Wind turbine yawing power collection ring intelligent temperature control method and system

Abstract

The application relates to the technical field of data processing, and discloses a yaw power collection ring intelligent temperature control method and system for a wind turbine generator set. The method comprises the following steps: collecting power collection ring temperature data and extracting multi-time scale characteristic parameters, obtaining a comprehensive thermal state evaluation index through weighted fusion according to working condition parameters, predicting a future temperature curve to generate a hierarchical intervention instruction, starting a cooling system and monitoring a temperature gradient, and switching to a pulse intermittent cooling mode and dynamically adjusting cooling parameters when the gradient is out of limit. The application improves the accuracy and reliability of power collection ring temperature control.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for intelligent temperature control of the yaw collector ring of a wind turbine generator set. Background Technology

[0002] The slip ring in the yaw system of a wind turbine generator is a key electrical transmission component, responsible for the continuous transmission of power and control signals during nacelle rotation. Its operating temperature directly affects the reliability of the yaw system and the generator's power generation efficiency. Existing slip ring temperature control methods typically employ temperature sensors to monitor the slip ring temperature in real time. When the temperature exceeds a preset threshold, a cooling system is activated to lower the temperature. Cooling methods include air cooling, water cooling, or oil cooling. Some technical solutions use algorithms such as fuzzy control, PID control, or expert systems to adjust the cooling system's operating state based on the current temperature value, achieving automatic slip ring temperature control.

[0003] However, current temperature monitoring and control strategies primarily rely on a single current temperature value or a simple rate of temperature change for judgment. They fail to adequately consider the combined effects of various factors on the collector ring's temperature rise across different time scales, such as current load, mechanical friction, and contact condition deterioration. This results in incomplete temperature assessments and difficulty in accurately identifying the true thermal state risks of the collector ring. Secondly, existing methods often employ fixed thresholds or simple control algorithms, lacking the ability to adapt to the differences in collector ring temperature rise characteristics under various operating conditions, such as high load, frequent yaw, and long-term operation. Consequently, their control performance is unsatisfactory in complex operating environments. Thirdly, most methods rely solely on feedback control based on the current temperature value, lacking the ability to accurately predict temperature trends. This prevents early intervention and preventative control, easily leading to situations where cooling measures are only initiated after the temperature exceeds limits, resulting in damage to the collector ring. Summary of the Invention

[0004] This application provides a method and system for intelligent temperature control of the yaw collector ring of a wind turbine generator set. It addresses the technical problems of existing yaw collector ring temperature control methods, such as incomplete temperature assessment, lack of adaptive operating condition capability, inability to predict temperature risks in advance, and excessive temperature gradients caused by cooling actions leading to deterioration of the contact state. This improves the accuracy and reliability of the collector ring temperature control. Firstly, this application provides an intelligent temperature control method for the yaw collector ring of a wind turbine generator set, comprising: Step S1: Collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set; Step S2: Weight and fuse the feature parameter set according to the operating condition parameters to obtain a comprehensive thermal state evaluation index; Step S3: Based on the comprehensive thermal state assessment index, predict the future temperature curve and generate graded early intervention instructions; Step S4: Start the cooling system according to the graded early intervention command, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. Calculate the temperature gradient value after each pulse cooling cycle, and dynamically adjust the strong cooling duration and cooling stop duration of the next cycle according to the deviation between the temperature gradient value and the target temperature gradient.

[0005] Secondly, this application provides an intelligent temperature control system for the yaw collector ring of a wind turbine generator set, the intelligent temperature control system for the yaw collector ring of the wind turbine generator set comprising: The extraction module is used to collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set. The weighting module is used to perform weighted fusion of the feature parameter set according to the operating condition parameters to obtain a comprehensive thermal state evaluation index; The generation module is used to predict future temperature curves based on the comprehensive thermal state assessment index and generate graded early intervention instructions. The switching module is used to start the cooling system according to the graded early intervention command, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to the pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. After each pulse cooling cycle, the temperature gradient value is calculated, and the strong cooling duration and cooling stop duration of the next cycle are dynamically adjusted according to the deviation between the temperature gradient value and the target temperature gradient.

[0006] The technical solution provided in this application collects surface and internal temperature data of the collector ring and calculates short-term temperature change rate, medium-term temperature accumulation, and long-term temperature degradation coefficient to construct a three-time-scale feature parameter set. This set can comprehensively characterize the thermal state evolution of the collector ring from three different dimensions: instantaneous thermal shock, continuous heat accumulation, and long-term contact degradation. Compared with the existing technology that relies on a single temperature threshold for judgment, the multi-time-scale feature extraction mechanism reveals the dynamic evolution process and potential degradation trend of the collector ring temperature rise, laying a data foundation for accurate temperature prediction. The fusion weight coefficients of each characteristic parameter are dynamically determined based on operating condition parameters such as current through the collector ring, yaw rate, and cumulative operating time of the unit. The comprehensive thermal state evaluation index is obtained through weighted fusion calculation, which enables the control strategy to adapt to the complex and ever-changing operating conditions of the wind turbine generator. Under high load conditions, the weight of short-term temperature change rate is increased to capture rapid thermal shocks. Under frequent yaw conditions, the weight of medium-term temperature accumulation is increased to assess continuous thermal load. Under long-term operating conditions, the weight of long-term temperature degradation coefficient is increased to identify contact degradation. The operating condition adaptive weighted fusion mechanism accurately distinguishes between two different thermal state types: instantaneous high temperature that can be naturally cooled and temperature that is not high but continues to deteriorate and requires early intervention. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of an embodiment of the intelligent temperature control method for the yaw collector ring of a wind turbine generator set in this application. Figure 2 This is a schematic diagram comparing the comprehensive thermal state evaluation indicators under different operating conditions in the embodiments of this application; Figure 3 This is a schematic diagram illustrating the verification of the prediction performance of the long short-term memory neural network temperature prediction model in the embodiments of this application. Detailed Implementation

[0009] This application provides a method and system for intelligent temperature control of the yaw collector ring of a wind turbine generator. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0010] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent temperature control method for the yaw collector ring of a wind turbine generator set in this application includes: Step S1: Collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set; Specifically, temperature data is collected by placing temperature sensors at different locations on the collector ring, and a historical temperature data buffer is established to store the complete temperature sequence of the past 1800 seconds. Multi-timescale temperature characteristic parameters include three categories: short-term temperature change rate, medium-term temperature accumulation, and long-term temperature degradation coefficient, which respectively reflect the thermal state evolution of the collector ring under different time windows. The short-term temperature change rate captures the instantaneous temperature rise rate within 1 minute, the medium-term temperature accumulation assesses the heat accumulation effect within 10 minutes, and the long-term temperature degradation coefficient reveals the changing trend of the temperature peak decay capability within 30 minutes. These three parameters constitute a complete characteristic parameter set that characterizes the multi-dimensional thermal state information of the collector ring from instantaneous thermal shock to long-term contact degradation.

[0011] Step S2: Weight and fuse the feature parameter set according to the operating condition parameters to obtain the comprehensive thermal state evaluation index; Specifically, based on operating parameters such as slip ring current, yaw rate, and cumulative unit operating time, the fusion weighting coefficients of each parameter in the characteristic parameter set are dynamically calculated. When the slip ring current exceeds 1.2 times the rated value, the weight of the short-term temperature change rate increases to 0.6; when the yaw rate is greater than 3 degrees per second and frequent operation occurs, the weight of the medium-term temperature accumulation increases to 0.5; and when the unit operating time exceeds 15,000 hours, the weight of the long-term temperature degradation coefficient increases to 0.4. By normalizing each characteristic parameter, multiplying it by its corresponding weighting coefficient, and then summing the results, a comprehensive thermal state assessment index is obtained. This index integrates temperature evolution characteristics at different time scales and heat generation factors from different physical mechanisms.

[0012] Step S3: Based on the comprehensive thermal state assessment index, predict the future temperature curve and generate graded early intervention instructions; Specifically, a 6-dimensional feature vector is composed of comprehensive thermal state assessment indicators, current temperature data, and operating parameters, and input into a long short-term memory neural network temperature prediction model. The temperature prediction model includes an input layer, two LSTM hidden layers, and a fully connected output layer. The first hidden layer, with 64 neurons, extracts the temporal dependency features of the temperature sequence. The second hidden layer, with 32 neurons, performs feature compression and abstraction. The output layer generates a sequence of predicted temperature values ​​every 60 seconds from the next 300 to 1800 seconds. The highest temperature peak and its occurrence time are extracted from the predicted value sequence. Based on the temperature range of the peak and the time interval from the current moment, a graded early intervention instruction including cooling intensity level parameters is generated.

[0013] Step S4: Start the cooling system according to the graded early intervention instruction, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. Calculate the temperature gradient value after each pulse cooling cycle, and dynamically adjust the strong cooling duration and cooling stop duration of the next cycle according to the deviation between the temperature gradient value and the target temperature gradient.

[0014] Specifically, the cooling system with the corresponding power is activated according to the cooling intensity level parameter in the graded early intervention instruction. In continuous cooling mode, the difference between the surface temperature drop rate and the internal temperature drop rate is calculated every second. When the difference exceeds 7 degrees Celsius per minute or the absolute temperature difference between the surface and the interior exceeds 15 degrees Celsius, the system switches to pulsed intermittent cooling mode and operates in the sequence of strong cooling, stop, medium cooling, and stop. The total duration of a single pulse cycle is 16 seconds. After switching to pulsed intermittent cooling mode, the calculation every second continues, only used to monitor the temperature gradient change status, and no longer triggers repeated switching of cooling modes. The pulsed intermittent cooling mode runs continuously with a complete cycle of 16 seconds, and is not interrupted or reset due to the calculation results every second within the cycle. After each pulse cycle, the temperature difference between the surface and the interior is calculated as the temperature gradient value. The temperature gradient value is compared with the target temperature gradient of 8 degrees Celsius. When the temperature gradient value is greater than the target value, the cooling stop time of the next cycle is extended and the strong cooling time is shortened. When the temperature gradient value is less than the target value and the instantaneous difference between the surface temperature drop rate and the internal temperature drop rate at the end of the pulse cooling cycle is less than 3 degrees Celsius per minute, the cooling stop time is shortened. Through iterative adjustments over multiple cycles, the temperature gradient is brought to a safe range.

[0015] In one specific embodiment, step S1 includes: The surface temperature at the current moment is collected by a temperature sensor placed on the contact surface of the slip ring brush, and the internal temperature at the current moment is collected by a temperature sensor placed inside the axial center line of the slip ring. A historical temperature data buffer is established to store the temperature sequence data of the past 1800 seconds. The short-term temperature change rate is obtained by calculating the ratio of the temperature change in the 60 seconds prior to the current moment to the time interval based on temperature sequence data. Integrate and sum the temperature values ​​within the previous 600 seconds to obtain the mid-term temperature accumulation. Identify all local temperature peaks in the temperature sequence data, calculate the attenuation ratio sequence between adjacent peaks, perform linear fitting on the attenuation ratio sequence to obtain the slope, and calculate the long-term temperature degradation coefficient based on the slope. The short-term temperature change rate, medium-term temperature accumulation, and long-term temperature degradation coefficient are combined to form a set of characteristic parameters.

[0016] Specifically, a temperature sensor on the contact surface of the slip ring brush collects surface temperature data, while a temperature sensor inside the axial centerline collects internal temperature data. Both types of sensors have a sampling frequency of 1Hz. The historical temperature data buffer stores all temperature sampling values ​​from the current moment forward 1800 seconds in chronological order. The short-term temperature change rate is calculated by subtracting the temperature value from 60 seconds ago from the current temperature value and then dividing by the 60-second time interval, reflecting the rate of temperature rise or fall of the slip ring within a 1-minute time window. The medium-term temperature accumulation is calculated by multiplying each temperature sampling value within the 600-second range from the current moment by a 1-second sampling interval and then summing the results, reflecting the total heat accumulation of the slip ring within a 10-minute time window.

[0017] The calculation of the long-term temperature degradation coefficient requires first identifying all local temperature peaks in the 1800-second temperature sequence data. The criterion for identifying a local peak is that the temperature value at that point is greater than the temperature values ​​of the five sampling points before and after it. After identifying all local peaks, they are arranged in chronological order. The temperature difference between the first and second peaks is calculated and divided by the temperature of the first peak to obtain the first attenuation ratio. The attenuation ratio between all adjacent peaks is calculated sequentially to form an attenuation ratio sequence. The attenuation ratio sequence is linearly fitted to obtain the slope of the attenuation ratio as a function of time. Finally, the slope value is subtracted from 1 to obtain the long-term temperature degradation coefficient. A degradation coefficient close to 1 indicates that the temperature peak attenuation capability is normal, while a degradation coefficient less than 0.85 indicates that the contact condition of the collector ring has deteriorated, resulting in a decrease in heat dissipation capability.

[0018] In one specific embodiment, step S2 includes: The current current through the slip ring, yaw rate, ambient temperature, and cumulative operating time of the unit at the current moment are collected as operating condition parameters. When the current through the collector ring is greater than 1.2 times the rated current, the fusion weighting coefficient of the short-term temperature change rate is set to 0.6. When the yaw rate is greater than 3 degrees per second and the number of yaw actions exceeds 5 in the past 300 seconds, the fusion weighting coefficient of the mid-term temperature accumulation is set to 0.5. When the cumulative operating time of the unit exceeds 15,000 hours, the fusion weighting coefficient of the long-term temperature degradation coefficient will be set to 0.4. The comprehensive thermal state assessment index is obtained by dividing the short-term temperature change rate by the reference value and multiplying it by the corresponding fusion weight coefficient; dividing the medium-term temperature accumulation by the reference value and multiplying it by the corresponding fusion weight coefficient; multiplying the difference between 1 and the long-term temperature degradation coefficient by the corresponding fusion weight coefficient; and summing the results of the three products.

[0019] Specifically, the current in the slip ring is collected in real time by a current sensor; the yaw rate is obtained by calculating the angle change per unit time using the angle encoder of the yaw system; the ambient temperature is measured by an ambient temperature sensor inside the nacelle; and the cumulative operating time of the unit is read from the wind turbine's operation record system. The rules for setting the fusion weighting coefficient are based on the impact mechanism of different operating conditions on the temperature rise of the slip ring. Under high load conditions, the increased current leads to a surge in contact resistance and heat generation, causing a rapid temperature rise. The short-term temperature change rate can sensitively capture this instantaneous thermal shock, so its weighting coefficient is increased from the baseline value of 0.35 to 0.6. Under frequent yaw conditions, mechanical friction continuously generates heat accumulation. The medium-term temperature accumulation can accurately assess the continuous thermal load, so its weighting coefficient is increased from the baseline value of 0.35 to 0.5. Under long-term operating conditions, brush wear and contact pressure attenuation affect heat dissipation capacity. The long-term temperature degradation coefficient can reveal the deterioration trend of the contact condition, so its weighting coefficient is increased from the baseline value of 0.3 to 0.4.

[0020] The reference value for the short-term temperature change rate is set at 5 degrees Celsius per minute, and the reference value for the medium-term temperature accumulation is set at 25,000 degrees Celsius per second. These characteristic parameters of different dimensions are normalized to a value range of 0 to 1 by dividing by the reference values. The long-term temperature degradation coefficient itself ranges between 0 and 1. The difference between 1 and the long-term temperature degradation coefficient is calculated because a smaller degradation coefficient indicates more severe degradation, and a larger difference indicates more severe degradation, consistent with the trends of the other two characteristic parameters. When summing the three products, the normalized short-term temperature change rate is multiplied by a weighting coefficient α, the normalized medium-term temperature accumulation is multiplied by a weighting coefficient β, and the difference between 1 and the long-term temperature degradation coefficient is multiplied by a weighting coefficient γ. The sum of these three products is the comprehensive thermal state assessment index. This index integrates the temperature evolution characteristics at different time scales and the heat-generating factors of different physical mechanisms; a larger value indicates a more severe thermal state of the collector ring.

[0021] Figure 2This is a schematic diagram comparing the comprehensive thermal state assessment index under different operating conditions in the embodiments of this application. The horizontal axis represents five different operating conditions, the vertical axis represents the value of the comprehensive thermal state assessment index, and the height of the bars represents the magnitude of the comprehensive thermal state assessment index corresponding to each operating condition. As can be seen from the figure, the comprehensive thermal state assessment index under normal operating conditions is 0.42, which is at a relatively low level. Under high load and frequent yaw conditions, the indices are 0.68 and 0.61 respectively, indicating an increase in the severity of the thermal state. Under long-term operating conditions, the index is 0.55, reflecting the impact of contact degradation. Under combined operating conditions, the index reaches the highest value of 0.79, indicating that the combined effect of multiple adverse factors leads to the most severe thermal state of the collector ring. This verifies that dynamically adjusting the fusion weight coefficient based on operating condition parameters can accurately assess the thermal state risk of the collector ring under different operating conditions.

[0022] In one specific embodiment, step S3 includes: A temperature prediction model was constructed, consisting of an input layer, two long short-term memory neural network hidden layers, and a fully connected output layer. The first hidden layer contains 64 neuron units, and the second hidden layer contains 32 neuron units. The comprehensive thermal state assessment index, the current surface temperature, the current internal temperature, the current through the slip ring, the yaw rate, and the ambient temperature are combined to form a 6-dimensional feature vector; The 6-dimensional feature vector is input into the temperature prediction model. Through forward propagation calculation of the input layer, two long short-term memory neural network hidden layers and a fully connected output layer, the model outputs a sequence of predicted temperature values ​​every 60 seconds within the time range of 300 to 1800 seconds. Extract the highest temperature peak and its corresponding occurrence time from the temperature prediction value sequence; Based on the temperature range where the highest temperature peak is located and the time interval between the corresponding occurrence time and the current time, a graded advance intervention command containing cooling intensity level parameters is generated.

[0023] Specifically, the input layer of the long short-term memory neural network temperature prediction model receives a 6-dimensional feature vector. The 64 LSTM neurons in the first hidden layer process the input data through three gating mechanisms: the forget gate, the input gate, and the output gate. The forget gate determines which information to discard from the cell state at the previous time step, the input gate determines which information in the current input needs to be stored in the cell state, and the output gate determines which information in the current cell state needs to be output. The output of the 64 neurons in the first hidden layer serves as the input to the second hidden layer. The 32 LSTM neurons in the second hidden layer continue to perform the same gating mechanism for feature abstraction and compression. The output of the 32 neurons in the second hidden layer is passed to the fully connected output layer. The fully connected output layer contains 26 neuron nodes corresponding to the temperature prediction values ​​at 26 time points from 300 seconds, 360 seconds, 420 seconds to 1800 seconds in the future. The model training uses historical running data as training samples. Each sample contains a 6-dimensional feature vector at a certain moment as input and the actual temperature change sequence from 300 to 1800 seconds after that moment as label. During training, the mean square error between the model's predicted value and the actual label is calculated as the loss function. The weight parameters and bias parameters of each layer in the network are adjusted through the Adam optimizer. The learning rate is set to 0.001, and the training is iterated for 500 rounds until the loss on the validation set no longer decreases.

[0024] The temperature prediction sequence contains predicted temperatures at 26 time points. The highest predicted temperature is selected by comparing these values, and the corresponding time point is recorded as the peak occurrence time. The time interval between the peak occurrence time and the current time is calculated. The generation rules for tiered early intervention commands are as follows: when the highest temperature peak is between 70°C and 80°C and the time interval is greater than 600 seconds, the cooling intensity level parameter is set to 1; when the highest temperature peak is between 80°C and 90°C and the time interval is between 300 seconds and 600 seconds, the cooling intensity level parameter is set to 2; when the highest temperature peak exceeds 90°C or the time interval is less than 300 seconds, the cooling intensity level parameter is set to 3. The tiered early intervention command encapsulates the cooling intensity level parameter and transmits it to the cooling system execution module. The model's predicted future temperature curve and peak information enable the control system to initiate cooling measures before the temperature reaches a dangerous value. The higher the predicted peak or the shorter the arrival time, the greater the cooling intensity activated.

[0025] Figure 3This diagram illustrates the verification of the prediction performance of the Long Short-Term Memory (LSTM) neural network temperature prediction model in this embodiment. The horizontal axis represents future time, and the vertical axis represents the collector ring temperature. Solid lines mark the actual measured collector ring temperature change curves, while dashed triangular lines mark the predicted temperature change curves output by the temperature prediction model. As can be seen from the diagram, the predicted temperature curves and the actual temperature curves maintain a high degree of consistency in overall trend and peak position. The prediction model accurately captures the rapid temperature decrease phase between 300 and 900 seconds and the fluctuating recovery phase between 900 and 1800 seconds, verifying that the LSTM temperature prediction model based on a comprehensive thermal state assessment index and a 6-dimensional feature vector has high prediction accuracy.

[0026] In one specific embodiment, step S4, which involves activating the cooling system according to a graded early intervention instruction, includes: Receive the cooling intensity level parameter in the graded early intervention instruction; When the cooling intensity level parameter is 1, the axial flow fan is started at 30% of the rated power; When the cooling intensity level parameter is 2, the axial flow fan is started at 60% of the rated power; When the cooling intensity level parameter is 3, start the axial flow fan at 100% rated power and simultaneously turn on the water-cooled spray device.

[0027] Specifically, after receiving the graded early intervention command, the cooling system execution module controls the inverter output frequency of the axial flow fan according to the value of the cooling intensity level parameter. When the level parameter is 1, the inverter output frequency is set to 30% of the rated frequency to drive the fan at low speed. When the level parameter is 2, the inverter output frequency is set to 60% of the rated frequency to drive the fan at medium speed. When the level parameter is 3, the inverter output frequency is set to 100% to drive the fan at full speed. At the same time, the water supply pipeline of the water-cooled spray device is opened through the solenoid valve. The nozzle of the water-cooled spray device sprays atomized water onto the collector ring shell to cool it down. Different cooling intensity levels correspond to different cooling capacities to match the severity of the predicted temperature peak.

[0028] In one specific embodiment, step S4 involves real-time monitoring of the difference between the surface temperature decrease rate and the internal temperature decrease rate. When the difference exceeds a preset gradient threshold, the system switches to a pulsed intermittent cooling mode, including: Calculate the surface temperature decrease rate and the internal temperature decrease rate every second; Calculate the difference between the rate of temperature decrease of the surface and the rate of temperature decrease of the interior, and calculate the absolute temperature difference between the surface and interior temperatures at the current moment. When the temperature difference is greater than 7 degrees Celsius per minute or the absolute temperature difference is greater than 15 degrees Celsius, switch to pulse intermittent cooling mode. In pulsed intermittent cooling mode, strong cooling is performed for 3 seconds at the power corresponding to the current cooling intensity level, followed by a complete stop for 5 seconds. Then, moderate cooling is performed at 50% of the power corresponding to the current cooling intensity level for 5 seconds, followed by a complete stop for 3 seconds, completing one pulsed cooling cycle. After switching to pulsed intermittent cooling mode, the difference between the surface temperature drop rate and the internal temperature drop rate, as well as the absolute temperature difference, are calculated every second. This is only used to monitor the temperature gradient change and does not trigger repeated switching of the cooling mode. The pulsed intermittent cooling mode runs continuously in 16-second cycles and is not interrupted or reset due to the calculation results every second.

[0029] Specifically, the surface temperature decrease rate is calculated by subtracting the surface temperature from one second ago from the current surface temperature, and the internal temperature decrease rate is calculated by subtracting the internal temperature from one second ago from the current internal temperature. The units for both decrease rates are degrees Celsius per second. When calculating the difference between the surface temperature decrease rate and the internal temperature decrease rate, the two values ​​are directly subtracted. The physical meaning of the difference is the degree to which the surface temperature decreases faster than the internal temperature. When the cooling system is started, the surface temperature decreases rapidly due to direct contact with the cooling medium, while the internal temperature decreases slowly due to the delay in heat conduction. A difference greater than 7 degrees Celsius per minute indicates that the temperature decrease rates of the surface and the interior are too different. The absolute temperature difference is calculated by subtracting the absolute value of the current surface temperature from the current internal temperature. An absolute temperature difference greater than 15 degrees Celsius indicates that the surface has cooled to a lower temperature while the interior is still maintaining a high temperature. If either of the two judgment conditions is met, it is determined that the temperature gradient is too large and it is necessary to switch to the pulsed intermittent cooling mode.

[0030] In the pulsed intermittent cooling mode, the strong cooling action is executed according to the axial fan power or the working status of the water-cooled spray device corresponding to the current cooling intensity level. After 3 seconds, the controller issues a stop command to shut down the axial fan and the water-cooled spray device. During the cooling stop, the heat inside the collector ring has time to conduct and diffuse to the surface, making the temperature distribution more uniform. After 5 seconds, the controller issues a medium cooling command. During medium cooling, the inverter output frequency of the axial fan is reduced to 50% of the frequency corresponding to the current cooling intensity level, and the water-cooled spray device remains closed, using only air cooling. After 5 seconds of medium cooling, the cooling action is completely stopped again for 3 seconds. The strong cooling is 3 seconds plus... A complete pulse cooling cycle consists of 16 seconds, including a 5-second stop, a 5-second medium cooling cycle, and a 3-second stop. During the operation of the pulse intermittent cooling mode, the difference between the surface temperature drop rate and the internal temperature drop rate, as well as the absolute temperature difference between the surface and the internal environment, are calculated every second. However, in pulse mode, this calculation result is only used as real-time monitoring data of the temperature gradient state and is no longer used as the basis for triggering the cooling mode switch. The judgment of cooling mode switch is only valid in continuous cooling mode. After the system enters the pulse intermittent cooling mode, it runs continuously in a 16-second cycle and is not interrupted or reset by the calculation results every second.

[0031] At the end of each pulse cycle, the instantaneous difference between the surface temperature decrease rate and the internal temperature decrease rate at that moment is used as the basis for parameter adjustment. Together with the temperature gradient value, this determines the adjustment direction of the strong cooling duration and the cooling-off duration for the next cycle. By alternating between strong cooling and cooling-off, the extreme temperature gradient state of excessive surface cooling while the internal temperature remains high is avoided. During the cooling-off period, internal heat is conducted to the surface, causing the temperature gradient to gradually decrease.

[0032] In one specific embodiment, step S4 involves calculating the temperature gradient value after each pulse cooling cycle ends, and dynamically adjusting the duration of intense cooling and the duration of cooling stoppage for the next cycle based on the deviation between the temperature gradient value and the target temperature gradient. This includes: After each pulse cooling cycle, the difference between the surface temperature and the internal temperature at the current moment is calculated as the temperature gradient value. Compare the temperature gradient value with the target temperature gradient of 8 degrees Celsius; When the temperature gradient value is greater than the target temperature gradient, the duration of the first cooling stop in the next cycle will be increased by 2 seconds, and the duration of the strong cooling in the next cycle will be decreased by 1 second. When the temperature gradient value is less than the target temperature gradient and the instantaneous difference between the surface temperature drop rate and the internal temperature drop rate at the end of the pulse cooling cycle is less than 3 degrees Celsius per minute, the first cooling stop duration of the next cycle will be reduced by 2 seconds, while keeping the strong cooling duration of the next cycle unchanged.

[0033] Specifically, at the end of the pulse cooling cycle, i.e., at the end of the second 3-second cooling stop, the difference between the surface temperature and the internal temperature is calculated as the temperature gradient value. This temperature gradient value is compared with the target temperature gradient of 8 degrees Celsius. If the temperature gradient value is greater than 8 degrees Celsius, it indicates that the temperature difference between the surface and the internal temperature is still too large, and the cooling stop time needs to be extended to allow more time for internal heat to be conducted to the surface. Therefore, the first cooling stop time in the next cycle is increased from 5 seconds to 7 seconds, while the strong cooling time is reduced from 3 seconds to 2 seconds to reduce the surface cooling rate. When the temperature gradient value is less than 8 degrees Celsius and the instantaneous difference between the surface temperature drop rate and the internal temperature drop rate at the end of the pulse cooling cycle (i.e., at the end of the second 3-second cooling stop) is less than 3 degrees Celsius per minute, it indicates that the temperature gradient has become uniform and the two drop rates are close, which can accelerate the cooling process. Therefore, the first cooling stop time in the next cycle is reduced from 5 seconds to 3 seconds to shorten the heat dissipation uniformity waiting time, while the strong cooling time remains unchanged at 3 seconds to maintain cooling efficiency. Through iterative parameter adjustments over multiple pulse cycles, the temperature gradient gradually converges to the safe target value of 8 degrees Celsius.

[0034] The above describes the intelligent temperature control method for the yaw collector ring of a wind turbine generator set in the embodiments of this application. The following describes the intelligent temperature control system for the yaw collector ring of a wind turbine generator set in the embodiments of this application. One embodiment of the intelligent temperature control system for the yaw collector ring of a wind turbine generator set in the embodiments of this application includes: The extraction module is used to collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set. The weighting module is used to perform weighted fusion of the feature parameter set according to the operating condition parameters to obtain a comprehensive thermal state evaluation index; The generation module is used to predict future temperature curves based on the comprehensive thermal state assessment index and generate graded early intervention instructions. The switching module is used to start the cooling system according to the graded early intervention command, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to the pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. After each pulse cooling cycle, the temperature gradient value is calculated, and the strong cooling duration and cooling stop duration of the next cycle are dynamically adjusted according to the deviation between the temperature gradient value and the target temperature gradient.

[0035] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent temperature control of the yaw collector ring of a wind turbine generator set, characterized in that, The method includes: Step S1: Collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set; Step S2: Weight and fuse the feature parameter set according to the operating condition parameters to obtain a comprehensive thermal state evaluation index; Step S3: Based on the comprehensive thermal state assessment index, predict the future temperature curve and generate graded early intervention instructions; Step S4: Start the cooling system according to the graded early intervention command, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. Calculate the temperature gradient value after each pulse cooling cycle, and dynamically adjust the strong cooling duration and cooling stop duration of the next cycle according to the deviation between the temperature gradient value and the target temperature gradient.

2. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 1, characterized in that, Step S1 includes: The surface temperature at the current moment is collected by a temperature sensor placed on the contact surface of the slip ring brush, and the internal temperature at the current moment is collected by a temperature sensor placed inside the axial center line of the slip ring. A historical temperature data buffer is established to store the temperature sequence data of the past 1800 seconds. Based on the temperature sequence data, the ratio of the temperature change in the previous 60 seconds to the time interval is calculated to obtain the short-term temperature change rate. Integrate and sum the temperature values ​​within the previous 600 seconds to obtain the mid-term temperature accumulation. Identify all local temperature peaks in the temperature sequence data, calculate the attenuation ratio sequence between adjacent peaks, perform linear fitting on the attenuation ratio sequence to obtain the slope, and calculate the long-term temperature degradation coefficient based on the slope. The short-term temperature change rate, the medium-term temperature accumulation, and the long-term temperature degradation coefficient are combined to form the characteristic parameter set.

3. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 1, characterized in that, Step S2 includes: The current current through the slip ring, yaw rate, ambient temperature, and cumulative operating time of the unit at the current moment are collected as operating condition parameters. When the current through the collector ring is greater than 1.2 times the rated current, the fusion weighting coefficient of the short-term temperature change rate is set to 0.6; When the yaw rate is greater than 3 degrees per second and the number of yaw actions exceeds 5 in the past 300 seconds, the fusion weighting coefficient of the mid-term temperature accumulation is set to 0.

5. When the cumulative operating time of the unit exceeds 15,000 hours, the fusion weighting coefficient of the long-term temperature degradation coefficient will be set to 0.

4. The comprehensive thermal state assessment index is obtained by dividing the short-term temperature change rate by the reference value and multiplying it by the corresponding fusion weight coefficient; dividing the medium-term temperature accumulation by the reference value and multiplying it by the corresponding fusion weight coefficient; multiplying the difference between 1 and the long-term temperature degradation coefficient by the corresponding fusion weight coefficient; and summing the three products.

4. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 1, characterized in that, Step S3 includes: A temperature prediction model was constructed, consisting of an input layer, two long short-term memory neural network hidden layers, and a fully connected output layer. The first hidden layer contains 64 neuron units, and the second hidden layer contains 32 neuron units. The comprehensive thermal state assessment index, the current surface temperature, the current internal temperature, the current through the slip ring, the yaw rate, and the ambient temperature are combined to form a 6-dimensional feature vector; The 6-dimensional feature vector is input into the temperature prediction model. Through forward propagation calculation of the input layer, the two long short-term memory neural network hidden layers and the fully connected output layer, a sequence of predicted temperature values ​​every 60 seconds is output within the time range of the next 300 to 1800 seconds. Extract the highest temperature peak and its corresponding occurrence time from the temperature prediction value sequence; Based on the temperature range in which the highest temperature peak is located and the time interval between the corresponding occurrence time and the current time, a graded advance intervention command containing cooling intensity level parameters is generated.

5. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 1, characterized in that, Step S4, which involves activating the cooling system according to the graded early intervention command, includes: Receive the cooling intensity level parameter from the graded early intervention instruction; When the cooling intensity level parameter is 1, the axial flow fan is started at 30% of the rated power; When the cooling intensity level parameter is 2, the axial flow fan is started at 60% of the rated power; When the cooling intensity level parameter is 3, the axial flow fan is started at 100% rated power and the water-cooled spray device is turned on at the same time.

6. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 5, characterized in that, In step S4, the difference between the surface temperature drop rate and the internal temperature drop rate is monitored in real time. When the difference exceeds a preset gradient threshold, the system switches to a pulsed intermittent cooling mode, including: Calculate the surface temperature decrease rate and the internal temperature decrease rate every second; Calculate the difference between the rate of decrease of the surface temperature and the rate of decrease of the internal temperature, and calculate the absolute temperature difference between the surface temperature and the internal temperature at the current moment; When the difference is greater than 7 degrees Celsius per minute or the absolute temperature difference is greater than 15 degrees Celsius, switch to pulse intermittent cooling mode; In the pulsed intermittent cooling mode, a strong cooling action is performed for 3 seconds at the power corresponding to the current cooling intensity level, followed by a complete 5-second cooling stop. Then, a medium cooling action is performed at 50% of the power corresponding to the current cooling intensity level for 5 seconds, followed by another 3-second complete 3-second cooling stop, completing one pulsed cooling cycle. After switching to the pulsed intermittent cooling mode, the operation of calculating the difference between the surface temperature drop rate and the internal temperature drop rate, as well as the absolute temperature difference, is continuously performed every second. This is only used to monitor the temperature gradient change and does not trigger repeated switching of the cooling mode. The pulsed intermittent cooling mode runs continuously in a 16-second complete cycle and is not interrupted or reset during the cycle due to the calculation results every second.

7. The intelligent temperature control method for the yaw collector ring of a wind turbine generator set according to claim 6, characterized in that, In step S4, after each pulse cooling cycle ends, the temperature gradient value is calculated, and the duration of intense cooling and the duration of cooling stoppage in the next cycle are dynamically adjusted based on the deviation between the temperature gradient value and the target temperature gradient. This includes: After each pulse cooling cycle, the difference between the surface temperature and the internal temperature at the current moment is calculated as the temperature gradient value. The temperature gradient value is compared with the target temperature gradient of 8 degrees Celsius. When the temperature gradient value is greater than the target temperature gradient, the duration of the first cooling stop in the next cycle will be increased by 2 seconds, and the duration of the strong cooling in the next cycle will be decreased by 1 second. When the temperature gradient value is less than the target temperature gradient and the instantaneous difference between the surface temperature drop rate and the internal temperature drop rate at the end of the pulse cooling cycle is less than 3 degrees Celsius per minute, the first cooling stop duration of the next cycle will be reduced by 2 seconds, while keeping the strong cooling duration of the next cycle unchanged.

8. A smart temperature control system for the yaw collector ring of a wind turbine generator set, characterized in that, For implementing the intelligent temperature control method for the yaw collector ring of a wind turbine generator set as described in any one of claims 1-7, the intelligent temperature control system for the yaw collector ring of the wind turbine generator set includes: The extraction module is used to collect collector ring temperature data, extract temperature feature parameters at multiple time scales, and construct a feature parameter set. The weighting module is used to perform weighted fusion of the feature parameter set according to the operating condition parameters to obtain a comprehensive thermal state evaluation index; The generation module is used to predict future temperature curves based on the comprehensive thermal state assessment index and generate graded early intervention instructions. The switching module is used to start the cooling system according to the graded early intervention command, monitor the difference between the surface temperature drop rate and the internal temperature drop rate in real time, and switch to the pulse intermittent cooling mode when the difference exceeds the preset gradient threshold. After each pulse cooling cycle, the temperature gradient value is calculated, and the strong cooling duration and cooling stop duration of the next cycle are dynamically adjusted according to the deviation between the temperature gradient value and the target temperature gradient.

9. The system according to claim 8, characterized in that, Collect collector ring temperature data, extract multi-timescale temperature feature parameters, and construct a feature parameter set, including: The surface temperature at the current moment is collected by a temperature sensor placed on the contact surface of the slip ring brush, and the internal temperature at the current moment is collected by a temperature sensor placed inside the axial center line of the slip ring. A historical temperature data buffer is established to store the temperature sequence data of the past 1800 seconds. Based on the temperature sequence data, the ratio of the temperature change in the previous 60 seconds to the time interval is calculated to obtain the short-term temperature change rate. Integrate and sum the temperature values ​​within the previous 600 seconds to obtain the mid-term temperature accumulation. Identify all local temperature peaks in the temperature sequence data, calculate the attenuation ratio sequence between adjacent peaks, perform linear fitting on the attenuation ratio sequence to obtain the slope, and calculate the long-term temperature degradation coefficient based on the slope. The short-term temperature change rate, the medium-term temperature accumulation, and the long-term temperature degradation coefficient are combined to form the characteristic parameter set.

10. The system according to claim 8, characterized in that, The feature parameter set is weighted and fused based on the operating condition parameters to obtain a comprehensive thermal state evaluation index, including: The current current through the slip ring, yaw rate, ambient temperature, and cumulative operating time of the unit at the current moment are collected as operating condition parameters. When the current through the collector ring is greater than 1.2 times the rated current, the fusion weighting coefficient of the short-term temperature change rate is set to 0.6; When the yaw rate is greater than 3 degrees per second and the number of yaw actions exceeds 5 in the past 300 seconds, the fusion weighting coefficient of the mid-term temperature accumulation is set to 0.

5. When the cumulative operating time of the unit exceeds 15,000 hours, the fusion weighting coefficient of the long-term temperature degradation coefficient will be set to 0.

4. The comprehensive thermal state assessment index is obtained by dividing the short-term temperature change rate by the reference value and multiplying it by the corresponding fusion weight coefficient; dividing the medium-term temperature accumulation by the reference value and multiplying it by the corresponding fusion weight coefficient; multiplying the difference between 1 and the long-term temperature degradation coefficient by the corresponding fusion weight coefficient; and summing the three products.

Images