An intelligent switching control method of a multi-working-condition heat dissipation module

Abstract

The application discloses an intelligent switching control method of a multi-working-condition heat dissipation module, and the current working condition category is identified through operation state data of the heat dissipation module, and the weight coefficients of a temperature target item, a pressure drop target item and a noise target item in a multi-target optimization function are dynamically adjusted according to the working condition category, so that an optimal heat dissipation mode is solved; a feedforward-feedback compound control strategy is adopted to execute target opening degree configuration and target rotating speed adjustment according to a cooperative time sequence rule, and during a steady state operation, parameter fine adjustment and model updating are performed according to the deviation between actual performance and predicted performance, and the position of a heat source center is inversed according to the fin surface temperature field distribution, so that the section coverage range of a louver activated window is adjusted to match the forced turbulence area and the heat source hot air path, and the collaborative optimization of heat dissipation performance, pressure drop and noise under multiple working conditions is realized.

Description

Technical Field

[0001] This invention relates to the field of thermal management control technology, specifically to an intelligent switching control method for multi-condition heat dissipation modules. Background Technology

[0002] With the increasing power density of electronic devices, the design of heat dissipation modules faces an inherent contradiction between heat transfer efficiency and flow resistance. To improve heat dissipation capacity within limited space, the industry commonly employs forced turbulence structures such as louvered fins to enhance convective heat transfer by disrupting the laminar boundary layer. However, the turbulence enhancement effect of such structures is positively coupled with pressure drop loss, and their effective area has a significant efficiency-sensitive zone relative to the heat source location. Once the fixed structure is determined, it is impossible to simultaneously meet the energy efficiency requirements under low-load conditions and the heat dissipation limits under high-load conditions. Existing control methods are limited to fan speed regulation, essentially changing the single dimension of airflow under fixed flow channel characteristics. They cannot dynamically adjust the intensity and spatial distribution of turbulence generation based on real-time heat load, nor can they adaptively balance multiple objectives such as heat dissipation performance, pressure drop, and noise, let alone adapt to application scenarios where the heat source location dynamically shifts. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent switching control method for multi-condition heat dissipation modules to solve the problems mentioned in the background art.

[0004] A method for intelligent switching control of a multi-condition heat dissipation module includes: Acquire the operating status data of the heat dissipation module, including heat source power consumption, base temperature, fin surface temperature field distribution, ambient temperature, inlet and outlet pressure difference, and fan speed; The temperature change rate and heat dissipation efficiency are calculated based on the operating status data. The current operating condition category is identified through a hierarchical rule tree. The operating condition categories include steady-state low load, steady-state medium load, steady-state high load, transient rise, transient fall, and heat accumulation warning. The weight coefficients of temperature, pressure drop and noise objectives in the multi-objective optimization function are dynamically adjusted according to the operating condition category. The multi-objective optimization function is solved based on the heat dissipation mode-performance mapping database to obtain the optimal heat dissipation mode. The optimal heat dissipation mode includes the target opening configuration of each louver section and the target fan speed. A feedforward-feedback composite control strategy is adopted, and the adjustment of the target opening configuration and the target speed are executed synchronously according to the preset coordinated timing rules. The coordinated timing rules set the order of action and the rate of change of the fan and the louvers respectively according to the scenario of increasing or decreasing the opening. During steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance. The location of the heat source center is inverted based on the temperature field distribution on the fin surface. The coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control result.

[0005] Furthermore, based on operational status data, the temperature change rate and heat dissipation efficiency are calculated, and the current operating condition category is identified through a hierarchical rule tree, including: The base temperature is subjected to a first-order digital low-pass filter, the inlet and outlet pressure difference is subjected to a cascaded process of median filtering and low-pass filtering, the heat source power consumption is subjected to a moving average filter, and the temperature change rate is calculated based on the filtered base temperature data using the difference method. The heat flux density is calculated based on the ratio of heat source power consumption to the contact area between the heat-conducting substrate and the heat source. The heat dissipation efficiency is calculated based on the ratio of heat source power consumption to the difference between the substrate temperature and the ambient temperature. The rate of change of heat dissipation efficiency is calculated based on historical data of heat dissipation efficiency. The apparent velocity is inversely calculated and the effective Reynolds number is calculated based on the pressure difference between the inlet and outlet using the Darcy-Fochheimer equation. According to the hierarchical rule tree, the heat accumulation warning judgment, transient condition judgment and steady-state condition subdivision judgment are executed in sequence. The heat accumulation warning judgment is triggered when the base temperature is higher than the warning threshold and the temperature change rate is positive. The transient condition judgment is triggered when the temperature change rate exceeds the rising threshold or falling threshold. The steady-state condition subdivision judgment determines the condition category based on the comparison result between the heat source power consumption and the preset threshold when the absolute value of the temperature change rate is less than the steady-state threshold. A working condition confidence counter is introduced to continuously confirm the candidate working condition category. The working condition is switched only when the candidate working condition category is maintained for a preset period. For transient rising working conditions, the future heat load change trend is predicted by linear extrapolation based on historical data of heat source power consumption. The heat dissipation mode level is adjusted in advance based on the prediction results.

[0006] Furthermore, a working condition confidence counter is introduced to continuously confirm candidate working condition categories. Working condition switching is only performed when a candidate working condition category remains continuously for a preset period. For transiently rising working conditions, future heat load change trends are predicted through linear extrapolation based on historical data of heat source power consumption. The heat dissipation mode level is adjusted in advance based on the prediction results, including: The operating condition confidence counter increments until it reaches the upper limit when the current candidate operating condition category matches the confirmed operating condition category, and decrements until it reaches the lower limit when the candidate operating condition category does not match the confirmed operating condition category. The confirmed operating condition category is switched to the candidate operating condition category only when the operating condition confidence counter reaches the upper limit. The preset period is the product of the upper limit and a single control period. Linear extrapolation predicts the heat load value within a preset time window based on the regression slope of historical heat source power consumption data. When the predicted heat load value exceeds the sum of the current heat source power consumption and the preset difference, it is determined that the heat load is showing a continuous upward trend, and the heat dissipation mode level is upgraded by one level in advance. The extent to which the cooling mode level can be increased is limited by the dominant role of the temperature weighting coefficient under transient rise conditions, so that the temperature target item occupies a dominant weight in the mode decision-making, in order to suppress temperature overshoot.

[0007] Furthermore, the weight coefficients of the temperature, pressure drop, and noise objective terms in the multi-objective optimization function are dynamically adjusted according to the operating condition category. The multi-objective optimization function is then solved using a heat dissipation mode-performance mapping database to obtain the optimal heat dissipation mode, including: Based on the Gaussian process regression model, a continuous mapping relationship is constructed from the louver opening configuration and fan speed to the shell temperature, inlet and outlet pressure drop and sound power level. The Gaussian process regression model uses the squared exponential kernel function as the covariance function, optimizes the hyperparameters through maximum likelihood estimation, and obtains the heat dissipation mode-performance mapping database by joint training based on experimental calibration data and computational fluid dynamics simulation data. The weighting coefficients for temperature, pressure drop, and noise targets are dynamically allocated according to the operating condition category. The weighting coefficients switch between quiet priority, balanced allocation, temperature dominance, and single target limit depending on the operating condition category. The model is traversed in the preset discrete heat dissipation model library and the objective function value is calculated. The model with the smallest objective function value is selected as the optimal model for coarse search. Based on the optimal louver opening configuration and fan speed in the coarse search mode, a multi-dimensional fine search is performed in the parameter neighborhood with a preset step size. The objective function value of each neighborhood sample point is quickly evaluated using a Gaussian process regression model, and the sample point with the smallest objective function value is selected as the optimal solution for the fine search. Perform hard boundary checks on candidate modes to ensure safety constraints. Safety constraints include upper limit of temperature safety, upper limit of pressure drop safety, physical range of fan speed, mechanical limit of louvers, and transient overshoot tolerance. Candidate modes that do not meet the constraints are eliminated. Calculate the improvement of the objective function value between the current operating mode and the fine search optimal solution. Only when the improvement exceeds a preset threshold and the steady-state or forced switching conditions are met will the fine search optimal solution be determined as the target heat dissipation mode.

[0008] Furthermore, the system iterates through each mode in the preset discrete heat dissipation mode library and calculates the objective function value. The mode with the smallest objective function value is selected as the optimal mode for coarse search, including: The objective function value is the weighted sum of squares of the temperature objective term, the pressure drop objective term, and the noise objective term. The temperature objective term is the normalized square of the difference between the predicted shell temperature and the target temperature. When the predicted shell temperature is not higher than the target temperature, the temperature objective term is zero. The pressure drop objective term is the normalized square of the ratio of the predicted inlet and outlet pressure drop to the maximum allowable pressure drop. The noise objective term is the normalized square of the ratio of the predicted sound power level to the maximum allowable noise level. The target temperature is determined based on the current operating condition category. Under transient rise conditions and heat accumulation warning conditions, the target temperature is lower than the target temperature under steady-state conditions, so as to reserve a larger temperature safety margin. Iterate through all modes in the preset discrete heat dissipation mode library, use the Gaussian process regression model to predict the shell temperature, inlet and outlet pressure drop and sound power level of each mode under the current operating conditions, and evaluate them one by one by substituting them into the objective function value calculation formula. The discrete pattern with the smallest objective function value is selected as the optimal pattern for coarse search, and the fine search stage of the parameter neighborhood is entered based on the optimal pattern for coarse search.

[0009] Furthermore, the target opening configuration and target speed are adjusted synchronously according to preset coordinated timing rules, including: When the target opening configuration shows an increasing trend relative to the current opening configuration, the fan advance timing is executed to control the fan to increase towards the target speed at a preset acceleration slope. After a first preset time delay, the opening increase action of the louver section is started, so that the fan can build up pressure head before the louver resistance increases. When the target opening configuration is decreasing relative to the current opening configuration, the louver advance timing is executed, and the louver section is controlled to perform the opening reduction action first. After a second preset time delay, the fan is controlled to reduce the speed with a preset deceleration slope, and the high air volume is maintained to accelerate heat dissipation by taking advantage of the window period when the resistance is reduced after the louver is closed. A rate limit is set for the angular velocity of the louver section. The rate limit is temporarily relaxed under transient rising conditions to accelerate the establishment of turbulent response. An acceleration slope limit and a deceleration slope limit are set for the slope of the fan speed change so that the louver opening change and the fan speed change are executed synchronously under the coupled rate constraint. Multiple louver sections are divided into two groups, and the opening adjustment of the two groups is staggered at preset time intervals to suppress mechanical crosstalk between adjacent sections. Based on the deviation between the measured value of the inlet and outlet pressure difference and the expected pressure drop predicted by the heat dissipation mode-performance mapping database, the fan speed compensation is calculated and superimposed on the target speed to compensate for the airflow loss caused by the system impedance shift due to the change in louver opening.

[0010] Furthermore, when the target opening configuration shows an increasing trend relative to the current opening configuration, a fan-first timing sequence is executed, controlling the fan to increase towards the target speed at a preset acceleration slope. After a first preset time delay, the opening increase action of the louver section is initiated, allowing the fan to establish pressure head before the louver resistance increases, including: The fan-first timing is triggered when the target opening configuration is detected to be greater than the current opening configuration. It controls the fan to start increasing the speed with a preset acceleration slope. After the fan speed increases for a first preset time, the opening of the louver section is increased. This makes the increase in flow resistance caused by the increase in louver opening match the pressure head already established by the fan, thus avoiding a temperature spike in the base plate due to a sudden drop in flow. The first preset time is determined based on the ratio of the acceleration time required for the fan to increase from the current speed to the target speed to the movement time required for the louvers to move from the current opening to the target opening, so that the lead time of the fan's pre-movement is sufficient to cover the changes in system impedance during the movement of the louvers; During the fan-first timing execution, if the base temperature change rate is detected to exceed the transient rise threshold, the preset acceleration slope is further increased, so that the fan builds up the pressure head at a faster rate and shortens the first preset time to cope with the emergency thermal load rise scenario.

[0011] Furthermore, during steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance, including: Once the system enters steady-state operation and maintains the current heat dissipation mode for more than a preset time, the steady-state performance evaluation program is started to calculate the steady-state temperature deviation between the actual steady-state temperature and the target temperature. When the steady-state temperature deviation exceeds the positive threshold, the fan speed is increased or the louver opening is increased by a preset step size, and the temperature response in subsequent control cycles is observed to confirm the adjustment effect. When the steady-state temperature deviation is below the negative threshold and the temperature safety margin is sufficient, the fan speed is reduced or the louver opening is reduced by a preset step size to reduce energy consumption and noise. Monitor the rate of change of heat dissipation efficiency; when the rate of change of heat dissipation efficiency shows a continuous downward trend, it is determined that the heat dissipation system has experienced performance degradation. Accumulate residual samples between the actual measured performance under each heat dissipation mode and the predicted values ​​of the Gaussian process regression model, perform outlier detection on the residual samples, and remove outlier samples that exceed the preset multiple of the historical residual standard deviation. When the number of effective residual samples exceeds the update threshold, the sparse approximation algorithm is used to incrementally update the posterior mean and variance of the Gaussian process regression model online, keeping the kernel function hyperparameters unchanged, so that the updated heat dissipation mode-performance mapping database reflects the time-varying performance characteristics of the heat dissipation module caused by dust accumulation or fin aging. Real-time anomaly diagnosis is performed, including fan failure diagnosis, air duct blockage diagnosis, temperature sensor failure diagnosis, louver actuator step loss diagnosis, and louver actuator jamming diagnosis. Based on the diagnosis results, corresponding protection actions are triggered, including switching to passive heat dissipation limit protection mode, adjusting the louver opening to find a low impedance path, enabling redundant sensors to replace failed sensors, performing zeroing and repositioning operations on the step loss section, and marking the jamming section as unusable and coordinating with other sections to compensate for heat dissipation loss.

[0012] Furthermore, based on the temperature field distribution on the fin surface, the location of the heat source center is inverted, and the coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control results, including: Based on the readings of multiple fin surface temperature sensors arranged along the airflow direction, the temperature gradient sequence between adjacent measuring points is calculated, and the inflection point segment where the temperature gradient sequence changes from positive to negative is located. The location of the heat source center is calculated using the weighted centroid method, with the temperature rise of each measuring point relative to the ambient temperature as the weight. Based on the center position of the heat source and the preset optimal offset, calculate the target center position of the louver activation window. The preset optimal offset points upstream of the heat source along the airflow direction. Map the target center location to the corresponding segment number, and set an activation window that covers a preset number of segments with that segment as the center; The target opening configuration of the louvers in the active window coverage area is maintained, while the opening of the louvers in the outer area of ​​the active window is attenuated to reduce the ineffective pressure drop loss in non-heat source areas. The heat source center location is inverted and the activation window is relocated according to the preset update cycle. Each relocation limits the window center to move no more than one segment interval. When multiple heat sources are detected to have similar temperature rise, the activation window is expanded to cover the corresponding areas of multiple heat sources at the same time. The active window's segment coverage area is integrated with the target heat dissipation mode, collaborative execution results, and feedback optimization results to generate and output the final switching control result.

[0013] Furthermore, the heat source center location is inverted and the activation window is relocated according to a preset update cycle. Each relocation limits the window center movement to no more than one segment interval. When multiple heat sources are detected to have similar temperature rises, the activation window is expanded to simultaneously cover the corresponding areas of multiple heat sources, including: The preset update cycle is an integer multiple of multiple control cycles. In each preset update cycle, the heat source center position is recalculated and compared with the heat source center position confirmed in the previous preset update cycle to calculate the window center offset. When the window center offset exceeds one segment spacing, the window center movement for this repositioning is limited to one segment spacing, and the remaining offset is carried over to the next preset update cycle to continue execution, so that the louver activated window gradually tracks the heat source migration path along the airflow direction. Before each relocation is executed, the surface temperature rise of the fins of each heat source is compared. When the temperature rise difference of multiple heat sources is less than the preset temperature difference threshold and is significantly higher than the ambient temperature, it is determined to be a multi-heat source balanced heating scenario. The activation window is then expanded to a global window that simultaneously covers the sections of multiple heat sources. In a scenario of balanced heating from multiple heat sources, the target opening configuration is maintained for the upstream sections of each heat source covered by the global window, while the louver opening of the external sections of the global window is attenuated, so that the heat plume paths of each heat source are subjected to forced turbulence enhancement.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention provides an intelligent switching control method for a multi-condition heat dissipation module. It identifies the current condition category by using the operating status data of the heat dissipation module, and dynamically adjusts the weight coefficients of the temperature target term, pressure drop target term, and noise target term in the multi-objective optimization function accordingly to solve for the optimal heat dissipation mode. It adopts a feedforward-feedback composite control strategy to execute the adjustment of the target opening configuration and target speed according to the cooperative timing rules. During steady-state operation, it performs parameter fine-tuning and model updates based on the deviation between actual performance and predicted performance. At the same time, it inverts the heat source center position based on the temperature field distribution on the fin surface and adjusts the segment coverage of the louver activation window to match the forced turbulence region with the heat source heat plume path, thereby realizing the cooperative optimization of heat dissipation performance, pressure drop, and noise under multiple conditions. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of steps S101 to S105 of the control method described in the specific implementation embodiment; Figure 2 This is a schematic diagram illustrating steps S201 to S204 of the control method described in the specific implementation embodiment; Figure 3 The diagram illustrates steps S301 to S306 of the control method described in the specific implementation. Detailed Implementation

[0016] Please see Figures 1 to 3 This invention provides a technical solution. The technical solution of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0017] Please see Figure 1 This embodiment provides an intelligent switching control method for a multi-condition heat dissipation module, including: S101. Obtain the operating status data of the heat dissipation module. The operating status data includes heat source power consumption, base temperature, fin surface temperature field distribution, ambient temperature, inlet and outlet pressure difference, and fan speed. S102. Calculate the temperature change rate and heat dissipation efficiency based on the operating status data, and identify the current operating condition category through a hierarchical rule tree. The operating condition categories include steady-state low load, steady-state medium load, steady-state high load, transient rise, transient fall, and heat accumulation warning. S103. Dynamically adjust the weight coefficients of the temperature target, pressure drop target and noise target in the multi-objective optimization function according to the working condition category, solve the multi-objective optimization function according to the heat dissipation mode-performance mapping database, and obtain the optimal heat dissipation mode. The optimal heat dissipation mode includes the target opening configuration of each louver section and the target speed of the fan. S104. A feedforward-feedback composite control strategy is adopted to synchronously execute the adjustment of the target opening configuration and the target speed according to the preset coordinated timing rules. The coordinated timing rules set the order of action and the rate of change of the fan and the louvers according to the scenario of increasing or decreasing the opening. S105. During steady-state operation, the parameters are fine-tuned and the model is updated online based on the deviation between the actual performance and the predicted performance. The location of the heat source center is inverted based on the temperature field distribution on the fin surface. The coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control result.

[0018] In step S101, the heat source power consumption characterizes the heat generation power level, which is obtained in real time by monitoring the current and voltage in the power supply circuit. Its value directly determines the lower limit of heat dissipation requirements. The substrate temperature characterizes the degree of heat accumulation at the interface between the heat source and the fin array, and is measured by a temperature sensing element attached to the surface of the heat-conducting substrate. It is a core indicator reflecting the heat dissipation margin. The fin surface temperature field distribution characterizes the spatial characteristics of heat transport along the airflow direction. It is collected by temperature measuring points distributed at different locations on the fin array, providing a spatial information basis for subsequent heat source location inference. The ambient temperature serves as the inlet temperature reference for the cooling medium and is used to assess the available margin of the heat dissipation temperature difference. The inlet and outlet pressure difference characterizes the flow resistance encountered by the airflow through the fin array, and its value change reflects the matching state between the flow channel structure and the airflow. The fan speed can be obtained by detecting the motor commutation frequency, characterizing the current ability to drive the airflow, and together with the pressure difference, determines the actual airflow level passing through the fin array. After the above data items are collected synchronously in the time dimension and formatted uniformly, a complete data set describing the operating status of the heat dissipation module is formed, providing comprehensive input information for subsequent operating condition determination.

[0019] In step S102, the temperature change rate is calculated from the change in base temperature over time. A positive value indicates accelerated heat accumulation, while a negative value indicates that heat dissipation is dominant. The magnitude of the value reflects the severity of the thermal state deviating from equilibrium. The heat dissipation efficiency is determined by the ratio of heat source power consumption to the difference between base temperature and ambient temperature. It represents the heat power that the module can dissipate per unit temperature difference and is a direct indicator of the current heat exchange capacity. The hierarchical rule tree is executed layer by layer from high to low priority. First, it checks whether the heat accumulation warning condition is met. Second, it determines whether it is in a transient rise or transient fall state. Finally, under steady-state conditions, it further subdivides the steady-state operating condition category based on the heat source power consumption level. Each operating condition corresponds to different heat dissipation requirements. Under steady-state low-load conditions, the heat generation is small and the temperature is stable, with the main heat dissipation requirement being to maintain basic cooling and reduce noise. Under steady-state medium-load conditions, the heat dissipation requirement is moderate, and both heat exchange performance and energy consumption level need to be considered. Under steady-state high-load conditions, the heat generation is continuous and high, with ensuring heat dissipation reliability as the primary goal. Under transient rising conditions, the heat load is rapidly increasing, requiring rapid enhancement of heat dissipation to suppress temperature overshoot. Under transient falling conditions, the heat load is rapidly decreasing, and it is necessary to prevent excessive heat dissipation and energy waste. Under heat accumulation warning conditions, the temperature is continuously rising and existing heat dissipation methods can no longer contain it, which is an abnormal state and requires triggering limit protection.

[0020] In step S103, the temperature objective term in the multi-objective optimization function characterizes the degree of deviation between the predicted shell temperature and the target temperature, reflecting the achievement of heat dissipation capacity; the pressure drop objective term characterizes the proportion of the predicted inlet and outlet pressure drop relative to the allowable upper limit, reflecting the level of flow resistance cost; and the noise objective term characterizes the proportion of the predicted sound power level relative to the allowable upper limit, reflecting the quality of acoustic comfort. The weighting coefficients are dynamically adjusted according to the operating condition category, causing the priority of the three objectives to change accordingly under different operating conditions. In transient high-temperature scenarios, the weight of the temperature objective term tends to dominate to ensure heat dissipation priority; in low-load scenarios, the weight of the noise objective term is relatively increased to meet the quietness requirement; and in normal load scenarios, the weights of the three tend to be balanced. The heat dissipation mode-performance mapping database is a pre-calibrated set of expected performance records for each candidate heat dissipation mode under different operating conditions. It is obtained through experimental measurements of each mode under different combinations of thermal load and ambient temperature, recording performance parameters such as the expected shell temperature, inlet and outlet pressure drop, and noise level for each mode. When solving the multi-objective optimization function, the performance prediction values ​​of each candidate mode are retrieved from the database based on the current operating conditions. The comprehensive evaluation value of each mode is calculated, and the mode with the best comprehensive performance is selected as the target heat dissipation mode. The target opening configuration of each louver section in the optimal heat dissipation mode is used to adjust the intensity and spatial distribution of turbulence at each location, and the target fan speed is used to adjust the overall airflow level. Together, they constitute the complete control instruction set of the heat dissipation module.

[0021] In step S104, the feedforward-feedback composite control strategy uses a feedforward channel to directly preset the initial target value of the control quantity based on the trend of operating condition changes, enabling the actuator to start operating before the deviation is fully manifested. The feedback channel performs closed-loop correction of the control quantity based on real-time temperature and pressure difference deviations, eliminating deviations caused by feedforward preset errors and external disturbances. The combination of the two approaches balances transient response speed and steady-state control accuracy. The coordinated timing rules are set to address the strong coupling characteristics between changes in louver opening and fan speed. Since changes in louver opening directly alter the flow channel resistance characteristics, they affect the fan operating point and the actual airflow through the fin array. When the opening increases, the fan speed must be increased first to establish pressure head, avoiding a sudden increase in resistance that causes a drop in flow and a temperature spike. When the opening decreases, the louvers must be closed first to utilize the high airflow window after the resistance decreases to accelerate heat dissipation, and then the fan speed is gradually reduced to return to the target operating state. The rate of change limits set upper limits for the slope of change in louver angular velocity and fan speed, ensuring smooth and controllable actuator operation and preventing temperature oscillations and mechanical shocks caused by sudden changes in airflow conditions. This step, by coordinating the timing and rate constraints of the two actuator channels, ensures flow stability and temperature stability during mode switching.

[0022] In step S105, during steady-state operation, parameter fine-tuning is performed based on the deviation between actual and predicted performance. When there is a systematic deviation between the actual and target temperatures, the fan speed or louver opening is slightly corrected, enabling the control parameters to be automatically corrected when the model mismatches or the environment changes. The online model update continuously corrects the performance prediction model by accumulating actual operating data, allowing the heat dissipation mode-performance mapping database to adapt to the time-varying performance characteristics caused by dust accumulation, fin oxidation, or fan wear during long-term operation of the heat dissipation module, thus avoiding the degradation of prediction accuracy over time. The location of the heat source center is inverted based on the temperature field distribution on the fin surface, and the spatial distribution characteristics of the temperature field along the airflow direction are used to infer the area where the heat source is located. Based on this, the coverage range of the louver activation window is adjusted to match the area where forced turbulence is generated with the concentrated path of the heat source plume, concentrating the benefits of turbulent heat transfer in the area directly opposite the heat source, while reducing the ineffective pressure drop loss caused by the opening of the louvers in non-heat source areas. The final switching control result integrates all the output information from mode decision-making, collaborative execution, and feedback optimization, and outputs it as a complete control command for the heat dissipation module to each actuator.

[0023] This embodiment utilizes the operating status data of the heat dissipation module to calculate the temperature change rate and heat dissipation efficiency. A hierarchical rule tree identifies the current operating condition category, enabling the control strategy to adaptively adjust based on the semantics of the operating condition. Weight coefficients for temperature, pressure drop, and noise are dynamically allocated in the multi-objective optimization function according to the operating condition category, and the optimal heat dissipation mode is solved based on the heat dissipation mode-performance mapping database. This ensures that the heat dissipation module obtains control parameters with optimal overall performance under different operating conditions. A feedforward-feedback composite control strategy is adopted, and the adjustment of louver opening and fan speed is executed synchronously according to cooperative timing rules, ensuring smooth and oscillating mode switching. During steady-state operation, control accuracy is maintained through parameter fine-tuning and online model updates. The coverage range of the louver activation window is adjusted by inverting the heat source center position, dynamically matching the turbulence enhancement region with the heat source plume path. This embodiment introduces new control degrees of freedom through the variable turbulence structure of the partitioned louvers, achieving multi-condition adaptive optimization of the heat dissipation module in terms of heat dissipation capacity, flow resistance, and noise level.

[0024] Please see Figure 2 In some embodiments, the temperature change rate and heat dissipation efficiency are calculated based on operating status data, and the current operating condition category is identified through a hierarchical rule tree, including: S201. Perform first-order digital low-pass filtering on the base temperature, perform cascaded processing of median filtering and low-pass filtering on the inlet and outlet pressure difference, perform moving average filtering on the heat source power consumption, and calculate the temperature change rate based on the filtered base temperature data using the difference method. S202. Calculate the heat flux density based on the ratio of heat source power consumption to the contact area between the heat-conducting substrate and the heat source; calculate the heat dissipation efficiency based on the ratio of heat source power consumption to the difference between substrate temperature and ambient temperature; calculate the rate of change of heat dissipation efficiency based on historical data of heat dissipation efficiency; and calculate the apparent velocity and effective Reynolds number by back-calculating the Darcy-Fochheimer equation based on the pressure difference between the inlet and outlet. S203. Based on the hierarchical rule tree, perform heat accumulation early warning judgment, transient condition judgment and steady-state condition subdivision judgment in sequence to ensure that high-risk anomalies and transient states are responded to first. Heat accumulation early warning judgment is triggered when the base temperature is higher than the early warning threshold and the temperature change rate is positive. Transient condition judgment is triggered when the temperature change rate exceeds the rising threshold or falling threshold. Steady-state condition subdivision judgment determines the condition category based on the comparison result of heat source power consumption and preset threshold when the absolute value of temperature change rate is less than the steady-state threshold. S204. Introduce a working condition confidence counter to continuously confirm the candidate working condition category. Only when the candidate working condition category is maintained for a preset period will the working condition be switched. For transient rising working conditions, predict the future heat load change trend by linear extrapolation based on historical data of heat source power consumption, and adjust the heat dissipation mode level in advance according to the prediction results.

[0025] In step S201, the first-order digital low-pass filter suppresses high-frequency noise in the base temperature signal by fusing the current sampled value with the previous filtered output value using weighted coefficients. These weighted coefficients can be determined based on the spectral characteristics of the sensor output signal, for example, by performing spectral analysis on the temperature signal to identify the boundary between the effective bandwidth and the noise band. The median filter sorts the inlet and outlet pressure difference signals within a sliding window and takes the median value as the output to eliminate potential pulse-type abnormal interference during sampling. The cascaded low-pass filtering further suppresses residual high-frequency fluctuations. The window width and cutoff frequency of the median filter are determined based on the fluctuation characteristics of the pressure difference signal. The moving average filter eliminates interference from transient power fluctuations in the operating condition determination by taking the arithmetic mean of the heat source power consumption signal over a preset time period. The length of the sliding window is determined based on the fluctuation period of the power consumption signal. When calculating the temperature change rate using the differential method, the difference between the current filtered value and the filtered value several sampling periods ago is divided by the corresponding time interval. The selected time interval is determined based on the control period and the bandwidth of the temperature signal.

[0026] In step S202, the heat source contact area can be directly read from the design parameters of the heat dissipation module. The rate of change of heat dissipation efficiency is calculated based on the change in heat dissipation efficiency over a continuous time period, reflecting the dynamic evolution trend of heat dissipation capacity. The permeability and inertia coefficient in the Darcy-Fochheimer equation are obtained through pressure drop calibration experiments on the heat dissipation module under different airflow rates. The measured pressure drop and flow velocity data are substituted into the equation for fitting and then determined. The measured inlet and outlet pressure difference is substituted into the calibrated Darcy-Fochheimer equation to solve for the apparent velocity of the airflow through the fin array. The ratio of the apparent velocity to the hydraulic diameter and the air kinematic viscosity is the effective Reynolds number. This dimensionless parameter is used to determine whether the current flow state is in the laminar or turbulent range.

[0027] In step S203, the warning threshold is determined based on the thermal design power and safe temperature limit of the heat dissipation module, and is set to the upper limit of the base temperature safety limit. The rising threshold and falling threshold are determined based on the historical distribution statistics of the temperature change rate, and are selected as critical values ​​that can distinguish between normal fluctuations and significant changing trends. The steady-state threshold is set as a small range of the temperature change rate, used to determine whether the system is in a thermal equilibrium state. Preferably, the preset thresholds include a first power consumption threshold and a second power consumption threshold. The first power consumption threshold corresponds to the boundary point of low load, and the second power consumption threshold corresponds to the boundary point of high load. The preset thresholds are determined proportionally based on the thermal design power of the heat dissipation module.

[0028] In step S204, the operating condition confidence counter is initially set to zero. The counter increments each time a candidate operating condition category output by the rule tree matches the currently confirmed operating condition, and decrements when they don't. The lower limit of the count is zero. When the count reaches the upper limit corresponding to a preset period, the currently confirmed operating condition is switched to a candidate operating condition category, and the counter is then reset and restarted. Linear extrapolation calculates the slope of change based on historical data of heat source power consumption. This slope is extrapolated to a short future time window to obtain the predicted power consumption value. When the predicted power consumption value exceeds the current power consumption value and the difference reaches a preset range, it is determined that the thermal load is showing a continuous upward trend, triggering an upgrade command for the heat dissipation mode level.

[0029] This embodiment performs targeted filtering on the raw sensor data to eliminate the impact of noise and interference on feature quantity calculation. It then extracts multi-dimensional features such as heat flux density, heat dissipation efficiency change rate, and effective Reynolds number from the pre-processed data, providing a comprehensive quantitative basis for operating condition determination. Based on a hierarchical rule tree, it sequentially performs heat accumulation early warning determination, transient operating condition determination, and steady-state operating condition subdivision determination according to risk priority, ensuring that different types of operating states can be accurately classified. An operating condition confidence counter is introduced to confirm the determination results in the time dimension, avoiding erroneous operating condition switching caused by noise. Simultaneously, for transient rising operating conditions, it uses the historical trend of heat source power consumption for linear extrapolation, enabling the heat dissipation module to have proactive adjustment capabilities, pre-enhancing heat dissipation capacity before temperature overshoot occurs.

[0030] In some embodiments, a working condition confidence counter is introduced to continuously confirm candidate working condition categories. Working condition switching is performed only when a candidate working condition category remains continuously for a preset period. Furthermore, for transiently rising working conditions, future heat load change trends are predicted through linear extrapolation based on historical data of heat source power consumption. The heat dissipation mode level is adjusted in advance based on the prediction results, including: The operating condition confidence counter increments until it reaches the upper limit when the current candidate operating condition category matches the confirmed operating condition category, and decrements until it reaches the lower limit when the candidate operating condition category does not match the confirmed operating condition category. The confirmed operating condition category is switched to the candidate operating condition category only when the operating condition confidence counter reaches the upper limit. The preset period is the product of the upper limit and a single control period. Linear extrapolation predicts the heat load value within a preset time window based on the regression slope of historical heat source power consumption data. When the predicted heat load value exceeds the sum of the current heat source power consumption and the preset difference, it is determined that the heat load is showing a continuous upward trend, and the heat dissipation mode level is upgraded by one level in advance. The extent to which the cooling mode level can be increased is limited by the dominant role of the temperature weighting coefficient under transient rise conditions, so that the temperature target item occupies a dominant weight in the mode decision-making, in order to suppress temperature overshoot.

[0031] In this embodiment, the upper limit is determined by a trade-off between the response speed requirement and the disturbance rejection capability requirement for the operating condition switch. The higher the response speed requirement, the smaller the upper limit; the higher the disturbance rejection capability requirement, the larger the upper limit. The specific value can be set according to the sensitivity of the electronic equipment served by the heat dissipation module to temperature fluctuations. The lower limit is zero or a small positive integer to prevent the counter from excessively decaying due to a single inconsistent judgment, thus losing the cumulative effect of continuous confirmation. A single control cycle is the time base for the system to execute a complete control process once. The preset cycle is the product of the upper limit and the single control cycle, representing the total continuous confirmation time required from the initial determination of the candidate operating condition category to the formal execution of the operating condition switch. After the counter reaches the upper limit and the switch is executed, it is reset to zero and re-accumulated to start counting again for the confirmation process of the next operating condition switch. If an inconsistent judgment occurs during the counter increment process, the count is decremented but will not completely return to zero due to a single inconsistency. It continues to increment when an consistent judgment occurs again, thus ensuring that accidental interference will not completely erase the accumulated confirmation time.

[0032] The regression slope is obtained by performing linear regression calculations on historical heat source power consumption data. The amount of historical data involved in the regression calculation is determined based on the fluctuation of the power consumption signal; the more severe the fluctuation, the larger the amount of historical data required to smooth the impact of short-term disturbances. The preset time window is the length of the predicted future time period, determined based on the thermal inertia time constant of the heat dissipation module. The greater the thermal inertia, the slower the temperature response, and the preset time window is extended accordingly to match the response rhythm of the heat dissipation system. The preset difference is the power consumption increase threshold required to trigger the upward adjustment judgment. It is determined based on the normal fluctuation range of heat source power consumption. The value should be greater than the normal fluctuation range to avoid frequent upward adjustments, and less than the power consumption increase when the heat load increases significantly to ensure timely response. Linear extrapolation is only activated under transient upward conditions. When the predicted heat load value exceeds the sum of the current heat source power consumption and the preset difference, it is determined that the heat load is showing a continuous upward trend, and the heat dissipation mode level is upgraded by one level in advance, so that the heat dissipation module completes the pre-enhancement of heat dissipation capacity before the temperature rises sufficiently.

[0033] Under transient surge conditions, the temperature weighting coefficient is significantly higher than that under steady-state conditions, giving the temperature objective a dominant weight in mode decision-making. With the temperature objective having a dominant weight, mode decision-making prioritizes temperature control effectiveness, while the weights of voltage drop and noise objectives are relatively reduced. Even if a candidate mode has a high voltage drop or noise level, it will be selected as the target cooling mode as long as its temperature control effect is optimal, thereby suppressing temperature overshoot.

[0034] This embodiment continuously confirms candidate operating condition categories using an operating condition confidence counter, filters out misclassifications caused by noise using a preset period of time hysteresis, and ensures the reliability of operating condition switching decisions. It predicts future heat load changes based on historical heat source power consumption data through linear extrapolation, enabling the heat dissipation module to pre-enhance its heat dissipation capacity before temperature overshoot occurs. By having the temperature weighting coefficient dominate under transient rising operating conditions, it constrains the upward adjustment of the heat dissipation mode level, ensuring that mode decisions prioritize temperature control, thereby achieving a balance between rapid response and temperature overshoot suppression under transient operating conditions.

[0035] Please see Figure 3 In some embodiments, the weight coefficients of the temperature, pressure drop, and noise objective terms in the multi-objective optimization function are dynamically adjusted according to the operating condition category. The multi-objective optimization function is solved based on the heat dissipation mode-performance mapping database to obtain the optimal heat dissipation mode, including: S301. Based on the Gaussian process regression model, a continuous mapping relationship is constructed from the louver opening configuration and fan speed to the shell temperature, inlet and outlet pressure drop and sound power level. The Gaussian process regression model uses the squared exponential kernel function as the covariance function, optimizes the hyperparameters through maximum likelihood estimation, and obtains the heat dissipation mode-performance mapping database by joint training based on experimental calibration data and computational fluid dynamics simulation data. S302. Dynamically allocate the weight coefficients of temperature target items, pressure drop target items and noise target items according to the operating condition category. The weight coefficients switch between quiet priority, balanced allocation, temperature dominance and single target limit according to the operating condition category. S303. Traverse each mode in the preset discrete heat dissipation mode library and calculate the objective function value. Select the mode with the smallest objective function value as the optimal mode for coarse search. S304. Based on the louver opening configuration and fan speed of the coarse search optimal mode, perform a multi-dimensional fine search within the parameter neighborhood with a preset step size, use the Gaussian process regression model to quickly evaluate the objective function value of each neighborhood sample point, and select the sample point with the smallest objective function value as the fine search optimal solution. S305. Perform hard boundary checks on safety constraints for candidate modes. Safety constraints include upper limit of temperature safety, upper limit of pressure drop safety, physical range of fan speed, mechanical limit of louvers and transient overshoot tolerance. Eliminate candidate modes that do not meet the constraints. S306. Calculate the improvement of the objective function value between the current operating mode and the optimal solution of the fine search. Only when the improvement exceeds the preset threshold and the steady-state or forced switching conditions are met will the optimal solution of the fine search be determined as the target heat dissipation mode.

[0036] In step S301, experimental calibration data is obtained by measuring the heat dissipation module under different combinations of heat load and ambient temperature. The measured objects include the shell temperature, inlet and outlet pressure drop, and sound power level of each candidate mode. Computational fluid dynamics simulation data is obtained by establishing a three-dimensional geometric model of the heat dissipation module, dividing the computational mesh, and solving the flow field and temperature field. The simulation conditions cover extreme boundary conditions that are difficult to reproduce experimentally. The Gaussian process regression model takes the louver opening configuration and fan speed as input features and the shell temperature, inlet and outlet pressure drop, and sound power level as output targets. The experimental calibration data and computational fluid dynamics simulation data are combined as the training sample set. Maximum likelihood estimation is used to optimize the hyperparameters of the squared exponential kernel function, so that the model accurately fits at the training sample points and smoothly interpolates between sample points. After training, the Gaussian process regression model can quickly predict the corresponding shell temperature, inlet and outlet pressure drop, and sound power level for any given combination of louver opening configuration and fan speed, forming the core prediction engine of the heat dissipation mode-performance mapping database.

[0037] In step S302, the silent priority mode corresponds to the steady-state low-load condition, where the weight coefficient of the noise target item is the highest, while the weight coefficients of the temperature and voltage drop target items are relatively lower; the balanced distribution mode corresponds to the steady-state medium-load condition, where the weight coefficients of the three target items are close; the temperature-dominant mode corresponds to the steady-state high-load condition and the transient rise condition, where the weight coefficient of the temperature target item is the highest; and the single-target limit mode corresponds to the heat accumulation warning condition, where the weight coefficient of the temperature target item is full, while the weight coefficients of the voltage drop and noise target items are zero. Optionally, the specific values ​​of the weight coefficients are calibrated through simulation and experimentation during the system design phase. For each type of operating condition, a set of weight coefficient combinations that meet the priority of heat dissipation requirements for that condition is determined, and the calibration results are stored in the controller's non-volatile memory. During runtime, the corresponding weight coefficient combination is directly read according to the current operating condition category.

[0038] In step S303, a preset discrete heat dissipation mode library is constructed before the system is put into use. Based on the physical characteristics and typical application scenarios of the heat dissipation module, the continuous adjustment range of the louver opening configuration and fan speed is discretized into several levels. Each level corresponds to a specific combination of louver opening configuration and fan speed. Each level combination covers the heat dissipation capacity range from silent to extreme performance. When traversing the preset discrete heat dissipation mode library, a Gaussian process regression model is used to predict the shell temperature, inlet and outlet pressure drop, and sound power level of each mode under the current operating conditions. The predicted values ​​are substituted into a multi-objective optimization function to calculate the objective function value of each mode. The mode with the smallest objective function value is selected as the optimal mode for coarse search.

[0039] In step S304, the parameter neighborhood is based on the louver opening configuration and fan speed of the coarse search optimal mode. It expands in both positive and negative directions along each parameter dimension with a preset step size, the value of which is determined according to the control accuracy requirements. Neighborhood sample points are generated by combining and enumerating the expansion amounts of each parameter dimension. A Gaussian process regression model is used to quickly evaluate the objective function value of each neighborhood sample point, and the sample point with the smallest objective function value is selected as the optimal solution for fine search.

[0040] In step S305, the upper limit of temperature safety can be determined by subtracting the temperature drop along the heat dissipation path from the highest permissible junction temperature of the electronic device; the upper limit of voltage drop safety can be determined by the maximum head capacity of the fan at its highest speed; the physical range of fan speed can be determined by the minimum and maximum speeds given in the fan specifications; the mechanical limit of the louvers can be determined by the stroke range of the actuator; and the transient overshoot tolerance can be determined based on the permissible range of the temperature change rate. When performing hard boundary checks on the candidate modes to ensure safety constraints, any candidate mode that violates any of the above constraints is eliminated and no longer participates in the subsequent comparison of objective function values.

[0041] In step S306, the improvement magnitude is calculated by comparing the objective function value of the current operating mode with that of the optimal solution obtained through fine-grained search. The preset threshold is determined based on the system's trade-off between switching frequency and performance improvement. In steady-state or forced switching conditions, the steady-state condition requires the current operating condition to be in a steady-state category and the improvement magnitude to exceed the preset threshold. The forced switching condition allows for a more relaxed improvement magnitude requirement under transient operating conditions, enabling the heat dissipation module to quickly respond to drastic changes in thermal load.

[0042] This embodiment constructs a heat dissipation mode-performance mapping database using a Gaussian process regression model, enabling rapid prediction of performance indicators under any combination of louver opening configurations and fan speeds, providing a continuous mapping basis for mode decision-making. By dynamically assigning weight coefficients based on operating condition categories, the decision priority under different operating conditions is matched with heat dissipation requirements. A two-stage strategy of coarse search and fine search is employed to quickly lock the optimal region in a preset discrete mode library and then accurately locate the optimal solution within the parameter neighborhood, balancing search efficiency and accuracy. Hard boundary checks with safety constraints eliminate infeasible candidate modes, ensuring that the output mode is within the physically feasible range. Finally, through dual judgment of improvement magnitude and switching conditions, mechanical wear and system oscillations caused by frequent switching are avoided, resulting in the target heat dissipation mode with optimal overall performance and meeting safety constraints.

[0043] In some embodiments, traversing each mode in a preset discrete heat dissipation mode library and calculating the objective function value, selecting the mode with the smallest objective function value as the optimal mode for coarse search, includes: The objective function value is the weighted sum of squares of the temperature objective term, the pressure drop objective term, and the noise objective term. The temperature objective term is the normalized square of the difference between the predicted shell temperature and the target temperature. When the predicted shell temperature is not higher than the target temperature, the temperature objective term is zero. The pressure drop objective term is the normalized square of the ratio of the predicted inlet and outlet pressure drop to the maximum allowable pressure drop. The noise objective term is the normalized square of the ratio of the predicted sound power level to the maximum allowable noise level. The target temperature is determined based on the current operating condition category. Under transient rise conditions and heat accumulation warning conditions, the target temperature is lower than the target temperature under steady-state conditions, so as to reserve a larger temperature safety margin. Iterate through all modes in the preset discrete heat dissipation mode library, use the Gaussian process regression model to predict the shell temperature, inlet and outlet pressure drop and sound power level of each mode under the current operating conditions, and evaluate them one by one by substituting them into the objective function value calculation formula. The discrete pattern with the smallest objective function value is selected as the optimal pattern for coarse search, and the fine search stage of the parameter neighborhood is entered based on the optimal pattern for coarse search.

[0044] In this embodiment, the target temperature is determined based on the thermal design power of the heat dissipation module and the allowable operating temperature of the electronic components. Under steady-state conditions, the target temperature is set to a value near the allowable operating temperature. Under transient rise conditions and thermal accumulation warning conditions, the target temperature is offset downwards by a preset amount based on the allowable operating temperature. The offset amount is determined according to the allowable range of temperature overshoot. When traversing all modes in the preset discrete heat dissipation mode library, it can start from the mode with the lowest heat dissipation capacity and sequentially visit the mode with the highest heat dissipation capacity, or it can be traversed in reverse order. After the objective function value of each mode has been calculated, the discrete mode with the smallest objective function value is selected as the coarse search optimal mode. Then, the louver opening configuration and fan speed of the coarse search optimal mode are used as the basis to enter the parameter neighborhood fine search stage.

[0045] This embodiment improves the sensitivity of the temperature target item to temperature deviation under transient rise conditions and heat accumulation warning conditions by setting different operating conditions for the target temperature. The driving mode decision prioritizes candidate modes with stronger heat dissipation capabilities. By traversing the preset discrete heat dissipation mode library and using a Gaussian process regression model to evaluate the objective function value of each mode, the mode with the smallest objective function value is selected as the optimal mode for coarse search, providing a reasonable initial benchmark for subsequent fine search of the parameter neighborhood.

[0046] In some embodiments, the adjustment of the target opening configuration and the target rotational speed are performed synchronously according to a preset coordinated timing rule, including: When the target opening configuration shows an increasing trend relative to the current opening configuration, the fan advance timing is executed to control the fan to increase towards the target speed at a preset acceleration slope. After a first preset time delay, the opening increase action of the louver section is started, so that the fan can build up pressure head before the louver resistance increases. When the target opening configuration is decreasing relative to the current opening configuration, the louver advance timing is executed, and the louver section is controlled to perform the opening reduction action first. After a second preset time delay, the fan is controlled to reduce the speed with a preset deceleration slope, and the high air volume is maintained to accelerate heat dissipation by taking advantage of the window period when the resistance is reduced after the louver is closed. A rate limit is set for the angular velocity of the louver section. The rate limit is temporarily relaxed under transient rising conditions to accelerate the establishment of turbulent response. An acceleration slope limit and a deceleration slope limit are set for the slope of the fan speed change so that the louver opening change and the fan speed change are executed synchronously under the coupled rate constraint. Multiple louver sections are divided into two groups, and the opening adjustment of the two groups is staggered at preset time intervals to suppress mechanical crosstalk between adjacent sections. Based on the deviation between the measured value of the inlet and outlet pressure difference and the expected pressure drop predicted by the heat dissipation mode-performance mapping database, the fan speed compensation is calculated and superimposed on the target speed to compensate for the airflow loss caused by the system impedance shift due to the change in louver opening.

[0047] In this embodiment, the preset acceleration slope can be calibrated based on the maximum output capacity of the fan driver and the system's allowable temperature rise rate. Preferably, the first preset time is determined based on the difference between the time required for the fan speed to increase from the current value to the target value and the time required for the louver actuator to start rotating from startup; the second preset time is determined based on the difference between the time required for the louver actuator to complete the opening reduction action from startup and the inertial delay for the fan speed to begin decreasing from the current value. The preset deceleration slope can be calibrated based on the minimum controllable rate of change of the fan driver in deceleration mode.

[0048] The upper limit of the speed can be determined by combining the mechanical limit angular velocity of the louver actuator with the allowable temperature change rate of the system. The upper limits of the acceleration slope and deceleration slope can be determined separately based on the rated current and thermal protection requirements of the fan motor.

[0049] Multiple louver sections are divided into two groups according to the parity of the section number, and the preset time interval can be calibrated based on the decay time of mechanical vibration.

[0050] The expected pressure drop is obtained by querying the heat dissipation mode-performance mapping database to find the predicted inlet and outlet pressure drops under the current mode. The fan speed compensation is calculated by multiplying the deviation between the measured inlet and outlet pressure difference and the expected pressure drop by the preset compensation gain. The compensation gain is calibrated based on the sensitivity of the pressure difference deviation to the air volume.

[0051] This embodiment sets the fan-first and louver-first action sequences for two scenarios: increasing and decreasing the opening degree. This ensures that the fan pressure head builds up before the louver resistance rises, and the louver resistance decreases before the fan speed drops, thus maintaining continuous and stable airflow. Upper limits for speed and slope constrain the action rates of the louvers and fan, preventing temperature oscillations and mechanical shocks caused by sudden changes in airflow state. Two sets of staggered start-up sections cut off the synchronous vibration transmission path, ensuring the positioning accuracy of the opening degree adjustment. Differential pressure compensation corrects the fan speed based on the deviation between the measured inlet and outlet pressure difference and the desired pressure drop, offsetting the airflow loss caused by system impedance offset and ensuring a smooth transition of airflow during mode switching.

[0052] In some embodiments, when the target opening configuration shows an increasing trend relative to the current opening configuration, a fan-first timing sequence is executed, controlling the fan to increase towards the target speed at a preset acceleration slope, and after a first preset time delay, initiating the opening increase action of the louver section, so that the fan establishes pressure head before the louver resistance increases, including: The fan-first timing is triggered when the target opening configuration is detected to be greater than the current opening configuration. It controls the fan to start increasing the speed with a preset acceleration slope. After the fan speed increases for a first preset time, the opening of the louver section is increased. This makes the increase in flow resistance caused by the increase in louver opening match the pressure head already established by the fan, thus avoiding a temperature spike in the base plate due to a sudden drop in flow. The first preset time is determined based on the ratio of the acceleration time required for the fan to increase from the current speed to the target speed to the movement time required for the louvers to move from the current opening to the target opening, so that the lead time of the fan's pre-movement is sufficient to cover the changes in system impedance during the movement of the louvers; During the fan-first timing execution, if the base temperature change rate is detected to exceed the transient rise threshold, the preset acceleration slope is further increased, so that the fan builds up the pressure head at a faster rate and shortens the first preset time to cope with the emergency thermal load rise scenario.

[0053] In this embodiment, the determination that the target opening configuration is greater than the current opening configuration is achieved by comparing each segment. The target opening value of each louver segment is compared with the current opening value one by one. When the target opening value of any segment is greater than the current opening value, the fan-first timing is triggered.

[0054] The acceleration time is obtained by dividing the difference in fan speed required to increase from the current value to the target value by a preset acceleration slope. The motion time is obtained by dividing the difference in angle required for the louvers to move from the current opening to the target opening by the louver's motion speed. The first preset time is determined based on the ratio between the motion time and the acceleration time, and can be set to a certain proportion of the motion time. This proportion is calibrated according to the response characteristics of the system impedance change, and the proportion value is chosen so that the lead time of the fan's pre-action is sufficient to cover the system impedance change during the louver's movement.

[0055] The transient rise threshold is determined based on the historical distribution of the base temperature change rate, selecting a critical value that can distinguish between normal fluctuations and a significant upward trend. During the fan-first timing execution, the base temperature change rate is continuously monitored. When the base temperature change rate exceeds the transient rise threshold, an increment is added to the original preset acceleration slope. The magnitude of the increment is determined by the extent to which the base temperature change rate exceeds the transient rise threshold; the greater the exceedance, the larger the increment, enabling the fan to build up pressure head at a faster rate while correspondingly shortening the first preset time to cope with emergency thermal load increases.

[0056] This embodiment accurately determines the increasing trend of the target opening configuration and triggers the fan advance timing by comparing the target opening value with the current opening value segment by segment. The first preset time is determined by the ratio of acceleration time to movement time, so that the advance of the fan advance action exactly covers the system impedance change during the louver movement, avoiding insufficient advance leading to a sudden drop in flow or excessive advance leading to energy waste. The dynamic increase of the acceleration slope is triggered by monitoring the base temperature change rate, so that the fan can build up the pressure head at a faster rate in emergency hot load scenarios, while shortening the first preset time to speed up the response speed, thereby effectively suppressing the base temperature spike under transient conditions.

[0057] In some embodiments, during steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance, including: Once the system enters steady-state operation and maintains the current heat dissipation mode for more than a preset time, the steady-state performance evaluation program is started to calculate the steady-state temperature deviation between the actual steady-state temperature and the target temperature. When the steady-state temperature deviation exceeds the positive threshold, the fan speed is increased or the louver opening is increased by a preset step size, and the temperature response in subsequent control cycles is observed to confirm the adjustment effect. When the steady-state temperature deviation is below the negative threshold and the temperature safety margin is sufficient, the fan speed is reduced or the louver opening is reduced by a preset step size to reduce energy consumption and noise. Monitor the rate of change of heat dissipation efficiency; when the rate of change of heat dissipation efficiency shows a continuous downward trend, it is determined that the heat dissipation system has experienced performance degradation. Accumulate residual samples between the actual measured performance under each heat dissipation mode and the predicted values ​​of the Gaussian process regression model, perform outlier detection on the residual samples, and remove outlier samples that exceed the preset multiple of the historical residual standard deviation. When the number of effective residual samples exceeds the update threshold, the sparse approximation algorithm is used to incrementally update the posterior mean and variance of the Gaussian process regression model online, keeping the kernel function hyperparameters unchanged, so that the updated heat dissipation mode-performance mapping database reflects the time-varying performance characteristics of the heat dissipation module caused by dust accumulation or fin aging. Real-time anomaly diagnosis is performed, including fan failure diagnosis, air duct blockage diagnosis, temperature sensor failure diagnosis, louver actuator step loss diagnosis, and louver actuator jamming diagnosis. Based on the diagnosis results, corresponding protection actions are triggered, including switching to passive heat dissipation limit protection mode, adjusting the louver opening to find a low impedance path, enabling redundant sensors to replace failed sensors, performing zeroing and repositioning operations on the step loss section, and marking the jamming section as unusable and coordinating with other sections to compensate for heat dissipation loss.

[0058] In this embodiment, the preset duration is calibrated based on the time required for the heat dissipation module to complete mode switching and reach temperature stability. This duration can be set after obtaining the transition time to steady-state temperature through a step response experiment on the heat dissipation module. The actual steady-state temperature is obtained by taking the arithmetic mean of the sampled base temperature values ​​during the steady-state period to eliminate the influence of temperature fluctuations.

[0059] The positive threshold is set according to the temperature control accuracy requirements, and the preset step size is calibrated according to the temperature response sensitivity of a single adjustment. After adjustment, the temperature change trend is continuously monitored in subsequent control cycles. If the temperature falls back to the target range, the adjustment is confirmed to be effective. If the temperature is still higher than the target range, the preset step size is continued to be increased until the temperature enters the target range or the upper limit of the number of adjustments is reached.

[0060] The negative threshold is set according to the temperature control accuracy requirements. Sufficient temperature safety margin is determined when the actual steady-state temperature is lower than the target temperature and the difference is greater than the preset safety margin. After reducing the fan speed or decreasing the louver opening, the temperature response is continuously monitored. If the temperature remains within the target range and the safety margin is still sufficient, the current parameters are maintained; if the temperature rises back to the target range but the safety margin is insufficient, the adjustment is stopped.

[0061] The rate of change of heat dissipation efficiency is obtained by performing linear regression on the heat dissipation efficiency values ​​over a continuous time period to obtain the slope of change. A continuous downward trend is determined when the slope of change is negative in multiple consecutive time periods and the absolute value gradually increases. This trend indicates that the heat exchange capacity of the heat dissipation module is gradually decreasing.

[0062] The residual samples are obtained by subtracting the actual measured performance values ​​under each heat dissipation mode from the predicted values ​​of the Gaussian process regression model. The historical residual standard deviation is calculated based on the historically accumulated residual samples. The preset multiple is set according to the sensitivity requirements for outlier determination. Residual samples exceeding the preset multiple are considered outliers and are removed.

[0063] The update threshold is set based on the statistical significance requirements for model updates. Incremental updates are triggered when the number of valid residual samples exceeds the update threshold. The sparse approximation algorithm only updates the local parameters in the posterior mean and variance of the Gaussian process regression model, keeping the kernel function hyperparameters unchanged. This ensures that the updated heat dissipation mode-performance mapping database reflects the time-varying performance characteristics of the heat dissipation module caused by dust accumulation or fin aging.

[0064] Preferably, the fan fault diagnosis determines that the PWM duty cycle has been increased to a high level but the fan speed has not increased accordingly and the inlet and outlet pressure difference is lower than the normal operating value, triggering a switch to passive heat dissipation limit protection mode; the air duct blockage diagnosis determines that the inlet and outlet pressure difference suddenly rises to a set proportion of the historical normal value when the louver opening configuration has not been changed, triggering adjustment of the louver opening to find a low impedance path; the temperature sensor failure diagnosis determines that the temperature reading has not changed for a long time or exceeds the physically reasonable range when the heat dissipation mode has been enhanced, triggering the activation of a redundant sensor to replace the failed sensor. The redundant sensor can be arranged in the adjacent position of the failed sensor. When replacing, the temperature value of the section where the failed sensor is located is replaced with the interpolated estimate of the temperature value of the adjacent section; the louver actuator step loss diagnosis determines that the motor has received the motion command but the position change fed back by the encoder is less than the set value, triggering a zeroing and repositioning operation for the step loss section; the louver actuator jamming diagnosis determines that the motor has continuously received the motion command for a set time but the position feedback has not changed, triggering marking the jamming section as unusable and coordinating with other sections to compensate for heat dissipation loss.

[0065] In this embodiment, thermal equilibrium is ensured for a preset time before steady-state temperature deviation is evaluated. Parameters are fine-tuned based on the deviation direction and verified by temperature response, so that the heat dissipation module maintains a balance between control accuracy and energy consumption noise. A continuous decrease in the rate of change of heat dissipation efficiency triggers performance degradation judgment, thereby accumulating residual samples and eliminating outliers. When the effective samples exceed the threshold, the Gaussian process regression model is incrementally updated through a sparse approximation algorithm, so that the heat dissipation mode-performance mapping database maintains prediction accuracy as performance changes over time. Each anomaly diagnosis is executed in real time and triggers corresponding protection actions to ensure that the heat dissipation module can still maintain basic heat dissipation capacity under abnormal conditions.

[0066] In some embodiments, the location of the heat source center is inverted based on the temperature field distribution on the fin surface, and the coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control result, including: Based on the readings of multiple fin surface temperature sensors arranged along the airflow direction, the temperature gradient sequence between adjacent measuring points is calculated, and the inflection point segment where the temperature gradient sequence changes from positive to negative is located. The location of the heat source center is calculated using the weighted centroid method, with the temperature rise of each measuring point relative to the ambient temperature as the weight. Based on the center position of the heat source and the preset optimal offset, calculate the target center position of the louver activation window. The preset optimal offset points upstream of the heat source along the airflow direction. Map the target center location to the corresponding segment number, and set an activation window that covers a preset number of segments with that segment as the center; The target opening configuration of the louvers in the active window coverage area is maintained, while the opening of the louvers in the outer area of ​​the active window is attenuated to reduce the ineffective pressure drop loss in non-heat source areas. The heat source center location is inverted and the activation window is relocated according to the preset update cycle. Each relocation limits the window center to move no more than one segment interval. When multiple heat sources are detected to have similar temperature rise, the activation window is expanded to cover the corresponding areas of multiple heat sources at the same time. The active window's segment coverage area is integrated with the target heat dissipation mode, collaborative execution results, and feedback optimization results to generate and output the final switching control result.

[0067] In this embodiment, the temperature gradient sequence is obtained by dividing the difference in temperature readings of adjacent measuring points along the airflow direction by the distance between the measuring points, thus obtaining the rate of temperature change between adjacent measuring points. The inflection point segment is located by comparing the positive and negative signs of adjacent gradient values ​​sequentially along the airflow direction. The measuring point interval corresponding to the gradient value changing from positive to negative is the inflection point segment, which characterizes the peak region of the influence of the heat source plume on the fin surface temperature.

[0068] The temperature rise is obtained by subtracting the ambient temperature from the temperature readings at each measuring point. The weighted centroid method is used to multiply the segment position number of each measuring point by the corresponding temperature rise, sum these values, and then divide by the total temperature rise to obtain an estimate of the heat source center position. The preset optimal offset is obtained through experimental calibration. The heat transfer performance of the heat dissipation module is measured at different offsets, and the offset with the best heat transfer performance is selected as the preset optimal offset. The target center position is obtained by adding the heat source center position to the preset optimal offset, ensuring that the center of the louvered active window is located upstream of the heat source center.

[0069] When mapping the target center location to the corresponding segment number, the range of the segment into which the target center location falls is calculated, and the segment number is taken as the mapping result. The activation window is centered on this segment and extends to the upstream and downstream sides by a preset number of adjacent segments, so that the activation window covers the main area traversed by the heat source plume path.

[0070] The attenuation process is achieved by multiplying the target opening of the louvers in the external section by an attenuation coefficient less than 1. The value of the attenuation coefficient is determined based on the distance between the external section and the activation window. The greater the distance, the smaller the attenuation coefficient, which greatly reduces the opening of the louvers in sections far from the heat source plume path, thereby reducing ineffective pressure drop losses.

[0071] The preset update cycle is set according to the rate of change of the heat source position. The segment spacing is determined by the segment division size of the fin array. Each repositioning restricts the window center to move no more than one segment spacing, allowing the active window to gradually track the heat source migration path along the airflow direction. The determination of similar temperature rises of multiple heat sources can be achieved by comparing the temperature rise values ​​of the corresponding areas of each heat source. When the temperature rise difference of the corresponding areas of multiple heat sources is less than a set threshold, it is determined to be a scenario of balanced heating of multiple heat sources, and the active window is expanded into a global window that simultaneously covers the corresponding areas of multiple heat sources.

[0072] During integration, the coverage area of ​​the activated window is used as a constraint to correct the opening degree of each louver section in the target heat dissipation mode. The collaborative execution result and the feedback optimization result are superimposed to generate the final switching control result and output.

[0073] This embodiment locates inflection point segments using a temperature gradient sequence and calculates the center position of the heat source using a weighted centroid method with temperature rise as the weight. This ensures that the estimated heat source position takes into account both the peak position and the overall temperature rise distribution. A preset optimal offset ensures that the center of the louver activation window is located upstream of the heat source, thereby applying turbulence enhancement before the thermal plume fully develops. The segment coverage constraint of the activation window ensures that the louver opening adjustment focuses on the path of the heat source thermal plume, and the external segment is attenuated to reduce ineffective pressure drop loss. Preset update cycle and single-step movement limit enable the activation window to smoothly track the migration of the heat source. When the temperature rise of multiple heat sources is similar, the window is expanded to achieve multi-region coverage. Finally, the results are integrated to output switching control commands.

[0074] In some embodiments, the inversion of the heat source center location and the relocation of the activation window are performed according to a preset update cycle. Each relocation limits the movement of the window center to no more than one segment interval. When multiple heat sources are detected to have similar temperature rises, the activation window is expanded to simultaneously cover the corresponding regions of multiple heat sources, including: The preset update cycle is an integer multiple of multiple control cycles. In each preset update cycle, the heat source center position is recalculated and compared with the heat source center position confirmed in the previous preset update cycle to calculate the window center offset. When the window center offset exceeds one segment spacing, the window center movement for this repositioning is limited to one segment spacing, and the movement direction is consistent with the sign direction of the window center offset. The remaining offset is carried over to the subsequent preset update cycle to continue execution, so that the louver activated window gradually tracks the heat source migration path along the airflow direction. Before each relocation is executed, the surface temperature rise of the fins of each heat source is compared. When the temperature rise difference of multiple heat sources is less than the preset temperature difference threshold and is significantly higher than the ambient temperature, it is determined to be a multi-heat source balanced heating scenario. The activation window is then expanded to a global window that simultaneously covers the sections of multiple heat sources. In a scenario of balanced heating from multiple heat sources, the upstream section of each heat source is obtained by extending a preset number of sections upstream along the airflow direction, based on the corresponding section of each heat source. The target opening configuration is maintained for each upstream section of the heat source covered by the global window, so that the heat plume path of each heat source is subjected to forced turbulence enhancement. The louver opening of the external section of the global window is attenuated, so that the heat plume path of each heat source is subjected to forced turbulence enhancement.

[0075] In this embodiment, the number of control cycles included in the preset update cycle is set according to the expected rate of change of the heat source position. When the rate of change of the heat source position is fast, a smaller integer multiple is used to improve the response speed of heat source tracking; when the rate of change of the heat source position is slow, a larger integer multiple is used to reduce the execution frequency of inversion calculation and window repositioning. The difference between the recalculated heat source center position in the current preset update cycle and the heat source center position confirmed in the previous preset update cycle is the window center offset. This offset is a signed quantity, and its positive or negative sign indicates the migration direction of the heat source along the airflow direction.

[0076] The segment spacing is determined by the segment division size of the fin array, which is obtained by dividing the total length of the fin array along the airflow direction by the number of segments. The preset temperature difference threshold is calibrated based on the allowable temperature difference range between heat sources, and its value is less than the temperature difference limit required to trigger the equalization heating scenario. The temperature rise of the fin surface of each heat source corresponding to a segment is obtained by subtracting the ambient temperature from the temperature readings at each measuring point.

[0077] The attenuation process involves multiplying the louver opening of the outer section of the global window by an attenuation coefficient less than 1. The value of the attenuation coefficient is determined based on the distance between the outer section and the global window; the greater the distance, the smaller the attenuation coefficient, in order to reduce the ineffective pressure drop loss in non-heat source areas.

[0078] This embodiment balances the response speed and computational cost of heat source tracking by setting the number of control cycles included in the preset update cycle according to the rate of change of the heat source position. The signed comparison of the window center offset and the single-step movement limit enable the active window to gradually track the heat source migration path along the airflow direction, avoiding a large jump in the window center that would cause a temporary mismatch between the turbulence enhancement region and the thermal plume path. The dual conditions of a preset temperature difference threshold and a temperature rise significantly higher than the ambient temperature determine the balanced heating scenario of multiple heat sources, so that the active window expands into a global window when multiple heat sources are heating at the same time. The upstream section of each heat source maintains the target opening configuration, and the outer section of the global window is attenuated according to the distance. Thus, in the multi-heat source scenario, a balance is achieved between the comprehensive turbulence enhancement and pressure drop cost of each heat source's thermal plume path.

[0079] By adopting the above technical solutions, this invention differs from existing technologies and has the following beneficial effects: By acquiring the operating status data of the heat dissipation module and calculating the temperature change rate and heat dissipation efficiency, the current operating condition category is identified through a hierarchical rule tree. Based on this, the weight coefficients of the temperature target term, pressure drop target term, and noise target term in the multi-objective optimization function are dynamically adjusted. The optimal heat dissipation mode is solved based on the heat dissipation mode-performance mapping database, enabling the heat dissipation module to obtain control parameters that match the heat dissipation requirements under different operating conditions. Furthermore, by employing a feedforward-feedback composite control strategy and synchronously executing the adjustment of the target opening degree and target speed according to preset coordinated timing rules, the airflow transitions smoothly during mode switching, avoiding temperature oscillations and mechanical shocks. During steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance. This ensures that the heat dissipation mode-performance mapping database maintains prediction accuracy by following the time-varying performance characteristics of the heat dissipation module caused by dust accumulation or fin aging. The location of the heat source center is inverted based on the temperature field distribution on the fin surface, and the coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path. This concentrates the benefits of enhanced turbulence heat transfer in the area directly opposite the heat source, while reducing ineffective pressure drop losses in non-heat source areas. This achieves multi-condition adaptive optimization of the heat dissipation module in terms of heat dissipation capacity, flow resistance, and noise level, while also taking into account control accuracy during long-term operation and heat dissipation reliability under abnormal conditions.

Claims

1. A method for intelligent switching control of a multi-condition heat dissipation module, characterized in that, include: Acquire the operating status data of the heat dissipation module, including heat source power consumption, base temperature, fin surface temperature field distribution, ambient temperature, inlet and outlet pressure difference, and fan speed; Based on the operating status data, the temperature change rate and heat dissipation efficiency are calculated, and the current operating condition category is identified through a hierarchical rule tree. The operating condition categories include steady-state low load, steady-state medium load, steady-state high load, transient increase, transient decrease, and heat accumulation warning. The weight coefficients of the temperature target, pressure drop target, and noise target in the multi-objective optimization function are dynamically adjusted according to the operating condition category. The multi-objective optimization function is solved according to the heat dissipation mode-performance mapping database to obtain the optimal heat dissipation mode. The optimal heat dissipation mode includes the target opening configuration of each louver section and the target fan speed. A feedforward-feedback composite control strategy is adopted to synchronously execute the adjustment of the target opening configuration and the target rotation speed according to a preset coordinated timing rule. The coordinated timing rule sets the order of action and the rate of change of the fan and the louvers according to the scenario of increasing or decreasing the opening. During steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance. The location of the heat source center is inverted based on the temperature field distribution on the fin surface. The coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control result.

2. The intelligent switching control method for a multi-condition heat dissipation module according to claim 1, characterized in that, Based on the aforementioned operating status data, the temperature change rate and heat dissipation efficiency are calculated. The current operating condition category is identified using a hierarchical rule tree, including: The base temperature is subjected to a first-order digital low-pass filter, the inlet and outlet pressure difference is subjected to a cascaded process of median filtering and low-pass filtering, the heat source power consumption is subjected to a moving average filter, and the temperature change rate is calculated using the difference method based on the filtered base temperature data. The heat flux density is calculated based on the ratio of the heat source power consumption to the contact area between the heat-conducting substrate and the heat source. The heat dissipation efficiency is calculated based on the ratio of the heat source power consumption to the difference between the substrate temperature and the ambient temperature. The rate of change of heat dissipation efficiency is calculated based on historical data of the heat dissipation efficiency. The apparent velocity is inversely calculated and the effective Reynolds number is calculated based on the pressure difference between the inlet and outlet using the Darcy-Fochheimer equation. According to the hierarchical rule tree, the heat accumulation early warning judgment, transient condition judgment and steady-state condition subdivision judgment are executed in sequence. The heat accumulation early warning judgment is triggered when the base temperature is higher than the warning threshold and the temperature change rate is positive. The transient condition judgment is triggered when the temperature change rate exceeds the rising threshold or falling threshold. The steady-state condition subdivision judgment determines the condition category based on the comparison result of the heat source power consumption and the preset threshold when the absolute value of the temperature change rate is less than the steady-state threshold. A working condition confidence counter is introduced to continuously confirm the candidate working condition category. The working condition is switched only when the candidate working condition category is maintained for a preset period. For transient rising working conditions, the future heat load change trend is predicted by linear extrapolation based on historical data of heat source power consumption. The heat dissipation mode level is adjusted in advance based on the prediction results.

3. The intelligent switching control method for a multi-condition heat dissipation module according to claim 2, characterized in that, A working condition confidence counter is introduced to continuously confirm candidate working condition categories. Working condition switching is only performed when the candidate working condition category remains continuously for a preset period. For transiently rising working conditions, future heat load change trends are predicted through linear extrapolation based on historical heat source power consumption data. The heat dissipation mode level is adjusted in advance based on the prediction results, including: The operating condition confidence counter increments until it reaches an upper limit when the current candidate operating condition category matches the confirmed operating condition category, and decrements until it reaches a lower limit when the candidate operating condition category does not match the confirmed operating condition category. The confirmed operating condition category is switched to the candidate operating condition category only when the operating condition confidence counter reaches the upper limit. The preset period is the product of the upper limit and a single control period. The linear extrapolation predicts the heat load value within a preset time window based on the regression slope of historical heat source power consumption data. When the predicted heat load value exceeds the sum of the current heat source power consumption and the preset difference, it is determined that the heat load is showing a continuous upward trend, and the heat dissipation mode level is upgraded by one level in advance. The extent to which the heat dissipation mode level is increased is limited by the dominant position of the temperature weight coefficient under the transient rise condition, so that the temperature target item occupies a dominant weight in the mode decision, thereby suppressing temperature overshoot.

4. The intelligent switching control method for a multi-condition heat dissipation module according to claim 1, characterized in that, The weight coefficients of the temperature, pressure drop, and noise objectives in the multi-objective optimization function are dynamically adjusted according to the operating condition category. The multi-objective optimization function is then solved using a heat dissipation mode-performance mapping database to obtain the optimal heat dissipation mode, including: Based on the Gaussian process regression model, a continuous mapping relationship is constructed from the louver opening configuration and fan speed to the shell temperature, inlet and outlet pressure drop and sound power level. The Gaussian process regression model uses the squared exponential kernel function as the covariance function, optimizes the hyperparameters through maximum likelihood estimation, and obtains the heat dissipation mode-performance mapping database by joint training based on experimental calibration data and computational fluid dynamics simulation data. The weighting coefficients of the temperature target item, pressure drop target item, and noise target item are dynamically allocated according to the operating condition category. The weighting coefficients switch between quiet priority, balanced allocation, temperature dominance, and single target limit as the operating condition category changes. The model is traversed in the preset discrete heat dissipation model library and the objective function value is calculated. The model with the smallest objective function value is selected as the optimal model for coarse search. Based on the louver opening configuration and fan speed of the coarse search optimal mode, a multi-dimensional fine search is performed in the parameter neighborhood with a preset step size. The objective function value of each neighborhood sample point is quickly evaluated using the Gaussian process regression model, and the sample point with the smallest objective function value is selected as the fine search optimal solution. Perform hard boundary checks on candidate modes for safety constraints, including upper limits for temperature safety, upper limits for pressure drop safety, physical range of fan speed, mechanical limits for louvers, and transient overshoot tolerance; and eliminate candidate modes that do not meet the constraints. Calculate the improvement of the objective function value between the current operating mode and the fine search optimal solution. Only when the improvement exceeds a preset threshold and the steady-state or forced switching conditions are met, the fine search optimal solution is determined as the target heat dissipation mode.

5. The intelligent switching control method for a multi-condition heat dissipation module according to claim 4, characterized in that, The system iterates through a pre-defined discrete heat dissipation mode library, calculates the objective function value for each mode, and selects the mode with the smallest objective function value as the optimal mode for coarse search. This includes: The objective function value is the weighted sum of squares of the temperature objective term, the pressure drop objective term, and the noise objective term. The temperature objective term is the normalized square of the difference between the predicted shell temperature and the target temperature. When the predicted shell temperature is not higher than the target temperature, the temperature objective term is zero. The pressure drop objective term is the normalized square of the ratio of the predicted inlet and outlet pressure drop to the maximum allowable pressure drop. The noise objective term is the normalized square of the ratio of the predicted sound power level to the maximum allowable noise level. The target temperature is determined according to the current operating condition category. Under transient rise conditions and heat accumulation warning conditions, the target temperature is lower than the target temperature under steady-state conditions, so as to reserve a larger temperature safety margin. The system iterates through all modes in the preset discrete heat dissipation mode library, uses the Gaussian process regression model to predict the shell temperature, inlet and outlet pressure drop and sound power level of each mode under the current operating conditions, and substitutes them into the objective function value calculation formula to evaluate them one by one. The discrete pattern with the smallest objective function value is selected as the optimal pattern for coarse search, and the fine search stage of the parameter neighborhood is entered based on the optimal pattern for coarse search.

6. The intelligent switching control method for a multi-condition heat dissipation module according to claim 1, characterized in that, The adjustment of the target opening configuration and the target rotation speed are executed synchronously according to a preset coordinated timing rule, including: When the target opening configuration shows an increasing trend relative to the current opening configuration, the fan advance timing is executed, and the fan is controlled to increase towards the target speed at a preset acceleration slope. After a first preset time delay, the opening increase action of the louver section is started, so that the fan can establish a pressure head before the louver resistance increases. When the target opening configuration is decreasing relative to the current opening configuration, the louver priority timing is executed, and the louver section is controlled to perform the opening reduction action first. After a second preset time delay, the fan is controlled to reduce the speed with a preset deceleration slope, and the high air volume is maintained to accelerate heat dissipation by taking advantage of the window period when the resistance is reduced after the louver is closed. An upper limit is set for the angular velocity of the louver section. The upper limit is temporarily relaxed under transient rising conditions to accelerate the establishment of turbulent response. An upper limit for the acceleration slope and an upper limit for the deceleration slope are set for the slope of the fan speed change, so that the change of louver opening and the change of fan speed are executed synchronously under the coupling rate constraint. Multiple louver sections are divided into two groups, and the opening adjustment of the two groups is staggered at preset time intervals to suppress mechanical crosstalk between adjacent sections. Based on the deviation between the measured value of the inlet and outlet pressure difference and the expected pressure drop predicted by the heat dissipation mode-performance mapping database, the fan speed compensation is calculated and superimposed on the target speed to compensate for the airflow loss caused by the system impedance shift due to the change in louver opening.

7. The intelligent switching control method for a multi-condition heat dissipation module according to claim 6, characterized in that, When the target opening configuration shows an increasing trend relative to the current opening configuration, a fan-first timing sequence is executed, controlling the fan to increase towards the target speed at a preset acceleration slope. After a first preset time delay, the opening increase action of the louver section is initiated, allowing the fan to establish pressure head before the louver resistance increases, including: The fan advance timing is triggered when the target opening configuration is detected to be greater than the current opening configuration. The fan is controlled to start increasing the speed with a preset acceleration slope. After the fan speed increase continues for a first preset time, the opening of the louver section is increased. This makes the increase in flow resistance caused by the increase in louver opening match the pressure head already established by the fan, thus avoiding a temperature spike in the base due to a sudden drop in flow. The first preset time is determined based on the ratio of the acceleration time required for the fan to increase from the current speed to the target speed to the movement time required for the louvers to move from the current opening to the target opening, so that the lead time of the fan's pre-movement is sufficient to cover the system impedance changes during the louver's movement; During the execution of the fan-first timing sequence, if the base temperature change rate is detected to exceed the transient rise threshold, the preset acceleration slope is further increased, so that the fan builds up the pressure head at a faster rate and shortens the first preset time to cope with the emergency thermal load rise scenario.

8. The intelligent switching control method for a multi-condition heat dissipation module according to claim 1, characterized in that, During steady-state operation, parameter fine-tuning and online model updates are performed based on the deviation between actual and predicted performance, including: Once the system enters steady-state operation and maintains the current heat dissipation mode for more than a preset time, the steady-state performance evaluation program is started to calculate the steady-state temperature deviation between the actual steady-state temperature and the target temperature. When the steady-state temperature deviation exceeds the positive threshold, the fan speed is increased or the louver opening is increased by a preset step size, and the temperature response in subsequent control cycles is observed to confirm the adjustment effect. When the steady-state temperature deviation is below the negative threshold and the temperature safety margin is sufficient, the fan speed is reduced or the louver opening is reduced by a preset step size to reduce energy consumption and noise. Monitor the rate of change of heat dissipation efficiency, and determine that the heat dissipation system has experienced performance degradation when the rate of change of heat dissipation efficiency shows a continuous downward trend; Accumulate residual samples between the actual measured performance under each heat dissipation mode and the predicted value of the Gaussian process regression model, perform outlier detection on the residual samples, and remove outlier samples that exceed a preset multiple of the historical residual standard deviation; When the number of effective residual samples exceeds the update threshold, the sparse approximation algorithm is used to incrementally update the posterior mean and variance of the Gaussian process regression model online, keeping the kernel function hyperparameter unchanged, so that the updated heat dissipation mode-performance mapping database reflects the time-varying performance characteristics of the heat dissipation module caused by dust accumulation or fin aging. Real-time anomaly diagnosis is performed, including fan failure diagnosis, air duct blockage diagnosis, temperature sensor failure diagnosis, louver actuator step loss diagnosis, and louver actuator jamming diagnosis. Based on the diagnosis results, corresponding protection actions are triggered. The protection actions include switching to passive heat dissipation limit protection mode, adjusting the louver opening to find a low impedance path, enabling redundant sensors to replace failed sensors, performing zeroing and repositioning operations on the step loss section, and marking the jamming section as unusable and coordinating with other sections to compensate for heat dissipation loss.

9. The intelligent switching control method for a multi-condition heat dissipation module according to claim 1, characterized in that, Based on the temperature field distribution on the fin surface, the location of the heat source center is inverted. The coverage range of the louver activation window is adjusted to match the forced turbulence region with the heat source plume path, generating and outputting the final switching control result, including: Based on the readings of multiple fin surface temperature sensors arranged along the airflow direction, the temperature gradient sequence between adjacent measuring points is calculated, and the inflection point segment where the temperature gradient sequence changes from positive to negative is located. The location of the heat source center is calculated using the weighted centroid method, with the temperature rise of each measuring point relative to the ambient temperature as the weight. Based on the center position of the heat source and the preset optimal offset, the target center position of the louver activation window is calculated, and the preset optimal offset points upstream of the heat source along the airflow direction. Map the target center location to the corresponding segment number, and set an activation window that covers a preset number of segments with that segment as the center; The target opening configuration is maintained for the louvers in the area covered by the active window, and the opening of the louvers in the area outside the active window is attenuated to reduce the ineffective pressure drop loss in the non-heat source area. The heat source center location is inverted and the activation window is repositioned according to a preset update cycle. Each repositioning restricts the window center to move no more than one segment interval. When multiple heat sources are detected to have similar temperature rises, the activation window is expanded to cover the corresponding areas of multiple heat sources at the same time. The segment coverage area of ​​the activated window is integrated with the target heat dissipation mode, collaborative execution results, and feedback optimization results to generate and output the final switching control result.

10. The intelligent switching control method for a multi-condition heat dissipation module according to claim 9, characterized in that, The heat source center location is inverted and the activation window is repositioned according to a preset update cycle. Each repositioning restricts the window center to move no more than one segment distance. When multiple heat sources are detected to have similar temperature rises, the activation window is expanded to simultaneously cover the corresponding areas of multiple heat sources, including: The preset update cycle is an integer multiple of multiple control cycles. In each preset update cycle, the heat source center position is recalculated, and the heat source center position is compared with the heat source center position confirmed in the previous preset update cycle to calculate the window center offset. When the window center offset exceeds one segment spacing, the window center movement for this repositioning is limited to one segment spacing, and the remaining offset is carried over to a subsequent preset update cycle to continue execution, so that the louver activated window gradually tracks the heat source migration path along the airflow direction. Before each relocation is executed, the surface temperature rise of the fins of each heat source is compared. When the temperature rise difference of multiple heat sources is less than the preset temperature difference threshold and is significantly higher than the ambient temperature, it is determined to be a multi-heat source balanced heating scenario. The activation window is then expanded to a global window that simultaneously covers the sections of multiple heat sources. In a scenario of balanced heating from multiple heat sources, the target opening configuration is maintained for the upstream sections of each heat source covered by the global window, and the louver opening of the outer section of the global window is attenuated, so that the heat plume paths of each heat source are subjected to forced turbulence enhancement.

Images