Wind turbine generator set abnormality early warning method, device, equipment and storage medium

Abstract

The application discloses a wind turbine abnormal early warning method, device, equipment and storage medium, and relates to the technical field of data processing, which comprises the following steps: based on environment sensing data and a preset gear mapping table, autonomously deciding an optimal gear state under a current environment; then, calling a mathematical model parameter group matched with the gear to perform local simulation, and obtaining simulation parameters; using a multi-dimensional analysis intelligent agent to compare actual monitoring data with the simulation parameters, and generating local multi-dimensional residual characteristics associated with the current gear; finally, through a preset cooperation mechanism between the multi-dimensional analysis intelligent agents, the residual characteristics are fused and analyzed, a global abnormal diagnosis result is obtained, and a prediction alarm is performed, so that the application realizes full-chain adaptive monitoring based on the environment, the gear, the model, the residual, the cooperation and the early warning, and significantly improves the accuracy of abnormal detection and the refinement degree of fault positioning.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, device, equipment and storage medium for early warning of anomalies in wind turbine generator sets. Background Technology

[0002] Wind energy is abundant in my country, and its environmental friendliness and low cost make it highly valuable for development. Wind turbines are crucial for converting wind energy into electricity, and the turbine blades are a key component, their performance directly impacting power generation efficiency. Existing methods for wind turbine inspection are time-consuming and labor-intensive, inefficient, unable to achieve precise monitoring with adaptive speed settings, and suffer from poor anomaly detection accuracy, leading to delayed warnings and ultimately affecting site efficiency. Summary of the Invention

[0003] The main purpose of this application is to provide a method, device, equipment and storage medium for early warning of abnormalities in wind turbine generator sets, which aims to solve the technical problems of existing methods for accurate monitoring of wind turbine generator sets that cannot achieve gear self-adaptation and poor accuracy of abnormality detection.

[0004] To achieve the above objectives, this application proposes a method for early warning of anomalies in wind turbine generator sets, the method comprising: The optimal gear position is determined based on the environmental perception data corresponding to the wind turbine generator set and the preset gear mapping table. Based on the optimal gear position, the mathematical model parameter group matching the gear position is invoked to perform local simulation and obtain simulation parameters; The multidimensional analysis agent compares the actual monitoring data with the simulation parameters to obtain the local multidimensional residual features associated with the current gear. Based on a multidimensional analysis agent and a preset collaborative mechanism, the local multidimensional residual features are fused and analyzed to obtain global anomaly diagnosis results. Predictive alerts are generated based on the global anomaly diagnosis results.

[0005] Optionally, after determining the optimal gear position under the current environment based on the environmental perception data corresponding to the wind turbine generator and a preset gear mapping table, the method further includes: Obtain the wind speed time series from the environmental perception data, and predict the wind speed state within the target time window based on the preset wind speed prediction model and the wind speed time series; If the wind speed state is about to enter the environmental parameter range corresponding to different gears, a pre-shift command is generated before the wind speed state reaches the shift threshold. The pre-shift command is used to adjust the mathematical model parameter group that matches the target gear in advance, and load the adjusted parameters into the local simulation model before the shift is executed.

[0006] Optionally, determining the optimal gear position under the current environment based on environmental perception data corresponding to the wind turbine generator and a preset gear mapping table includes: If the average wind speed in the environmental perception data is in the first wind speed range and the turbulence intensity is lower than the first threshold, the optimal gear state is determined to be the high gear ratio gear. If the average wind speed is in the second wind speed range or the turbulence intensity exceeds the second threshold, then the optimal gear state is determined to be the low gear ratio gear. If the wind speed is within the preset hysteresis range, the current gear setting will remain unchanged.

[0007] Optionally, the multidimensional analysis agent includes an aerodynamic agent for the aerodynamic system, a transmission agent for the transmission system, and a pitch agent for the pitch system. The multidimensional analysis agent compares actual monitoring data with the simulation parameters to obtain local multidimensional residual features associated with the current gear position, including: Collect multi-source sensor monitoring data of the aerodynamic system, transmission system and pitch system of the wind turbine generator set during actual operation; The monitoring data from the multi-source sensors are compared with the simulation parameters generated by the aerodynamic intelligent agent, transmission intelligent agent, and pitch intelligent agent at the same time to obtain the comparison results. The local multidimensional residual characteristics are determined based on the comparison results.

[0008] Optionally, the step of fusing and analyzing the local multidimensional residual features based on a multidimensional analysis agent and a preset collaborative mechanism to obtain a global anomaly diagnosis result includes: Calculate the matching degree between the local multidimensional residual features and each fault mode in the preset fault feature map; The feature fusion weight coefficient is determined based on the matching degree and the preset collaboration mechanism. The preset collaboration mechanism is based on the collaborative agent receiving the local multidimensional residual features output by each multidimensional analysis agent, as well as the current environmental perception data and gear status, and outputting the dynamic fusion weight of each agent. Based on the feature fusion weight coefficients, the local multidimensional residual features are fused and analyzed to obtain global anomaly diagnosis results.

[0009] Optionally, the multidimensional analysis agent further includes a gear shift agent; the gear shift agent further performs the following steps: After the shift command is issued, monitor the changes in the actual gear status and record the shift duration and the fluctuations in speed and torque during the shift process; The actual shift response curve is compared with the preset gear shift transient model to calculate the shift process residual; If the residual error of the shifting process exceeds a preset threshold, a fault diagnosis result of the shifting mechanism is generated. The fault diagnosis result includes shift fork jamming, synchronizer wear, or insufficient hydraulic pressure.

[0010] Optionally, the step of predicting alarms based on the global anomaly diagnosis results includes: The time series of local multidimensional residual features output by the multidimensional analysis agent, the current environmental perception data, and the current gear status are input into the time series prediction model to determine the residual prediction information within the target time period. If the residual prediction information is to generate alarm information including fault mode, fault location, remaining effective life and recommended maintenance actions when the alarm threshold is exceeded at the target time point.

[0011] In addition, to achieve the above objectives, this application also proposes a wind turbine generator abnormality early warning device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the wind turbine generator abnormality early warning method described above.

[0012] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the wind turbine generator abnormality early warning method described above.

[0013] Furthermore, to achieve the above objectives, this application also proposes a wind turbine generator abnormality early warning device, which includes: The gear decision module is used to determine the optimal gear state under the current environment based on the environmental perception data corresponding to the wind turbine generator and the preset gear mapping table. The simulation module is used to perform local simulation based on the optimal gear state by calling the mathematical model parameter group that matches the gear, and to obtain simulation parameters. The intelligent agent analysis module is used to compare the actual monitoring data with the simulation parameters based on the multidimensional analysis intelligent agent to obtain the local multidimensional residual features associated with the current gear position; The collaborative fusion module is used to perform fusion analysis on the local multidimensional residual features based on the multidimensional analysis agent and the preset collaborative mechanism to obtain global anomaly diagnosis results. The predictive alarm module is used to predict alarms based on the global anomaly diagnosis results.

[0014] One or more technical solutions proposed in this application have at least the following technical effects: This application autonomously determines the optimal gear state under the current environment based on environmental perception data and a preset gear mapping table. Then, it calls the mathematical model parameter group matched with the gear to perform local simulation and obtain simulation parameters. Using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear. Finally, through a preset collaborative mechanism among the multidimensional analysis agents, it fuses and analyzes the residual features to obtain global anomaly diagnosis results and make predictions and alarms. This solves the technical problems of existing wind turbines' inability to achieve accurate gear adaptation monitoring and poor anomaly detection accuracy. This application, based on the full-chain adaptive monitoring of environment, gear, model, residual, collaboration, and early warning, significantly improves the accuracy of anomaly detection and the refinement of fault location. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating an embodiment of the wind turbine generator abnormality early warning method of this application. Figure 2 This is a schematic diagram of the scheme provided in Embodiment 1 of the wind turbine generator abnormality early warning method of this application; Figure 3 The flowchart provided in Embodiment 3 of the wind turbine generator abnormality early warning method of this application is as follows: Figure 4 This is a schematic diagram of the module structure of the wind turbine generator abnormality early warning method device according to an embodiment of this application; Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the wind turbine generator abnormality early warning method in this application embodiment.

[0018] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0020] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of this application embodiment is as follows: based on environmental perception data and a preset gear mapping table, the system autonomously decides the optimal gear state under the current environment; then, it calls the mathematical model parameter group matching the gear to perform local simulation and obtain simulation parameters; using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear; finally, through a preset collaborative mechanism among multidimensional analysis agents, it performs fusion analysis on the residual features to obtain global anomaly diagnosis results and make predictions and alarms.

[0022] In this embodiment, for ease of description, the following description uses a computing service device as the execution subject.

[0023] The existing methods for accurately monitoring wind turbines, which cannot achieve gear-adaptive operation, suffer from poor accuracy in anomaly detection.

[0024] This application provides a solution that can autonomously decide the optimal gear state in the current environment based on environmental perception data and a preset gear mapping table; then, it calls the mathematical model parameter group matching the gear to perform local simulation and obtain simulation parameters; using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear; finally, through a preset collaborative mechanism among the multidimensional analysis agents, it performs fusion analysis on the residual features to obtain global anomaly diagnosis results and provide predictive alarms.

[0025] As can be seen from the above embodiments, this application autonomously decides the optimal gear state under the current environment based on environmental perception data and a preset gear mapping table; then, it calls the mathematical model parameter group matched with the gear to perform local simulation and obtain simulation parameters; it uses a multidimensional analysis agent to compare the actual monitoring data with the simulation parameters and generate local multidimensional residual features associated with the current gear; finally, it uses a preset collaborative mechanism among multidimensional analysis agents to perform fusion analysis on the residual features, obtain global anomaly diagnosis results, and make predictions and alarms. Compared with the existing precise monitoring of wind turbines that cannot achieve gear adaptation and has poor anomaly detection accuracy, this application significantly improves the accuracy of anomaly detection and the refinement of fault location based on the full-chain adaptive monitoring of environment, gear, model, residual, collaboration and early warning.

[0026] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, including a wind turbine generator abnormality early warning system. The following description uses a computer as an example to illustrate this embodiment and the subsequent embodiments.

[0027] Based on this, this application provides a method for early warning of anomalies in wind turbine generator sets, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the wind turbine generator abnormality early warning method of this application.

[0028] In this embodiment, the wind turbine generator abnormality early warning method includes steps S10 to S50: Step S10: Determine the optimal gear state under the current environment based on the environmental perception data corresponding to the wind turbine generator set and the preset gear mapping table; It should be noted that this embodiment provides an anomaly early warning method for wind turbine generator sets, which can be applied to a 2.5MW doubly-fed wind turbine generator set equipped with a two-speed planetary gearbox. The generator set can switch between a high transmission ratio gear, such as 1:90, suitable for low wind speed ranges, and a low transmission ratio gear, such as 1:45, suitable for high wind speed ranges. The following sensors are deployed on the generator set: The aerodynamic system includes blade root strain gauges for calculating aerodynamic bending moments and a pitch angle encoder. The transmission system includes a low-speed shaft torque meter, a gearbox housing triaxial accelerometer, a high-speed shaft speed encoder, and a generator stator current Hall sensor. The pitch system includes pitch position sensors for each of the three blades and a pitch motor current sensor. The gearing system includes a shift fork proximity switch within the gearbox for identifying the actual gear position and a synchronizer ring strain sensor. All sensor data can be aggregated to the nacelle edge computing gateway via an industrial Ethernet network.

[0029] Understandably, by collecting physical parameters of the environment in which the unit is located in real time, such as wind speed and turbulence intensity, and then according to the pre-designed mapping rules that correspond environmental conditions to transmission gears, i.e., the preset gear mapping table, the unit can automatically determine and select the gear state that is most conducive to the safe and efficient operation of the unit under the current environmental conditions. The optimal gear state includes: high transmission ratio gear or low transmission ratio gear.

[0030] After step S10, the method further includes: acquiring the wind speed time series in the environmental perception data, and predicting the wind speed state within the target time window based on the preset wind speed prediction model and the wind speed time series; if the wind speed state is about to enter the environmental parameter range corresponding to different gears, then generating a pre-shift command before the wind speed state reaches the shift threshold, the pre-shift command is used to adjust the mathematical model parameter group matching the target gear in advance, and load the adjusted parameters into the local simulation model before the shift is executed.

[0031] It should be noted that the system does not wait until the environmental parameters actually reach the shift threshold before switching gears. Instead, it uses a wind speed prediction model to predict the wind speed trend over a future period. Once it predicts that it will soon enter the environmental parameter range corresponding to a different gear, it generates a pre-shift command before the actual shift occurs. This command loads the mathematical model parameter set of the target gear into the local simulation model in advance. The wind speed time series is a data sequence of continuously collected wind speed values ​​arranged in chronological order, for example, one sampling point per second over the past 10 minutes. The preset wind speed prediction model uses a time-series prediction algorithm, such as a model trained with an LSTM model. This model takes the historical wind speed sequence as input and outputs the predicted wind speed value within a specified future time window (e.g., 30 seconds, 1 minute, 5 minutes). The predicted future wind speed state is compared with the environmental parameter range in the preset gear mapping table. If the prediction indicates that the average wind speed at a certain future moment (e.g., 30 seconds later) will rise from the current low-speed range (e.g., 6 m / s) to a high-speed range (e.g., above 8 m / s), it is determined that "it is about to enter the environmental parameter range corresponding to a different gear". The wind speed prediction model employs a two-layer LSTM network with 64 hidden units per layer. The input is the wind speed time series of the past 10 minutes, with a sampling frequency of 1Hz (600 points). The output is the wind speed series of the next 1 minute, with 60 points. The model is pre-trained on historical wind field data, and the root mean square error is less than 0.3 m / s.

[0032] Understandably, the wind speed prediction module uses a pre-trained LSTM network to predict the wind speed for the next minute based on the wind speed time series of the past 10 minutes. If the prediction shows that V_avg will exceed 8 m / s within the next 30 seconds, a pre-shift instruction is generated to switch the parameter set of the transmission mathematical model from high-gear parameters to low-gear parameters in advance, reducing shift shock.

[0033] Step S20: Based on the optimal gear state, perform local simulation using the mathematical model parameter group that matches the gear to obtain simulation parameters.

[0034] It should be noted that this is for reference only. Figure 2The schematic diagram shows that after determining the optimal gear state in the current environment through environmental perception and gear mapping table, the system dynamically calls the set of parameters that perfectly matches the current gear from the pre-stored mathematical model parameter set corresponding to each gear, based on the gear state (high transmission ratio or low transmission ratio). Then, the system uses the parameter set to drive the physical mathematical model of the corresponding subsystem to perform local simulation, and finally calculates the simulation parameters that can reflect the ideal operating state of the subsystem, such as aerodynamic torque, transmission torque, and propeller pitch angle response.

[0035] Understandably, each subsystem of a wind turbine generator (aerodynamics, transmission, pitch) employs mathematical models. These models contain numerous physical parameters, such as: Transmission system model: moment of inertia J (rotor, gearbox, generator), torsional stiffness K, damping coefficient C, and transmission ratio i. These parameters differ significantly at different speeds. For example, at higher speeds (larger transmission ratios), the gearbox has a larger equivalent inertia and different stiffness characteristics. Aerodynamic system model: blade lift-drag coefficient curve, induction factor correction coefficient, dynamic inflow time constant, etc. The aerodynamic characteristics differ between low-wind-speed (high speed) and high-wind-speed (low speed) regions, requiring different coefficient corrections. Pitch system model: gain, time constant, dead zone, saturation limit, etc., of the pitch actuator. Pitch control strategies, such as pitch rate and angle limit, often differ at different speeds.

[0036] It should be understood that local simulation refers to digital simulation performed independently for a single subsystem, not a global joint simulation of the entire unit. Each agent runs its own simulation model, taking into account sensor data and current gear parameters related to that subsystem, and outputting the desired state of that subsystem, i.e., the simulation parameters.

[0037] Step S30: Based on the multidimensional analysis agent, the actual monitoring data is compared with the simulation parameters to obtain the local multidimensional residual features associated with the current gear.

[0038] It should be noted that after obtaining the simulation parameters of each subsystem (pneumatics, transmission, pitch, etc.), the system utilizes multiple deployed distributed analytical agents (each agent corresponding to a subsystem) to compare the monitoring data collected by the actual sensors of that subsystem with the simulation parameters, i.e., the expected values ​​under ideal conditions, item by item at the same time. By calculating the relative differences, a set of local multidimensional residual features strongly correlated with the current gear position is obtained. The multidimensional analytical agents include: a pneumatic agent responsible for calculating the residuals of the pneumatic system (blades, hub, main shaft, etc.); a transmission agent responsible for calculating the residuals of the transmission system (gearbox, bearings, couplings, etc.); a pitch agent responsible for calculating the residuals of the pitch system (pitch motor, hydraulic cylinder, bearings, etc.); and a gear agent responsible for calculating the residuals of the gear shifting mechanism.

[0039] Understandably, the simulation parameters are generated by calling the matching mathematical model parameter set for local simulation, with the same time step. The acquisition time of the actual monitoring data strictly corresponds to the calculation time of the simulation parameters, for example, t=0.02s, 0.04s, ... The simulation parameters are calculated under model parameters that perfectly match the current actual gear, thus accurately reflecting the normal expectation under the current gear.

[0040] It should be understood that local multidimensional residual characteristics are determined by calculating relative residuals. Specifically, the local multidimensional residual characteristics corresponding to the aerodynamic agent include aerodynamic torque residuals, thrust residuals, and blade root bending moment residuals; those corresponding to the transmission agent include torque residuals, vibration acceleration residuals (X / Y / Z axes), speed residuals, and gear meshing frequency sideband energy residuals; and those corresponding to the pitch agent include pitch angle residuals, pitch rate residuals, and motor current residuals. The local multidimensional residual characteristics corresponding to the gear-shifting agent include shift process time residuals, speed fluctuation residuals, and consistency residuals between actual and commanded gears. These residuals are assigned a correlation with the current gear, and the calculation results are tagged with a gear label (e.g., gear=HIGH or LOW). Subsequent collaborative fusion steps will employ different processing strategies or establish gear-specific fault maps based on the gear label.

[0041] Understandably, the aerodynamic residual = (actual aerodynamic torque - simulated aerodynamic torque) / rated torque; the transmission residual = (actual gearbox vibration amplitude - simulated vibration amplitude) / rated vibration reference value; and the pitch residual = actual pitch angle - simulated pitch angle. Each residual is also labeled with the current gear position.

[0042] Step S40: Based on the multidimensional analysis agent and the preset collaborative mechanism, the local multidimensional residual features are fused and analyzed to obtain the global anomaly diagnosis result.

[0043] It should be noted that after obtaining the local multidimensional residual features of each subsystem (aerodynamics, transmission, pitch, etc.), the system uses multiple distributed analytical agents and a pre-designed collaborative mechanism (such as weighted voting, attention network, reinforcement learning decision-making, etc.) to fuse and analyze these residual features from different subsystems and different dimensions. That is, it comprehensively considers the local judgments of each agent, the magnitude and trend of the residuals, and the coupling relationship between the subsystems, and finally outputs a global anomaly diagnosis result (such as determining which subsystem or component has experienced what type of failure and giving the confidence level).

[0044] Step S50: Make a prediction alarm based on the global anomaly diagnosis results.

[0045] It should be noted that after obtaining the global anomaly diagnosis results (e.g., determining the existence of a certain fault mode, fault location, severity, etc.), this application does not stop at the current fault identification and alarm, but further utilizes a time-series prediction model, combined with information such as historical residual sequences, environmental parameter evolution trends, and gear status changes, to predict the development trend of the anomaly in the future (e.g., whether the residual will continue to increase, when it will exceed the emergency threshold), and generates forward-looking alarm information accordingly. This alarm information typically includes the fault mode, fault location, remaining useful life (RUL), and recommended maintenance actions, thereby supporting the upgrade from "passive alarm" to "proactive predictive maintenance".

[0046] In its implementation, the time-series prediction model of this application can employ a three-layer LSTM network. The input feature vector includes: aerodynamic residuals, transmission residuals, and pitch residuals from the past hour (360 points, one sampling point every 10 seconds), as well as synchronously recorded average wind speed, turbulence intensity, and gear code. The model outputs a residual prediction sequence for the next hour. When the predicted residuals will exceed the warning threshold of 0.10 (relative to the rated value) within the next 30 minutes, a Level 1 warning is generated; when the predicted residuals will exceed the alarm threshold of 0.18 within the next 2 hours, a Level 2 alarm is generated, and the remaining effective lifetime is estimated. The remaining effective lifetime is calculated using linear interpolation: RUL = (threshold - current value) / average slope. The generated Level 2 alarm information is as follows: {"Alarm Level": "Orange", "Fault Mode": "Gear Tooth Surface Wear", "Fault Location": "Gearbox - Intermediate Shaft", "Current Anomaly Index": 0.15, "Remaining Effective Life": "Approximately 18 hours", "Recommended Action": "It is recommended to schedule a planned shutdown within 12 hours to prepare for endoscopy and oil analysis"}. This information is pushed to the field-level monitoring system and the maintenance personnel's mobile app via the MQTT protocol.

[0047] This embodiment provides an anomaly early warning method for wind turbine generator sets. Based on environmental perception data and a preset gear mapping table, it autonomously decides the optimal gear state under the current environment. Then, it calls the mathematical model parameter set matched with the gear to perform local simulation and obtain simulation parameters. Using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear. Finally, through a preset collaborative mechanism among the multidimensional analysis agents, it fuses and analyzes the residual features to obtain a global anomaly diagnosis result and makes a prediction and alarm. Compared with existing precise monitoring methods for wind turbine generator sets that cannot achieve gear adaptation and have poor anomaly detection accuracy, this embodiment, based on the full-chain adaptive monitoring of environment, gear, model, residual, collaboration, and early warning, significantly improves the accuracy of anomaly detection and the refinement of fault location.

[0048] Based on the above Figure 1The first embodiment shown presents a second embodiment of the wind turbine generator abnormality early warning method of this application; based on the first embodiment of this application, the same or similar content as the first embodiment above can be referred to the above description, and will not be repeated hereafter.

[0049] In this embodiment, step S10 further includes: if the average wind speed in the environmental perception data is in the first wind speed range and the turbulence intensity is lower than the first threshold, the optimal gear state is determined to be the high gear ratio gear; if the average wind speed is in the second wind speed range or the turbulence intensity exceeds the second threshold, the optimal gear state is determined to be the low gear ratio gear; if the wind speed is within the preset hysteresis range, the current gear state is maintained unchanged.

[0050] It should be noted that when the average wind speed is in a low range, such as the first wind speed range and the turbulence intensity is below the first threshold, a high transmission ratio gear is selected. When the average wind speed is in a high range, such as the second wind speed range or the turbulence intensity exceeds the second threshold, a low transmission ratio gear is selected. When the wind speed is within the preset hysteresis range, the gear position remains unchanged to avoid frequent gear changes. By switching the gear selection decision of the wind turbine from a simple single-parameter threshold to a smart decision based on a two-factor (wind speed + turbulence) joint judgment + hysteresis protection, the unit can achieve stable and efficient operation in complex environments.

[0051] Understandably, the first wind speed range is a low-wind-speed area, requiring a high transmission ratio to increase the rotational speed, such as [3 m / s, 8 m / s]. The second wind speed range is a high-wind-speed area, requiring a low transmission ratio to limit the rotational speed, such as [8 m / s, 25 m / s]. The first turbulence threshold is used to determine whether turbulence is acceptable below this value, such as 0.10 ~ 0.12. The second turbulence threshold is used to determine whether a lower transmission ratio is forced above this value, such as 0.12 ~ 0.15. For doubly-fed or permanent magnet synchronous wind turbines, the generator speed is limited by the converter capacity and grid frequency, typically having a rated speed range. To allow the generator to operate within a better speed range even in low-wind-speed areas, the transmission ratio (i.e., the gearbox speed increase ratio) needs to be increased so that the generator can still reach the grid-connected speed when the wind turbine is at a lower speed. Conversely, in high-wind-speed areas, the wind turbine speed is already high enough; if a high transmission ratio is still used, the generator speed will exceed the rated value, therefore the transmission ratio needs to be reduced. The "first wind speed range" is typically a low wind speed range (e.g., 3 m / s ~ 8 m / s), where a high gear ratio (e.g., i=1:90) is used to increase the generator speed. The second wind speed range is a high wind speed range (e.g., ≥8 m / s), where a low gear ratio (e.g., i=1:45) is used to limit the generator speed to near its rated value. Turbulence intensity is the ratio of the standard deviation of wind speed to the average wind speed, characterizing the degree of wind pulsation. Under the same average wind speed, if a high gear ratio is used during high turbulence, the rotor speed will change drastically with the wind speed, causing significant impact on the transmission chain. Therefore, when the turbulence intensity exceeds the second threshold (e.g., TI ≥ 0.12), even if the average wind speed is still within the first range, a forced switch to a low gear ratio is implemented to reduce the gear ratio, increase the buffering effect of mechanical inertia, and reduce load impact. This reflects the principle of prioritizing safety over efficiency. Set a hysteresis range, for example, the upper limit of the first range is 8 m / s, but switching back to a lower setting requires a wind speed > 8.5 m / s; or switching back from a lower setting to a higher setting requires a wind speed < 7.5 m / s. In this way, within the range of 7.5 ~ 8.5 m / s, the setting remains unchanged, effectively suppressing oscillations.

[0052] In practical implementation, when the average wind speed is in a low range (first wind speed range) and the turbulence intensity is below the first threshold, a high gear ratio is selected (usually corresponding to a larger torque amplification factor and a lower speed). When the average wind speed is in a high range (second wind speed range) or the turbulence intensity exceeds the second threshold, a low gear ratio is selected (usually corresponding to a smaller torque amplification factor and a higher speed). When the wind speed is within a preset hysteresis range, the gear position remains unchanged to avoid frequent gear changes. The gear decision rule based on wind speed range, turbulence intensity, and hysteresis range is a concrete implementation of the environmental adaptive gear control in this application. By integrating wind speed level and wind speed mass turbulence, and introducing a hysteresis range to ensure stability, it achieves optimal gear selection for wind turbines under different environmental conditions. This rule not only improves the power generation efficiency in low wind speed areas but also strengthens the protection performance in high turbulence and high wind speed areas, while avoiding frequent gear changes, providing a stable and accurate gear benchmark for subsequent gear matching simulation and multi-agent early warning.

[0053] In this embodiment, step S30 further includes: collecting multi-source sensor monitoring data of the aerodynamic system, transmission system, and pitch system of the wind turbine generator set during actual operation; comparing the multi-source sensor monitoring data with the simulation parameters generated by the aerodynamic intelligent agent, transmission intelligent agent, and pitch intelligent agent at the same time to obtain the comparison results; and determining the local multidimensional residual features based on the comparison results.

[0054] It should be noted that various sensors deployed on the aerodynamic, transmission, and pitch systems of the wind turbine generator acquire real-time data reflecting the actual operating status of each subsystem (such as speed, torque, vibration, pitch angle, and current). The collected actual monitoring data is time-aligned with the simulation parameters (i.e., the expected values ​​output by the model) independently generated by the aerodynamic agent, transmission agent, and pitch agent at the same time, so that the residuals can be calculated and compared point by point later.

[0055] Understandably, various sensors deployed on the aerodynamic, transmission, and pitch systems of wind turbine generators acquire real-time data reflecting the actual operating status of each subsystem (such as speed, torque, vibration, pitch angle, and current). The collected actual monitoring data is then time-aligned with the simulation parameters (i.e., the expected values ​​output by the model) independently generated by the aerodynamic agent, transmission agent, and pitch agent at the same time, so that residuals can be calculated and compared point by point later.

[0056] It should be understood that the aerodynamic system sensors include ultrasonic / mechanical anemometers for monitoring real-time wind speed and direction, pitch angle encoders for monitoring the pitch angle of each blade, rotor speed encoders for monitoring rotor speed, and blade root strain gauges for monitoring blade root bending moment and thrust. The transmission system sensors include low-speed shaft torque meters for monitoring main shaft torque, gearbox accelerometers for monitoring housing vibration acceleration, and high-speed shaft speed encoders for monitoring generator input shaft speed. The pitch system sensors include pitch position sensors for monitoring the actual pitch angle, pitch motor current sensors for monitoring motor current, and hydraulic pitch pressure sensors for monitoring cylinder pressure. All sensor signals, after conditioning (filtering, amplification, A / D conversion), are collected via industrial Ethernet at a fixed sampling period to an edge computing gateway or programmable logic controller (PLC). Each data point is precisely timestamped. Each agent (aerodynamic, transmission, pitch) independently runs its physical mathematical model, performs simulation calculations at the same sampling period, and outputs simulation parameters at the corresponding time. To ensure the effectiveness of the comparison, all agents and the data acquisition system use the same system clock, and the simulation step size is the same as or an integer multiple of the data acquisition sampling period, so that each actual data point has a corresponding simulation output point. "Same moment" refers to the same point in time when the actual physical process occurs, where the actual sensor measurement value is paired with the expected value calculated by the simulation. This can be achieved through synchronous triggering: after each sampling period, the data acquisition system sends a new data arrival signal to each agent, and the agent immediately executes a simulation step, outputting a simulation value matching the current moment. Alternatively, offline alignment can be used: all data is timestamped, and the data processing module pairs them according to the timestamp. After the comparison is completed, the obtained local multidimensional residual features are labeled with the corresponding subsystem, timestamp, and current gear, for subsequent fusion analysis.

[0057] In this embodiment, step S40 further includes: calculating the matching degree between the local multidimensional residual features and each fault mode in the preset fault feature map; determining the feature fusion weight coefficient based on the matching degree and the preset collaborative mechanism, wherein the preset collaborative mechanism is based on the collaborative agent receiving the local multidimensional residual features output by each multidimensional analysis agent, as well as the current environmental perception data and gear status, and outputting the dynamic fusion weight of each agent; and performing fusion analysis on the local multidimensional residual features based on the feature fusion weight coefficient to obtain the global anomaly diagnosis result.

[0058] It should be noted that the matching degree between the local multidimensional residual features and each fault mode in the preset fault feature map is calculated; the feature fusion weight coefficient is determined based on the matching degree and the preset collaborative mechanism. The preset collaborative mechanism is based on the collaborative agent receiving the local multidimensional residual features output by each multidimensional analysis agent, as well as the current environmental perception data and gear status, and outputting the dynamic fusion weight of each agent; the local multidimensional residual features are fused and analyzed based on the feature fusion weight coefficient to obtain the global anomaly diagnosis result.

[0059] Understandably, this embodiment can employ a weighted voting collaborative mechanism. The weights are pre-set as follows: aerodynamic agent weight 0.25, transmission agent weight 0.45, pitch agent weight 0.20, and gear agent weight 0.10. Each agent outputs a local fault type and its confidence level based on the matching degree between its residual and the fault map. After weighted summation, the fault type with the highest score is selected as the global anomaly diagnosis result. For example, if the transmission agent outputs "gear wear" with a confidence level of 0.8, the aerodynamic agent outputs "normal" with a confidence level of 0.9, and the pitch agent outputs "normal" with a confidence level of 0.95, then the weighted scores are: gear wear = 0.45 × 0.8 = 0.36, normal = 0.25 × 0.9 + 0.20 × 0.95 = 0.225 + 0.19 = 0.415, resulting in a final diagnosis of "normal," but the global anomaly index is 0.36, indicating a need for attention.

[0060] Furthermore, the multidimensional analysis agent also includes a gear shift agent; the gear shift agent further performs the following steps: after the gear shift command is issued, it monitors the changes in the actual gear position, records the shift duration and the speed and torque fluctuations during the shift process; it compares the actual shift response curve with the preset gear shift transient model and calculates the shift process residual; if the shift process residual exceeds a preset threshold, it generates a fault diagnosis result for the shift mechanism, the fault diagnosis result including shift fork jamming, synchronizer wear, or insufficient hydraulic pressure.

[0061] It should be noted that after a shift command is issued, the gear shifting agent does not simply wait for the shift to complete, but actively performs the following operations: monitoring the actual gear position changes, recording the shift duration, and recording dynamic characteristics such as speed and torque fluctuations during the shift. The actual shift response curve (i.e., the gear position's trajectory over time and the speed / torque waveform) is compared point-by-point or feature-by-feature with the system's built-in gear shifting transient model (i.e., the ideal / nominal shift dynamic characteristics) to calculate the shift process residual. If the shift process residual exceeds a preset threshold, a fault is determined in the shift mechanism, and the specific fault type is output, such as shift fork sticking, synchronizer wear, or insufficient hydraulic pressure.

[0062] In the specific implementation, after the gear shifting command is issued, the gear shifting agent records the signals from the shift fork position sensor and the synchronizer gear ring strain sensor at a sampling frequency of 200Hz, forming the actual shifting position curve s_actual(t) and the strain curve ε(t). s_actual(t) is compared point-by-point with the ideal position curve s_ideal(t) output by the built-in gear shifting transient model to calculate the shifting process residual R_shift = ∫|s_actual(t)-s_ideal(t)|dt / T_shift. If R_shift>0.15, the residual pattern is further analyzed: when the shifting time exceeds the nominal value by more than 40% and there is no abnormal strain, it is diagnosed as insufficient hydraulic pressure; when a strain pulse occurs during the shifting process and the position rebounds, it is diagnosed as synchronizer wear.

[0063] This embodiment provides an anomaly early warning method for wind turbine generator sets. Based on environmental perception data and a preset gear mapping table, it autonomously decides the optimal gear state under the current environment. Then, it calls the mathematical model parameter set matched with the gear to perform local simulation and obtain simulation parameters. Using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear. Finally, through a preset collaborative mechanism among the multidimensional analysis agents, it fuses and analyzes the residual features to obtain a global anomaly diagnosis result and makes a prediction and alarm. Compared with existing precise monitoring methods for wind turbine generator sets that cannot achieve gear adaptation and have poor anomaly detection accuracy, this embodiment, based on the full-chain adaptive monitoring of environment, gear, model, residual, collaboration, and early warning, significantly improves the accuracy of anomaly detection and the refinement of fault location.

[0064] Based on the above Figure 1 The first embodiment shown presents a third embodiment of the wind turbine generator abnormality early warning method of this application; refer to Figure 3 , Figure 3 This is a flowchart illustrating the third embodiment of the wind turbine generator abnormality early warning method of this application. Based on the first embodiment of this application, the same or similar content as the first embodiment described above can be referred to the above description and will not be repeated hereafter.

[0065] In this embodiment, as Figure 3 As shown, step S50 further includes: S501: Input the time series of local multidimensional residual features output by the multidimensional analysis agent, the current environmental perception data, and the current gear status into the time series prediction model to determine the residual prediction information within the target time period.

[0066] It should be noted that the time series of local multidimensional residual features output by the multidimensional analysis agent, the current environmental perception data, and the current gear status are input into the time series prediction model to determine the residual prediction information within the target time period. The target time period can be a future time period.

[0067] Understandably, residual prediction information can be a residual prediction sequence for a future target time period.

[0068] S502: If the residual prediction information is such that the alarm threshold will be exceeded at the target time point, generate alarm information including fault mode, fault location, remaining effective life and recommended maintenance actions.

[0069] It should be noted that the time series of local multidimensional residual features output by the multidimensional analysis agent (i.e., the trajectory of residual changes in each subsystem over a past period), along with current environmental perception data (such as wind speed and turbulence intensity) and the current gear status, are input into a time-series prediction model. This model calculates residual prediction information (such as residual change trends and the time point when thresholds are reached) for the target future time period. If the prediction result indicates that the residual will exceed a preset alarm threshold at a certain future time point, the system generates a structured alarm message, which includes at least: fault mode, fault location, remaining effective lifespan, and recommended maintenance actions.

[0070] Understandably, the system presets multiple alarm thresholds, which may include warning thresholds, alarm thresholds, and emergency thresholds. For example, a warning threshold indicating that a residual exceeding a certain value indicates a significant abnormal trend, but has not yet jeopardized safety. An alarm threshold indicating that a residual exceeding a certain value indicates a planned shutdown is required. An emergency threshold indicating that a residual exceeding a certain value indicates an immediate shutdown is necessary. Remaining effective lifetime refers to the time remaining from the current moment until the residual prediction reaches the emergency threshold (or critical failure threshold).

[0071] It should be understood that after the time-series prediction model outputs the future residual trajectory, it determines whether this trajectory will exceed a certain threshold at a future time point. If so, an alarm of the corresponding level is triggered, and the time when the threshold is reached is recorded. For example: Input: Transmission vibration residual sequence for the past 24 hours (one point every 10 minutes), current wind speed = 6.5 m / s, turbulence intensity = 0.09, gear = high gear ratio. Time-series prediction model: LSTM network, input 60 time steps (10 hours), output residual prediction for the next 120 time steps (20 hours). Prediction result: The model predicts that the vibration residual will slowly increase from 0.07 m / s² to 0.10 m / s² (warning threshold) in the next 5 hours, and reach 0.15 m / s² (emergency threshold) in the next 18 hours. Alarm generation: Trigger a level 2 alarm (orange), output: Fault mode: Gear tooth surface wear. Fault location: Transmission system - gearbox - intermediate shaft gear. Remaining effective life: Approximately 18 hours. Recommended maintenance actions: It is recommended to schedule a planned shutdown within 12 hours to prepare for endoscopy and oil analysis, and to inspect the gearbox. This embodiment provides a wind turbine generator abnormality early warning method. Based on environmental perception data and a preset gear mapping table, it autonomously decides the optimal gear state under the current environment; then, it calls the mathematical model parameter group matching the gear to perform local simulation and obtain simulation parameters; using a multidimensional analysis agent, it compares the actual monitoring data with the simulation parameters to generate local multidimensional residual features associated with the current gear; finally, through a preset collaborative mechanism between multidimensional analysis agents, it fuses and analyzes the residual features to obtain global abnormality diagnosis results and predict alarms. Compared with existing methods for wind turbine generators that cannot achieve gear adaptive accurate monitoring and have poor abnormality detection accuracy, this embodiment integrates historical residuals, current environment, and gear state into a time series model to predict future residual evolution and generates alarm information including fault mode, location, remaining effective life, and maintenance suggestions based on threshold triggering timing. This feature enables operations and maintenance teams to shift from reactive to proactive maintenance, significantly reducing unplanned downtime and operating costs, and improving the overall life-cycle economics of wind turbines.

[0072] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the wind turbine generator abnormality early warning method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0073] This application also provides an early warning device for abnormal wind turbine generator sets; please refer to... Figure 4 The wind turbine generator abnormality early warning device includes: The gear decision module 10 is used to determine the optimal gear state under the current environment based on the environmental perception data corresponding to the wind turbine generator set and the preset gear mapping table. Simulation module 20 is used to call the mathematical model parameter group matching the optimal gear state to perform local simulation and obtain simulation parameters. The intelligent agent analysis module 30 is used to compare the actual monitoring data with the simulation parameters based on the multidimensional analysis intelligent agent to obtain the local multidimensional residual features associated with the current gear position; The collaborative fusion module 40 is used to perform fusion analysis on the local multidimensional residual features based on the multidimensional analysis agent and the preset collaborative mechanism to obtain the global anomaly diagnosis result. The predictive alarm module 50 is used to predict alarms based on the global anomaly diagnosis results.

[0074] The wind turbine generator abnormality early warning device provided in this application adopts the wind turbine generator abnormality early warning method in the above embodiments, which can solve the problem of poor accuracy in existing wind turbine generators due to the inability to achieve gear adaptive precise monitoring. Compared with the prior art, the beneficial effects of the wind turbine generator abnormality early warning device provided in this application are the same as those of the wind turbine generator abnormality early warning method provided in the above embodiments, and other technical features in the wind turbine generator abnormality early warning device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0075] This application provides a wind turbine generator abnormality early warning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the wind turbine generator abnormality early warning method in the first embodiment described above.

[0076] The following is for reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing the wind turbine generator abnormality early warning device in the embodiments of this application. The wind turbine generator abnormality early warning device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The wind turbine generator abnormality early warning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0077] like Figure 5As shown, the wind turbine generator abnormality early warning device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the wind turbine generator abnormality early warning device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the wind turbine generator anomaly early warning device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows wind turbine generator anomaly early warning devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0078] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0079] The wind turbine generator abnormality early warning device provided in this application, employing the wind turbine generator abnormality early warning method described in the above embodiments, can solve the technical problems of existing methods for accurately monitoring wind turbine generators that cannot achieve gear adaptation and having poor accuracy in abnormality detection. Compared with the prior art, the beneficial effects of the wind turbine generator abnormality early warning device provided in this application are the same as those of the wind turbine generator abnormality early warning method provided in the above embodiments, and other technical features of this wind turbine generator abnormality early warning device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0080] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0081] 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.

[0082] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the wind turbine generator abnormality early warning method in the above embodiments.

[0083] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0084] The aforementioned computer-readable storage medium may be included in the wind turbine generator abnormality early warning device; or it may exist independently and not be installed in the wind turbine generator abnormality early warning device.

[0085] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the wind turbine generator abnormality early warning device, the wind turbine generator abnormality early warning device: autonomously decides the optimal gear state under the current environment based on environmental perception data and a preset gear mapping table; then calls the mathematical model parameter set matched with the gear to perform local simulation and obtain simulation parameters; uses a multidimensional analysis agent to compare the actual monitoring data with the simulation parameters and generates local multidimensional residual features associated with the current gear; finally, it performs fusion analysis of the residual features through a preset collaborative mechanism among the multidimensional analysis agents to obtain a global abnormality diagnosis result and issue a prediction alarm.

[0086] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0087] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0088] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0089] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described wind turbine generator abnormality early warning method. This solves the technical problem of existing methods failing to achieve accurate monitoring of wind turbine generators with adaptive gear positions and exhibiting poor accuracy in abnormality detection. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the wind turbine generator abnormality early warning method provided in the above embodiments, and will not be elaborated upon here.

[0090] The above description is only a part of the embodiments of this application and does not limit the scope of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included within the protection scope of this application.

Claims

1. A method for early warning of anomalies in wind turbine generator sets, characterized in that, The wind turbine generator abnormality early warning method includes: The optimal gear position is determined based on the environmental perception data corresponding to the wind turbine generator set and the preset gear mapping table. Based on the optimal gear position, the mathematical model parameter group matching the gear position is invoked to perform local simulation and obtain simulation parameters; The multidimensional analysis agent compares the actual monitoring data with the simulation parameters to obtain the local multidimensional residual features associated with the current gear. Based on a multidimensional analysis agent and a preset collaborative mechanism, the local multidimensional residual features are fused and analyzed to obtain global anomaly diagnosis results. Predictive alerts are generated based on the global anomaly diagnosis results.

2. The method as described in claim 1, characterized in that, After determining the optimal gear position under the current environment based on the environmental perception data corresponding to the wind turbine generator set and the preset gear mapping table, the process also includes: Obtain the wind speed time series from the environmental perception data, and predict the wind speed state within the target time window based on the preset wind speed prediction model and the wind speed time series; If the wind speed state is about to enter the environmental parameter range corresponding to different gears, a pre-shift command is generated before the wind speed state reaches the shift threshold. The pre-shift command is used to adjust the mathematical model parameter group that matches the target gear in advance, and load the adjusted parameters into the local simulation model before the shift is executed.

3. The method as described in claim 1, characterized in that, The process of determining the optimal gear position under the current environment based on environmental perception data corresponding to the wind turbine generator set and a preset gear mapping table includes: If the average wind speed in the environmental perception data is in the first wind speed range and the turbulence intensity is lower than the first threshold, the optimal gear state is determined to be the high gear ratio gear. If the average wind speed is in the second wind speed range or the turbulence intensity exceeds the second threshold, then the optimal gear state is determined to be the low gear ratio gear. If the wind speed is within the preset hysteresis range, the current gear setting will remain unchanged.

4. The method as described in claim 1, characterized in that, The multidimensional analysis agent includes an aerodynamic agent for the pneumatic system, a transmission agent for the transmission system, and a pitch agent for the pitch system. The multidimensional analysis agent compares actual monitoring data with the simulation parameters to obtain local multidimensional residual features associated with the current gear position, including: Collect multi-source sensor monitoring data of the aerodynamic system, transmission system and pitch system of the wind turbine generator set during actual operation; The monitoring data from the multi-source sensors are compared with the simulation parameters generated by the aerodynamic intelligent agent, transmission intelligent agent, and pitch intelligent agent at the same time to obtain the comparison results. The local multidimensional residual characteristics are determined based on the comparison results.

5. The method as described in claim 1, characterized in that, The method of fusing and analyzing the local multidimensional residual features based on a multidimensional analysis agent and a preset collaborative mechanism to obtain global anomaly diagnosis results includes: Calculate the matching degree between the local multidimensional residual features and each fault mode in the preset fault feature map; The feature fusion weight coefficient is determined based on the matching degree and the preset collaboration mechanism. The preset collaboration mechanism is based on the collaborative agent receiving the local multidimensional residual features output by each multidimensional analysis agent, as well as the current environmental perception data and gear status, and outputting the dynamic fusion weight of each agent. Based on the feature fusion weight coefficients, the local multidimensional residual features are fused and analyzed to obtain global anomaly diagnosis results.

6. The method as described in claim 5, characterized in that, The multidimensional analysis agent also includes a gear shift agent; the gear shift agent further performs the following steps: After the shift command is issued, monitor the changes in the actual gear status and record the shift duration and the fluctuations in speed and torque during the shift process; The actual shift response curve is compared with the preset gear shift transient model to calculate the shift process residual; If the residual error of the shifting process exceeds a preset threshold, a fault diagnosis result of the shifting mechanism is generated. The fault diagnosis result includes shift fork jamming, synchronizer wear, or insufficient hydraulic pressure.

7. The method as described in claim 1, characterized in that, The step of predicting and alerting based on the global anomaly diagnosis results includes: The time series of local multidimensional residual features output by the multidimensional analysis agent, the current environmental perception data, and the current gear status are input into the time series prediction model to determine the residual prediction information within the target time period. If the residual prediction information is to generate alarm information including fault mode, fault location, remaining effective life and recommended maintenance actions when the alarm threshold is exceeded at the target time point.

8. A wind turbine generator abnormality early warning device, characterized in that, The wind turbine generator abnormality early warning device includes: The gear decision module is used to determine the optimal gear state under the current environment based on the environmental perception data corresponding to the wind turbine generator and the preset gear mapping table. The simulation module is used to perform local simulation based on the optimal gear state by calling the mathematical model parameter group that matches the gear, and to obtain simulation parameters. The intelligent agent analysis module is used to compare the actual monitoring data with the simulation parameters based on the multidimensional analysis intelligent agent to obtain the local multidimensional residual features associated with the current gear position; The collaborative fusion module is used to perform fusion analysis on the local multidimensional residual features based on the multidimensional analysis agent and the preset collaborative mechanism to obtain global anomaly diagnosis results. The predictive alarm module is used to predict alarms based on the global anomaly diagnosis results.

9. A wind turbine generator abnormality early warning device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the wind turbine generator abnormality early warning method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the wind turbine generator abnormality early warning method as described in any one of claims 1 to 7.

Images