A cooling control method and system for wind turbine generator sets

By combining real-time data acquisition and multi-dimensional feature analysis with historical operating conditions, dynamically dividing the thermal phase and setting the prediction time domain, the problems of heat fluctuation interference and health status misjudgment in the cooling control of wind turbine generator sets are solved, and precise cooling system management and energy consumption optimization are achieved.

CN123109984APending Publication Date: 2026-11-17ZHANGJIAGANG HENGQIANG COOLING EQUIP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611148372.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing wind turbine cooling control technologies rely on local temperature data of a single target unit, fail to combine horizontal comparisons with neighboring units and multidimensional environmental characteristics, making it difficult to isolate heat fluctuation interference, misjudge the health status of the cooling system, fail to accurately match the heat load, and lack predictive intervention and closed-loop feedback for energy consumption assessment.

Method used

Real-time acquisition of operating and environmental data of the target and neighboring units, multi-dimensional feature extraction, combined with historical operating condition analysis, dynamic division of thermal stages, setting of prediction time domain for temperature trend prediction, execution of adaptive cooling strategy, and self-learning through control effect evaluation coefficient.

Benefits of technology

It enables accurate assessment of the cooling system's health status, avoids thermal stress shocks, ensures that the control strategy operates within the optimal range, and improves the matching between heat dissipation capacity and energy efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123109984A_ABST
    Figure CN123109984A_ABST
Patent Text Reader

Abstract

This invention relates to the field of cooling control for wind turbine generator sets, and discloses a cooling control method and system for wind turbine generator sets. The method involves collecting and preprocessing operational and environmental data from the target and neighboring units, extracting and analyzing features to obtain a comprehensive cooling thermal state evaluation coefficient, and assessing the health status. It then combines historical operating conditions with matching analysis to obtain operating condition deviation and cooling efficiency factors, thereby dividing the thermal phase and invoking a matched temperature rise rate safety threshold and temperature response lag time. Based on the lag time, a prediction time domain is set, the temperature rise rate is predicted, and compared with the safety threshold to determine the cooling safety status level and execute the corresponding control strategy. Finally, data after strategy execution is collected, and temperature changes and energy consumption are compared and analyzed to obtain a control effect evaluation coefficient. The comprehensive cooling thermal state evaluation coefficient is then adaptively corrected. This method achieves accurate thermal state assessment, proactive preventative control, and closed-loop adaptive optimization.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of wind turbine generator cooling control technology, and more specifically to a cooling control method and system for wind turbine generators. Background Technology

[0002] During wind turbine operation, key components such as the generator, gearbox, and converter generate significant amounts of heat. The thermal changes of these components are influenced by a combination of factors, including the unit's active power load, external environmental conditions (such as wind speed and ambient temperature), and the operating status of the cooling system itself, exhibiting strong coupling and high dynamic characteristics. To ensure the safe and stable operation of the unit and extend its service life, the cooling system needs to continuously adjust its heat dissipation capacity to maintain the operation of key components within safe temperature ranges. This thermal management process involves a complex mapping relationship between operating data, environmental parameters, and thermal response status. It relies on continuous monitoring of the unit's real-time operating conditions and external environmental information, multi-dimensional feature extraction, and dynamic assessment of the cooling system's health status and cooling efficiency to achieve precise matching and dynamic response between heat dissipation capacity and the unit's real-time heat load.

[0003] However, existing cooling control technologies for wind turbine generators still have the following drawbacks: Firstly, existing technologies typically rely solely on local temperature data from a single target unit as the basis for cooling start-up and shutdown, without comprehensively combining comparative data from neighboring units and multi-dimensional operating environment characteristics for feature extraction, and without incorporating historical operating conditions for matching analysis. This makes it difficult for the system to accurately isolate the heat fluctuation interference caused by sudden changes in the external environment (such as a sudden increase in wind speed or abnormal ambient temperature), and it is impossible to accurately quantify the deviation of the current operating conditions and the actual cooling efficiency reduction of the cooling system, which can easily lead to misjudgment of the health status. Secondly, most existing cooling controls use fixed temperature thresholds or single PID parameters, failing to dynamically divide the operating thermal phases based on the unit's real-time operating condition deviation and cooling efficiency factor. This static parameter setting cannot adapt to the unit's highly dynamic and strongly coupled heat load changes, and under complex operating conditions, it is easy to cause lag in cooling intervention or over-cooling, making it difficult to achieve accurate matching between heat dissipation capacity and real-time heat load. Third, existing methods are mostly reactive mechanisms, that is, cooling commands are only triggered after the temperature of critical components actually exceeds the limit. This method does not consider the "temperature response lag time" of the cooling system itself and does not set a prediction time domain to predict trends in advance. Because the cooling safety status level cannot be determined in advance, the system cannot take preventive intervention before the temperature runs out of control, which can easily cause critical components to be subjected to severe thermal stress shocks and affect the life of the equipment. Fourth, existing technologies often only focus on whether the temperature drops after performing cooling control, while ignoring the additional energy consumption cost brought by the high-load operation of fans and water pumps. They do not comprehensively evaluate the cost-effectiveness of the control between the "temperature suppression effect" and the "increase in cooling energy consumption", and lack a closed-loop feedback mechanism to adaptively correct the front-end health assessment baseline based on the control effect evaluation coefficient. As a result, the system cannot learn and evolve its parameters through historical execution effects in long-term operation, and the control strategy gradually deviates from the optimal operating range. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a cooling control method and system for wind turbine generator sets to solve the problems existing in the background art.

[0005] This invention provides the following technical solution: a cooling control method for a wind turbine generator set, comprising: S1: Real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent units, and preprocessing of the acquired operating data and external environmental data; S2: Based on operational data and external environmental data, feature extraction and feature analysis are performed on the target unit to obtain the comprehensive evaluation coefficient of cooling thermal state. Based on the comprehensive evaluation coefficient of cooling thermal state, the health status of the cooling system of the target unit is evaluated. S3: Based on the health assessment results and historical operating data, perform operating condition matching analysis to obtain the operating condition deviation and cooling efficiency factor. S4: Divide the current operating thermal stage according to the deviation of the operating condition and the cooling efficiency factor, and call the control parameter group that matches the current thermal stage from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time. S5: Set the prediction time domain according to the temperature response lag time, predict the trend of the temperature of key components, obtain the predicted temperature rise rate, compare the predicted temperature rise rate with the temperature rise rate safety threshold, and determine the cooling safety status level. S6: Execute the corresponding cooling control strategy according to the cooling safety status level; S7: By collecting temperature response data after the control strategy is executed, comparing and analyzing the temperature changes and energy consumption before and after the control strategy is executed, the control effect evaluation coefficient is obtained, and the cooling thermal state comprehensive evaluation coefficient is adaptively corrected based on the control effect evaluation coefficient.

[0006] Preferably, the real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent turbines by S1 specifically includes: Using data acquisition terminals deployed on the wind turbine generator side or in the control center, the monitoring and data acquisition systems of the target wind turbine generator and neighboring units are accessed in real time via industrial communication protocols. Operating status parameters and external environmental parameters are collected from the corresponding databases, and these parameters are synchronously sampled according to a preset sampling frequency to obtain raw time-series data. The operating status parameters include active power, generator temperature, gearbox oil temperature, converter temperature, cooling medium temperature, cooling fan speed, and cooling pump current. The external environmental parameters include wind speed at hub height, ambient temperature, and wind direction. The preprocessing of the collected operational data and external environment data specifically includes: The operational data and external environment data acquired as raw time series data are cleaned and quality controlled to detect and remove outliers and missing data in the raw time series data; the raw time series data of the target unit and neighboring units are time aligned using a unified timestamp as a reference; and the time-aligned data are processed using a preset smoothing filtering algorithm to remove high-frequency noise interference, finally obtaining standard time series data.

[0007] Preferably, S2 specifically includes: The initial ratio of the current power to the rated power is calculated based on the active power in the standard time series data. The moving average value of the initial ratio within a preset sliding time window is obtained, and the deviation of the moving average value from the steady-state operating reference load is calculated. The deviation is used as the power load rate feature. Based on the generator temperature, gearbox oil temperature and converter temperature in the standard time series data, the temperature change at adjacent time points is calculated for each sampling point to obtain the instantaneous temperature rise rate of each component. The multiple instantaneous temperature rise rates are summed according to preset weights to obtain the initial comprehensive temperature rise rate. Then, the initial comprehensive temperature rise rate is compensated and corrected by combining the change in ambient temperature at the same time to obtain the compensated comprehensive temperature rise rate characteristic value. The compensated and corrected comprehensive temperature rise rate is used as the temperature rise rate characteristic of key components. The theoretical heat dissipation capacity characteristic value of the current cooling system is calculated based on the cooling fan speed and cooling pump current in the standard time series data. The actual heat dissipation effect characteristic value of the current cooling system is calculated based on the time series change of the cooling medium temperature. The difference between the theoretical heat dissipation capacity characteristic value and the actual heat dissipation effect characteristic value is calculated, and the difference is smoothed within the time window to obtain the performance deviation evaluation value. The performance deviation evaluation value is used as the characteristic of the heat transfer efficiency of the cooling medium. The power load rate characteristics, key component temperature rise rate characteristics, and cooling medium heat transfer efficiency characteristics are constructed into a multi-dimensional feature vector. The deviation analysis is performed between this vector and the standard operating feature vector of neighboring units under similar environmental conditions to generate a comprehensive evaluation coefficient of the target unit's cooling thermal state. The overall cooling thermal state evaluation coefficient is compared with the preset overall cooling thermal state evaluation threshold to determine the current health status of the cooling system, specifically including: If the cooling thermal state comprehensive evaluation coefficient is less than the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be abnormal, and the abnormal attribution and operating condition matching analysis process is triggered; if the cooling thermal state comprehensive evaluation coefficient is greater than or equal to the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be normal, and the analysis process of S5 continues.

[0008] Preferably, S3 specifically includes: When the current cooling system's health status is determined to be abnormal, based on the external environmental parameters in the current standard time series data, a set of historical samples with environmental similarity meeting preset conditions is retrieved from the historical operating condition database. The power load rate feature sequence and the key component temperature rise rate feature sequence are extracted from this historical sample set, and the mean and standard deviation of these two feature sequences are calculated respectively. Then, the power difference between the current target unit's power load rate feature and the corresponding historical mean, and the temperature rise difference between the key component temperature rise rate feature and the corresponding historical mean are calculated respectively. The power difference is divided by the corresponding historical power standard deviation to obtain the power relative deviation rate, and the temperature rise difference is divided by the corresponding historical temperature rise standard deviation to obtain the temperature rise relative deviation rate. The power relative deviation rate and the temperature rise relative deviation rate are weighted and summed according to preset weights to obtain the operating condition deviation degree. The actual energy consumption index of the cooling system is obtained by calculating the product of the current cooling fan speed and the cooling pump current. The current initial cooling efficiency index is obtained by calculating the ratio of the current temperature rise rate characteristic of the key components to the actual energy consumption index. Historical initial cooling efficiency indices with the same actual energy consumption index are extracted from the historical sample set and their average values ​​are calculated. The efficiency decay is obtained by calculating the difference between the current initial cooling efficiency index and the historical average value. The relative ratio of the efficiency decay to the historical average value is calculated. Finally, the relative ratio is nonlinearly weighted and fused with the operating condition deviation, and the fusion result is mapped to a preset interval to generate a cooling efficiency factor.

[0009] Preferably, S4 specifically includes: Based on the deviation of the operating condition and the cooling efficiency factor, the ratio of the deviation of the operating condition to the deviation of the preset operating condition benchmark is calculated to obtain the deviation ratio, and the difference between the cooling efficiency factor and the preset safety efficiency threshold is calculated to obtain the efficiency difference. The thermal stage discrimination index is obtained by calculating the ratio of the deviation ratio to the efficiency difference. This index is then compared with multiple preset thermal stage judgment intervals to determine the current operating thermal stage, specifically including: Multiple thermal stage judgment intervals are pre-defined and sorted in order of numerical value. Each thermal stage judgment interval is pre-bound with a unique thermal stage level. The thermal stage discrimination index is compared with the upper and lower boundary values ​​of each thermal stage judgment interval in turn to determine the target thermal stage judgment interval into which the thermal stage discrimination index falls. The thermal stage level bound to the target thermal stage judgment interval is then determined as the current running thermal stage. Based on a defined thermal stage, the baseline temperature rise rate safety threshold and baseline temperature response hysteresis time that match the thermal stage are retrieved from the preset parameter library. By calculating the dynamic ratio of the current temperature rise rate characteristic of key components to the basic temperature rise rate safety threshold, the basic temperature response lag time is dynamically reduced and corrected according to the preset exponential decay function based on this dynamic ratio, and the target temperature response lag time is obtained. The basic temperature rise rate safety threshold is used as the temperature rise rate safety threshold, and the target temperature response lag time is used as the temperature response lag time. Together, they constitute the current control parameter set.

[0010] Preferably, S5 specifically includes: Obtain the current health status assessment results of the cooling system to determine the target control parameters: if the health status is normal, the preset reference temperature response lag time and reference temperature rise rate safety threshold are used as the target control parameters; if the health status is abnormal, the temperature response lag time and temperature rise rate safety threshold in the current control parameter group are used as the target control parameters. Set the time length of the temperature response lag time in the target control parameters to the prediction time domain; The generator temperature, gearbox oil temperature, and converter temperature at the current moment are obtained from standard time series data, and the temperature rise rate characteristics of key components are obtained. Based on the generator temperature, gearbox oil temperature, and converter temperature at the current moment, temperature prediction analysis is performed by combining the temperature rise rate characteristics of key components with the time length of the prediction time domain to obtain the predicted temperature value at the end of the prediction time domain. The predicted temperature change is obtained by calculating the difference between the predicted temperature value and the current generator temperature, gearbox oil temperature and converter temperature, respectively. The ratio of the predicted temperature change to the time length of the predicted time domain is used as the predicted temperature rise rate. The temperature rise risk coefficient is obtained by calculating the ratio of the predicted temperature rise rate to the safe threshold temperature rise rate in the target control parameters. Multiple consecutive cooling safety status judgment intervals are preset, and each cooling safety status judgment interval is pre-bound with a unique safety status level. The temperature rise risk coefficient is compared with the upper and lower boundary values ​​of each cooling safety status judgment interval in turn to determine the target status judgment interval into which the temperature rise risk coefficient falls, and the safety status level bound to the target status judgment interval is determined as the current cooling safety status level.

[0011] Preferably, S6 specifically includes: The mapping relationship between multiple cooling safety status levels and different levels of cooling control command sets is preset, and the target cooling control command set that matches the current cooling safety status level is obtained. If the current cooling safety status level is a safe status level, then a linear fine-tuning command is generated based on the current cooling fan speed and cooling pump current in the standard timing data to maintain the current heat dissipation status of the cooling system. If the current cooling safety status level is risky or dangerous, then the forced heat dissipation increment is calculated by combining the temperature rise risk coefficient and the temperature rise rate safety threshold in the target control parameter group, and a forced cooling command is generated to increase the cooling fan speed and cooling pump current to the corresponding preset operating range. At the same time, a corresponding active power load reduction command is generated based on the temperature rise risk coefficient. Finally, the generated linear fine-tuning command, forced cooling command, or active power load reduction command is sent to the actuator of the target wind turbine to dynamically adjust the heat dissipation capacity of the cooling system and limit the power generation load of the unit.

[0012] Preferably, S7 specifically includes: Within a preset time window after the cooling control strategy is executed, the generator temperature, gearbox oil temperature and converter temperature of the target wind turbine are re-collected as post-execution temperature response data. Based on this post-execution temperature response data, the actual temperature rise rate is extracted, and the difference between the temperature rise rate safety threshold and the actual temperature rise rate is calculated to obtain the effective temperature suppression amount. Simultaneously, the cooling fan speed and cooling pump current of the cooling system are collected after execution. Based on the cooling fan speed and cooling pump current after execution, the cooling energy consumption increment is calculated relative to the cooling fan speed and cooling pump current before execution. The control effect evaluation coefficient is obtained by calculating the ratio of the effective amount of temperature suppression to the increase in cooling energy consumption. The control effect evaluation coefficient and the cooling thermal state comprehensive evaluation coefficient are weighted and fused to generate the corrected cooling thermal state comprehensive evaluation coefficient, which is then used as the input benchmark for assessing the health status of the target unit's cooling system in the next control cycle.

[0013] To achieve the above objectives, the present invention provides the following technical solution: a cooling control system for a wind turbine generator set, comprising the following steps for implementing the aforementioned cooling control method for a wind turbine generator set: Data acquisition and preprocessing module: Real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent units, and preprocessing of the acquired operating data and external environmental data; Thermal health assessment module: Based on operating data and external environmental data, feature extraction and feature analysis are performed on the target unit to obtain the comprehensive thermal evaluation coefficient of cooling, and the health status of the cooling system of the target unit is assessed based on the comprehensive thermal evaluation coefficient of cooling. Operating condition deviation analysis module: Based on the health assessment results and historical operating condition data, it performs operating condition matching analysis to obtain the operating condition deviation degree and cooling efficiency factor; Thermal Stage and Parameter Adjustment Module: The current operating thermal stage is divided according to the deviation of the operating condition and the cooling efficiency factor, and the control parameter group matching the current thermal stage is called from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time. Situation prediction and judgment module: Based on the temperature response lag time, the prediction time domain is set, the temperature trend of key components is predicted, the predicted temperature rise rate is obtained, and the predicted temperature rise rate is compared with the temperature rise rate safety threshold to determine the cooling safety situation level. Control strategy execution module: Executes the corresponding cooling control strategy based on the cooling safety status level; Effect evaluation and correction module: By collecting temperature response data after the control strategy is executed, comparing and analyzing the temperature changes and energy consumption before and after the control strategy is executed, the control effect evaluation coefficient is obtained, and the cooling thermal state comprehensive evaluation coefficient is adaptively corrected based on the control effect evaluation coefficient.

[0014] The technical effects and advantages of this invention are as follows: By collecting real-time operating and environmental data of the target unit and neighboring units, and extracting multi-dimensional features to obtain the comprehensive evaluation coefficient of the cooling thermal state, and combining it with historical operating data for operating condition matching analysis, this process effectively introduces horizontal comparison with neighboring units and historical longitudinal benchmarks. It can accurately isolate the heat fluctuation interference caused by sudden changes in the external environment, accurately quantify the deviation of the current operating condition and the actual cooling efficiency decay of the cooling system, thereby avoiding misjudgment of the health status caused by single local data and achieving accurate assessment of the health status of the cooling system.

[0015] The current operating thermal stage is dynamically divided based on the deviation of operating conditions and the cooling efficiency factor, and the control parameter group matching the current thermal stage is called from the preset parameter library. This process breaks the limitations of traditional fixed temperature thresholds or single PID parameters, enabling the control parameters to adapt to the high dynamic and strongly coupled heat load changes of the unit. This fundamentally avoids the problem of delayed or excessive cooling intervention under complex operating conditions, and achieves precise matching between heat dissipation capacity and real-time heat load.

[0016] By using the "temperature response lag time" in the control parameter group to set the prediction time domain, the temperature trend of key components is predicted to obtain the predicted temperature rise rate. Combined with the temperature rise rate safety threshold, the cooling safety status level is determined in advance, and then the corresponding strategy is executed. This can transform the traditional "post-event passive response" into "pre-event proactive prevention". It is beneficial to carry out preventive intervention before the temperature actually runs out of control, effectively avoid the severe thermal stress impact on key components due to the rapid temperature exceeding the limit, and significantly extend the service life of the equipment.

[0017] By collecting temperature response data after the control strategy is executed, and comprehensively comparing and analyzing the "temperature change" and "energy consumption" before and after execution, a control effect evaluation coefficient reflecting the cost-effectiveness of "temperature suppression effect" and "cooling energy consumption increment" is obtained. Based on this, the front-end cooling thermal state comprehensive evaluation coefficient is adaptively corrected. This process not only overcomes the problem of existing technologies ignoring the cost of cooling energy consumption, but also constructs a complete closed-loop feedback channel, enabling the system to learn from historical execution results and dynamically evolve the evaluation baseline, ensuring that the control strategy remains in the optimal operating range for a long time. Attached Figure Description

[0018] Figure 1 This is a diagram illustrating the method steps of the present invention.

[0019] Figure 2 This is a system structure block diagram of the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The cooling control method and system for wind turbine generator sets involved in the present invention are not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 The embodiment shown provides a cooling control method for a wind turbine generator set, including: S1: Real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent units, and preprocessing of the acquired operating data and external environmental data.

[0022] In this embodiment, the real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent turbines by S1 specifically includes: Using data acquisition terminals deployed on the wind turbine generator side or in the control center, the monitoring and data acquisition systems of the target wind turbine generator and neighboring units are accessed in real time via industrial communication protocols. Operating status parameters and external environmental parameters are collected from the corresponding databases, and these parameters are synchronously sampled according to a preset sampling frequency to obtain raw time-series data. The operating status parameters include active power, generator temperature, gearbox oil temperature, converter temperature, cooling medium temperature, cooling fan speed, and cooling pump current. The external environmental parameters include wind speed at hub height, ambient temperature, and wind direction. The preprocessing of the collected operational data and external environment data specifically includes: The operational data and external environment data acquired as raw time series data are cleaned and quality controlled to detect and remove outliers and missing data in the raw time series data; the raw time series data of the target unit and neighboring units are time aligned using a unified timestamp as a reference; and the time-aligned data are processed using a preset smoothing filtering algorithm to remove high-frequency noise interference, finally obtaining standard time series data.

[0023] It should be specifically noted that, in actual execution, the data acquisition terminal can access the SCADA system of the wind turbine generator set via industrial communication protocols such as Modbus TCP or OPCUA, and synchronously acquire operating status parameters such as active power and generator temperature of the target unit and neighboring units, as well as external environmental parameters such as wind speed and ambient temperature, at a preset sampling frequency of, for example, 1Hz, as raw time series data. Subsequently, this raw time series data is cleaned and quality controlled, specifically using the Raida criterion (…). (Guideline) Detects and eliminates outliers caused by sensor malfunctions, and sets differentiated judgment criteria for different types of parameters: For parameters such as active power with large magnitudes and drastic fluctuations, the dynamic mean and standard deviation within a sliding time window are used for calculation. The interval is used to eliminate abrupt outliers; for parameters such as generator temperature, which have small magnitudes and change slowly, a static method based on global historical statistical characteristics is used. Intervals are eliminated, and linear interpolation is used to fill in missing data caused by communication packet loss. Next, using a unified absolute timestamp as a benchmark, the time series data of the target unit and neighboring units are time-aligned. During this process, a time tolerance threshold of, for example, 200 milliseconds is set. If the timestamp deviation of a unit exceeds this tolerance threshold due to communication delay, the data at that moment is considered out of sync, marked as missing, and resampled and aligned using the same interpolation algorithm to ensure the time synchronization of cross-unit comparison. Finally, moving average filtering or Butterworth low-pass filtering algorithms are used to remove high-frequency noise interference caused by high-frequency wind speed fluctuations from the time-aligned data. The window length or cutoff frequency is dynamically calibrated according to the turbulence intensity of the current wind field. In high-turbulence wind fields, the moving average window length is appropriately increased or the cutoff frequency of Butterworth low-pass filtering is decreased to enhance the smoothing effect. Finally, clean and time-synchronized standard time series data is output for subsequent feature extraction.

[0024] S2: Based on the operating data and external environment data, feature extraction and feature analysis are performed on the target unit to obtain the comprehensive evaluation coefficient of cooling thermal state. Based on the comprehensive evaluation coefficient of cooling thermal state, the health status of the cooling system of the target unit is evaluated.

[0025] In this embodiment, S2 specifically includes: The initial ratio of the current power to the rated power is calculated based on the active power in the standard time series data. The moving average value of the initial ratio within a preset sliding time window is obtained, and the deviation of the moving average value from the steady-state operating reference load is calculated. The deviation is used as the power load rate feature. Based on the generator temperature, gearbox oil temperature and converter temperature in the standard time series data, the temperature change at adjacent time points is calculated for each sampling point to obtain the instantaneous temperature rise rate of each component. The multiple instantaneous temperature rise rates are summed according to preset weights to obtain the initial comprehensive temperature rise rate. Then, the initial comprehensive temperature rise rate is compensated and corrected by combining the change in ambient temperature at the same time to obtain the compensated comprehensive temperature rise rate characteristic value. The compensated and corrected comprehensive temperature rise rate is used as the temperature rise rate characteristic of key components. The theoretical heat dissipation capacity characteristic value of the current cooling system is calculated based on the cooling fan speed and cooling pump current in the standard time series data. The actual heat dissipation effect characteristic value of the current cooling system is calculated based on the time series change of the cooling medium temperature. The difference between the theoretical heat dissipation capacity characteristic value and the actual heat dissipation effect characteristic value is calculated, and the difference is smoothed within the time window to obtain the performance deviation evaluation value. The performance deviation evaluation value is used as the characteristic of the heat transfer efficiency of the cooling medium. The power load rate characteristics, key component temperature rise rate characteristics, and cooling medium heat transfer efficiency characteristics are constructed into a multi-dimensional feature vector. The deviation analysis is performed between this vector and the standard operating feature vector of neighboring units under similar environmental conditions to generate a comprehensive evaluation coefficient of the target unit's cooling thermal state. The overall cooling thermal state evaluation coefficient is compared with the preset overall cooling thermal state evaluation threshold to determine the current health status of the cooling system, specifically including: If the cooling thermal state comprehensive evaluation coefficient is less than the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be abnormal, and the abnormal attribution and operating condition matching analysis process is triggered; if the cooling thermal state comprehensive evaluation coefficient is greater than or equal to the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be normal, and the analysis process of S5 continues.

[0026] It should be specifically noted that during feature extraction and analysis, a 5-minute preset sliding time window is used to calculate the moving average of the initial ratio of the target unit's current active power to its rated power within this window. This is then subtracted from the 0.5 times benchmark load statistically obtained during historical steady-state operation, and the resulting difference is used as the power load rate feature. Subsequently, the temperature difference between adjacent moments of the generator, gearbox, and converter is calculated second by second to obtain the instantaneous temperature rise rate of each component. These are then weighted and summed using preset weights of 0.4, 0.3, and 0.3 to obtain the initial comprehensive temperature rise rate. Simultaneously, the ambient temperature change rate is introduced as a compensation term. The proportionality coefficient of this compensation term is dynamically calibrated based on the linear regression slope of the ambient temperature rise rate and the component's natural temperature rise rate recorded under historical no-cooling-intervention conditions. If the ambient temperature rises, the initial comprehensive temperature rise rate is increased by this calibration coefficient to characterize the negative impact of backflow of ambient heat, thereby obtaining the temperature rise rate characteristics of key components. Next, the cooling fan... The rotational speed and cooling pump current are normalized separately and then multiplied. This product is used as the theoretical heat dissipation capacity feature value. The temperature drop gradient of the cooling medium per unit time is calculated as the actual heat dissipation effect feature value. The difference between the two is calculated and smoothed using a low-pass filter with a cutoff frequency of 0.05Hz. The filtered output is used as the cooling medium heat transfer efficiency feature value. Subsequently, the above three features are constructed into a three-dimensional vector. Similar operating condition samples of neighboring units with wind speed difference less than 0.5m / s and ambient temperature difference less than 2℃ are retrieved from the database. When the number of samples meeting the conditions exceeds the preset upper limit of 20, the 20 nearest samples are selected according to the principle of closest time. The feature vectors corresponding to these samples are averaged and aggregated to generate a standard operating feature vector. The Euclidean distance between the target unit feature vector and the standard operating feature vector is calculated as the deviation measure. This Euclidean distance is input into a Logistic Sigmoid function. The mapping is performed in the form of x, where x is the Euclidean distance and the parameter is... The offset center point is set at 1.5 times the historical average Euclidean distance during normal operation. The parameter k is set to an empirical value of 2.0 to control the steepness of the curve near the threshold, thus nonlinearly mapping the Euclidean distance to a numerical range of 0 to 1. The output value is used as the comprehensive evaluation coefficient for cooling thermal status. Finally, the comprehensive evaluation coefficient for cooling thermal status is compared with a preset safety threshold of 0.8. If the coefficient is less than 0.8, the current health status of the cooling system is determined to be abnormal, and the operating condition matching analysis process is triggered. If the coefficient is greater than or equal to 0.8, it is determined to be normal, and the subsequent trend prediction analysis process continues. S3: Based on the health assessment results and historical operating data, perform operating condition matching analysis to obtain the operating condition deviation and cooling efficiency factor.

[0027] In this embodiment, S3 specifically includes: When the current cooling system's health status is determined to be abnormal, based on the external environmental parameters in the current standard time series data, a set of historical samples with environmental similarity meeting preset conditions is retrieved from the historical operating condition database. The power load rate feature sequence and the key component temperature rise rate feature sequence are extracted from this historical sample set, and the mean and standard deviation of these two feature sequences are calculated respectively. Then, the power difference between the current target unit's power load rate feature and the corresponding historical mean, and the temperature rise difference between the key component temperature rise rate feature and the corresponding historical mean are calculated respectively. The power difference is divided by the corresponding historical power standard deviation to obtain the power relative deviation rate, and the temperature rise difference is divided by the corresponding historical temperature rise standard deviation to obtain the temperature rise relative deviation rate. The power relative deviation rate and the temperature rise relative deviation rate are weighted and summed according to preset weights to obtain the operating condition deviation degree. The actual energy consumption index of the cooling system is obtained by calculating the product of the current cooling fan speed and the cooling pump current. The current initial cooling efficiency index is obtained by calculating the ratio of the current temperature rise rate characteristic of the key components to the actual energy consumption index. Historical initial cooling efficiency indices with the same actual energy consumption index are extracted from the historical sample set and their average values ​​are calculated. The efficiency decay is obtained by calculating the difference between the current initial cooling efficiency index and the historical average value. The relative ratio of the efficiency decay to the historical average value is calculated. Finally, the relative ratio is nonlinearly weighted and fused with the operating condition deviation, and the fusion result is mapped to a preset interval to generate a cooling efficiency factor.

[0028] It should be specifically noted that during the performance matching analysis, when the current cooling system's health status is determined to be abnormal, based on the external environmental parameters in the current standard time series data, a preset similarity condition is used: the absolute value of the wind speed difference is less than 0.8 m / s and the absolute value of the ambient temperature difference is less than 3℃. Historical samples meeting these conditions are retrieved from the historical performance database to form a historical sample set. Then, the power load rate feature sequence and the key component temperature rise rate feature sequence are extracted from this historical sample set. The mean and standard deviation of these two feature sequences are calculated respectively, and the power load rate of the current target unit is then used to determine the similarity. The power relative deviation rate is obtained by dividing the difference between the load rate characteristic and the corresponding historical mean by the corresponding historical power standard deviation. Similarly, the temperature rise difference is obtained by dividing the temperature rise difference by the corresponding historical temperature rise standard deviation. These are weighted and summed with weights of 0.6 and 0.4 respectively to obtain the operating condition deviation. This weighting ratio is calibrated based on the historical operating data where power load, as the driving force of heat source, has a dominant contribution to heat load changes (calibrated to 60%), while temperature rise, as a passive response result, has a relatively minor contribution (calibrated to 40%). Subsequently, the current cooling fan speed and cooling pump power are calculated... The product of the flow rates yields the actual energy consumption index of the cooling system. To avoid negative division results due to negative temperature rise rates (i.e., when components are cooling down), which could interfere with the cooling efficiency calculation, the absolute value of the current temperature rise rate characteristic of the key components is divided by the actual energy consumption index. Simultaneously, an anti-overflow boundary condition is set: when the absolute value of the temperature rise rate is less than 0.001℃ / s, it is forcibly assigned a value of 0.001℃ / s to prevent division by zero errors, thus obtaining the current initial cooling efficiency index. Furthermore, a sample with the smallest absolute difference from the current actual energy consumption index, within an error range of no more than 5%, is selected from the historical sample set. A subset of historical samples is used to calculate the average of the historical initial cooling efficiency index within this subset. The efficiency degradation is obtained by subtracting this historical average from the current initial cooling efficiency index, and the relative ratio of the efficiency degradation to the historical average is calculated. Finally, this relative ratio and the operating condition deviation are used as input variables for nonlinear weighted fusion. Specifically, the relative ratio and the operating condition deviation are assigned weights of 0.6 and 0.4 respectively based on their significance in representing the cooling system efficiency degradation, and then summed to obtain the fused input value. This fused input value is set as variable z, and variable z is substituted into a Sigmoid function. The calculation is performed within the function, where z represents the overall attenuation level after weighted fusion. A larger value for z indicates a more severe deviation from the current operating condition and a greater degree of efficiency degradation. The output value is determined after mapping by this function. The smaller the value, the smoother the fusion result will be mapped to a preset range of 0 to 1, generating a cooling efficiency factor with a lower value, which represents a more severe decline in cooling efficiency.

[0029] S4: Divide the current operating thermal stage according to the deviation of the operating condition and the cooling efficiency factor, and call the control parameter group that matches the current thermal stage from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time.

[0030] In this embodiment, S4 specifically includes: Based on the deviation of the operating condition and the cooling efficiency factor, the ratio of the deviation of the operating condition to the deviation of the preset operating condition benchmark is calculated to obtain the deviation ratio, and the difference between the cooling efficiency factor and the preset safety efficiency threshold is calculated to obtain the efficiency difference. The thermal stage discrimination index is obtained by calculating the ratio of the deviation ratio to the efficiency difference. This index is then compared with multiple preset thermal stage judgment intervals to determine the current operating thermal stage, specifically including: Multiple thermal stage judgment intervals are pre-defined and sorted in order of numerical value. Each thermal stage judgment interval is pre-bound with a unique thermal stage level. The thermal stage discrimination index is compared with the upper and lower boundary values ​​of each thermal stage judgment interval in turn to determine the target thermal stage judgment interval into which the thermal stage discrimination index falls. The thermal stage level bound to the target thermal stage judgment interval is then determined as the current running thermal stage. Based on a defined thermal stage, the baseline temperature rise rate safety threshold and baseline temperature response hysteresis time that match the thermal stage are retrieved from the preset parameter library. By calculating the dynamic ratio of the current temperature rise rate characteristic of key components to the basic temperature rise rate safety threshold, the basic temperature response lag time is dynamically reduced and corrected according to the preset exponential decay function based on this dynamic ratio, and the target temperature response lag time is obtained. The basic temperature rise rate safety threshold is used as the temperature rise rate safety threshold, and the target temperature response lag time is used as the temperature response lag time. Together, they constitute the current control parameter set.

[0031] It should be specifically noted that when performing thermal stage division and parameter calling, the preset operating condition baseline deviation is set to 0.01 to prevent division by zero errors. When the current operating condition deviation is less than 0.01, it is determined that the current operating condition has no significant deviation and the deviation ratio is forcibly assigned to 0. When the current operating condition deviation is greater than or equal to 0.01, the deviation ratio is calculated by dividing the operating condition deviation by 0.01. Simultaneously, considering that a lower cooling efficiency factor value represents a more severe degradation, the difference between the preset safety efficiency threshold of 0.7 and the current cooling efficiency factor is calculated to obtain the efficiency difference, ensuring that this difference monotonically increases with the degree of deterioration. Subsequently, the ratio of the deviation ratio to the efficiency difference is calculated to obtain the thermal stage discrimination index, and presets, for example, [0, 0.5) for the first-level warning thermal stage, [0.5, 1.5) for the second-level dangerous thermal stage, and [1.5, ... The three judgment intervals, ordered by numerical value, represent the three levels of emergency thermal stages. The calculated thermal stage discrimination index is compared sequentially with the upper and lower boundaries of these intervals to determine the target interval and extract the corresponding thermal stage level. Next, the baseline temperature rise rate safety threshold and baseline temperature response lag time matching the level are retrieved from the preset parameter library. Specifically, the values ​​are: Level 1 warning thermal stage corresponds to a baseline temperature rise rate safety threshold of 0.5℃ / s and a baseline temperature response lag time of 10s; Level 2 dangerous thermal stage corresponds to a baseline temperature rise rate safety threshold of 0.8℃ / s and a baseline temperature response lag time of 5s; and Level 3 emergency thermal stage corresponds to a baseline temperature rise rate safety threshold of 1.2℃ / s and a baseline temperature response lag time of 2s. Finally, the dynamic ratio of the current critical component temperature rise rate characteristic to the baseline temperature rise rate safety threshold is calculated, and this dynamic ratio is substituted into the complete exponential decay function. Dynamic reduction and correction are performed in this process, where The target temperature response lag time. The base temperature response lag time is given by r, which is the dynamic ratio. With a preset attenuation coefficient of 0.5, the function logic indicates that the higher the current temperature rise rate relative to the safety threshold, the greater the risk of thermal runaway. At this time, the increase in the dynamic ratio r leads to a decrease in the calculation result of the exponential term, thereby shortening the target temperature response lag time to require the system to make a cooling response faster. Finally, the basic temperature rise rate safety threshold is directly used as the temperature rise rate safety threshold, which together with the target temperature response lag time constitutes the current control parameter group for subsequent calls.

[0032] S5: Based on the temperature response lag time, the prediction time domain is set, the temperature trend of key components is predicted, the predicted temperature rise rate is obtained, and the predicted temperature rise rate is compared with the temperature rise rate safety threshold to determine the cooling safety status level.

[0033] In this embodiment, S5 specifically includes: Obtain the current health status assessment results of the cooling system to determine the target control parameters: if the health status is normal, the preset reference temperature response lag time and reference temperature rise rate safety threshold are used as the target control parameters; if the health status is abnormal, the temperature response lag time and temperature rise rate safety threshold in the current control parameter group are used as the target control parameters. Set the time length of the temperature response lag time in the target control parameters to the prediction time domain; The generator temperature, gearbox oil temperature, and converter temperature at the current moment are obtained from standard time series data, and the temperature rise rate characteristics of key components are obtained. Based on the generator temperature, gearbox oil temperature, and converter temperature at the current moment, temperature prediction analysis is performed by combining the temperature rise rate characteristics of key components with the time length of the prediction time domain to obtain the predicted temperature value at the end of the prediction time domain. The predicted temperature change is obtained by calculating the difference between the predicted temperature value and the current generator temperature, gearbox oil temperature and converter temperature, respectively. The ratio of the predicted temperature change to the time length of the predicted time domain is used as the predicted temperature rise rate. The temperature rise risk coefficient is obtained by calculating the ratio of the predicted temperature rise rate to the safe threshold temperature rise rate in the target control parameters. Multiple consecutive cooling safety status judgment intervals are preset, and each cooling safety status judgment interval is pre-bound with a unique safety status level. The temperature rise risk coefficient is compared with the upper and lower boundary values ​​of each cooling safety status judgment interval in turn to determine the target status judgment interval into which the temperature rise risk coefficient falls, and the safety status level bound to the target status judgment interval is determined as the current cooling safety status level.

[0034] It should be specifically noted that when performing temperature trend prediction and safety status assessment, if the current health status assessment result of the cooling system is normal, the preset 15-second reference temperature response lag time and the reference temperature rise rate safety threshold of 0.3℃ / s will be used as the target control parameters. This 15-second fixed prediction time domain is calibrated based on the sum of the typical mechanical response delay of the actuators of the wind turbine cooling system, including the cooling fan inverter and the cooling pump start / stop relay, and the physical inertia time required to establish effective cooling medium circulation, to ensure that the prediction time domain can cover the complete response process of the cooling system from receiving the command to actually outputting the rated heat dissipation. If the assessment result is abnormal, the aforementioned... The temperature response lag time and temperature rise rate safety threshold in the target control parameter group obtained in the first step are used as target control parameters. Then, the length of the temperature response lag time in the target control parameters is set as the prediction time domain. The generator temperature, gearbox oil temperature, and converter temperature at the current moment are obtained from standard time-series data, and the instantaneous temperature rise rate corresponding to each component is also obtained. When performing temperature prediction analysis, a wind speed change rate of 1 m / s² is set as the applicable boundary for linear extrapolation. When the current wind speed change rate is less than or equal to 1 m / s² and the system is in steady-state condition, the linear extrapolation formula "Predicted final temperature value of each component = Current temperature of each component + Instantaneous temperature rise rate of the corresponding component" is used. The predicted temperature values ​​of each component at the end of the predicted time domain are calculated separately. When the wind speed change rate is greater than 1 m / s², it is determined to be an extreme condition of short-term strong gusts. Linear extrapolation is no longer applicable to constant temperature rise rate. A dynamic gain coefficient positively correlated with the wind speed change rate needs to be introduced to amplify and correct the instantaneous temperature rise rate of the corresponding component before being entered into the above formula to characterize the sudden increase in heat load caused by strong gusts. Then, the difference between the predicted temperature value at the end of each component and the corresponding temperature at the current time is calculated to obtain the predicted temperature change of the generator, gearbox and converter respectively. The ratio of the predicted temperature change of each component to the length of the predicted time domain is used as the predicted temperature rise rate of each component. Subsequently, the ratio of the predicted temperature rise rate of each component to the temperature rise rate safety threshold in the target control parameters is calculated to obtain the independent temperature rise risk coefficient of each component. The maximum value is extracted as the overall temperature rise risk coefficient of the current cooling system. For example, [0, 0.8) is the safety status level, [0.8, 1.0) is the warning status level, and [1.0, The three consecutive cooling safety status judgment intervals are used to determine the danger status level. The overall temperature rise risk coefficient is compared with the upper and lower boundary values ​​of each interval in turn to determine the target status judgment interval into which the coefficient falls, and the safety status level bound to the target status judgment interval is determined as the current cooling safety status level.

[0035] S6: Execute the corresponding cooling control strategy based on the cooling safety status level.

[0036] In this embodiment, S6 specifically includes: The mapping relationship between multiple cooling safety status levels and different levels of cooling control command sets is preset, and the target cooling control command set that matches the current cooling safety status level is obtained. If the current cooling safety status level is a safe status level, then a linear fine-tuning command is generated based on the current cooling fan speed and cooling pump current in the standard timing data to maintain the current heat dissipation status of the cooling system. If the current cooling safety status level is risky or dangerous, then the forced heat dissipation increment is calculated by combining the temperature rise risk coefficient and the temperature rise rate safety threshold in the target control parameter group, and a forced cooling command is generated to increase the cooling fan speed and cooling pump current to the corresponding preset operating range. At the same time, a corresponding active power load reduction command is generated based on the temperature rise risk coefficient. Finally, the generated linear fine-tuning command, forced cooling command, or active power load reduction command is sent to the actuator of the target wind turbine to dynamically adjust the heat dissipation capacity of the cooling system and limit the power generation load of the unit.

[0037] It should be specifically noted that when executing the cooling control strategy, if the current cooling safety status level is at the safe status level, the current cooling fan speed and cooling pump current in the standard timing data are obtained, and the proportional fine-tuning formula "target speed or current = current speed or current + " is applied. The command "(temperature rise rate safety threshold - current temperature rise rate)" generates a linear fine-tuning instruction, where the scaling factor is... The value is set to 0.1, and output limiting rules are set: if the calculated target speed or current exceeds 100% of the rated speed or current, it is forcibly cut off to 100%; if it is lower than 20% of the rated speed or current, it is increased to 20%, in order to prevent the actuator from overloading or stopping while maintaining the current heat dissipation state and eliminating small temperature rise fluctuations; if the current cooling safety status level is a risk or dangerous status level, the forced heat dissipation increment is calculated by combining the temperature rise risk coefficient and the temperature rise rate safety threshold in the target control parameter group. The specific calculation formula is "Forced heat dissipation increment = Rated heat dissipation power". (Risk factor for temperature rise -1) The gain coefficient is set to 0.5. Based on this forced cooling increment, a forced cooling command is generated to increase the cooling fan speed and cooling pump current. When the situation is at a risk level, the cooling fan speed is increased to 80% to 90% of the rated speed and the cooling pump current is increased to 75% to 85% of the rated current. Within this range, a linear interpolation is performed based on the current temperature rise risk coefficient to accurately determine the value. The higher the temperature rise risk coefficient, the closer the value is to the upper limit of the range. When the situation is at a dangerous level, the cooling fan speed is directly increased to 100% of the rated speed and the cooling pump current is increased to 100% of the rated current. At the same time, a corresponding active power reduction command is generated based on the temperature rise risk coefficient, specifically according to the formula: "Active power reduction ratio = (temperature rise risk coefficient - 1)". The calculation is performed at 50%, and the upper limit of the load reduction ratio is truncated to 50% to prevent the unit from disconnecting from the grid. When implementing load reduction, a step-by-step load reduction strategy is adopted, with the single-step load reduction step size set at 5% of the unit's rated power and a load reduction command issued every 10 seconds until the actual cumulative load reduction reaches the target load reduction ratio, in order to avoid large fluctuations in the unit's active power. Finally, the generated linear fine-tuning command, forced cooling command, or active power load reduction command is sent to the fan inverter, water pump controller, and main control power regulation module corresponding to the target wind turbine to dynamically adjust the heat dissipation capacity of the cooling system and limit the unit's power generation load.

[0038] S7: By collecting temperature response data after the control strategy is executed, comparing and analyzing the temperature changes and energy consumption before and after the control strategy is executed, the control effect evaluation coefficient is obtained, and the cooling thermal state comprehensive evaluation coefficient is adaptively corrected based on the control effect evaluation coefficient.

[0039] In this embodiment, S7 specifically includes: Within a preset time window after the cooling control strategy is executed, the generator temperature, gearbox oil temperature and converter temperature of the target wind turbine are re-collected as post-execution temperature response data. Based on this post-execution temperature response data, the actual temperature rise rate is extracted, and the difference between the temperature rise rate safety threshold and the actual temperature rise rate is calculated to obtain the effective temperature suppression amount. Simultaneously, the cooling fan speed and cooling pump current of the cooling system are collected after execution. Based on the cooling fan speed and cooling pump current after execution, the cooling energy consumption increment is calculated relative to the cooling fan speed and cooling pump current before execution. The control effect evaluation coefficient is obtained by calculating the ratio of the effective amount of temperature suppression to the increase in cooling energy consumption. The control effect evaluation coefficient and the cooling thermal state comprehensive evaluation coefficient are weighted and fused to generate the corrected cooling thermal state comprehensive evaluation coefficient, which is then used as the input benchmark for assessing the health status of the target unit's cooling system in the next control cycle.

[0040] Specifically, within a preset 60-second time window after the cooling control strategy is executed, the generator temperature, gearbox oil temperature, and converter temperature of the target wind turbine are re-collected as post-execution temperature response data. The maximum temperature rise rate between adjacent sampling points within this time window is extracted as the actual temperature rise rate. This avoids the problem of using only the temperature difference between the first and last temperatures to calculate the temperature rise rate when the temperature fluctuates first and then rises, thus masking the instantaneous temperature surge peak. The difference between the temperature rise rate safety threshold in the target control parameters and this actual temperature rise rate is calculated to obtain the effective temperature suppression quantity. A non-negative cutoff rule is set: when the actual temperature rise rate is greater than the temperature rise rate safety threshold, resulting in a negative difference, the effective temperature suppression quantity is forcibly assigned to 0 to indicate that the current control strategy has failed to produce an effective cooling effect. Simultaneously, the cooling fan speed and cooling pump current of the cooling system after execution are collected. The instantaneous power after execution is obtained by calculating the product of the cooling fan speed and cooling pump current after execution, and then subtracting the execution... The instantaneous power before execution is obtained by multiplying the cooling fan speed and cooling pump current, and the difference is used as the cooling energy consumption increment. To prevent division by zero errors, the value is forcibly assigned to 0.1 when the cooling energy consumption increment is less than or equal to 0. The control effect evaluation coefficient is obtained by calculating the ratio of the effective temperature suppression to the cooling energy consumption increment. Then, the control effect evaluation coefficient is weighted and fused with the cooling thermal state comprehensive evaluation coefficient of the previous control cycle. The cooling thermal state comprehensive evaluation coefficient of the previous cycle is assigned a weight of 0.7 and the control effect evaluation coefficient is assigned a weight of 0.3 for weighted summation to generate the corrected cooling thermal state comprehensive evaluation coefficient. In the first control cycle, since there is no historical data from the previous control cycle, the initial cooling thermal state comprehensive evaluation coefficient is calibrated to the preset empirical benchmark value of 1.0 as the starting benchmark. Finally, the corrected cooling thermal state comprehensive evaluation coefficient is used as the input benchmark for the health status evaluation of the target unit's cooling system in the next control cycle.

[0041] like Figure 2 The embodiment shown provides an implementation system corresponding to a cooling control method for a wind turbine generator set, including a data acquisition and preprocessing module, a thermal health assessment module, an operating condition deviation analysis module, a thermal stage and parameter adjustment module, a situation prediction and judgment module, a control strategy execution module, and an effect evaluation and correction module. The data acquisition and preprocessing module is connected to the thermal health assessment module, the thermal health assessment module is connected to the operating condition deviation analysis module, the operating condition deviation analysis module is connected to the thermal stage and parameter adjustment module, the thermal health assessment module and the thermal stage and parameter adjustment module are both connected to the situation prediction and judgment module, the situation prediction and judgment module is connected to the control strategy execution module, the control strategy execution module is connected to the effect evaluation and correction module, and the effect evaluation and correction module is connected to the thermal health assessment module.

[0042] The data acquisition and preprocessing module collects real-time operating data and external environmental data of the target wind turbine and adjacent units, and preprocesses the collected operating data and external environmental data. The thermal health assessment module extracts and analyzes features of the target unit based on operating data and external environmental data to obtain a comprehensive cooling thermal assessment coefficient. Based on the comprehensive cooling thermal assessment coefficient, the module assesses the health status of the target unit's cooling system. The operating condition deviation analysis module performs operating condition matching analysis based on the health assessment results and historical operating condition data to obtain the operating condition deviation degree and cooling efficiency factor. The thermal stage and parameter adjustment module divides the current operating thermal stage according to the deviation of the operating condition and the cooling efficiency factor, and calls the control parameter group that matches the current thermal stage from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time. The situation prediction and determination module sets the prediction time domain according to the temperature response lag time, performs trend prediction of the temperature of key components, obtains the temperature rise rate prediction value, compares the temperature rise rate prediction value with the temperature rise rate safety threshold, and determines the cooling safety situation level. The control strategy execution module executes the corresponding cooling control strategy according to the cooling safety status level; The effect evaluation and correction module collects temperature response data after the control strategy is executed, compares and analyzes the temperature changes and energy consumption before and after the control strategy is executed, obtains the control effect evaluation coefficient, and adaptively corrects the cooling thermal state comprehensive evaluation coefficient based on the control effect evaluation coefficient.

[0043] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0044] 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 cooling control method of a wind turbine generator system, characterized by, include: S1: Real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent units, and preprocessing of the acquired operating data and external environmental data; S2: Based on operational data and external environment data, feature extraction and feature analysis are performed on the target unit to obtain the comprehensive evaluation coefficient of cooling thermal state. Based on the comprehensive evaluation coefficient of cooling thermal state, the health status of the cooling system of the target unit is evaluated. S3: Based on the health assessment results and historical operating data, perform operating condition matching analysis to obtain the operating condition deviation and cooling efficiency factor; S4: Divide the current operating thermal stage according to the deviation of the operating condition and the cooling efficiency factor, and call the control parameter group that matches the current thermal stage from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time. S5: Set the prediction time domain based on the temperature response lag time, predict the trend of the temperature of key components, obtain the predicted temperature rise rate, compare the predicted temperature rise rate with the temperature rise rate safety threshold, and determine the cooling safety status level. S6: Execute the corresponding cooling control strategy according to the cooling safety status level; S7: By collecting temperature response data after the control strategy is executed, comparing and analyzing the temperature changes and energy consumption before and after the control strategy is executed, the control effect evaluation coefficient is obtained, and the cooling thermal state comprehensive evaluation coefficient is adaptively corrected based on the control effect evaluation coefficient.

2. A cooling control method of a wind turbine generator system according to claim 1, characterized by, The S1 collects real-time operating data and external environmental data of the target wind turbine and adjacent turbines, specifically including: Using data acquisition terminals deployed on the wind turbine generator side or in the control center, the monitoring and data acquisition systems of the target wind turbine generator and neighboring units are accessed in real time via industrial communication protocols. Operating status parameters and external environmental parameters are collected from the corresponding databases, and these parameters are synchronously sampled according to a preset sampling frequency to obtain raw time-series data. The operating status parameters include active power, generator temperature, gearbox oil temperature, converter temperature, cooling medium temperature, cooling fan speed, and cooling pump current. The external environmental parameters include wind speed at hub height, ambient temperature, and wind direction. The preprocessing of the collected operational data and external environment data specifically includes: The operational data and external environment data acquired as raw time series data are cleaned and quality controlled to detect and remove outliers and missing data in the raw time series data; the raw time series data of the target unit and neighboring units are time aligned using a unified timestamp as a reference; and the time-aligned data are processed using a preset smoothing filtering algorithm to remove high-frequency noise interference, finally obtaining standard time series data.

3. A cooling control method of a wind turbine generator system according to claim 2, characterized by, S2 specifically includes: The initial ratio of the current power to the rated power is calculated based on the active power in the standard time series data. The moving average value of the initial ratio within a preset sliding time window is obtained, and the deviation of the moving average value from the steady-state operating reference load is calculated. The deviation is used as the power load rate feature. Based on the generator temperature, gearbox oil temperature and converter temperature in the standard time series data, the temperature change at adjacent time points is calculated for each sampling point to obtain the instantaneous temperature rise rate of each component. The multiple instantaneous temperature rise rates are summed according to preset weights to obtain the initial comprehensive temperature rise rate. Then, the initial comprehensive temperature rise rate is compensated and corrected by combining the change in ambient temperature at the same time to obtain the compensated comprehensive temperature rise rate characteristic value. The compensated and corrected comprehensive temperature rise rate is used as the temperature rise rate characteristic of key components. The theoretical heat dissipation capacity characteristic value of the current cooling system is calculated based on the cooling fan speed and cooling pump current in the standard time series data. The actual heat dissipation effect characteristic value of the current cooling system is calculated based on the time series change of the cooling medium temperature. The difference between the theoretical heat dissipation capacity characteristic value and the actual heat dissipation effect characteristic value is calculated, and the difference is smoothed within the time window to obtain the performance deviation evaluation value. The performance deviation evaluation value is used as the characteristic of the heat transfer efficiency of the cooling medium. The power load rate characteristics, key component temperature rise rate characteristics, and cooling medium heat transfer efficiency characteristics are constructed into a multi-dimensional feature vector. The deviation analysis is performed between this vector and the standard operating feature vector of neighboring units under similar environmental conditions to generate a comprehensive evaluation coefficient of the target unit's cooling thermal state. The overall cooling thermal state evaluation coefficient is compared with the preset overall cooling thermal state evaluation threshold to determine the current health status of the cooling system, specifically including: If the cooling thermal state comprehensive evaluation coefficient is less than the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be abnormal, and the abnormal attribution and operating condition matching analysis process is triggered; if the cooling thermal state comprehensive evaluation coefficient is greater than or equal to the preset cooling thermal state comprehensive evaluation threshold, the current health status of the cooling system is determined to be normal, and the analysis process of S5 continues.

4. A cooling control method of a wind turbine generator system according to claim 3, characterized by, S3 specifically includes: When the current cooling system's health status is determined to be abnormal, based on the external environmental parameters in the current standard time series data, a set of historical samples with environmental similarity meeting preset conditions is retrieved from the historical operating condition database. The power load rate feature sequence and the key component temperature rise rate feature sequence are extracted from this historical sample set, and the mean and standard deviation of these two feature sequences are calculated respectively. Then, the power difference between the current target unit's power load rate feature and the corresponding historical mean, and the temperature rise difference between the key component temperature rise rate feature and the corresponding historical mean are calculated respectively. The power difference is divided by the corresponding historical power standard deviation to obtain the power relative deviation rate, and the temperature rise difference is divided by the corresponding historical temperature rise standard deviation to obtain the temperature rise relative deviation rate. The power relative deviation rate and the temperature rise relative deviation rate are weighted and summed according to preset weights to obtain the operating condition deviation degree. The actual energy consumption index of the cooling system is obtained by calculating the product of the current cooling fan speed and the cooling pump current. The current initial cooling efficiency index is obtained by calculating the ratio of the current temperature rise rate characteristic of the key components to the actual energy consumption index. Historical initial cooling efficiency indices with the same actual energy consumption index are extracted from the historical sample set and their average values ​​are calculated. The efficiency decay is obtained by calculating the difference between the current initial cooling efficiency index and the historical average value. The relative ratio of the efficiency decay to the historical average value is calculated. Finally, the relative ratio is nonlinearly weighted and fused with the operating condition deviation, and the fusion result is mapped to a preset interval to generate a cooling efficiency factor.

5. A cooling control method of a wind turbine generator system according to claim 4, characterized by, S4 specifically includes: Based on the deviation of operating conditions and the cooling efficiency factor, the ratio of the deviation of operating conditions to the preset operating condition benchmark is calculated to obtain the deviation ratio, and the difference between the cooling efficiency factor and the preset safety efficiency threshold is calculated to obtain the efficiency difference. The thermal stage discrimination index is obtained by calculating the ratio of the deviation ratio to the efficiency difference. This index is then compared with multiple preset thermal stage judgment intervals to determine the current operating thermal stage, specifically including: Multiple thermal stage judgment intervals are pre-defined and sorted in order of numerical value. Each thermal stage judgment interval is pre-bound with a unique thermal stage level. The thermal stage discrimination index is compared with the upper and lower boundary values ​​of each thermal stage judgment interval in turn to determine the target thermal stage judgment interval into which the thermal stage discrimination index falls. The thermal stage level bound to the target thermal stage judgment interval is then determined as the current running thermal stage. Based on a defined thermal stage, the baseline temperature rise rate safety threshold and baseline temperature response hysteresis time that match the thermal stage are retrieved from the preset parameter library. By calculating the dynamic ratio between the current temperature rise rate characteristics of key components and the basic temperature rise rate safety threshold, the basic temperature response lag time is dynamically reduced and corrected according to the dynamic ratio using a preset exponential decay function to obtain the target temperature response lag time. The basic temperature rise rate safety threshold is used as the temperature rise rate safety threshold, and the target temperature response lag time is used as the temperature response lag time. Together, they constitute the current control parameter set.

6. A cooling control method of a wind turbine generator system according to claim 5, characterized by, S5 specifically includes: Obtain the current health status assessment results of the cooling system to determine the target control parameters: if the health status is normal, the preset reference temperature response lag time and reference temperature rise rate safety threshold are used as the target control parameters; if the health status is abnormal, the temperature response lag time and temperature rise rate safety threshold in the current control parameter group are used as the target control parameters. Set the time length of the temperature response lag time in the target control parameters to the prediction time domain; The generator temperature, gearbox oil temperature, and converter temperature at the current moment are obtained from standard time series data, and the temperature rise rate characteristics of key components are obtained. Based on the generator temperature, gearbox oil temperature, and converter temperature at the current moment, temperature prediction analysis is performed by combining the temperature rise rate characteristics of key components with the time length of the prediction time domain to obtain the predicted temperature value at the end of the prediction time domain. The predicted temperature change is obtained by calculating the difference between the predicted temperature value and the current generator temperature, gearbox oil temperature and converter temperature, respectively. The ratio of the predicted temperature change to the time length of the predicted time domain is used as the predicted temperature rise rate. The temperature rise risk coefficient is obtained by calculating the ratio of the predicted temperature rise rate to the safe threshold temperature rise rate in the target control parameters. Multiple consecutive cooling safety status judgment intervals are preset, and each cooling safety status judgment interval is pre-bound with a unique safety status level. The temperature rise risk coefficient is compared with the upper and lower boundary values ​​of each cooling safety status judgment interval in turn to determine the target status judgment interval into which the temperature rise risk coefficient falls, and the safety status level bound to the target status judgment interval is determined as the current cooling safety status level.

7. The cooling control method for a wind turbine generator set according to claim 6, characterized in that, S6 specifically includes: The mapping relationship between multiple cooling safety status levels and different levels of cooling control command sets is preset, and the target cooling control command set that matches the current cooling safety status level is obtained. If the current cooling safety status level is a safe status level, then a linear fine-tuning command is generated based on the current cooling fan speed and cooling pump current in the standard timing data to maintain the current heat dissipation status of the cooling system. If the current cooling safety status level is a risk or danger level, then the forced heat dissipation increment is calculated by combining the temperature rise risk coefficient and the temperature rise rate safety threshold in the target control parameter group, and a forced cooling command is generated to increase the cooling fan speed and cooling pump current to the corresponding preset operating range. At the same time, a corresponding active power load reduction command is generated based on the temperature rise risk coefficient. Finally, the generated linear fine-tuning command, forced cooling command, or active power load reduction command is sent to the actuator of the target wind turbine to dynamically adjust the heat dissipation capacity of the cooling system and limit the generator load.

8. The cooling control method for a wind turbine generator set according to claim 7, characterized in that, Specifically, S7 includes: Within a preset time window after the cooling control strategy is executed, the generator temperature, gearbox oil temperature and converter temperature of the target wind turbine are re-collected as post-execution temperature response data. Based on this post-execution temperature response data, the actual temperature rise rate is extracted, and the difference between the temperature rise rate safety threshold and the actual temperature rise rate is calculated to obtain the effective temperature suppression amount. Simultaneously, the cooling fan speed and cooling pump current of the cooling system are collected after execution. Based on the cooling fan speed and cooling pump current after execution, the cooling energy consumption increment is calculated relative to the cooling fan speed and cooling pump current before execution. The control effect evaluation coefficient is obtained by calculating the ratio of the effective amount of temperature suppression to the increase in cooling energy consumption. The control effect evaluation coefficient and the cooling thermal state comprehensive evaluation coefficient are weighted and fused to generate the corrected cooling thermal state comprehensive evaluation coefficient, which is then used as the input benchmark for assessing the health status of the target unit's cooling system in the next control cycle.

9. A cooling control system of a wind turbine generator system, which implements a cooling control method of a wind turbine generator system according to any one of claims 1 to 8, characterized by include: Data acquisition and preprocessing module: Real-time acquisition of operating data and external environmental data of the target wind turbine and adjacent units, and preprocessing of the acquired operating data and external environmental data; Thermal health assessment module: Based on operating data and external environmental data, feature extraction and feature analysis are performed on the target unit to obtain the comprehensive thermal assessment coefficient of cooling, and the health status of the cooling system of the target unit is assessed based on the comprehensive thermal assessment coefficient of cooling. Operating condition deviation analysis module: Based on the health assessment results and historical operating condition data, it performs operating condition matching analysis to obtain the operating condition deviation degree and cooling efficiency factor; Thermal Stage and Parameter Adjustment Module: The current operating thermal stage is divided according to the deviation of the operating condition and the cooling efficiency factor, and the control parameter group matching the current thermal stage is called from the preset parameter library. The control parameter group includes the temperature rise rate safety threshold and the temperature response lag time. Situation prediction and judgment module: Based on the temperature response lag time, the prediction time domain is set, the temperature trend of key components is predicted, the predicted temperature rise rate is obtained, and the predicted temperature rise rate is compared with the temperature rise rate safety threshold to determine the cooling safety situation level. Control strategy execution module: Executes the corresponding cooling control strategy based on the cooling safety status level; Effect evaluation and correction module: By collecting temperature response data after the control strategy is executed, comparing and analyzing the temperature changes and energy consumption before and after the control strategy is executed, the control effect evaluation coefficient is obtained, and the cooling thermal state comprehensive evaluation coefficient is adaptively corrected based on the control effect evaluation coefficient.