is a flowchart illustrating the wind turbine pitch fault pre-diagnosis system and method based on digital twins of the present invention.

[0051] Figure 2 This is a mind map of the method of the wind power pitch fault pre-diagnosis system and method based on digital twins of the present invention.

[0052] Figure 3 This is a schematic diagram of the modules of the wind power pitch fault pre-diagnosis system and method based on digital twins of the present invention. Detailed Implementation

[0053] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0054] This invention provides, for example Figure 1 and Figure 2 The wind turbine pitch fault pre-diagnosis method based on digital twins, as shown, specifically includes the following steps:

[0055] Real-time monitoring of the operation of the pitch control equipment of the wind turbine to determine whether the pitch control equipment has entered a fault state;

[0056] The system can collect various real-time operational data from the wind turbine's pitch control equipment during operation, including but not limited to blade angle changes, pitch motor speed, current load response, control signal feedback, wind speed input and execution deviation, and pitch response delay. This data is then fed into the data processing module in real time. The data processing module uses multiple operational status analysis models and extracts normal state characteristic curves from the equipment's historical operating modes to perform feature comparison analysis on the real-time collected data. The system sets a stable dynamic envelope interval for each key indicator. When continuous real-time data deviates from the envelope interval simultaneously across multiple indicators and fails to revert within a set time threshold, the state discrimination model automatically classifies the operational state as a fault state. This process is implemented through pre-trained software algorithms such as threshold comparison, trend slope analysis, and state variation rate calculation, requiring no external intervention and possessing high real-time performance and automation capabilities.

[0057] During the operation of wind turbines, the pitch control system, as a core component in wind energy conversion control, directly impacts the overall wind energy capture efficiency and load control safety. Failures in the pitch control system, such as response delays, drive motor malfunctions, or abnormal blade angles, can lead to wind energy regulation failures, and in severe cases, cause turbine shutdowns or structural damage. Therefore, continuous, accurate, and real-time monitoring of the equipment's status is crucial for early fault detection, identification, and response. Automated fault state identification via software not only avoids the lag of manual judgment but also completes the judgment process in a very short time, providing reliable time-point anchoring and data foundation for subsequent fault recovery monitoring, virtual model construction, and diagnostic response. This approach standardizes and automates the trigger points for fault diagnosis, serving as the logical prerequisite and fundamental guarantee for initiating the entire pre-diagnosis process.

[0058] When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and a digital twin virtual model is constructed to determine whether the pitch control equipment exhibits frequent fault state switching.

[0059] In this embodiment, when the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and a digital twin virtual model is constructed to determine whether the pitch control equipment exhibits frequent fault state switching. Specifically:

[0060] When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and the recovery status information of the pitch control equipment during the fault recovery process is continuously acquired during the monitoring process.

[0061] To achieve the goal of "continuously monitoring the fault recovery process of the pitch control equipment when it enters a fault state, and continuously acquiring recovery status information during the monitoring process," this can be achieved through real-time data acquisition and processing via software. Specifically, firstly, various sensors installed on the pitch control equipment (such as angle sensors, current sensors, and speed sensors) collect key operating parameters (such as blade angle, motor speed, and current load) in real time, and transmit this information to the data processing system via a data acquisition unit. The software system monitors this sensor data in real time, promptly identifying and marking the fault state when the equipment malfunctions. Once the equipment enters a fault state, the system automatically switches to continuous monitoring mode, periodically or in real-time acquiring current operating data from the sensors and comparing it with standard data under normal conditions. This data is dynamically updated in the virtual model, which reflects the equipment's recovery process and status changes in real time. The continuously acquired recovery status information includes, but is not limited to, the fluctuation range of various equipment parameters, the response time during fault recovery, and the differences from the pre-fault state. This information is used to determine whether the equipment has fully recovered or is still in the fault recovery process, ensuring comprehensive and accurate monitoring of the equipment's fault recovery process.

[0062] Based on the acquired recovery status information, a digital twin virtual model is constructed to simulate the dynamic behavior of the pitch equipment during the fault recovery process. The pitch equipment status of the virtual model is updated in real time to reflect the different state changes of the pitch equipment during the recovery process.

[0063] To achieve the goal of "building a digital twin virtual model based on acquired recovery status information, simulating the dynamic behavior of pitch control equipment during fault recovery, and updating the pitch control equipment status in the virtual model in real time," the following software approach can be used: First, the software system collects real-time data (such as blade angle, current load, and speed) of the pitch control equipment during fault recovery and inputs this data into the virtual model construction process. The digital twin virtual model is built based on the physical attributes and operating characteristics of the equipment, describing its behavior by establishing multi-dimensional models including dynamics, mechanics, and control systems. During model construction, the software compares each piece of data during the fault recovery process with the equipment's historical operating modes, combining physical laws and control algorithms to simulate the entire process from fault to recovery. When the equipment status changes, the virtual model is updated through real-time data input to ensure that the equipment status in the model is consistent with the actual equipment. For example, when the blade angle of the equipment is adjusted, the software calculates the impact of this adjustment on other parameters such as current, motor load, and speed, and updates the equipment status in the virtual model accordingly. This process is dynamic. The system continuously updates the virtual model through real-time data streams to ensure that every state change is reflected, and the model can accurately simulate all dynamic changes during equipment recovery. In this way, the digital twin virtual model can accurately simulate and track the behavior of pitch equipment during fault recovery, updating the equipment status in real time to reflect the different state changes of the actual equipment during the recovery process.

[0064] The purpose of constructing a digital twin virtual model and updating the pitch control equipment status in real time is to ensure accurate reflection of the equipment's real-time operating status and precise simulation of its dynamic behavior during fault recovery. The changes in the pitch control equipment's status during fault recovery are highly complex and can be influenced by various factors, such as load variations, wind speed fluctuations, and mechanical adjustments. By constructing a digital twin virtual model, these changes can be captured in real time and mapped onto the virtual model, thereby obtaining a precise simulation of the equipment's dynamic behavior during the recovery process. This ensures that every state change in the model reflects the actual situation of the equipment in a timely manner, making subsequent fault diagnosis more accurate and timely. Furthermore, real-time updates to the virtual model's status can help the system identify potential problems during the recovery process, such as frequent fault switching, thereby providing optimization adjustments and early warning suggestions to prevent the equipment from continuing to operate under suboptimal conditions due to untimely adjustments, ensuring the stability and operational efficiency of the wind turbine.

[0065] Based on the constructed digital twin virtual model, the system updates and analyzes the equipment status in real time to determine whether the pitch equipment experiences frequent fault state switching during the fault recovery process. The specific judgment logic is as follows: within a preset time period, if the frequency of the transition between the fault and non-fault states of the pitch equipment exceeds a preset frequency threshold, and after each entry into the non-fault state, the pitch equipment fails to maintain the non-fault state for more than a set stable time period, then it is determined that the pitch equipment is experiencing frequent fault state switching.

[0066] To achieve the goal of "determining whether pitch control equipment experiences frequent fault state switching during fault recovery by analyzing and updating equipment status in real time based on a constructed digital twin virtual model," the following approach can be used: First, by collecting multiple key parameters of the equipment in real time (such as blade angle, current load, rotational speed, and wind speed), the software system compares and analyzes this data with preset parameters in the virtual model. The digital twin model is continuously updated, reflecting the current status of the equipment in real time. The system monitors the transition between the equipment's fault and non-fault states, records the time point of each state transition, and calculates whether the equipment can maintain a stable non-fault state for more than a set stable period after each recovery from a fault state. If, within the set time period, the frequency of transitions between the equipment's fault and non-fault states exceeds a preset threshold, and the equipment cannot maintain a stable non-fault state after each state transition, the system will determine that the equipment has experienced frequent fault state switching. This judgment is achieved through an automated calculation model and analysis algorithm. The system automatically executes this judgment logic based on the equipment's real-time data stream, without manual intervention.

[0067] The purpose of this approach is to identify instability in equipment during fault recovery in real time, especially frequent fault switching, to ensure the long-term reliability of wind turbines. During wind turbine operation, pitch control equipment may frequently transition between fault and non-fault states during fault recovery. This frequent switching indicates that the equipment has not fully recovered and may harbor potential problems. If such fault switching is not detected and addressed promptly, the equipment may continue to operate in an unstable state, leading to more faults, energy efficiency losses, and even equipment damage. Therefore, by monitoring and automatically analyzing frequent equipment state switching in real time, signs of incomplete recovery can be detected early, allowing for timely corrective measures. This digital twin-based monitoring and analysis method not only improves the accuracy of fault diagnosis but also effectively prevents equipment from operating in suboptimal conditions for extended periods, thereby improving overall system efficiency, reducing failure risks, and extending equipment lifespan.

[0068] When the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time, analyzed, and the switching amplitude of the pitch control equipment when frequent fault switching occurs is determined.

[0069] In this embodiment, when the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and analyzed to determine the switching amplitude of the pitch control equipment when frequent fault state switching occurs. Specifically, this includes the following steps:

[0070] When the pitch control equipment experiences frequent fault state switching, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and preprocessed after acquisition.

[0071] To achieve real-time acquisition of behavior switching information during frequent fault state transitions in pitch control equipment, the software system needs to continuously access the real-time operational data streams of key sub-components in the pitch control system via a data acquisition module. Specifically, after the equipment enters a frequent fault state transition, the system automatically activates the information acquisition channel based on control logic, collecting dynamic parameter data highly correlated with the fault state change from sources such as angle sensors, current sensors, speed encoders, and load sensors. This data includes, but is not limited to, actual and target blade angles at different times, current load, speed changes, angle adjustment range, and load torque. The software, through a built-in synchronous sampling and scheduling mechanism, ensures that various sensor data are packaged into a behavior switching data frame with a unified timestamp, forming complete behavior switching information. This information is then transmitted in real-time to the data buffer and sent to the analysis module via a message queue system, achieving zero-latency, high-precision, and structured data input.

[0072] Directly importing behavior switching information into the analysis process after raw acquisition can lead to problems such as noise interference, inconsistent units, and outlier effects. Therefore, preprocessing at the software level is necessary to improve the accuracy and stability of subsequent parameter calculations. The preprocessing process mainly includes four types of operations: first, denoising, which removes instantaneous spikes and sensor jitter from the data using filter algorithms (such as moving average and median filtering); second, data normalization, which adjusts the dimensions of different physical quantities (such as angle, current, and torque) to ensure they participate in subsequent modeling within a uniform scale; third, missing data completion, which fills in missing data in certain sampling periods using interpolation algorithms (such as linear interpolation and spline interpolation); and fourth, dynamic window segmentation, which divides continuous data into multiple analysis segments based on the set switching behavior detection window for subsequent feature extraction. These operations are all automatically executed by the software before the data enters the analysis module, and the processing effect is monitored by a data quality detection mechanism to ensure that the dataset sent to the parameter calculation logic is clean, stable, and structurally complete.

[0073] Multidimensional operational fluctuation information and execution response offset information are extracted from the preprocessed behavior switching information, and analyzed after extraction to generate fluctuation expansion coefficient and state alienation index, respectively.

[0074] To extract multidimensional operational fluctuation information and execution response offset information from preprocessed behavior switching information, the software system needs to perform archiving and clustering extraction on different physical quantities based on a structured feature separation and data field classification mechanism, following a predefined index mapping logic. Specifically, the system first identifies various physical parameters contained in the preprocessed data stream, such as blade angle, current load, rotational speed, angle adjustment range, deviation between the target and actual blade angles, and load torque, and classifies and groups them using a field label matching mechanism. After classification, the software encapsulates the sets of parameters such as blade angle, current load, and rotational speed into multidimensional operational fluctuation information, and encapsulates the sets of parameters such as angle deviation, adjustment range, and torque into execution response offset information, according to feature mapping rules. To ensure the accuracy and consistency of the extraction, the system uses a sliding analysis window to extract time series data from various parameters and calculates their distribution characteristics (such as variance, kurtosis, and slope range). A feature transformer further encapsulates these time-series distribution characteristics into the two information structures. Finally, the system outputs these two pieces of information as independent feature objects for subsequent coefficient calculation and evaluation model analysis. The entire extraction process is automatically driven by software logic, requiring no manual intervention, and is reentrant and real-time.

[0075] A switching amplitude assessment model is constructed based on the generated fluctuation expansion coefficient and state alienation index to generate switching assessment coefficients;

[0076] Determine the pre-set threshold range of the switching evaluation coefficient, and compare it with the generated switching evaluation coefficient. Based on the comparison results, determine the switching amplitude of the pitch equipment when frequent fault switching occurs.

[0077] To determine the pre-defined threshold range for the handover evaluation coefficient, the software system can adaptively learn and dynamically set the threshold based on historical fault data analysis and multi-scenario simulation archiving. Specifically, the system first retrieves historical frequent fault handover sample data stored in the database. This data covers the fluctuation expansion coefficient and state alienation index generated by the device at multiple operational stages under different handover amplitude levels. By performing cluster analysis and distribution modeling on these sample data, the software system can statistically determine the typical value range of the handover evaluation coefficient in actual operation and associate it with the corresponding operation and maintenance results (such as successful device recovery, delayed response, secondary system failure, etc.) for labeling. Subsequently, the system uses methods such as density estimation or K-means clustering to identify the boundaries of the numerical range of the evaluation coefficient, thereby forming three level-division intervals: mild, moderate, and severe. In addition, during operation, the system will periodically update the sample boundaries in the model and dynamically adjust the upper and lower limits of the threshold interval based on newly added data to ensure consistency with the current device operating environment and response characteristics. The entire threshold setting process is executed automatically by the software, without relying on human experience, ensuring that the evaluation results are adaptable, accurate, and engineering feasible.

[0078] In this embodiment, the logic for obtaining the fluctuation expansion coefficient is as follows:

[0079] Multidimensional operational fluctuation information is extracted from the preprocessed behavior switching information. Specifically, this includes the blade angle change value, current load value, and pitch motor speed value at different times over a period of time during frequent fault state switching of the pitch control equipment, and these values ​​are denoted as θ. m I m and ω m θ m I represents the blade angle change at time m over a period of time during frequent fault state switching of the pitch control equipment. m ω represents the current load value at time m during a period of time when the pitch control equipment is undergoing frequent fault state switching. m This represents the rotational speed of the pitch motor at time m during a period of time when the pitch equipment is undergoing frequent fault state switching, where m = 1, 2, 3, ..., g, and g is a positive integer;

[0080] To achieve real-time acquisition of three types of data—blade angle change, current load, and pitch motor speed—during frequent fault state switching of the pitch control equipment, the software system needs to establish a data interface with the wind turbine's underlying monitoring platform (such as a SCADA system or an independent PLC control module) and interact with the sensor layer in real time based on standard industrial communication protocols (such as Modbus, CAN, or IEC 61400-25). The system will call multiple sensor node data sources deployed within the pitch control system. Specifically: the blade angle change can be measured in real time by an angle encoder installed at the blade root. By calculating the difference between the actual blade angle at each moment and the angle at the previous moment, a continuous angle change sequence is obtained; the current load value comes from the current sensor in the pitch motor drive circuit. The system directly reads the current channel output at different time points and records it as a quantitative value; the pitch motor speed value is obtained by a speed encoder connected to the motor shaft. This encoder is based on the number of rotation pulses per unit time, which is converted into instantaneous speed by the controller. The software system only extracts the numerical value without considering the unit, and uses it to evaluate the speed of the adjustment action and its fluctuation trend. The three types of data mentioned above are uniformly archived and packaged at each time point to form a multi-dimensional feature data frame, and stored in a cache queue in the form of a sliding window for subsequent average calculation and fluctuation analysis modules to call, thereby ensuring that the acquired data has both temporal integrity and supports high-frequency and high-precision dynamic analysis.

[0081] The average value calculation formula is used to determine the blade angle change θ at all times over a period of time during frequent fault state switching of the pitch control equipment. m The average value θ between _ Current load value I at all times m The average value between I _ and the speed ω of the pitch motor at all times. m The average value ω between _ According to the formula:

[0082] The fluctuation spread factor (FEC) is calculated by taking the blade angle change value θ at each moment during a period of time when the pitch control system is undergoing frequent fault state switching. m Its average θ - The absolute value of the difference between them is used to calculate the exponential function; the current load value I at each moment during a period of time in the pitch control equipment during frequent fault state switching is calculated. m Its average value I - The square root function is calculated by taking the absolute value of the difference between them; the speed ω of the pitch motor at each moment during a period of time during frequent fault state switching of the pitch equipment is calculated. m Its average value ω_ The absolute value of the difference between the values ​​is taken, and then one is added to calculate the logarithmic function. The three results obtained are summed, and the arithmetic mean of the summation results at all times is taken. Then, one is added to the mean and the natural logarithm is taken. The result is the fluctuation spread coefficient FEC.

[0083] The specific calculation formula is as follows:

[0084]

[0085] In the formula, FEC is the volatility expansion coefficient.

[0086] The formula for calculating the fluctuation expansion factor (FEC) is used to mathematically and comprehensively characterize the multidimensional fluctuation intensity of the pitch control system's operating state during frequent fault state switching. The specific calculation structure integrates multiple nonlinear functions such as exponential, square root, and logarithmic functions, exhibiting strong robustness and sensitivity. First, by calculating the absolute deviation between the blade angle change, current load, and pitch motor speed at each moment and their corresponding global average values, the fluctuation amplitude of each parameter at each moment is quantified. Then, an exponential function is used... Nonlinear amplification of angle fluctuations minimizes the impact of small changes on the results, while significantly amplifying drastic fluctuations, thus aiding in the highly sensitive identification of unstable behavior; the square root is used for the current load fluctuation portion. While maintaining the trend of change, extreme values ​​are appropriately compressed to prevent them from dominating the overall evaluation; the speed fluctuation term introduces a double logarithm ln(1+|ω m -ω - The variable |) is used to suppress the impact of sudden peak disturbances and maintain a smooth response to speed changes. The three factors are combined and averaged to ensure a relatively balanced contribution of different dimensions of data to the coefficient. Finally, 1 is added and the natural logarithm is taken to maintain numerical stability and a controllable range, avoiding negative or zero values ​​that could lead to calculation anomalies. This ensures that the entire fluctuation assessment coefficient has both a keen ability to identify anomalies and stable output under complex operating conditions, making it suitable for highly dynamic wind power control environments.

[0087] The fluctuation expansion coefficient (FEC) is positively correlated with the switching amplitude of pitch control equipment during frequent fault switching. A higher FEC value indicates more severe overall fluctuations in the equipment's operating state during the switching process. Since this coefficient comprehensively measures the deviation of three key parameters—blade angle, current load, and pitch motor speed—from their steady-state state, and amplifies severe fluctuations through a nonlinear function, the FEC value increases significantly when the equipment frequently switches between fault and non-fault states, accompanied by large-scale angle adjustments, load changes, and motor speed fluctuations. A higher FEC value indicates stronger instability and high dynamic response pressure during switching, reflecting a larger switching amplitude that may trigger secondary misalignment or control failure. Conversely, a lower FEC value indicates smaller equipment state fluctuations and a more moderate switching behavior. Therefore, by using the fluctuation expansion coefficient as input to the switching amplitude assessment model, the degree of fluctuation in the equipment's operating state during fault switching can be effectively quantified and graded, thereby accurately determining the level of switching amplitude.

[0088] In this embodiment, the logic for obtaining the state alienation index is as follows:

[0089] Execution response offset information is extracted from the preprocessed behavior switching information. Specifically, this includes the deviation of the actual blade angle from the preset target angle at different times during a period of time when the pitch control equipment is undergoing frequent fault state switching, the blade angle adjustment range, and the impedance load torque borne by the pitch motor, and is denoted as Δθ. m α m and T m , Δθ m α represents the deviation of the actual blade angle from the preset target angle at time m within a certain period of time during frequent fault state switching of the pitch control equipment. m T represents the blade angle adjustment range at time m within a certain period of time during frequent fault state switching of the pitch control equipment. m This represents the impedance load torque borne by the pitch motor at time m during a period of time when the pitch equipment is undergoing frequent fault state switching, where m = 1, 2, 3, ..., g, and g is a positive integer;

[0090] To achieve real-time acquisition of three types of data—the deviation between the actual blade angle and the preset target angle at different times during frequent fault state switching of the pitch control equipment, the blade angle adjustment range, and the impedance load torque borne by the pitch motor—the software system needs to establish a data acquisition channel with the pitch control system and monitoring platform. It automatically compares and calculates data based on the real-time output of operating commands from the internal controller and sensor feedback data. Specifically, the actual blade angle is acquired in real-time by a high-resolution rotary encoder installed on the blade shaft, while the preset target angle is calculated and generated by the pitch control system according to the current wind speed and operating conditions. The two are compared in the software at corresponding moments to obtain the angle deviation at that moment; the numerical value represents the magnitude of the deviation, but the unit is not collected. The angle adjustment range is calculated by the software from the difference between the actual blade angles at two consecutive moments, representing the intensity of each adjustment action; the system only extracts the change value for calculation and analysis. The impedance load torque borne by the pitch motor is dynamically calculated in the software by reading parameters such as the motor's internal current, voltage, and angular velocity, combined with the motor model formula. This reflects the torque required by the motor to counteract external blade drag and wind load in its current actual output. The system directly extracts the numerical results from this calculation process. All the above data is transmitted in real time to the upper-level analysis module via the data bus and used as numerical (unitless) input for the state alienation index calculation module, ensuring response accuracy and processing efficiency.

[0091] The State Dissimilarity Index (SDI) is calculated by taking the blade angle adjustment magnitude α at each moment. m As the exponent of the exponential function, it represents the deviation Δθ between the actual blade angle and the preset target angle at each moment. m Adding 1 and using it as the independent variable of the logarithmic function, the square root of the product of the two is used as the first part; the impedance load torque T borne by the pitch motor at each moment is... m The square of the result plus 1 is used as the independent variable of the logarithmic function, and its natural logarithm is calculated as the second part. The results of the two parts are summed in the range of m = 1 to m = g, and divided by the total number of time points g. The average value is the state alienation index SDI.

[0092] The specific calculation formula is as follows:

[0093]

[0094] In the formula, SDI is the state alienation index.

[0095] The State Dissimilarity Index (SDI) is used to accurately measure the deviation and instability trend between the control response behavior and the physical execution state of the pitch control system during frequent fault state transitions using a mathematical model. First, the blade angle adjustment magnitude α at each moment m is... m The exponent in the exponential operation is used to amplify the impact of drastic adjustments on system stability; simultaneously, it represents the deviation Δθ between the actual blade angle and the target angle at that moment. m Adding 1 and taking the natural logarithm serves as a nonlinear measure of the adjustment deviation, enhancing the ability to identify the cumulative effects of small errors. Multiplying these two parts and then taking the square root constructs a highly sensitive response index that jointly reflects both the "adjustment amplitude" and the "degree of deviation." Subsequently, the impedance load torque T borne by the pitch motor at that moment is calculated. m The square of the result plus 1, followed by the natural logarithm, characterizes the mechanical load stress encountered by the system during the response process. This approach avoids extreme value dominance and maintains a continuous response trend. Finally, the results from the two parts are averaged over all times from m=1 to g to obtain the global comprehensive offset intensity. The overall calculation process integrates a composite structure of exponential, logarithmic, and square root methods, which can effectively characterize the degree of inconsistency between control and execution during high-dynamic regulation and is a core indicator for identifying abnormal behavior.

[0096] The State Distortion Index (SDI) is positively correlated with the switching amplitude of pitch control equipment during frequent fault switching. A higher SDI value indicates a greater deviation from expected control response during frequent switching, indicating system instability and a larger switching amplitude. This index integrates three key factors: the actual deviation of blade angle, the severity of adjustment, and the load pressure on the pitch motor. When the equipment frequently makes high-amplitude adjustments, and there is a significant deviation between control commands and actual responses, accompanied by high load impedance, the index value will rise significantly, reflecting a higher degree of dynamic anomaly during switching, and a correspondingly larger switching amplitude. Conversely, when the equipment operates stably with small deviations and a stable load, the SDI is lower, indicating good response consistency and a smaller switching amplitude during frequent switching. Therefore, the value of the SDI can effectively determine the degree of deviation between the actual response behavior of the pitch control equipment and the expected control during fault switching, thereby determining the level of switching amplitude.

[0097] In this embodiment, a switching amplitude assessment model is constructed based on the generated fluctuation expansion coefficient (FEC) and state alienation index (SDI). The switching assessment coefficient (SEC) is generated by weighted summation, and the specific calculation formula is as follows:

[0098] SEC = β1 * FEC + β2 * SDI

[0099] In the formula, SEC is the switching evaluation coefficient, β1 and β2 are the non-zero weight coefficients of the volatility expansion coefficient FEC and the state alienation index SDI, respectively, and β1+β2=1.

[0100] In practical implementation, the switching evaluation coefficients are generated by the software system through a weighted combination of the fluctuation expansion coefficient (FEC) and the state alienation index (SDI). Specifically, the evaluation model module is invoked, with the two coefficients used as input variables and multiplied by preset non-zero weighting coefficients β1 and β2, respectively. The products are then summed to generate the final switching evaluation coefficients (SEC). The weighting coefficients β1 and β2 control the relative contribution of FEC and SDI to the evaluation results. Both are non-zero values ​​and satisfy β1 + β2 = 1, ensuring that the model provides a normative constraint on the influence of these two types of parameters. In practical applications, these two weights can be set based on historical operating data, empirical rules, or model training results. For example, β1 can be appropriately increased in scenarios where equipment state fluctuations have a greater impact on diagnostic results, while β2 can be increased in scenarios where control deviations are more likely to lead to misdiagnosis. This makes the generated switching evaluation coefficients more closely reflect the actual operating characteristics of the equipment, improving the accuracy and adaptability of switching magnitude judgment.

[0101] In this embodiment, a pre-set threshold range for the switching evaluation coefficient [SEC] is determined. min SEC max After determination, it is compared with the generated switching evaluation coefficient SEC. Based on the comparison results, the switching amplitude of the pitch equipment when frequent fault switching occurs is determined. The specific comparison analysis is as follows:

[0102] If SEC <SEC min When the pitch control equipment experiences frequent fault switching, the switching amplitude is low.

[0103] This situation indicates that during frequent fault switching, the overall operating status of the pitch control equipment fluctuates relatively little, its adjustment behavior is stable, its control response deviation is manageable, and the load resistance it experiences during execution is not significant. This means that although the equipment experiences frequent state switching, the intensity of each switch is low and insufficient to exert significant stress on the equipment's structure or control system. Such low-amplitude switching can usually be considered as fine-tuning actions during the equipment's adaptive adjustment process, requiring no additional intervention; only continuous monitoring is needed. The system can maintain normal operation and the current control strategy without triggering warnings or corrective actions.

[0104] If SEC min ≤SEC≤SEC max The switching amplitude of the pitch control equipment when frequent fault switching occurs is of medium magnitude.

[0105] This situation indicates that the pitch control equipment exhibits a certain degree of fluctuation during frequent fault switching. Although the adjustment process is not out of control, there are obvious execution deviations, increased adjustment drasticities, or excessively high load torque. These moderate-amplitude switching events indicate that the equipment has begun to deviate from its stable operating trajectory. Continued accumulation may adversely affect the motor, transmission structure, or control accuracy of the pitch control system. Therefore, in this state, the system should activate its early warning mechanism, adjust the diagnostic sensitivity of the virtual model to improve its ability to detect subsequent fault symptoms, and it is recommended that maintenance personnel monitor the execution frequency and load status of the pitch control equipment, and perform strategy fine-tuning or troubleshooting maintenance as necessary.

[0106] If seC > SEc max When the pitch control equipment experiences frequent fault switching, the switching amplitude is high.

[0107] This situation indicates that the pitch control equipment has exhibited highly abnormal dynamic behavior during frequent fault switching, including excessive adjustment amplitude, severely lagging control response, or drastic fluctuations in load torque, reflecting that the equipment's current operating state has lost its controllability. Such high-amplitude switching represents a significant imbalance state, which can easily lead to serious faults such as pitch motor overheating, mechanical jamming, and control synchronization failure, thereby affecting overall machine operating efficiency or causing unplanned downtime. In this case, the system should immediately adjust the fault diagnosis strategy of the virtual model, switch to a high-priority operating mode, and issue a strong warning to maintenance personnel, recommending rapid intervention to check the equipment's mechanical connections, control logic, and sensor accuracy. If necessary, mandatory load limiting or shutdown protection measures should be taken to prevent further escalation of the fault.

[0108] Based on the determined results, the fault diagnosis process of the virtual model is dynamically adjusted to adapt to the fault switching mode of the pitch equipment;

[0109] In this embodiment, the fault diagnosis process of the virtual model is dynamically adjusted based on the determined results, specifically as follows:

[0110] When the result is determined to be low, keep the fault diagnosis threshold parameters of the current virtual model unchanged and continuously monitor the operating status of the pitch control equipment;

[0111] When the result is determined to be low, the system maintains the fault diagnosis threshold parameters in the current virtual model unchanged through software and continuously monitors the operating status of the pitch control equipment. This is mainly achieved through a combination of model parameter freezing and real-time data tracking mechanisms. Specifically, after the condition of low-amplitude switching is triggered, the system first locks the status of various threshold parameters of the current diagnostic model through the model management module, that is, it maintains the judgment boundaries used to identify features such as angle deviation, current anomaly, and speed fluctuation without adjustment. Subsequently, the data acquisition module continues to acquire key operating data, including blade angle, motor current, and response delay, from the pitch control equipment sensors at a set frequency, and inputs this data into the virtual model in real time through the diagnostic interface module for performing routine diagnostic calculations that match the original threshold parameters. The reason for keeping the parameters unchanged is that low-amplitude switching means that the device's behavior fluctuates less during frequent state changes, and the overall stability is good. The existing recognition sensitivity and judgment range of the diagnostic model are sufficient to deal with such working conditions. If the parameters are adjusted excessively in this state, it may introduce misjudgment or false alarms caused by excessive sensitivity. Therefore, the strategy of freezing parameters and strengthening real-time monitoring is adopted to ensure the stability and resource efficiency of system operation while maintaining diagnostic accuracy.

[0112] When the result is determined to be of moderate magnitude, the increase of the fault diagnosis sensitive parameters in the virtual model is adjusted, and the fault identification frequency is increased.

[0113] When the result is determined to be of moderate magnitude, the system adjusts the amplification of fault diagnosis sensitive parameters in the virtual model via software and increases the fault identification frequency to enhance the model's responsiveness to potential abnormal states. Specifically, the system first identifies the current switching magnitude as moderate by the decision logic module, then sends parameter adjustment instructions to the diagnostic engine. Through the parameter control interface, it amplifies the values ​​of key sensitive parameters related to fault identification in the virtual model. For example, it increases the sensitivity range for variables such as blade angle deviation, current disturbance, and speed change, enabling the model to trigger the diagnostic mechanism promptly even with small fluctuations. Simultaneously, the system also adjusts the diagnostic cycle scheduler, increasing the fault identification frequency from the original base frequency to a shorter-cycle fast scanning mode, thereby capturing fault signals under moderate fluctuation conditions at a higher time resolution. This strategy is adopted because moderate-magnitude switching indicates a trend of declining equipment stability. Although it has not yet reached a level of severe loss of control, failing to promptly improve the model's monitoring accuracy and diagnostic frequency may cause the crucial window for identifying early abnormal signs to be missed. By appropriately enhancing parameter sensitivity and data response frequency, it is possible to intervene in advance when the system may evolve into highly abnormal behavior, while maintaining overall system stability, thereby improving the preventiveness and adaptability of the entire diagnostic process.

[0114] When the result is determined to be high, the diagnostic sensitive parameters in the virtual model are enhanced by a high amplitude, and the fault mode recognition weight configuration and diagnostic judgment range are adjusted simultaneously to improve the adaptability to abnormal states and the diagnostic accuracy.

[0115] When the result is determined to be high, the system significantly enhances the diagnostic sensitive parameters in the virtual model through software, and simultaneously adjusts the weight configuration and diagnostic judgment range of fault mode recognition to significantly improve the model's response capability and recognition accuracy to complex abnormal states. The specific implementation process includes three levels: First, after identifying a high-amplitude switching, the system immediately calls the model control module to significantly adjust the gain of key diagnostic sensitive parameters, such as drastically reducing the allowable error range for variables like angle offset, current fluctuation, and torque anomaly, enabling the model to respond instantly to any subtle deviations. Second, through the model weight adjustment mechanism, the weights of diagnostic channels corresponding to high-risk fault modes (such as blade runaway, execution synchronization failure, and load impact) in the virtual model are increased, prioritizing fault judgment for high-severity anomaly types. Finally, the diagnostic judgment range is also reset, making the model's fault judgment threshold more stringent and increasing the likelihood of triggering the model in the early stages of anomaly evolution. The reason for adopting this high-intensity adjustment strategy is that a high-amplitude switch indicates that the pitch control equipment is on the verge of systemic instability. If conventional diagnostic logic is continued, faults are likely to go unidentified or be misclassified, leading to serious consequences such as control lag, protection failure, or equipment damage. By significantly enhancing sensitive parameters, focusing on high-risk channels, and tightening judgment thresholds, the system can perform fault identification with higher alertness, priority, and discrimination capabilities, effectively ensuring the safe and stable operation of wind power equipment.

[0116] Based on the fault diagnosis results, fault prediction is performed and early warnings are issued to operators. At the same time, historical fault data and diagnosis results of the pitch equipment are recorded to optimize the diagnosis strategy of the virtual model.

[0117] To enable fault prediction based on fault diagnosis results and issue early warnings to operators, the software system can achieve this by constructing a predictive logic model and an early warning triggering mechanism that work in tandem. Specifically, after obtaining the fault diagnosis results for each round, the system inputs the current diagnostic status into a historical state evolution model. This model, based on time series analysis or trend recognition algorithms, combines current diagnostic data with recent continuous diagnostic result changes to predict whether the pitch equipment is likely to deteriorate further into a functional fault or structural anomaly. Once the predicted value reaches the predicted threshold range, the system immediately issues a multi-channel early warning to operators via the control interface, maintenance terminal, or alarm module, including visual and audible alerts, along with numerical descriptions of key influencing factors in this prediction, indicating the possible risk type and recommended response level. This mechanism can capture potential abnormal evolution trends before the equipment enters the manifest fault stage, giving maintenance personnel greater initiative in response and effectively reducing operational risks such as unplanned downtime, secondary damage, or maintenance delays.

[0118] Meanwhile, the system also utilizes a built-in data archiving and model adaptation mechanism to record each fault diagnosis result, prediction conclusion, equipment response status, and subsequent manual intervention as a structured dataset, automatically writing it into the pitch equipment's historical fault database. During data archiving, the model update module periodically calls upon this historical record, comparing and analyzing the deviation between the predicted results and the actual state to identify the error distribution in the model's judgment, thereby dynamically optimizing the sensitivity settings, weight coefficients, threshold ranges, and response strategies of each parameter in the virtual model. This optimization process is entirely software-driven, enabling the virtual model to continuously learn and adjust as the equipment's operating status and fault modes evolve, enhancing its adaptability and judgment accuracy over long-term use, forming a self-closing diagnostic optimization mechanism. This not only improves the model's ability to predict future faults but also endows the entire diagnostic system with continuously evolving intelligent characteristics.

[0119] like Figure 3 The wind turbine pitch fault pre-diagnosis system based on digital twin shown includes an operation status perception module, a recovery process modeling module, a behavior switching evaluation module, a diagnostic logic control module, and a prediction and early warning optimization module.

[0120] The operation status sensing module monitors the operation of the pitch control equipment of the wind turbine in real time and determines whether the pitch control equipment has entered a fault state.

[0121] The recovery process modeling module continuously monitors the fault recovery process of the pitch equipment when it enters a fault state, and builds a digital twin virtual model to determine whether the pitch equipment is experiencing frequent fault state switching.

[0122] The behavior switching assessment module acquires and analyzes the behavior switching information of the pitch equipment during the frequent fault state switching when the pitch equipment experiences frequent fault state switching, and determines the switching amplitude of the pitch equipment when frequent fault state switching occurs.

[0123] The diagnostic logic control module dynamically adjusts the fault diagnosis process of the virtual model based on the determined results to adapt to the fault switching mode of the pitch equipment.

[0124] The prediction and early warning optimization module predicts faults based on fault diagnosis results and issues early warnings to operators. It also records historical fault data and diagnosis results of the pitch equipment and optimizes the diagnosis strategy of the virtual model.

[0125] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0126] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0127] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0128] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0130] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0131] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A wind turbine variable pitch fault pre-diagnosis method based on digital twinning, characterized in that, Specifically, the following steps are included: Real-time monitoring of the operation of the pitch control equipment of the wind turbine to determine whether the pitch control equipment has entered a fault state; When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and a digital twin virtual model is constructed to determine whether the pitch control equipment is experiencing frequent fault switching. Specifically: When the pitch control equipment enters a fault state, the fault recovery process of the pitch control equipment is continuously monitored, and the recovery status information of the pitch control equipment during the fault recovery process is continuously acquired during the monitoring process. Based on the acquired recovery status information, a digital twin virtual model is constructed to simulate the dynamic behavior of the pitch equipment during the fault recovery process, and the pitch equipment status of the virtual model is updated in real time. Based on the constructed digital twin virtual model, the system updates and analyzes the equipment status in real time to determine whether the pitch equipment experiences frequent fault switching during the fault recovery process. The specific judgment logic is as follows: within a preset time period, if the frequency of the transition between the fault and non-fault states of the pitch equipment exceeds the preset frequency threshold, and after each entry into the non-fault state, it fails to maintain the non-fault state for more than the set stable time period, then it is determined that the pitch equipment has experienced frequent fault switching. When frequent fault switching occurs in the pitch control equipment, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and analyzed to determine the switching amplitude of the pitch control equipment when frequent fault switching occurs. The specific steps include: When frequent fault switching occurs in the pitch control equipment, the behavior switching information of the pitch control equipment during the frequent fault state switching process is acquired in real time and preprocessed after acquisition. Multidimensional operational fluctuation information and execution response offset information are extracted from the preprocessed behavior switching information, and analyzed after extraction to generate fluctuation expansion coefficient and state alienation index, respectively. A switching amplitude assessment model is constructed based on the generated fluctuation expansion coefficient and state alienation index to generate switching assessment coefficients; Determine the pre-set threshold range of the switching evaluation coefficient, and compare it with the generated switching evaluation coefficient after determination. Based on the comparison results, determine the switching amplitude of the pitch equipment when frequent fault switching occurs. Based on the determined results, the fault diagnosis process of the virtual model is dynamically adjusted to adapt to the fault switching mode of the pitch equipment; Based on the fault diagnosis results, fault prediction is performed and early warnings are issued to operators. At the same time, historical fault data and diagnosis results of the pitch equipment are recorded to optimize the diagnosis strategy of the virtual model.

2. The wind power variable-pitch fault pre-diagnosis method based on digital twinning according to claim 1, characterized in that, The logic for obtaining the fluctuation expansion coefficient is as follows: Multidimensional operational fluctuation information is extracted from the preprocessed behavior switching information. Specifically, this includes blade angle changes, current load values, and pitch motor speed values ​​at different times within a certain period during frequent fault state switching of the pitch control equipment. These values ​​are then calibrated as follows: , and , This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... The change in blade angle at time t. This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... Current load value at time t, This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... The constant speed of the pitch motor. , It is a positive integer; The average value calculation formula is used to determine the blade angle variation values ​​at all times over a period of time during frequent fault state switching of the pitch control equipment. average between Current load values ​​at all times average between and the speed values ​​of the pitch motor at all times. average between ; Calculate the fluctuation expansion coefficient The specific calculation method is as follows: The blade angle change value is calculated at each moment within a certain period of time during the frequent fault state switching process of the pitch control equipment. Its average The absolute value of the difference between them is used to calculate the exponential function; the current load value of the pitch control equipment at each moment during a period of time during frequent fault state switching is calculated. Its average The square root function is calculated by taking the absolute value of the difference between them; the speed of the pitch motor is calculated at each moment during a period of time when the pitch equipment is undergoing frequent fault state switching. Its average The absolute value of the difference between the values ​​is taken, then one is added to calculate the logarithmic function. The three results obtained separately are summed, and the arithmetic mean of the summations over all time points is taken. Then, one is added to this mean, and the natural logarithm is taken. The result is the fluctuation spread coefficient. .

3. The wind turbine pitch fault pre-diagnosis method based on digital twin according to claim 2, characterized in that, The logic for obtaining the state alienation index is as follows: Execution response offset information is extracted from the preprocessed behavior switching information. Specifically, this includes the deviation of the actual blade angle from the preset target angle at different times within a certain period during frequent fault state switching of the pitch control equipment, the blade angle adjustment range, and the impedance load torque borne by the pitch motor. These are then calibrated as follows: , and , This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... The deviation between the actual blade angle and the preset target angle at any given time. This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... The range of blade angle adjustment at any given time. This indicates that during a period of time when the pitch control equipment is undergoing frequent fault state switching... The impedance load torque that the pitch motor bears at any given time; Calculate the state alienation index The specific calculation method is as follows: adjust the blade angle by the amount at each moment. As the exponent of the exponential function, it represents the deviation between the actual blade angle and the preset target angle at each moment. Adding 1 and using it as the independent variable of the logarithmic function, the square root of the product of the two is used as the first part; the impedance load torque borne by the pitch motor at each moment is... The square of the product plus 1 is used as the independent variable of the logarithmic function, and its natural logarithm is calculated as the second part; the results of the above two parts are then combined in... to Summing is performed within the range of [a certain value], and then divided by the total number of time points. The average value obtained is the state alienation index. .

4. The wind power variable-pitch fault pre-diagnosis method based on digital twinning according to claim 3, characterized in that, The generated fluctuation expansion coefficient and state alienation index Construct a handover magnitude assessment model and generate handover assessment coefficients through weighted summation. .

5. The wind power variable-pitch fault pre-diagnosis method based on digital twinning according to claim 4, characterized in that, Determine the pre-set threshold range for the switching evaluation coefficient. And after determination, it is compared with the generated switching evaluation coefficients. A comparison was conducted, and the switching amplitude of the pitch control equipment when frequent fault switching occurred was determined based on the comparison results. The specific comparison analysis is as follows: like When the pitch control equipment experiences frequent fault switching, the switching amplitude is low. If , the switching range of the pitch device in the presence of frequent fault switching phenomena is a medium range; If , the switching amplitude of the pitch device in the presence of frequent failure switching phenomena is high.

6. The wind power variable-pitch fault pre-diagnosis method based on digital twinning according to claim 5, characterized in that, Based on the determined results, the fault diagnosis process of the virtual model is dynamically adjusted, specifically as follows: When the result is determined to be low, keep the fault diagnosis threshold parameters of the current virtual model unchanged and continuously monitor the operating status of the pitch control equipment; When the result is determined to be of moderate magnitude, the fault diagnosis threshold parameter in the virtual model is adjusted by increasing the magnitude, and the fault identification frequency is increased. When the result is determined to be high, the fault diagnosis threshold parameter in the virtual model is enhanced by a high amplitude, and the fault mode recognition weight configuration and the diagnostic judgment range are adjusted simultaneously to improve the adaptability to abnormal states and the diagnostic accuracy.

7. A wind turbine variable pitch fault pre-diagnosis system based on digital twinning, for implementing the wind turbine variable pitch fault pre-diagnosis method based on digital twinning in any one of claims 1-6, characterized in that, It includes a running status perception module, a recovery process modeling module, a behavior switching evaluation module, a diagnostic logic control module, and a prediction, early warning, and optimization module; The operation status sensing module monitors the operation of the pitch control equipment of the wind turbine in real time and determines whether the pitch control equipment has entered a fault state. The recovery process modeling module continuously monitors the fault recovery process of the pitch equipment when it enters a fault state, and builds a digital twin virtual model to determine whether the pitch equipment is experiencing frequent fault switching. The behavior switching assessment module acquires and analyzes the behavior switching information of the pitch equipment during the frequent fault switching process when the pitch equipment experiences frequent fault switching, and determines the switching amplitude of the pitch equipment when frequent fault switching occurs. The diagnostic logic control module dynamically adjusts the fault diagnosis process of the virtual model based on the determined results to adapt to the fault switching mode of the pitch equipment. The prediction and early warning optimization module predicts faults based on fault diagnosis results and issues early warnings to operators. It also records historical fault data and diagnosis results of the pitch equipment and optimizes the diagnosis strategy of the virtual model.

Images