Thermal management method and system for wide-temperature-range Micro LED display module

Abstract

The invention relates to the technical field of Micro LED display modules, and discloses a thermal management method and system for a wide-temperature-range Micro LED display module, and the method comprises the steps: dividing the Micro LED display module into a high-temperature region, a medium-temperature region and a low-temperature region, carrying out the temperature data collection, and obtaining target temperature data and temperature gradient data; performing deep learning training to obtain a time re-attention long and short-term memory network model and a temperature zone adaptive re-attention mechanism parameter; constructing a three-layer cascade control system comprising a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller; the temperature equalization control is executed through the three-layer cascade control system to obtain the temperature uniformity index and the temperature buffer zone parameter of each temperature zone, so that the temperature sudden change phenomenon between the adjacent temperature zones is effectively relieved, and the temperature uniformity of the display module is improved.

Description

Technical Field

[0001] The present application relates to the technical field of Micro LED display modules, and in particular to a thermal management method and system for a Micro LED display module with a wide temperature range. Background Art

[0002] With the rapid development of Micro LED display technology, large-size Micro LED display modules are widely used in various application scenarios. However, Micro LED display modules generate a lot of heat during operation. If the heat cannot be dissipated in time and effectively, the device temperature will be too high, which will not only reduce the display performance and service life, but also may cause safety hazards. Especially in a wide temperature range working environment, the problem of uneven temperature distribution is more prominent, which brings huge challenges to thermal management.

[0003] The commonly used thermal management methods for Micro LED display modules currently rely mainly on traditional PID control and simple temperature threshold adjustment, which are difficult to adapt to complex and changing temperature environments. These methods often control the temperature of the display module as a whole, ignoring the temperature differences and heat transfer characteristics between different areas, resulting in local overheating or overcooling, affecting the uniformity and stability of the display effect. Summary of the invention

[0004] The present application provides a thermal management method and system for a wide temperature range Micro LED display module, thereby effectively alleviating the temperature mutation phenomenon between adjacent temperature zones and improving the temperature uniformity of the display module.

[0005] In a first aspect, the present application provides a thermal management method for a wide temperature range Micro LED display module, the thermal management method for a wide temperature range Micro LED display module comprising: Divide the Micro LED display module into high temperature zone, medium temperature zone and low temperature zone, collect temperature data, and obtain target temperature data and temperature gradient data; Perform deep learning training based on the target temperature data and the temperature gradient data to obtain a time-re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; A three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller is constructed according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; Temperature balance control is performed through the three-layer cascade control system to obtain the temperature uniformity index and temperature buffer zone parameters of each temperature zone.

[0006] A second aspect of the present application provides a thermal management system for a wide temperature range Micro LED display module, the thermal management system for the wide temperature range Micro LED display module comprising: A data acquisition module is used to divide the Micro LED display module into high temperature zone, medium temperature zone and low temperature zone, and collect temperature data to obtain target temperature data and temperature gradient data; A training module, used for performing deep learning training based on the target temperature data and the temperature gradient data to obtain a time re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; A construction module, used to construct a three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; The balance control module is used to perform temperature balance control through the three-layer cascade control system to obtain the temperature uniformity index and temperature buffer zone parameters of each temperature zone.

[0007] Compared with the prior art, the present application has the following beneficial effects: the time-re-attention long short-term memory network model obtained through deep learning training can accurately predict the temperature change trend of each temperature zone, and improve the foresight and accuracy of temperature control. The temperature zone adaptive re-attention mechanism parameters can dynamically adjust the control weights of different temperature zones to achieve more accurate temperature management. The design of the three-layer cascade control system realizes coordinated control from global to local. The top-level temperature zone coordination controller ensures the rationality of the overall heat dissipation strategy, the middle-level temperature zone independent controller makes fine adjustments according to the characteristics of each temperature zone, and the bottom-level execution controller ensures the accurate execution of control instructions, forming a complete control closed loop. The establishment of the temperature zone thermal network model accurately describes the heat transfer relationship between each temperature zone and provides a theoretical basis for temperature balance control. By setting a temperature buffer zone, the temperature mutation phenomenon between adjacent temperature zones is effectively alleviated, and the temperature uniformity of the display module is improved. Based on the energy consumption optimization scheme of ambient temperature adaptation, the particle swarm algorithm is used to dynamically balance the temperature control and energy consumption, which reduces the system energy consumption while ensuring the temperature control effect, and improves the economy of the thermal management system. The optimization design of the multi-task loss function comprehensively considers the temperature prediction accuracy, temperature gradient control and the rationality of attention distribution. The learning rate is adjusted through the cosine annealing strategy to improve the training effect and generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0009] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0010] Figure 1 1 is a flow chart of a thermal management method for a wide temperature range Micro LED display module provided by an embodiment of the present invention; Figure 2 It is a schematic block diagram of the structure of the thermal management system of the wide temperature range Micro LED display module provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0013] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0014] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1In the embodiment of the present application, an embodiment of the thermal management method of the wide temperature range Micro LED display module includes: Step 100: Divide the Micro LED display module into a high temperature zone, a medium temperature zone, and a low temperature zone, and collect temperature data to obtain target temperature data and temperature gradient data; It is understandable that the execution subject of the present application can be a thermal management system of a wide temperature range Micro LED display module, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0015] Specifically, the overall temperature distribution of the Micro LED display module is analyzed. According to the heat distribution characteristics of the Micro LED display module under different working conditions, combined with thermal simulation and actual test data, the temperature division standard applicable to the module is set, so that the entire Micro LED display module is divided into high temperature zone, medium temperature zone and low temperature zone. The high temperature zone includes areas with obvious heat accumulation, such as areas with concentrated distribution of high-power Micro LEDs or areas with greater environmental impact. The medium temperature zone covers parts with relatively uniform heat distribution, while the low temperature zone is mainly areas with faster heat diffusion, stronger heat dissipation capacity or far away from heat sources. Corresponding temperature sensors are arranged for different temperature zones, and temperature data is collected. For the high temperature zone, due to its large temperature fluctuation and concentrated heat, temperature sensors are arranged at a high density to ensure accurate monitoring of temperature changes in the area. High-precision thermocouples or integrated temperature sensors are used, combined with non-contact measuring equipment such as thermal imagers, to obtain instantaneous temperature data and temperature change trends in the area. Redundant sensors are arranged according to the characteristics of the high temperature zone to ensure the reliability of data collection and avoid affecting the accuracy of the overall data due to the failure of a single sensor. In the medium temperature zone, due to its relatively uniform temperature distribution, a sensor layout scheme with appropriately reduced density is adopted to reduce hardware costs and data transmission burden while ensuring data acquisition accuracy. Through evenly distributed temperature sensors, this area achieves relatively stable temperature acquisition. In the low temperature zone, due to the low temperature and small changes in this area, the sensor layout further reduces the density, and a low-power, remote transmission sensor solution is adopted to reduce energy consumption and improve the overall efficiency of the system. For temperature data acquisition in the low temperature zone, a wireless sensor network is used to avoid the installation complexity caused by wiring and ensure real-time data transmission. The temperature data of the high temperature zone, the temperature data of the medium temperature zone, and the temperature data of the low temperature zone are input into the CAN bus for sampling and transmission to obtain the target temperature data. As an efficient industrial-grade communication protocol, the CAN bus can ensure stable data transmission in complex environments and ensure efficient acquisition of temperature data. In the data acquisition system, the data of multiple temperature sensors are preprocessed by the acquisition module and then uniformly transmitted to the data processing center by the CAN bus. After receiving the temperature data from different temperature zones, the data processing center performs data preprocessing, including denoising, error correction, and outlier detection, to ensure the accuracy and reliability of the data. The target temperature data is stored and used for further temperature gradient calculation. Based on the target temperature data, the temperature gradient data between adjacent temperature zones is calculated for subsequent thermal management optimization. During the calculation process, numerical differentiation or interpolation algorithms are used to analyze the spatial temperature variation trend and calculate the temperature gradient between each temperature zone. By measuring the temperature difference between two adjacent temperature sensors and calculating the temperature gradient based on the distance between the two, the temperature gradient distribution of the entire Micro LED display module is obtained.By analyzing the temperature gradient data, the direction and rate of heat conduction can be identified, thereby optimizing the heat dissipation design of the Micro LED display module.

[0016] Step 200: Perform deep learning training based on target temperature data and temperature gradient data to obtain a time-re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; Specifically, the target temperature data is divided into high temperature zone data, medium temperature zone data and low temperature zone data according to the temperature zone, and the temperature gradient data is separated to extract the gradient characteristics between temperature zones. This division process depends on the data collected by the sensor. The system stores the temperature data of different regions independently according to the temperature distribution law, and calculates the temperature gradient of adjacent temperature zones through temperature change trend analysis to obtain the characteristic information of heat conduction between temperature zones. A temperature zone feature branch is constructed, which extracts features from the data of high temperature zone, medium temperature zone and low temperature zone respectively. Therefore, three parallel long short-term memory networks (LSTM) are used, and each LSTM contains 256 hidden layer nodes to ensure that the temperature change patterns inside different temperature zones can be captured. The advantage of this structure is that it can learn the thermal characteristics of each temperature zone separately, while avoiding mutual interference between different temperature zones, so that the model can be optimized independently for different temperature control requirements. At the same time, in order to enhance the understanding of the heat transfer characteristics between temperature zones, a temperature gradient feature branch is constructed, which is specifically used to process the gradient characteristics between temperature zones, and a convolutional neural network (CNN) is used for feature extraction. The advantage of CNN is that it can automatically extract local temperature change patterns and identify temperature gradient trends between different regions. Through the operation of multiple convolutional layers and pooling layers, the model effectively learns the spatial distribution law of heat diffusion, thereby improving the adaptability of the overall thermal management system. Since temperature data has time series characteristics, a time feature branch is constructed. This feature branch extracts time series features from the target temperature data, and a bidirectional long short-term memory network (Bi-LSTM) is used. The network structure contains two hidden states, forward and reverse, so that it can simultaneously capture the historical temperature change trend and the potential impact of future temperature changes, improving the prediction accuracy of the model and the reliability of the temperature control strategy. Attention fusion is performed on the temperature zone feature branch, the temperature gradient feature branch, and the time feature branch to generate a multi-temperature zone coupling feature vector. The attention mechanism highlights important information and ignores irrelevant information by dynamically assigning weights. In this process, the importance of each feature branch is calculated based on the feature weight of the temperature data, and this information is fused to form a feature vector. The multi-temperature zone coupling feature vector contains temperature zone features, heat conduction features between temperature zones, and time series change features of temperature. The multi-temperature zone coupling feature vector is input into the long short-term memory decoder network with temperature zone adaptive gating mechanism for feature decoding. The decoder reconstructs and trains the features of different temperature zones by dynamically adjusting the gating parameters, so that the network can more accurately capture the temperature change law of each temperature zone and form a time-re-attention long short-term memory network model. The design of the gating mechanism enables the temperature control strategy of different temperature zones to be adaptively adjusted, thereby effectively improving the flexibility and intelligence level of the thermal management system. Based on the temperature zone feature branch, temperature gradient feature branch and time feature branch, the attention weight of the temperature zone division information is calculated to obtain the temperature zone adaptive re-attention mechanism parameters.The attention mechanism is used to re-weight the temperature zone features, so that the temperature control system can dynamically adjust the temperature control strategy of each temperature zone according to the changes in real-time data.

[0017] The weight coefficients are assigned to the temperature zone division information to obtain the temperature zone weight coefficient matrix. The construction of this matrix depends on the temperature change characteristics of each temperature zone, and is combined with historical temperature data for statistics to determine the relative importance of each temperature zone in temperature control. The first output feature of the temperature zone feature branch is subjected to a matrix transposition operation, and a matrix multiplication operation is performed with the target temperature data. The temperature zone features are weighted using the target temperature data, and the most representative temperature control features are extracted by means of matrix operations. The matrix transposition operation can adjust the arrangement of the feature data so that the target temperature data and the temperature zone feature branch are operated on the same dimension, and the matrix multiplication method helps to capture the correlation between the temperature zone features and the actual temperature, thereby constructing a preliminary attention distribution pattern, namely the basic attention weight matrix. The Hadamard product operation is performed on the temperature zone weight coefficient matrix and the basic attention weight matrix, that is, the two matrices are fused by element-by-element multiplication to obtain the temperature zone weighted attention matrix. The introduction of the Hadamard product ensures that the temperature zone weight information will not be lost during the calculation process, and can effectively enhance the differentiated features between different temperature zones. For example, in the case of uneven temperature distribution, the Hadamard product operation enhances the attention to the high temperature area, while in the scenario where the temperature is relatively balanced, the impact on the temperature zone division is correspondingly reduced, forming a more flexible attention adjustment strategy. Based on the second output feature of the temperature gradient feature branch, a temperature gradient threshold is set for the temperature interval gradient feature, and a temperature gradient correction matrix is ​​constructed. The setting of the temperature gradient threshold is based on the thermal diffusion characteristics between temperature zones. For example, if the temperature gradient of a certain area exceeds the set threshold, it means that the heat transfer speed of the area is faster, and extra attention is given in the attention allocation process. By setting the temperature gradient threshold and constructing the corresponding correction matrix, the system fully considers the influence of the temperature gradient during the temperature control calculation process to ensure that the final attention distribution is more in line with the actual thermal management needs. The temperature gradient correction matrix is ​​combined with the temperature zone weighted attention matrix to obtain the corrected attention weight. The corrected attention weight is optimized by multi-task loss function. The temperature prediction loss term, temperature gradient prediction loss term and attention distribution loss term are introduced in a multi-task learning manner to measure the performance of the model. The temperature prediction loss term is used to measure the system's prediction accuracy for future temperature changes, while the temperature gradient prediction loss term is used to evaluate the model's ability to estimate temperature gradients between different temperature zones. At the same time, the attention distribution loss term is used to ensure the stability and rationality of the attention mechanism. In order to improve training efficiency and optimize convergence speed, the cosine annealing strategy is used to dynamically adjust the learning rate, so that larger step lengths are explored in the early stages of the training process, and gradually converge to the optimal solution in the later stages. Through the above calculation and optimization steps, the temperature zone adaptive re-attention mechanism parameters are finally obtained.

[0018] Step 300: construct a three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; It should be noted that the real-time temperature data of the Micro LED display module is input into the time re-attention long short-term memory (TAR-LSTM) network model. The model predicts the temperature by capturing the temporal characteristics of the temperature data, combining the historical temperature data with the current ambient temperature changes, and obtains the initial temperature prediction data. The initial temperature prediction data is weighted based on the re-attention mechanism parameters to obtain the temperature prediction data of the temperature zone that is more in line with the dynamic changes of the actual temperature zone. The top-level temperature zone coordination controller is constructed based on the temperature prediction data of the temperature zone. The main task of the controller is to formulate a global heat dissipation strategy and calculate the target temperature of each temperature zone. The top-level temperature zone coordination controller analyzes the temperature distribution of the entire Micro LED module, and calculates the optimal temperature balance strategy based on the ambient temperature, power input and the working mode of the device. For example, when the temperature in a certain area rises abnormally, the coordination controller adjusts the target temperature of the surrounding temperature zone to avoid local overheating problems, and optimizes the heat dissipation path to improve the overall heat dissipation efficiency. The controller combines the energy consumption optimization requirements of the thermal management system to balance the power consumption of the cooling equipment and the accuracy of temperature control to ensure that the system reduces unnecessary energy consumption while meeting thermal stability. Independent controllers for the middle temperature zone are constructed for the high temperature zone, the middle temperature zone, and the low temperature zone. These independent controllers use the model predictive control (MPC) algorithm to dynamically adjust the local heat dissipation strategy according to the temperature prediction data of the temperature zone. The MPC algorithm predicts future temperature changes through mathematical models and selects the optimal control input based on the optimization target, so that the temperature control system can maintain stability under different load conditions. The independent controller of each temperature zone calculates the best adjustment scheme based on the target temperature set by the top temperature zone coordination controller and the temperature state of the current temperature zone, and adjusts the fan speed, heat pipe power, Micro LED power consumption and other heat dissipation methods in real time to ensure that the temperature of the temperature zone is stable within the target range. Due to the different thermal inertia of different temperature zones, the MPC controller performs adaptive adjustment according to the heat conduction characteristics. For example, in the high temperature zone, the MPC controller needs a faster response speed, while in the low temperature zone, a milder adjustment strategy is adopted to reduce the impact of temperature fluctuations on the display effect. The bottom-level execution controller is constructed, which is mainly composed of a heat pipe temperature controller and a fan speed controller, and a PID control algorithm is used to achieve specific temperature adjustment. The PID control algorithm is a feedback control method that adjusts the control variables in real time by performing proportional (P), integral (I) and differential (D) calculations on the error signal, so that the temperature control system can respond quickly to temperature changes and eliminate steady-state errors. For example, in high-temperature areas, the heat pipe temperature controller adjusts the heat transfer rate of the heat pipe according to the output signal of the MPC controller, while the fan speed controller adjusts the speed of the fan to enhance the air convection cooling effect. In actual applications, in order to improve control accuracy, the PID parameters are systematically identified and adaptively adjusted to ensure that the control effect under different ambient temperatures and load conditions is always maintained at the best state.In order to improve the stability of the system, the bottom-level execution controller combines fuzzy control strategies to enhance the adaptability to nonlinear temperature changes, so that it can still maintain good thermal management performance under complex working conditions. The top-level temperature zone coordination controller, the middle-level temperature zone independent controller and the bottom-level execution controller are cascaded, and temperature zone constraints are added at the connection to finally form a complete three-layer cascade control system. The purpose of the cascade connection is to ensure the seamless connection of temperature control from global strategy to local execution. The top-level coordination controller formulates the global strategy and assigns the target temperature to the middle-level temperature zone independent controller, while the middle-level controller further refines the control strategy and issues specific instructions to the bottom-level execution controller to ensure that the temperature control strategy can be accurately executed. The setting of temperature zone constraints is used to ensure that the temperature gradient between each temperature zone does not exceed the safety range and avoid drastic fluctuations in local temperature due to over-adjustment. During the operation of the entire system, the temperature zone coordination controller continuously monitors the operating status of each control layer and dynamically adjusts according to the feedback data to optimize the overall thermal management effect and improve the long-term stability and reliability of the Micro LED display module.

[0019] Step 400: Perform temperature balance control through a three-layer cascade control system to obtain temperature uniformity indicators and temperature buffer zone parameters for each temperature zone.

[0020] Specifically, the heat transfer channels between the high temperature zone, the medium temperature zone and the low temperature zone are modeled to construct a temperature zone thermal network model. The model is established based on the heat transfer equation, the heat diffusion equation and the heat dissipation characteristics of each zone. Through numerical analysis methods, combined with finite element analysis or computational fluid dynamics simulation, the heat exchange path, heat flow rate and temperature gradient between each temperature zone are determined. Especially for the high temperature zone, due to its large heat dissipation demand, it is necessary to describe the heat source distribution and heat transfer path in detail in the thermal network model to ensure that the model can accurately reflect the actual heat diffusion process. Based on the temperature zone thermal network model, the temperature balance target of the temperature zone is formulated to optimize the temperature distribution of the entire Micro LED module. The first temperature uniformity control target is set for the high temperature zone to maintain temperature uniformity while efficiently dissipating heat and avoid the occurrence of local overheating; at the same time, the second temperature uniformity control target is set for the medium temperature zone to ensure that the temperature distribution in this area is stable and acts as a thermal buffer layer between the high temperature zone and the low temperature zone to reduce the impact of excessive temperature gradients; the third temperature uniformity control target is set for the low temperature zone to ensure that while ensuring the heat dissipation capacity, it avoids the problem of too low temperature causing the Micro LED luminous efficiency to decrease or the performance fluctuation of electronic components. The setting of these goals needs to be combined with the working environment, power density, heat dissipation capacity and ambient temperature range of the Micro LED module to ensure that the temperature balance strategy can adapt to different working conditions. The temperature balance target of the temperature zone is input into the three-layer cascade control system, and the global heat dissipation power allocation is performed on the temperature zone thermal network model through the top-level temperature zone coordination controller. The role of the top-level controller is to comprehensively consider the thermal management requirements of the entire Micro LED module and calculate the optimal heat dissipation power allocation parameters for each temperature zone based on the temperature balance target of the temperature zone. The controller analyzes the heat flow distribution, evaluates the heat accumulation of each temperature zone, and dynamically adjusts the heat dissipation strategy based on real-time temperature data. The controller also needs to optimize the heat dissipation power allocation in combination with the ambient temperature change trend so that the Micro LED module can maintain stable operation under different environmental conditions. Based on the heat dissipation power allocation parameters of the temperature zone, the local radiator efficiency of the hot spot area is adjusted through the middle-level temperature zone independent controller, and the temperature gradient of the adjacent temperature zone is adjusted to obtain the temperature adjustment parameters of the temperature zone. The middle-level temperature zone independent controller dynamically adjusts the heat dissipation equipment of each temperature zone, such as increasing the fan speed or increasing the working efficiency of the heat pipe in the high temperature zone, and optimizing the heat conduction path through passive heat dissipation in the middle temperature zone. The controller adjusts the temperature difference between adjacent temperature zones through temperature gradient analysis to reduce the uneven display of Micro LEDs caused by sudden temperature changes. For example, if the temperature in a certain area changes too quickly, the middle-level controller adjusts the heat dissipation rate of the adjacent areas to make the temperature change more gradual, thereby improving the stability of the overall thermal management system. The temperature adjustment parameters of the temperature zones are set with a buffer zone to ensure the smoothness of the temperature change process and form the temperature buffer zone parameters.Temperature buffer zones are added between the high temperature zone and the medium temperature zone, and between the medium temperature zone and the low temperature zone, respectively, to reduce the temperature gradient between different temperature zones, thereby avoiding the influence of thermal stress concentration and temperature change on the Micro LED display effect. The setting of the temperature buffer zone is achieved by adjusting the heat transfer medium, optimizing the heat sink layout or using phase change materials to ensure that the heat transfer between different temperature zones is smoother. The specific parameters of the buffer zone need to be dynamically adjusted in combination with the temperature change rate, the performance of the heat dissipation equipment and the change of the ambient temperature to ensure that it can play an effective role under various working conditions. Based on the temperature buffer zone parameters, the temperature distribution within each temperature zone and between the temperature zones is calculated to obtain the temperature uniformity index. The statistical analysis method is used to evaluate the temperature balance of the entire Micro LED module by calculating the temperature standard deviation, the maximum temperature difference and the temperature gradient change rate of each temperature zone. The temperature distribution of the Micro LED module is visualized and analyzed by thermal imaging analysis technology to verify the effectiveness of the temperature control strategy. If the temperature uniformity index does not reach the expected target, the parameters of the control system are optimized and adjusted, such as adjusting the heat dissipation power allocation strategy, optimizing the local heat dissipation efficiency, or reconfiguring the temperature buffer zone parameters to further improve the temperature control accuracy of the system.

[0021] According to the temperature uniformity index, the energy consumption functions of the high temperature zone, the medium temperature zone and the low temperature zone are constructed respectively, and the energy consumption model of the high temperature zone, the medium temperature zone and the low temperature zone are obtained. The construction of the energy consumption function depends on the heat dissipation mode, power consumption characteristics and temperature regulation strategy of each temperature zone. For example, the high temperature zone usually requires stronger heat dissipation measures, so its energy consumption model considers the power consumption of fans, heat pipes and active cooling equipment, while the energy consumption model of the medium temperature zone is more related to the efficiency of passive heat dissipation and heat conduction, and the low temperature zone only needs to maintain the minimum energy consumption. When constructing the energy consumption function, the corresponding mathematical model is established in combination with the actual power consumption characteristics of the heat dissipation equipment. The energy consumption model of the high temperature zone, the medium temperature zone and the low temperature zone are adaptively weighted according to the ambient temperature to obtain the energy consumption weight coefficient of the temperature zone. The energy consumption proportion of each temperature zone is dynamically adjusted according to the change of ambient temperature. For example, at a higher ambient temperature, the heat dissipation demand of the high temperature zone is greater, so its energy consumption weight should be increased, while in a low temperature environment, the energy consumption of the medium temperature zone and the low temperature zone needs to be increased to prevent the temperature from being too low and affecting the performance of the Micro LED display. An adaptive weight calculation method based on ambient temperature is adopted. The historical temperature data and real-time ambient temperature data are used to calculate the optimal energy consumption allocation ratio of each temperature zone under different working conditions, and form a temperature zone energy consumption weight coefficient matrix, so that energy consumption optimization can be adaptively adjusted according to changes in the external environment. An energy consumption optimization function is constructed based on the temperature zone energy consumption weight coefficient, and the particle swarm is initialized through the optimization function to obtain the particle swarm initialization parameters. Particle swarm optimization is an efficient global optimization algorithm that quickly searches for the optimal solution in complex multi-objective optimization problems. Particle swarm optimization is used to optimize the balance between energy consumption and temperature control. The energy consumption optimization function is defined, and its goal is to minimize the overall energy consumption while satisfying the temperature balance constraint. The function includes three core parts, namely, the temperature control item, the energy consumption item, and the temperature gradient item. The temperature control item is used to ensure that the temperature distribution of the Micro LED display module meets the set target, the energy consumption item is used to minimize the power consumption of the heat dissipation equipment, and the temperature gradient item is used to constrain the temperature difference between different temperature zones to prevent local overheating or overcooling. After constructing the optimization function, the particle swarm is initialized, that is, the initial particles are generated in the search space, and their initial positions and speeds are set according to the temperature zone energy consumption weight coefficients, so as to ensure that the algorithm can quickly converge to the optimal solution. Based on the particle swarm initialization parameters, the weight coefficients of the temperature control items, energy consumption items, and temperature gradient items in the energy consumption optimization function are adjusted, and a dynamic balance analysis of temperature control is performed to obtain the balance parameters of temperature control and energy consumption. The weight coefficients of the temperature control items, energy consumption items, and temperature gradient items are adaptively adjusted to ensure the optimal temperature control strategy under different working conditions. For example, in high-load mode, the weight of the temperature control item is increased to ensure that the temperature does not exceed the safety range, while in low-load mode, the weight of the temperature control item is reduced and the weight of the energy consumption item is increased to minimize energy consumption.In the weight adjustment process, the influence of the temperature gradient term is considered to ensure that the temperature difference between different temperature zones is not too large, thereby avoiding structural damage or display performance degradation caused by thermal stress. A dynamic balance analysis of temperature control is performed to evaluate the temperature adjustment requirements of different temperature zones under various working conditions and optimize the energy consumption allocation strategy. This analysis process relies on real-time temperature data and historical temperature change trends. Through simulation calculations of different temperature control strategies, their impact on system energy consumption and temperature balance is evaluated. By continuously adjusting the weight parameters and combining the global search capability of the particle swarm optimization algorithm, the optimal temperature control and energy consumption balance parameters are finally obtained, so that the Micro LED display module can maintain a stable operating state under various environmental conditions and achieve the lowest energy consumption while ensuring temperature balance.

[0022] In the embodiment of the present application, the time-re-attention long short-term memory network model obtained through deep learning training can accurately predict the temperature change trend of each temperature zone, and improve the foresight and accuracy of temperature control. The temperature zone adaptive re-attention mechanism parameters can dynamically adjust the control weights of different temperature zones to achieve more accurate temperature management. The design of the three-layer cascade control system realizes coordinated control from global to local. The top-level temperature zone coordination controller ensures the rationality of the overall heat dissipation strategy, the middle-level temperature zone independent controller makes fine adjustments according to the characteristics of each temperature zone, and the bottom-level execution controller ensures the accurate execution of control instructions, forming a complete control closed loop. The establishment of the temperature zone thermal network model accurately describes the heat transfer relationship between each temperature zone and provides a theoretical basis for temperature balance control. By setting a temperature buffer zone, the temperature mutation phenomenon between adjacent temperature zones is effectively alleviated, and the temperature uniformity of the display module is improved. Based on the energy consumption optimization scheme of ambient temperature adaptation, the particle swarm algorithm is used to dynamically balance temperature control and energy consumption, which reduces the system energy consumption while ensuring the temperature control effect, and improves the economy of the thermal management system. The optimization design of the multi-task loss function comprehensively considers the temperature prediction accuracy, temperature gradient control and the rationality of attention distribution. The learning rate is adjusted through the cosine annealing strategy to improve the training effect and generalization ability of the model.

[0023] In a specific embodiment, the process of executing step 100 may specifically include the following steps: The temperature range of the Micro LED display module is divided into a high temperature zone, a medium temperature zone and a low temperature zone; Arrange sensors and collect temperature data in the high temperature area to obtain temperature data in the high temperature area; Arrange sensors and collect temperature data in the medium temperature zone to obtain temperature data in the medium temperature zone; Arrange sensors and collect temperature data in the low temperature zone to obtain temperature data in the low temperature zone; The temperature data of the high temperature zone, the temperature data of the medium temperature zone and the temperature data of the low temperature zone are input into the CAN bus for sampling and transmission to obtain the target temperature data; Based on the target temperature data, the temperature gradient data between adjacent temperature zones is calculated.

[0024] Specifically, the thermal distribution characteristics of the Micro LED display module are analyzed. The temperature distribution of the Micro LED display module is affected by current drive, luminous power, ambient temperature and heat dissipation design. An infrared thermal imager, thermocouple sensor or integrated temperature sensor is used to perform thermal analysis on the entire Micro LED module to obtain its temperature distribution. The method of setting the temperature division threshold is based on statistical methods, such as performing mean and variance analysis on the temperature data on the surface of the Micro LED display module to determine the division boundaries of different temperature ranges. Assuming that the mean of the temperature data is , the standard deviation is , the temperature ranges of the high temperature zone, medium temperature zone and low temperature zone are defined as follows: in, and is an empirical coefficient, which is adjusted according to the specific application scenario. Sensors are arranged in different temperature zones and the corresponding temperature data are collected. In the high temperature zone, due to the concentration of heat, the temperature changes rapidly, so high-density temperature sensors are arranged, such as thermocouple arrays or high-precision integrated temperature sensors, to achieve high temporal resolution and high spatial resolution data collection. Assume that the high temperature zone is arranged with The temperature data collected by the temperature sensor is expressed as: in, Indicates The high temperature zone sensor is Temperature data at the moment. Similarly, sensors in the medium temperature zone and low temperature zone collect data respectively: in, and The number of sensors in the medium temperature zone and the low temperature zone respectively. The collected temperature data is sampled and transmitted through the CAN bus to achieve efficient data transmission and processing. The CAN bus adopts a distributed data transmission protocol. The data packet format of each sensor includes the sensor ID, temperature value and timestamp. For example, the format of a temperature data packet is expressed as: Where ID represents the sensor number, Represents the temperature value, Represents the timestamp. After all temperature data are transmitted via the CAN bus, they are aggregated to the central processing unit for storage and calculation to obtain the target temperature data. : After obtaining the target temperature data, calculate the temperature gradient data between adjacent temperature zones to analyze the trend of heat transfer. Assume that the temperature gradient between the high temperature zone and the medium temperature zone is recorded as , and the temperature gradient between the medium temperature zone and the low temperature zone is recorded as , then it is calculated by numerical differentiation method: in, and They respectively represent the distance from the high temperature zone to the medium temperature zone and from the medium temperature zone to the low temperature zone. These distances are measured through the physical dimensions of the Micro LED module.

[0025] In a specific embodiment, the process of executing step 200 may specifically include the following steps: The target temperature data is divided into high temperature zone data, medium temperature zone data and low temperature zone data according to the temperature zone, and the temperature gradient data is feature separated to obtain the gradient feature between the temperature zones; Construct a temperature zone feature branch, which is used to extract features from high temperature zone data, medium temperature zone data, and low temperature zone data respectively. The temperature zone feature branch includes three parallel long short-term memory networks, each of which contains 256 hidden layer nodes. Construct a temperature gradient feature branch, which is used to extract the gradient features between temperature intervals. The temperature gradient feature branch is a convolutional neural network. Construct a time feature branch, which is used to extract time series features of target temperature data. The time feature branch is a bidirectional long short-term memory network, which contains two hidden states, forward and reverse. Attention fusion is performed on the temperature zone feature branch, temperature gradient feature branch and time feature branch to generate a multi-temperature zone coupling feature vector; The multi-temperature zone coupling feature vector is input into the long short-term memory decoder network with temperature zone adaptive gating mechanism for feature decoding. The features of different temperature zones are reconstructed and trained by dynamically adjusting the gating parameters to obtain the time-re-attention long short-term memory network model. Based on the temperature zone feature branch, temperature gradient feature branch and time feature branch, the attention weights of the temperature zone division information are calculated to obtain the temperature zone adaptive re-attention mechanism parameters.

[0026] Specifically, the target temperature data is divided into high temperature zone data, medium temperature zone data and low temperature zone data according to the temperature zone, and the temperature gradient data is feature separated to obtain the gradient feature between temperature zones. The target temperature data is collected by multiple sensors, each of which is detected at different times. The collected temperature data is expressed as ,in Represents the sensor number Represents the timestamp. Normalize the temperature data and classify it according to the preset temperature range. Assume that the mean of the temperature data is , the standard deviation is , then the high temperature area data satisfies , the data in the medium temperature zone meets , while the low temperature zone data meets ,in and is the coefficient for adjusting the range of different temperature zones. After completing the temperature zone division, calculate the temperature gradient between the temperature zones to extract the characteristics of heat transfer. Assume that the average temperature of the high temperature zone is The mean temperature in the medium temperature zone is The mean temperature of the low temperature zone is , and the distance from the high temperature zone to the medium temperature zone is , the distance from the medium temperature zone to the low temperature zone is , the temperature gradient is calculated as follows: These gradient data are used to measure the heat transfer between different temperature zones. A temperature zone feature branch is constructed to extract the temperature features of the high temperature zone, the medium temperature zone, and the low temperature zone. Since the temperature data has time series characteristics and the temperature distribution patterns of different temperature zones are different, three parallel long short-term memory (LSTM) networks are used for feature extraction, and each LSTM network processes the data of one temperature zone. Assume that the temperature data sequence of a certain temperature zone is expressed as , then the hidden state update equation of the LSTM network is as follows: in, Indicates time The hidden state of the moment, and are the weight matrices of the hidden layer and the input layer, respectively. is the bias, Represents the activation function. Each LSTM network contains 256 hidden layer nodes to ensure that the dynamic temperature change characteristics within the temperature zone are effectively captured. At the same time, a temperature gradient feature branch is constructed to extract the gradient characteristics between temperature zones. Since the gradient data is essentially spatially distributed, a convolutional neural network is used for feature extraction. The calculation of the convolutional layer is expressed as: in, represents the input temperature gradient data, represents the convolution kernel, * represents the convolution operation, represents the bias, is a nonlinear activation function. Through multi-layer convolution and pooling operations, the convolutional neural network learns the thermal diffusion pattern between different temperature zones. A time feature branch is constructed to extract the temporal characteristics of the temperature data. Since the unidirectional LSTM will lose some historical information, a bidirectional long short-term memory (Bi-LSTM) network is used for modeling. The network contains two hidden states, forward and reverse. The calculation method is as follows: The final hidden state is the concatenation of the forward and backward hidden states: The Bi-LSTM network effectively captures the long-term dependency of temperature data and improves the prediction ability of the model. The temperature zone feature branch, temperature gradient feature branch, and time feature branch are fused to generate a multi-temperature zone coupling feature vector. The calculation method of the attention mechanism is: in, is the attention weight matrix, is the attention weight at each time step, is the fused feature vector. Through this step, the model automatically adjusts the degree of attention to different temperature zones to improve the accuracy of prediction. The multi-temperature zone coupling feature vector is input into the LSTM decoder network with temperature zone adaptive gating mechanism for feature decoding. The function of the gating mechanism is to dynamically adjust the feature weights of different temperature zones to ensure the accuracy of temperature prediction. The calculation formula of the gating unit is as follows: in, is the gating factor, is the Sigmoid activation function, represents element-wise multiplication, is the decoded feature. By dynamically adjusting the gating parameters, adaptive learning of different temperature zone features is achieved to form a time-re-attention long-term memory network model. Based on the temperature zone feature branch, temperature gradient feature branch and time feature branch, the attention weight of the temperature zone division information is calculated to obtain the temperature zone adaptive re-attention mechanism parameters. These parameters are used to optimize the temperature control strategy so that the system can adaptively adjust the temperature management plan according to different working conditions.

[0027] In a specific embodiment, the execution step calculates the attention weights of the temperature zone division information based on the temperature zone feature branch, the temperature gradient feature branch, and the time feature branch to obtain the temperature zone adaptive re-attention mechanism parameters, which may specifically include the following steps: Assign weight coefficients to the temperature zone division information to obtain a temperature zone weight coefficient matrix; Perform a matrix transposition operation on the first output feature of the temperature zone feature branch and perform a matrix multiplication operation with the target temperature data to obtain a basic attention weight matrix; Perform Hadamard product operation on the temperature zone weight coefficient matrix and the basic attention weight matrix to obtain the temperature zone weighted attention matrix; Based on the second output feature of the temperature gradient feature branch, a temperature gradient threshold is set for the temperature interval gradient feature, and a temperature gradient correction matrix is ​​constructed; The temperature gradient correction matrix is ​​combined with the temperature zone weighted attention matrix to obtain the corrected attention weights; The multi-task loss function is optimized for the corrected attention weights. The multi-task loss function includes temperature prediction loss term, temperature gradient prediction loss term and attention distribution loss term. The learning rate is adjusted through the cosine annealing strategy to obtain the temperature zone adaptive re-attention mechanism parameters.

[0028] Specifically, the temperature zone division information is assigned weight coefficients to construct a temperature zone weight coefficient matrix. Assume that the temperature data of the high temperature zone, the medium temperature zone, and the low temperature zone are respectively expressed as , and , and its corresponding temperature distribution weight is , and , then the temperature zone weight coefficient matrix It is expressed as: Among them, each weight It is determined by the temperature characteristics, heat dissipation capacity and historical temperature fluctuation range of the temperature zone. For example, the weight of the high temperature zone is higher to ensure that the system can prioritize high temperature issues, while the weight of the low temperature zone is relatively low to reduce excessive intervention in the low temperature area. The first output feature of the temperature zone feature branch is matrix transposed and matrix multiplied with the target temperature data to obtain the basic attention weight matrix. Assume is the first output feature matrix of the temperature zone feature branch, is the target temperature data matrix, then the basic attention weight matrix The calculation is as follows: in, represents the transposed matrix of the temperature zone feature branch, and Represents the temperature data of each temperature zone at different time steps. The role of matrix multiplication is to calculate the relationship between the temperature zone characteristics and the target temperature data, and obtain the preliminary attention distribution. Perform Hadamard product operation on the temperature zone weight coefficient matrix and the basic attention weight matrix to obtain the temperature zone weighted attention matrix. Hadamard product is an element-by-element multiplication operation. Suppose the basic attention weight matrix The elements of , temperature zone weight coefficient matrix The elements of , then the temperature zone weighted attention matrix The calculation is as follows: in, represents the Hadamard product operation, that is The significance of this operation is to weight the basic attention distribution by temperature zone, so that the data in the high temperature zone has a higher weight in the calculation, while the data in the low temperature zone has a relatively lower weight, thereby ensuring that the temperature control strategy pays more attention to the temperature changes in the key areas. Based on the second output feature of the temperature gradient feature branch, a temperature gradient threshold is set for the temperature interval gradient feature to construct a temperature gradient correction matrix. Assume that the temperature gradient data It is calculated from the temperature change in each temperature range, namely: To prevent the temperature control system from over-regulating non-critical temperature zones, set the gradient threshold To construct the temperature gradient correction matrix , whose elements are calculated as follows: The purpose of this operation is to give higher weights to areas with drastic gradient changes, and lower weights to areas with smaller temperature changes, thereby improving the model's ability to regulate temperature in key areas. The temperature gradient correction matrix is ​​combined with the temperature zone weighted attention matrix to obtain the corrected attention weights. The calculation process is expressed as: The purpose of this operation is to optimize the attention distribution to make it more in line with the actual thermal management requirements. The multi-task loss function is optimized for the corrected attention weights to ensure the accuracy and stability of the temperature control strategy. Assume that the temperature prediction loss term The mean square error is used, that is: in, To predict the temperature value, is the true temperature value. Assume that the temperature gradient prediction loss term Using gradient error calculation: Adding attention distribution loss To ensure the stability of the attention mechanism, the loss term is calculated using KL divergence: The final total loss function Take the form of a weighted sum: in, , and are the weight coefficients of each loss term. During the training process, the cosine annealing strategy is used to dynamically adjust the learning rate : in, and are the minimum and maximum learning rates, respectively, is the total number of training steps, is the current step number. This strategy effectively improves the training stability and prevents falling into the local optimal solution. Through the optimization process, the parameters of the temperature zone adaptive re-attention mechanism are obtained, so that the temperature control system can more accurately adjust the thermal management strategy of the Micro LED display module and improve the temperature balance and energy efficiency.

[0029] In a specific embodiment, the process of executing step 300 may specifically include the following steps: The real-time temperature data of the Micro LED display module is input into the time re-attention long short-term memory network model for temperature prediction to obtain initial temperature prediction data, and the initial temperature prediction data is weighted based on the re-attention mechanism parameters to obtain the temperature zone temperature prediction data; A top-level temperature zone coordination controller is constructed based on the temperature prediction data of each temperature zone. The top-level temperature zone coordination controller is used to formulate a global heat dissipation strategy and calculate the target temperature of each temperature zone. An independent controller for the middle temperature zone is constructed for the high temperature zone, the middle temperature zone and the low temperature zone respectively, and the independent controller for the middle temperature zone adopts a model predictive control algorithm; Construct the bottom-level execution controller, which includes the heat pipe temperature controller and the fan speed controller. The bottom-level execution controller adopts the PID control algorithm. The top-level temperature zone coordination controller, the middle-level temperature zone independent controller and the bottom-level execution controller are cascaded, and temperature zone constraints are added at the connection points to obtain a three-layer cascade control system.

[0030] Specifically, the real-time temperature data of the Micro LED display module is input into the time-recorded long short-term memory (TAR-LSTM) network model for temperature prediction to obtain the initial temperature prediction data. Assume that the temperature data of the Micro LED display module is collected in real time by multiple sensors, and each time step The temperature data is expressed as ,in is the number of sensors, For the Sensors at time The collected temperature values. TAR-LSTM captures the time series features and uses the previous temperature data to predict future temperature values. Its hidden state update equation is as follows: in, For time The hidden state of the moment, and are the weight matrices of the hidden layer and the input layer, respectively. is the bias term, is the activation function. The output of the network As the initial temperature prediction data for the next moment. The initial temperature prediction data is weighted based on the re-attention mechanism parameters to obtain the temperature prediction data for the temperature zone. Suppose the weight matrix of the re-attention mechanism is , the weighted calculation is as follows: in, represents the weighted temperature prediction data of the temperature zone, It is used to adaptively adjust the temperature prediction data of different temperature zones to make the prediction more accurate. Based on the temperature prediction data of the temperature zones, a top-level temperature zone coordination controller is constructed. The function of this controller is to formulate a global heat dissipation strategy and calculate the target temperature of each temperature zone. .set up is the total heat dissipation power of the system, is the thermal resistance of each temperature zone, the target temperature calculation formula is as follows: in, is the ambient temperature, Dynamic adjustments are made through optimization algorithms to ensure that the temperature of each temperature zone is stable within a reasonable range. The top-level temperature zone coordination controller optimizes the global heat dissipation power distribution based on real-time temperature data and temperature prediction data to achieve uniform heat dissipation. After the top-level controller sets the target temperature, an independent middle-level temperature zone controller is constructed for the high-temperature zone, the medium-temperature zone, and the low-temperature zone. The controller uses a model predictive control algorithm for local temperature regulation. The model predictive control algorithm optimizes the control input at the current moment by predicting temperature changes in multiple time steps in the future. Assume that the temperature state vector is , the control input is , the optimization objective of the model predictive control algorithm is: in, To predict the time domain length, is the regularization coefficient of the control input to prevent over-adjustment. By solving the optimization problem, the optimal local temperature control strategy is obtained, so that the temperature of each temperature zone gradually approaches the target temperature. After the middle-level controller optimizes the heat dissipation strategy, the bottom-level execution controller actually performs the temperature control task. The bottom-level execution controller includes the heat pipe temperature controller and the fan speed controller, which are adjusted using the PID control algorithm. Assume that the actual temperature is , the error is , then the output control quantity of PID control is: in, , and are proportional, integral and differential gain coefficients respectively. The heat pipe temperature controller affects the heat conduction efficiency by adjusting the coolant flow rate or heat pipe power, while the fan speed controller adjusts the fan speed according to the PID calculation results to enhance the air convection heat dissipation capacity. The top temperature zone coordination controller, the middle temperature zone independent controller and the bottom execution controller are cascaded, and temperature zone constraints are added at the connection to build a complete three-layer cascade control system. is the temperature zone constraint matrix, and the constraints are as follows: This constraint ensures that the temperature difference between adjacent temperature zones is not too large to avoid thermal shock or local overheating.

[0031] Among them, cascading the top-layer temperature zone coordination controller, the middle-layer temperature zone independent controller and the bottom-layer execution controller also includes: setting adaptive duty cycle control parameters for each temperature zone, including: calculating the temperature change rate and temperature control error of each temperature zone based on real-time temperature data and predicted temperature data; establishing a temperature zone thermal inertia model according to the temperature change rate, and calculating the thermal response time constant of each temperature zone; setting a reference duty cycle period and a duty cycle adjustment range for each temperature zone based on the thermal response time constant; performing multi-objective optimization solution, including: establishing a multi-objective optimization function with temperature control accuracy, energy efficiency and temperature uniformity between temperature zones as the goals; setting the temperature fluctuation allowable range and energy consumption constraint conditions for each temperature zone; using an adaptive weight method to determine the dynamic weight coefficient of each optimization target; performing adaptive duty cycle control, including: dynamically adjusting the duty cycle value of each temperature zone according to the temperature control error; avoiding temperature overshoot by progressive duty cycle adjustment when the temperature of the temperature zone approaches the target value; coordinating the duty cycle switching timing of adjacent temperature zones based on the temperature gradient between temperature zones; establishing a temperature prediction compensation mechanism, including: using time to pay attention to long-term and short-term memory The memory network model predicts the temperature change trend; adjusts the duty cycle parameters in advance according to the prediction results to achieve advance compensation of temperature control; starts the rapid response mechanism when the temperature changes suddenly to temporarily increase the duty cycle adjustment frequency; builds an adaptive adjustment strategy, including: dynamically adjusts the response sensitivity of the duty cycle control based on the thermal characteristics of the temperature zone; sets the temperature fluctuation detection threshold to prevent frequent duty cycle adjustment; optimizes the overall energy consumption of the system under the premise of ensuring temperature uniformity; realizes multi-temperature zone collaborative control, including: establishes a heat transfer model and temperature coupling relationship between temperature zones; calculates the direction and intensity of heat transfer according to the temperature difference between the temperature zones; coordinates the duty cycle control rhythm of each temperature zone to maintain the stability of the temperature gradient; performs temperature overshoot suppression, including: sets the temperature change rate limit and the maximum adjustment step; reduces the duty cycle adjustment amplitude when the temperature approaches the target value; establishes a temperature overshoot warning mechanism and adjusts the control parameters in time; records and updates the control parameters, including: stores the duty cycle control effect and energy consumption data of each temperature zone; analyzes the relationship between temperature control accuracy and energy consumption; and regularly updates the optimization objectives and constraints of the duty cycle control strategy.

[0032] In a specific embodiment, the process of executing step 400 may specifically include the following steps: Model the heat transfer channels between the high temperature zone, the medium temperature zone and the low temperature zone to obtain the temperature zone thermal network model; According to the temperature zone thermal network model, the temperature balance target of the temperature zone is established, the first temperature uniformity control target is set for the high temperature zone, the second temperature uniformity control target is set for the medium temperature zone, and the third temperature uniformity control target is set for the low temperature zone; The temperature balance target of each temperature zone is input into the three-layer cascade control system, and the top-level temperature zone coordination controller is used to perform global heat dissipation power allocation on the temperature zone thermal network model to obtain the temperature zone heat dissipation power allocation parameters; Based on the heat dissipation power allocation parameters of the temperature zones, the local radiator efficiency of the hot spot area is adjusted through the independent controller of the middle temperature zone, and the temperature gradient of the adjacent temperature zones is adjusted to obtain the temperature adjustment parameters of the temperature zones; The buffer zone is set for the temperature adjustment parameters of the temperature zone, and the temperature buffer zone is added between the high temperature zone and the medium temperature zone, and between the medium temperature zone and the low temperature zone, respectively, to obtain the temperature buffer zone parameters; Based on the temperature buffer zone parameters, the temperature distribution within each temperature zone and between temperature zones is calculated to obtain the temperature uniformity index.

[0033] Specifically, the heat transfer channels between the high temperature zone, the medium temperature zone and the low temperature zone are modeled to establish a temperature zone thermal network model. During the operation of the Micro LED display module, the temperature field changes dynamically due to the uneven distribution of the heating elements and the different heat dissipation capabilities. It is assumed that the heat transfer of the Micro LED display module obeys Fourier's heat conduction law, that is, the heat flux It is proportional to the temperature gradient, and its mathematical expression is: in, is the thermal conductivity, is the temperature gradient, defined as in, , and are the average temperatures of the high temperature zone, medium temperature zone and low temperature zone respectively, and and They are the physical distances from the high temperature zone to the medium temperature zone, and from the medium temperature zone to the low temperature zone. After constructing the temperature zone thermal network model, set the temperature balance target of the temperature zone to ensure uniform temperature distribution in each temperature zone and avoid local overheating or overcooling. For the high temperature zone, its temperature uniformity target is mainly to reduce the maximum temperature deviation, that is: in, is the number of temperature sampling points in the high temperature area, For the Similarly, for the medium temperature zone and the low temperature zone, their equilibrium targets are defined as: make sure , and The temperature balance target of the temperature zone is input into the three-layer cascade control system, and the top-level temperature zone coordination controller is used to perform global heat dissipation power allocation on the temperature zone thermal network model to calculate the temperature zone heat dissipation power allocation parameters. Assume that the heat dissipation power distribution ratio of each temperature zone is , and , then the total heat dissipation power distribution is calculated as follows: in, ,and The top temperature zone coordination controller is used for dynamic adjustment to adapt to different working environments and power inputs. Based on the temperature zone heat dissipation power allocation parameters, the middle temperature zone independent controller is used to adjust the local radiator efficiency of the hot spot area and adjust the temperature gradient of the adjacent temperature zones to obtain the temperature adjustment parameters of the temperature zone. Among them, the local heat sink efficiency is calculated by the fan speed. and heat pipe power To make adjustments: in, To adjust the coefficient, optimize according to the system parameters. Adjust the heat dissipation parameters , so that the temperature changes in each temperature zone tend to be balanced, while ensuring that the system heat dissipation efficiency is maximized. Set a buffer zone for the temperature adjustment parameters of the temperature zone to reduce the impact of temperature fluctuations. The settings are as follows: in, and is the buffer zone ratio parameter, which is adjusted according to the experimental data to ensure a smooth transition of the temperature gradient between different temperature zones. Based on the temperature buffer zone parameters, the temperature distribution within each temperature zone and between temperature zones is calculated, and the temperature uniformity index is obtained. Assume that the temperature uniformity index The temperature standard deviation of each temperature zone is defined as follows: in, By calculating the temperature uniformity index, the effectiveness of the temperature control strategy is evaluated, and the heat dissipation power allocation parameters of the temperature zones are further optimized. And heat dissipation adjustment parameters , ultimately achieving efficient temperature balance control and ensuring that the Micro LED display module maintains stable operation under different environmental conditions.

[0034] In a specific embodiment, the thermal management method for a wide temperature range Micro LED display module further includes the following steps: Based on the temperature uniformity index, energy consumption functions are constructed for the high temperature zone, medium temperature zone and low temperature zone respectively, and the energy consumption model of the high temperature zone, the energy consumption model of the medium temperature zone and the energy consumption model of the low temperature zone are obtained; Performing ambient temperature adaptive weight allocation on the high temperature zone energy consumption model, the medium temperature zone energy consumption model and the low temperature zone energy consumption model to obtain the temperature zone energy consumption weight coefficient; An energy consumption optimization function is constructed based on the energy consumption weight coefficient of the temperature zone, and the particle swarm is initialized through the energy consumption optimization function to obtain the particle swarm initialization parameters; Based on the particle swarm initialization parameters, the weight coefficients of the temperature control term, energy consumption term and temperature gradient term in the energy consumption optimization function are adjusted and the dynamic balance analysis of temperature control is performed to obtain the balance parameters of temperature control and energy consumption.

[0035] Specifically, energy consumption functions are constructed according to the temperature uniformity indexes of the high temperature zone, the medium temperature zone, and the low temperature zone, respectively, to obtain the energy consumption model of each temperature zone. Assume that the temperature uniformity index of the high temperature zone is ,in Defined as the standard deviation of the temperature at each sampling point in the high temperature zone: in, It is the first The temperature of the sampling point, is the target temperature of the high temperature zone, and is the total number of sampling points in the high temperature zone. Similarly, the temperature uniformity indexes in the medium temperature zone and the low temperature zone are defined as: Based on these temperature uniformity indicators, energy consumption models for high temperature zone, medium temperature zone and low temperature zone are constructed respectively. Assume that the energy consumption function is composed of heat dissipation power And fan power consumption Composition, the energy consumption model of the high temperature zone is expressed as: in, is the heat dissipation power in the high temperature area, is the power consumption of the fan in the high temperature area, and and is the empirical weight coefficient. Similarly, the energy consumption model of the medium temperature zone and the low temperature zone is expressed as: In order to optimize energy consumption, the energy consumption model of each temperature zone is adaptively weighted according to the ambient temperature to obtain the energy consumption weight coefficient of the temperature zone. , then the temperature zone energy consumption weight coefficient It is expressed as: in, and To adjust the parameters, it is determined by experimental data. For example, in a high temperature environment, the energy consumption weight of the high temperature area is increased, while in a low temperature environment, the energy consumption weight of the high temperature area is reduced to optimize the overall heat dissipation strategy. The energy consumption weights of the high temperature area, medium temperature area and low temperature area are calculated as follows: Based on the energy consumption weight coefficient, an energy consumption optimization function is constructed, and the particle swarm is initialized through the optimization function to obtain the particle swarm initialization parameters. The goal of the energy consumption optimization function is to minimize the total energy consumption while maintaining temperature uniformity, which is defined as follows: in, is the overall temperature uniformity index, defined as: and Represents the overall temperature gradient change: In order to optimize the objective function, the particle swarm optimization is initialized to find the optimal solution by simulating the movement of particles in the search space. In time The location is , the speed is , then the update formula is as follows: in, is the inertia weight, is the learning factor, is a random number, For particles The individual optimal position of is the global optimal position. Through this optimization process, the optimal energy consumption allocation strategy is found to achieve the best balance between temperature balance and energy consumption. After the particle swarm is initialized, the weight coefficients of the temperature control term, energy consumption term and temperature gradient term in the energy consumption optimization function are adjusted based on the optimized particle swarm parameters, and the temperature control dynamic balance analysis is performed to obtain the balance parameters of temperature control and energy consumption. Suppose the final optimization objective function is: in, and After the particle swarm optimization, the parameters are adjusted to ensure that the optimization strategies of temperature balance and energy consumption can be dynamically adjusted under different ambient temperatures. Through the above optimization process, the optimal temperature control and energy consumption balance parameters are obtained, so that the Micro LED display module can maintain good temperature control effects under various working conditions, while reducing the overall energy consumption of the system and improving operating efficiency and stability.

[0036] Among them, the process of optimizing the energy consumption optimization function based on the particle swarm initialization parameters also includes: performing aggregation analysis on the energy consumption data of the temperature zones in different operating periods to establish a time series energy consumption prediction model; calculating the energy consumption probability distribution of each temperature zone based on the time series energy consumption prediction model to obtain the energy consumption fluctuation range; constructing a two-stage random optimization algorithm: in the first stage, setting energy consumption constraints based on the energy consumption fluctuation range, and calculating the energy consumption risk value of each temperature zone; in the second stage, adding the energy consumption risk value to the objective function, and performing probability constraint optimization on the energy consumption of the temperature zone; dynamically adjusting the search strategy of the particle swarm optimization algorithm according to the energy consumption risk value to achieve risk-adaptive energy consumption optimization; establishing a mapping relationship between the temperature zone scale and the energy consumption prediction accuracy, and automatically adjusting the prediction model parameters when the temperature zone area is greater than the preset threshold; and feeding back the optimized energy consumption control strategy to the three-layer cascade control system to achieve coordinated optimization of temperature control and energy consumption.

[0037] The thermal management method of the above-mentioned wide temperature range Micro LED display module also includes: based on the temperature balance target of the temperature zone, constructing a nonlinear differential algebraic equation model containing temperature dynamic characteristics, the nonlinear differential algebraic equation model includes a temperature state equation and a temperature algebraic constraint equation; linearizing the nonlinear differential algebraic equation model to obtain a local linearization model of the temperature zone node; constructing the temperature zone Lyapunov function based on the local linearization model, and calculating the Lyapunov exponent spectrum of the temperature zone node; converting the Lyapunov exponent spectrum into a temperature zone node stability index, and establishing a temperature zone node stability evaluation matrix; according to the temperature zone node stability evaluation matrix, the temperature control input of each temperature zone node is graded for stability to obtain the temperature zone node stability level; based on the temperature zone node stability level, a temperature control system stability constraint condition is established; the temperature control system stability constraint condition is added to the control algorithm of the three-layer cascade control system to obtain a temperature balance control strategy considering stability; the temperature balance control strategy considering stability is simulated and verified to obtain the stability guarantee range and temperature control limit of each temperature zone node.

[0038] The above describes the thermal management method of the wide temperature range Micro LED display module in the embodiment of the present application. The following describes the thermal management system 10 of the wide temperature range Micro LED display module in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the thermal management system 10 of the wide temperature range Micro LED display module includes: The data acquisition module 11 is used to divide the Micro LED display module into a high temperature zone, a medium temperature zone and a low temperature zone, and collect temperature data to obtain target temperature data and temperature gradient data; A training module 12 is used to perform deep learning training based on target temperature data and temperature gradient data to obtain a time re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; A construction module 13 is used to construct a three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; The balance control module 14 is used to perform temperature balance control through a three-layer cascade control system to obtain the temperature uniformity index and temperature buffer zone parameters of each temperature zone.

[0039] Through the collaboration of the above components, the time-re-attention long short-term memory network model obtained through deep learning training can accurately predict the temperature change trend of each temperature zone, improving the foresight and accuracy of temperature control. The temperature zone adaptive re-attention mechanism parameters can dynamically adjust the control weights of different temperature zones to achieve more accurate temperature management. The design of the three-layer cascade control system realizes coordinated control from global to local. The top-level temperature zone coordination controller ensures the rationality of the overall heat dissipation strategy, the middle-level temperature zone independent controller makes fine adjustments according to the characteristics of each temperature zone, and the bottom-level execution controller ensures the accurate execution of control instructions, forming a complete control closed loop. The establishment of the temperature zone thermal network model accurately describes the heat transfer relationship between each temperature zone and provides a theoretical basis for temperature balance control. By setting a temperature buffer zone, the temperature mutation phenomenon between adjacent temperature zones is effectively alleviated, and the temperature uniformity of the display module is improved. Based on the energy consumption optimization scheme of ambient temperature adaptation, the particle swarm algorithm is used to dynamically balance temperature control and energy consumption, which reduces system energy consumption while ensuring the temperature control effect, and improves the economy of the thermal management system. The optimization design of the multi-task loss function comprehensively considers the temperature prediction accuracy, temperature gradient control and the rationality of attention distribution. The learning rate is adjusted through the cosine annealing strategy to improve the training effect and generalization ability of the model.

[0040] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0041] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0042] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A thermal management method for a wide temperature range Micro LED display module, characterized in that: The method comprises: Divide the Micro LED display module into high temperature zone, medium temperature zone and low temperature zone, collect temperature data, and obtain target temperature data and temperature gradient data; Perform deep learning training based on the target temperature data and the temperature gradient data to obtain a time-re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; A three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller is constructed according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; Temperature balance control is performed through the three-layer cascade control system to obtain the temperature uniformity index and temperature buffer zone parameters of each temperature zone.

2. The thermal management method for a wide temperature range Micro LED display module according to claim 1, characterized in that: The method of dividing the Micro LED display module into a high temperature zone, a medium temperature zone and a low temperature zone, and collecting temperature data to obtain target temperature data and temperature gradient data includes: The temperature range of the Micro LED display module is divided into a high temperature zone, a medium temperature zone and a low temperature zone; Arrange sensors and collect temperature data in the high temperature zone to obtain temperature data of the high temperature zone; Arrange sensors and collect temperature data in the medium temperature zone to obtain temperature data of the medium temperature zone; Arranging sensors and collecting temperature data in the low temperature zone to obtain temperature data of the low temperature zone; Input the temperature data of the high temperature zone, the temperature data of the medium temperature zone and the temperature data of the low temperature zone into the CAN bus for sampling and transmission to obtain target temperature data; Based on the target temperature data, temperature gradient data between adjacent temperature zones are calculated.

3. The thermal management method of a wide temperature range Micro LED display module according to claim 1, characterized in that: The deep learning training is performed based on the target temperature data and the temperature gradient data to obtain the time re-attention long short-term memory network model and the temperature zone adaptive re-attention mechanism parameters, including: The target temperature data is divided into high temperature zone data, medium temperature zone data and low temperature zone data according to the temperature zone, and the temperature gradient data is feature separated to obtain the gradient feature between the temperature zones; Constructing a temperature zone feature branch, the temperature zone feature branch is used to extract features of the high temperature zone data, the medium temperature zone data and the low temperature zone data respectively, the temperature zone feature branch includes three parallel long short-term memory networks, each long short-term memory network includes 256 hidden layer nodes; Constructing a temperature gradient feature branch, wherein the temperature gradient feature branch is used to extract the feature of the temperature interval gradient feature, and the temperature gradient feature branch is a convolutional neural network; Constructing a time feature branch, the time feature branch is used to extract time series features of the target temperature data, the time feature branch is a bidirectional long short-term memory network, including two hidden states, forward and reverse; Attention fusion is performed on the temperature zone feature branch, the temperature gradient feature branch and the time feature branch to generate a multi-temperature zone coupling feature vector; Inputting the multi-temperature zone coupling feature vector into a long short-term memory decoder network with a temperature zone adaptive gating mechanism for feature decoding, reconstructing and training the features of different temperature zones by dynamically adjusting the gating parameters, and obtaining a time-re-attention long short-term memory network model; Attention weights are calculated for the temperature zone division information based on the temperature zone feature branch, the temperature gradient feature branch, and the time feature branch to obtain temperature zone adaptive re-attention mechanism parameters.

4. The thermal management method of a wide temperature range Micro LED display module according to claim 3, characterized in that: The attention weight calculation of the temperature zone division information based on the temperature zone feature branch, the temperature gradient feature branch and the time feature branch to obtain the temperature zone adaptive re-attention mechanism parameters includes: Assign weight coefficients to the temperature zone division information to obtain a temperature zone weight coefficient matrix; Performing a matrix transposition operation on the first output feature of the temperature zone feature branch and performing a matrix multiplication operation on the target temperature data to obtain a basic attention weight matrix; Performing a Hadamard product operation on the temperature zone weight coefficient matrix and the basic attention weight matrix to obtain a temperature zone weighted attention matrix; Based on the second output feature of the temperature gradient feature branch, setting a temperature gradient threshold for the temperature interval gradient feature, and constructing a temperature gradient correction matrix; Combining the temperature gradient correction matrix with the temperature zone weighted attention matrix to obtain a corrected attention weight; The modified attention weight is optimized by a multi-task loss function, wherein the multi-task loss function includes a temperature prediction loss term, a temperature gradient prediction loss term, and an attention distribution loss term. The learning rate is adjusted by a cosine annealing strategy to obtain temperature zone adaptive re-attention mechanism parameters.

5. The thermal management method of a wide temperature range Micro LED display module according to claim 1, characterized in that: The three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller is constructed according to the re-attention mechanism parameters and the time re-attention long short-term memory network model, including: Input the real-time temperature data of the Micro LED display module into the time re-attention long short-term memory network model to perform temperature prediction to obtain initial temperature prediction data, and perform weighted processing on the initial temperature prediction data based on the re-attention mechanism parameters to obtain temperature zone temperature prediction data; Constructing a top-level temperature zone coordination controller according to the temperature zone temperature prediction data, wherein the top-level temperature zone coordination controller is used to formulate a global heat dissipation strategy and calculate a target temperature for each temperature zone; constructing middle-layer temperature zone independent controllers for the high-temperature zone, the middle-temperature zone and the low-temperature zone respectively, wherein the middle-layer temperature zone independent controller adopts a model predictive control algorithm; Constructing a bottom-level execution controller, the bottom-level execution controller includes a heat pipe temperature controller and a fan speed controller, and the bottom-level execution controller adopts a PID control algorithm; The top-layer temperature zone coordination controller, the middle-layer temperature zone independent controller and the bottom-layer execution controller are cascade-connected, and temperature zone constraints are added at the connection points to obtain a three-layer cascade control system.

6. The thermal management method of a wide temperature range Micro LED display module according to claim 1, characterized in that: The temperature uniformity index and temperature buffer zone parameters of each temperature zone are obtained by performing temperature balance control through the three-layer cascade control system, including: Modeling the heat transfer channels between the high temperature zone, the medium temperature zone and the low temperature zone to obtain a temperature zone thermal network model; According to the temperature zone thermal network model, a temperature balance target for the temperature zone is established, a first temperature uniformity control target is set for the high temperature zone, a second temperature uniformity control target is set for the medium temperature zone, and a third temperature uniformity control target is set for the low temperature zone; The temperature balance target of the temperature zone is input into the three-layer cascade control system, and the global heat dissipation power allocation is performed on the temperature zone thermal network model through the top-level temperature zone coordination controller to obtain the temperature zone heat dissipation power allocation parameter; Based on the temperature zone heat dissipation power allocation parameters, the local radiator efficiency of the hot spot area is adjusted by the middle temperature zone independent controller, and the temperature gradient of the adjacent temperature zones is adjusted to obtain the temperature adjustment parameters of the temperature zone; Setting a buffer zone for the temperature adjustment parameters of the temperature zone, adding a temperature buffer zone between the high temperature zone and the medium temperature zone, and between the medium temperature zone and the low temperature zone, respectively, to obtain a temperature buffer zone parameter; Based on the temperature buffer zone parameters, the temperature distribution within each temperature zone and between temperature zones is calculated to obtain a temperature uniformity index.

7. The thermal management method of a wide temperature range Micro LED display module according to claim 6, characterized in that: The thermal management method of the wide temperature range Micro LED display module also includes: Based on the temperature uniformity index, energy consumption functions are constructed for the high temperature zone, the medium temperature zone and the low temperature zone respectively to obtain a high temperature zone energy consumption model, a medium temperature zone energy consumption model and a low temperature zone energy consumption model; Performing ambient temperature adaptive weight allocation on the high temperature zone energy consumption model, the medium temperature zone energy consumption model and the low temperature zone energy consumption model to obtain temperature zone energy consumption weight coefficients; Constructing an energy consumption optimization function based on the energy consumption weight coefficient of the temperature zone, and initializing the particle swarm through the energy consumption optimization function to obtain the particle swarm initialization parameters; Based on the particle swarm initialization parameters, the weight coefficients of the temperature control item, the energy consumption item and the temperature gradient item in the energy consumption optimization function are adjusted and the temperature control dynamic balance analysis is performed to obtain the balance parameters of the temperature control and the energy consumption.

8. A thermal management system for a wide temperature range Micro LED display module, characterized in that: A method for thermal management of a wide temperature range Micro LED display module according to any one of claims 1 to 7, the system comprising: A data acquisition module is used to divide the Micro LED display module into high temperature zone, medium temperature zone and low temperature zone, and collect temperature data to obtain target temperature data and temperature gradient data; A training module, used for performing deep learning training based on the target temperature data and the temperature gradient data to obtain a time re-attention long short-term memory network model and temperature zone adaptive re-attention mechanism parameters; A construction module, used to construct a three-layer cascade control system including a top-layer temperature zone coordination controller, a middle-layer temperature zone independent controller and a bottom-layer execution controller according to the re-attention mechanism parameters and the time re-attention long short-term memory network model; The balance control module is used to perform temperature balance control through the three-layer cascade control system to obtain the temperature uniformity index and temperature buffer zone parameters of each temperature zone.

Images