LED large screen heat dissipation method and system for achieving intelligent partition temperature control

Abstract

The invention provides an LED large screen heat dissipation method and system for achieving intelligent partition temperature control, and belongs to the technical field of LED large screen heat dissipation. The method comprises the steps that an LED large screen is divided into pixel units, and a micro thermoelectric refrigeration sheet and a micro-channel heat dissipation layer are arranged between an LED chip and a substrate of each pixel unit; dividing the LED large screen into partitions, collecting environment temperature and humidity values, brightness values of all the partitions and real-time temperature values, and obtaining temperature prediction values of all the partitions by using an LSTM (Long Short Term Memory) model; and the target temperature of each subarea is calculated by combining the real-time temperature value and the temperature prediction value of each subarea, and the power of the micro thermoelectric refrigeration sheet is controlled according to the target temperature of each subarea. Intelligent partition temperature control heat dissipation of the large LED screen is achieved, the heat dissipation power can be dynamically adjusted according to the actual conditions of different partitions, the situation of excessive heat dissipation or insufficient heat dissipation is avoided, and the heat dissipation structure is optimized by adopting the mode of combining the micro thermoelectric refrigeration sheet and the micro-channel heat dissipation layer.

Description

Technical Field

[0001] This application belongs to the technical field of LED large screen heat dissipation, and specifically relates to an LED large screen heat dissipation method and system that realizes intelligent partition temperature control. Background Technique

[0002] LED large screens are widely used in many fields, such as vehicle-mounted screens, medical equipment, etc. Since LED large screens generate a large amount of heat during operation, if heat cannot be dissipated in a timely and effective manner, it will seriously affect their performance and service life.

[0003] The heat dissipation of LED large screens has always been a difficult problem to solve and is a technical bottleneck restricting the development of LED large screens. The existing heat dissipation of LED large screens is mainly divided into two methods: passive heat dissipation and active heat dissipation. Passive heat dissipation usually relies on metal heat sinks, and then uses fans to dissipate heat from the metal heat sinks, but there are many limitations. On the one hand, both the heat sink and the fan require a large amount of space, and the space requirements for enclosed scenarios such as vehicle-mounted screens and medical equipment are relatively strict, making it difficult to meet safety requirements. On the other hand, for transient high-heat generation scenarios, such as when the LED is fully white, passive heat dissipation cannot respond in a timely manner, easily causing the chip junction temperature to exceed the safety threshold, which in turn leads to problems such as screen color deviation, brightness attenuation, and even dead lights.

[0004] For active heat dissipation, one is to directly perform forced convection through a cooling fan. Although the heat dissipation efficiency is improved to a certain extent, the noise problem is relatively prominent. Since the fan speed is positively correlated with the heat dissipation demand, it is difficult to balance heat dissipation and noise control in enclosed spaces such as conference rooms and vehicle-mounted environments. Excessive noise will have a serious impact on the user experience. The other is thermoelectric cooling. Although the heat dissipation effect is better than that of the cooling fan, it requires an external pump valve system, which not only increases the overall failure rate of the LED large screen system but also significantly increases the cost. Compared with the traditional heat dissipation solution, the cost increases by more than 200%. Summary of the Invention

[0005] In a first aspect, an embodiment of the present application provides an LED large screen heat dissipation method that realizes intelligent partition temperature control, including the following steps: S1. Divide the pixel units of the LED large screen, and set a micro thermoelectric cooler and a microchannel heat dissipation layer between the LED chips and the substrate of each pixel unit; S2. Divide the LED large screen into partitions, collect the ambient temperature and humidity values, the brightness values of each partition, and the real-time temperature values of each partition, and use the LSTM model to obtain the temperature prediction values of each partition; S3. Calculate the target temperature of each partition by combining the real-time temperature value and the temperature prediction value of each partition, and control the power of the micro thermoelectric cooler according to the target temperature of each partition.

[0006] Further, the specific steps of step S1 are as follows: S11. Determine the size of the pixel unit and divide the LED large screen into several pixel units; S12. Configure a micro thermoelectric cooler and a microchannel heat dissipation layer for each pixel unit.

[0007] Further, the specific steps of step S2 are as follows: S21. Determine the partition size of the LED large screen, divide the pixel units of the LED large screen according to the partition size, and obtain several partitions; S22. Set a temperature sensor for each partition on the back of the substrate; S23. Collect the historical working parameters of the LED large screen, unify the historical working parameters on the time axis, and construct a data set; the historical working parameters include historical ambient temperature and humidity values, historical brightness values of each partition, and historical temperature values of each partition; S24. Use the data set to train the LSTM model to obtain a temperature prediction model for LED large screen partitions; S25. Collect the real-time working parameters of the LED large screen, and input the real-time working parameters into the temperature prediction model for LED large screen partitions to obtain the temperature prediction values of each partition within a set time period; the real-time working parameters include real-time ambient temperature and humidity values, real-time brightness values of each partition, and real-time temperature values of each partition.

[0008] Further, the specific steps of step S24 are as follows: S241. Determine the input features from the data set; S242. Determine the output labels from the data set; S243. Divide the historical time series numbers in the data set into several sliding windows, and set the input data with n time steps and the output labels with k time steps in each window; S244. Construct an LSTM model, set the input layer of the LSTM model to receive historical ambient temperature and humidity values, historical brightness values, and historical temperature values, set the LSTM layer to capture the long-term dependencies in the time series, and set the fully connected layer to map the output of the LSTM layer to the temperature values at the next k moments; S245. Use the mean square error to construct a loss function for measuring the predicted temperature value and the actual temperature value of the LSTM model; S246. Use the data of each partition in the data set as shared data to iteratively train the LSTM model, and use the constructed loss function to adjust the LSTM model during the training process until the loss function is less than the set threshold or the maximum number of iterations is reached, and the training is completed to obtain a temperature prediction model for LED large screen partitions.

[0009] Further, the specific steps of step S3 are as follows: S31. Set a dynamic weight coefficient with the real-time temperature value as the main factor and the temperature prediction value as the auxiliary factor, and use the dynamic weight coefficient, the real-time temperature value, and the temperature prediction value of each zone to calculate the target temperature of each zone; S32. Divide the priority levels of the LED large screen according to the target temperature of each zone; If the target temperature of a zone is higher than the temperature upper limit threshold, mark the corresponding zone as high priority; If the target temperature of a zone is less than or equal to the temperature upper limit threshold and greater than or equal to the temperature lower limit threshold, mark the corresponding zone as medium priority; If the target temperature of a zone is lower than the temperature lower limit threshold, mark the corresponding zone as low priority; S33. Control the micro thermoelectric cooler of the high-priority zone to work forward, and increase the power of the micro thermoelectric cooler to the first set proportion of the maximum power; Control the micro thermoelectric cooler of the medium-priority zone to work forward, and increase the power of the micro thermoelectric cooler in a gradient manner according to the set proportion range of the maximum power; For the low-priority zone, determine whether the target temperature is lower than the temperature lower limit value; If not, control the corresponding micro thermoelectric cooler to work forward, and increase the power of the micro thermoelectric cooler to the second set proportion of the maximum power; If so, control the corresponding micro thermoelectric cooler to work in reverse.

[0010] Further, the following steps are also included: S34. Monitor the working states of the micro thermoelectric coolers of the LED large screen; If all the micro thermoelectric coolers are normal, return to step S25; If there is a faulty micro thermoelectric cooler, enter step S35; S35. Determine the zone to which the faulty micro thermoelectric cooler belongs, and increase the power of the normal micro thermoelectric coolers in this zone.

[0011] Further, in step S33, the specific steps of controlling the micro thermoelectric cooler of the medium-priority zone to work forward and increasing the power of the micro thermoelectric cooler in a gradient manner according to the set proportion range of the maximum power are as follows: SS1. Obtain the set temperature range [T1, T2] of the medium-priority zone; SS2. Obtain that the target power of the micro thermoelectric cooler in the medium-priority zone is within the set proportion range of the maximum power as [P1, P2]; SS3. Map the set ratio range [P1, P2] of the maximum power through the linear interpolation method to the set temperature range [T1, T2]; SS4. Calculate the ratio P of the target power to the maximum power for the medium-priority partition temperature value T through the following formula; 。

[0012] In a second aspect, the embodiments of the present application further provide an LED large-screen cooling system for realizing intelligent partition temperature control, including an LED large screen and a large-screen controller; The LED large screen includes a substrate and LED chips; The LED chips are divided into several pixel units, and a micro thermoelectric cooler and a microchannel heat dissipation layer are arranged between each pixel unit and the front of the substrate; A temperature sensor is arranged on the back of the substrate corresponding to each pixel unit; The large-screen controller collects the real-time temperature value of each pixel unit through the temperature sensor, predicts the temperature of each pixel unit through a pre-configured LED large-screen partition temperature prediction model, and then combines the real-time temperature value and the predicted temperature value of each pixel unit to calculate the target temperature value; The large-screen controller controls the power of the micro thermoelectric cooler of each pixel unit through a power control circuit according to the target temperature value of each pixel unit.

[0013] Further, the micro thermoelectric cooler is provided with a cold end and a hot end; The cold end is attached to the LED chip of the corresponding pixel unit, the hot end is attached to the microchannel heat dissipation layer, and the other end of the microchannel heat dissipation layer is attached to the front of the substrate.

[0014] Further, the power control circuit includes MOS transistor Q1, MOS transistor Q2, MOS transistor Q3, and MOS transistor Q4; The gates of MOS transistor Q1, MOS transistor Q2, MOS transistor Q3, and MOS transistor Q4 are connected to the large-screen controller M1; The drain of MOS transistor Q1 is connected to the power supply VCC, the source of MOS transistor Q1 is connected to the hot end of the micro thermoelectric cooler, the cold end of the micro thermoelectric cooler is connected to the drain of MOS transistor Q4, and the source of MOS transistor Q4 is grounded; The drain of MOS transistor Q3 is connected to the power supply VCC, the source of MOS transistor Q3 is connected to the cold end of the micro thermoelectric cooler, the hot end of the micro thermoelectric cooler is also connected to the drain of MOS transistor Q2, and the source of MOS transistor Q2 is grounded.

[0015] From the above technical solutions, it can be seen that the present application has the following advantages: In the LED large screen heat dissipation method and system for realizing intelligent partition temperature control provided by this application, intelligent partition temperature control heat dissipation of the LED large screen is realized. Through accurate temperature prediction and targeted power control, the accuracy and efficiency of heat dissipation are improved; the heat dissipation power can be dynamically adjusted according to the actual conditions of different partitions, avoiding the situations of overheat dissipation or insufficient heat dissipation, enhancing the stability and reliability of the LED large screen, and prolonging the service life of the large screen; the combination of a micro thermoelectric cooler and a microchannel heat dissipation layer is adopted to optimize the heat dissipation structure, realizing the light weight and miniaturization of the LED large screen while ensuring the heat dissipation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of this application, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic flow chart of the LED large screen heat dissipation method for realizing intelligent partition temperature control of the present invention.

[0018] Figure 2 It is a schematic diagram of the system for realizing intelligent partition temperature control of the LED large screen heat dissipation of the present invention.

[0019] Figure 3 It is a schematic circuit diagram of the power control circuit of the present invention.

[0020] Among them, 1 - large screen controller; 2 - substrate; 3 - LED chip; 4 - microchannel heat dissipation layer; 5 - temperature sensor; TEC - micro thermoelectric cooler. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In the following, the specific steps of the LED large screen heat dissipation method for realizing intelligent partition temperature control will be described in detail, and various embodiments of the present disclosure will be described more comprehensively. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents and / or alternative solutions falling within the spirit and scope of the various embodiments of the present disclosure.

[0022] Exemplarily, LED large screens have been widely used in multiple fields such as vehicle-mounted screens and medical devices. However, during the working process, the LED large screen will generate a large amount of heat. If the heat dissipation is not timely, its performance and service life will be seriously affected. The heat dissipation problem has always been a technical bottleneck in the development of LED large screens.

[0023] At present, the heat dissipation methods of LED large screens are mainly divided into two types: passive heat dissipation and active heat dissipation. Passive heat dissipation usually relies on metal heat sinks and dissipates heat from the heat sinks through fans. However, this method has many limitations. On the one hand, the heat sinks and fans occupy a large space and it is difficult to meet the stringent space requirements of enclosed scenarios such as in-vehicle screens and medical equipment. On the other hand, in transient high-heat generation scenarios (such as when the LED is fully white), passive heat dissipation cannot respond in time, easily leading to the chip junction temperature exceeding the safety threshold, which in turn causes problems such as color deviation of the screen, brightness attenuation, and even dead lights.

[0024] In the active heat dissipation method, one is to directly perform forced convection through a cooling fan. Although the heat dissipation efficiency is improved to a certain extent, the noise problem is relatively prominent. Since the fan speed is positively correlated with the heat dissipation demand, it is difficult to balance heat dissipation and noise control in enclosed spaces such as conference rooms and in-vehicle environments. Excessive noise will have a serious impact on the user experience. The other is thermoelectric cooling, whose heat dissipation effect is better than that of the cooling fan, but it requires an external pump valve system. This not only increases the overall failure rate of the LED large screen system but also significantly increases the cost, with the cost increasing by more than 200% compared to the traditional heat dissipation solution.

[0025] To address the above problems, this embodiment provides an LED large screen heat dissipation method that realizes intelligent partition temperature control. By dividing pixel units and partitions, configuring micro thermoelectric cooling chips and microchannel heat dissipation layers, combining the LSTM model to predict temperature, and dynamically adjusting the power of the micro thermoelectric cooling chips, efficient, precise, and intelligent partition temperature control is achieved, which is applicable to enclosed scenarios such as in-vehicle screens and medical equipment.

[0026] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0027] Please refer to Figure 1 The following is a flowchart of an LED large screen heat dissipation method that realizes intelligent partition temperature control in a specific embodiment. The method includes the following steps: S1. Set a micro thermoelectric cooling chip and a microchannel heat dissipation layer between the LED chips and the substrate of the LED large screen; It should be noted that by dividing pixel units and configuring micro thermoelectric cooling chips and microchannel heat dissipation layers, efficient heat dissipation is achieved and space occupation is reduced; S2. Divide the LED large screen into partitions, collect the ambient temperature value, the brightness value of each partition, and the real-time temperature value of each partition and input them into the LSTM model to obtain the temperature prediction value of each partition; It should be noted that by collecting the environmental temperature and humidity, brightness value and real-time temperature value, and combining with the LSTM model to predict the future temperature, data support is provided for temperature control; S3. Calculate the target temperature of each zone by combining the real-time temperature value and the temperature prediction value of each zone, and control the power of the micro thermoelectric cooler according to the target temperature of each zone; It should be noted that by calculating the target temperature by combining the real-time temperature and the predicted temperature, the power of the micro thermoelectric cooler is dynamically adjusted to achieve precise temperature control.

[0028] In this embodiment, by dividing pixel units and zones, and combining the LSTM model to predict the temperature, precise temperature control is achieved, and the heat dissipation efficiency and response speed are improved.

[0029] Furthermore, as a refinement and extension of the specific implementation manner of the above embodiment, in order to fully illustrate the specific implementation process in this embodiment, another method for cooling an LED large screen with intelligent zone temperature control is provided. The method includes the following steps: S1. Set a micro thermoelectric cooler and a microchannel heat dissipation layer between the LED chips and the substrate of the LED large screen; The specific steps of step S1 are as follows: S11. Determine the size of the pixel unit, and divide the LED large screen into several pixel units; Exemplarily, select a 6×6 LED chip as the size of the pixel unit; S12. Configure a micro thermoelectric cooler and a microchannel heat dissipation layer for each pixel unit; Specifically, set the cold end of the micro thermoelectric cooler to be in contact with each LED chip in the zone, and set the hot end of the micro thermoelectric cooler to be in contact with the front end face of the micro microchannel heat dissipation layer, and then set the rear end face of the microchannel heat dissipation layer to be in contact with the substrate; It should be noted that the cold end of the micro thermoelectric cooler is pasted to each LED chip in the zone through nano silver glue; The thickness of the nano silver glue is less than 0.01 mm; Exemplarily, the micro thermoelectric cooler refers to a 3 mm×3 mm thermoelectric cooler, and the thickness of the microchannel heat dissipation layer is 0.5 mm; The flow heat dissipation layer adopts a honeycomb structure; S2. Divide the LED large screen into zones, collect the environmental temperature value, the brightness value of each zone, and the real-time temperature value of each zone, and use the LSTM model to obtain the temperature prediction value of each zone; The specific steps of step S2 are as follows: S21. Determine the zone size of the LED large screen, divide the pixel units of the LED large screen according to the zone size, and obtain several zones; Exemplarily, the zone size can be set to 10 cm; S22. Set a temperature sensor for each partition on the back of the substrate; Specifically, drill holes at the center position corresponding to each partition on the back of the substrate, and set the temperature sensor inside the holes; Exemplarily, the temperature sensor can adopt a digital temperature sensor of the DS18B20 model, with an accuracy of ±0.5°C; S23. Collect the historical working parameters of the LED large screen, unify the historical working parameters on the time axis, and construct a data set; The historical working parameters include historical environmental temperature and humidity values, historical brightness values of each partition, and historical temperature values of each partition; S24. Use the data set to train the LSTM model to obtain a temperature prediction model for LED large screen partitions; It should be noted that convert the temperature prediction model for LED large screen partitions to the INT8 format and burn it into the large screen controller; The large screen controller adopts ARM Cortex-M7; S25. Collect the real-time working parameters of the LED large screen, and input the real-time working parameters into the temperature prediction model for LED large screen partitions to obtain the temperature prediction values of each partition within a set time period; The real-time working parameters include real-time environmental temperature and humidity values, real-time brightness values of each partition, and real-time temperature values of each partition; It should be noted that the historical and real-time environmental temperature and humidity values can be obtained through a thermometer and a hygrometer; The historical and real-time brightness values of each partition can be obtained by collecting the current values of the LED chips in the partition, or can be collected by arranging optical sensors on the front of the partition; The historical or real-time temperature values of each partition are collected by the temperature sensors arranged on the back of the partition substrate; S3. Calculate the target temperature of each partition by combining the real-time temperature value and the temperature prediction value of each partition, and control the power of the micro thermoelectric cooler according to the target temperature of each partition; The specific steps of step S3 are as follows: S31. Take the real-time temperature value as the main, and the temperature prediction value as the auxiliary, set a dynamic weight coefficient, and use the dynamic weight coefficient, the real-time temperature value and the temperature prediction value of each partition to calculate the target temperature of each partition; Among them, T 实时 is the real-time temperature value of the partition, T 预测 is the temperature prediction value of the partition, T 目标 represents the target temperature of the partition, α is the dynamic weight coefficient, α>0.5, for example, α = 0.7; S32. Divide the LED large screen into priorities according to the target temperature of each partition; If the target temperature of the partition is higher than the temperature upper limit threshold, mark the corresponding partition as high priority; If the target temperature of a partition is less than or equal to the temperature upper limit threshold and greater than or equal to the temperature lower limit threshold, mark the corresponding partition with high priority; If the target temperature of a partition is less than the temperature lower limit threshold, mark the corresponding partition with low priority; Exemplarily, the temperature upper limit threshold is set to 50 °C and the temperature lower limit threshold is set to 30 °C; S33. Control the micro thermoelectric cooler of the high-priority partition to work forward, and increase the power of the micro thermoelectric cooler to the first set ratio of the maximum power; Control the micro thermoelectric cooler of the medium-priority partition to work forward, and increase the power of the micro thermoelectric cooler in a gradient manner according to the set ratio range of the maximum power; It should be noted that the first set ratio is higher than the set ratio range; For the low-priority partition, judge whether the target temperature is less than the temperature lower limit value; If not, control the corresponding micro thermoelectric cooler to work forward, and increase the power of the micro thermoelectric cooler to the second set ratio of the maximum power; It should be noted that the second set ratio is lower than the set ratio range; If so, control the corresponding micro thermoelectric cooler to work in reverse; Specifically, the reverse power of the micro thermoelectric cooler is controlled according to the target temperature; for the partition with high priority in the high-temperature area, increase the PWM signal duty cycle to increase the power of the micro thermoelectric cooler; for the partition with low priority in the low-temperature area, reduce the PWM signal duty cycle or switch the current direction for heating; Exemplarily, the temperature lower limit value is 0 °C, the first set ratio is selected from 80% - 100%, such as 90%; the set ratio range is 40% - 70%; the second set ratio is selected from 10% - 20%, such as 15%; the reverse power of the micro thermoelectric cooler is set to 40% - 50% of the maximum power; The large-screen controller is configured with a CAN bus communication protocol. The large-screen controller collects the real-time working parameters of each partition through the CAN bus communication protocol, and sends the power increase instruction of each partition to the micro thermoelectric cooler of the corresponding partition through the CAN bus communication protocol.

[0030] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to completely illustrate the specific implementation process in this embodiment, another method for realizing intelligent partition temperature control of an LED large screen is provided. The following steps are further included in step S3 of this method: S34. Monitor the working states of the micro thermoelectric coolers of the LED large screen; If all the micro thermoelectric coolers are normal, return to step S25; If there is a faulty micro thermoelectric cooler, go to step S35; S35. Determine the partition to which the faulty micro thermoelectric cooler belongs, and increase the power of the normal micro thermoelectric cooling fins in this partition; It should be noted that since the faulty micro cooler is not necessarily located at the center of the partition, as an alternative solution, calculate the distances between each micro thermoelectric cooler and the faulty micro thermoelectric cooler, and increase the power of the micro thermoelectric coolers with distances less than the set range; Another alternative solution is to increase the power of the micro thermoelectric coolers in the adjacent partitions of the partition to which the faulty micro cooler belongs.

[0031] In an embodiment of the present invention, based on steps S23 and S24, a possible embodiment will be given below to non - restrictively elaborate on its specific implementation scheme.

[0032] The specific steps of step S23 are as follows: S231. Collect the historical ambient temperature and humidity values, historical brightness values, and historical temperature values of the LED large - screen in each partition according to the set collection period as historical working parameters; S232. After pre - processing various historical working parameters, align them according to the time stamp and store them as historical time - series numbers in chronological order; Specifically, the pre - processing process includes data cleaning, such as removing outliers and filling missing values, and normalization processing; S233. Use the historical time - series numbers to construct a data set; The specific steps of step S24 are as follows: S241. Determine the input features from the data set; Ambient temperature and humidity: E 温度 、E 湿度 ; Partition brightness value: B （i） , representing the brightness value of the i - th partition; Historical temperature value: , representing the n - time historical temperature values of partition i; S242. Determine the output labels from the data set; Future temperature value: , representing the temperature values of the future k time - steps of partition i, such as 10s; S243. Divide the historical time - series numbers in the data set into several sliding windows, and set that each window contains input data of n time - steps and output labels of k time - steps; S244. Construct an LSTM model. Set the input layer of the LSTM model to receive historical environmental temperature and humidity values, historical brightness values, and historical temperature values. Set the LSTM layer to capture long-term dependencies in the time series. Set the fully connected layer to map the output of the LSTM layer to the temperature values at the next k time steps. Specifically, the input data is represented as follows:

[0033] Where, and Environmental temperature and humidity, B （i） represents the brightness value of the i-th partition, represents the historical temperature value of partition i at time step t - k; The output of the LSTM model is represented as follows:

[0034] Where, is the temperature prediction value of partition i at the k th second in the future; S245. Use the mean squared error to construct a loss function for measuring the predicted temperature value and the actual temperature value of the LSTM model.

[0035] Where, N is the number of samples, k is the number of prediction time steps, is the actual temperature value of partition i at the j-th second in the future, is the predicted temperature value of partition i at the j-th second in the future; S246. Use the data of each partition in the dataset as shared data to iteratively train the LSTM model, and use the constructed loss function to adjust the LSTM model during the training process until the loss function is less than the set threshold or the maximum number of iterations is reached, completing the training to obtain the LED large screen partition temperature prediction model. Specifically, calculate the gradient of the loss function with respect to the model parameters through the backpropagation algorithm, and use the gradient descent method to update the parameters:

[0036] Where, θ is the parameter of the LSTM model, η is the learning rate, is the gradient of the loss function L with respect to the parameter θ.

[0037] In an embodiment of the present invention, based on step S33, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation.

[0038] In step S33, the specific steps for controlling the micro thermoelectric cooler in the medium-priority partition to operate in the forward direction and gradually increasing the power of the micro thermoelectric cooler within the set proportional range of the maximum power are as follows: SS1. Obtain the set temperature range [T1, T2] of the medium-priority partition; SS2. Obtain that the target power of the micro thermoelectric cooler in the medium-priority partition is within the set proportional range of the maximum power as [P1, P2]; SS3. Map the set temperature range [T1, T2] to the set proportional range [P1, P2] of the maximum power through linear interpolation; SS4. Calculate the ratio P of the target power to the maximum power by the following formula for the temperature value T of the medium-priority partition; .

[0039] Through the implementation of the present invention, through the improvement of heat dissipation performance, at the same power consumption, the junction temperature of the LED chip is reduced by 18 - 25 °C, and the service life is extended by 3 times; by replacing the heat dissipation fan with the heat dissipation structure of the micro thermoelectric cooler and the microchannel heat dissipation layer, the space occupied by the heat dissipation system can be reduced, and the thickness of the heat dissipation system is compressed from 50 mm in the traditional solution to 25 mm, which is suitable for in-vehicle embedded installation; through the large-screen controller, the response speed is optimized, and it only takes 0.1 second from detecting abnormal temperature to starting refrigeration, and the temperature fluctuation is controlled within ±2 °C; it has been verified that when playing 4K videos continuously for 8 hours at an ambient temperature of 60 °C, the highest temperature of the screen is stable at 48 °C; and compared with the traditional fan heat dissipation solution, the power consumption is reduced from 120 W to 72 W.

[0040] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0041] As Figure 2 shown, the following is an embodiment of an LED large-screen heat dissipation system for realizing intelligent partition temperature control provided by the embodiments of the present disclosure. This system and the method for realizing intelligent partition temperature control of an LED large screen in the above embodiments belong to the same inventive concept. For the details not described in detail in the embodiments of the LED large-screen heat dissipation system for realizing intelligent partition temperature control, reference can be made to the embodiments of the method for realizing intelligent partition temperature control of an LED large screen.

[0042] The system includes an LED large screen and a large-screen controller 1; The LED large screen includes a substrate 2 and an LED chip 3; The LED chip 3 is divided into several pixel units, and a micro thermoelectric cooler TEC and a microchannel heat dissipation layer 4 are provided between each pixel unit and the front of the substrate 2; A temperature sensor 5 is provided on the back of the substrate 2 corresponding to each pixel unit; The large-screen controller 1 collects the real-time temperature value of each pixel unit through the temperature sensor 5, predicts the temperature of each pixel unit through a pre-configured LED large-screen partition temperature prediction model, and then combines the real-time temperature value and the predicted temperature value of each pixel unit to calculate the target temperature value; According to the target temperature value of each pixel unit, the large-screen controller 1 controls the power of the micro thermoelectric cooler TEC of each pixel unit through a power control circuit.

[0043] In this embodiment, by integrating the TEC, the temperature sensor 5 and the large-screen controller 1, intelligent partition temperature control automation is realized.

[0044] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to completely illustrate the specific implementation process in this embodiment, another LED large-screen heat dissipation system for realizing intelligent partition temperature control is provided. This system includes an LED large screen and a large-screen controller 1; The LED large screen includes a substrate 2 and an LED chip 3; The LED chips 3 are divided into several pixel units, and a micro thermoelectric cooler TEC and a microchannel heat dissipation layer 4 are provided between each pixel unit and the front of the substrate 2; It should be noted that 6×6 LED chips are selected as pixel units; the micro thermoelectric cooler TEC is selected as a 3mm×3mm thermoelectric cooler, and the thickness of the microchannel heat dissipation layer 4 is selected as 0.5mm, and a honeycomb structure is adopted; A temperature sensor 5 is provided on the back of the substrate 2 corresponding to each pixel unit; the size of the pixel unit can be set to 10 cm, and the temperature sensor 5 is set at the center position of the corresponding pixel unit on the substrate 2, and the distance between the temperature sensors 5 corresponding to adjacent pixel units is 10 cm; Exemplarily, the temperature sensor can adopt a digital temperature sensor of the DS18B20 model, with an accuracy of ±0.5°C; The large-screen controller 1 collects the real-time temperature value of each pixel unit through the temperature sensor 5, predicts the temperature of each pixel unit through a pre-configured LED large-screen partition temperature prediction model, and then combines the real-time temperature value and the predicted temperature value of each pixel unit to calculate the target temperature value; It should be noted that the large-screen controller adopts ARM Cortex-M7; The generation process of the LED large-screen partition temperature prediction model includes data collection, model construction and model training; The specific data collection process is as follows: Collect the historical working parameters of the LED large screen, unify each historical working parameter on the time axis, and construct a data set; the historical working parameters include historical ambient temperature and humidity values, historical brightness values of each partition, and historical temperature values of each partition. Specifically, collect the historical ambient temperature and humidity values, historical brightness values, and historical temperature values of each partition of the LED large screen as historical working parameters according to the set collection period. After preprocessing various historical working parameters, align them according to the time stamp and store them as historical time series numbers in chronological order. The preprocessing process includes data cleaning, such as removing outliers and filling missing values, and normalization processing. Use the historical time series numbers to construct a data set. The specific model construction and training process are as follows: Determine the input features from the data set. Ambient temperature and humidity: E 温度 、E 湿度 ; Partition brightness value: B （i） , representing the brightness value of the i-th partition. Historical temperature value: , representing the n-time historical temperature values of partition i. Determine the output label from the data set. Future temperature value: , representing the temperature values of the future k time moments of partition i, such as 10s. Divide the historical time series numbers in the data set into several sliding windows, and set that each window contains input data of n time steps and output labels of k time steps. Construct an LSTM model, set the input layer of the LSTM model to receive historical ambient temperature and humidity values, historical brightness values, and historical temperature values, set the LSTM layer to capture long-term dependencies in the time series, and set the fully connected layer to map the output of the LSTM layer to the temperature values of the future k time moments. Specifically, the input data is represented as follows:

[0045] Among them, and Ambient temperature and humidity, B （i） represents the brightness value of the i-th partition, represents the historical temperature value of partition i at time step t-k; The output of the LSTM model is represented as follows:

[0046] Among them, is the partitioni The predicted temperature value at the k th second in the future; Construct a loss function using the mean squared error to measure the predicted temperature value and the actual temperature value of the LSTM model;

[0047] where N is the number of samples, k is the number of prediction time steps, is the actual temperature value of partition i at the jth second in the future, is the predicted temperature value of partition i at the jth second in the future; Use the data of each partition in the dataset as shared data to iteratively train the LSTM model, and use the constructed loss function to adjust the LSTM model during the training process until the loss function is less than the set threshold or reaches the maximum number of iterations, complete the training, and obtain the LED large screen partition temperature prediction model; Convert the obtained LED large screen partition temperature prediction model to INT8 format and burn it into the large screen controller 1; The large screen controller 1 is configured with a CAN bus communication protocol. The large screen controller 1 collects the real-time working parameters of each partition through the CAN bus communication protocol, and sends the power boost instruction of each partition to the micro thermoelectric cooler TEC of the corresponding partition through the CAN bus communication protocol; The large screen controller 1 performs power control on the micro thermoelectric cooler TEC of each pixel unit through a power control circuit according to the target temperature value of each pixel unit; The micro thermoelectric cooler TEC is provided with a cold end and a hot end; The cold end is attached to the LED chip 3 of the corresponding pixel unit, the hot end is attached to the microchannel heat dissipation layer 4, and the other end of the microchannel heat dissipation layer 4 is attached to the front side of the substrate 2; The cold end of the micro thermoelectric cooler TEC is attached to the LED chip 3 through nano silver glue, and the thickness of the nano silver glue is less than 0.01 mm; As Figure 3 shown, the power control circuit includes MOS transistor Q1, MOS transistor Q2, MOS transistor Q3, and MOS transistor Q4; The gate of MOS transistor Q1, the gate of MOS transistor Q2, the gate of MOS transistor Q3, and the gate of MOS transistor Q4 are connected to the large screen controller 1; The drain of MOS transistor Q1 is connected to the power supply VCC. The source of MOS transistor Q1 is connected to the hot end of the micro thermoelectric cooler TEC. The cold end of the micro thermoelectric cooler TEC is connected to the drain of MOS transistor Q4, and the source of MOS transistor Q4 is grounded; The drain of MOS transistor Q3 is connected to the power supply VCC, the source of MOS transistor Q3 is connected to the cold end of the micro thermoelectric cooler TEC, the hot end of the micro thermoelectric cooler TEC is also connected to the drain of MOS transistor Q2, and the source of MOS transistor Q2 is grounded; It should be noted that the large screen controller 1 controls the on and off of each MOS transistor through PWM signals. Specifically, the average power of the micro thermoelectric cooler TEC is controlled by adjusting the duty cycle of the PWM signal; The relationship between the duty cycle and power:

[0048] where, P TEC is the actual power of the micro thermoelectric cooler, D is the duty cycle of the PWM signal, P max is the maximum power of the micro thermoelectric cooler; The large screen controller 1 realizes the refrigeration or heating mode by controlling the switching of the current direction in each MOS transistor; Specifically, when refrigeration is required, the large screen controller 1 controls MOS transistors Q1 and Q4 to conduct, and MOS transistors Q2 and Q3 to turn off. The current path is: +Vcc → Q1 → TEC+ → TEC- → Q4 → GND; At this time, the micro thermoelectric cooler TEC cools the LED chip 3; When heating is required, the large screen controller 1 controls MOS transistors Q2 and Q3 to conduct, and MOS transistors Q1 and Q4 to turn off. The current path is: +Vcc → Q3 → TEC- → TEC+ → Q2 → GND; At this time, the micro thermoelectric cooler TEC heats the LED chip; Specifically, in a low-temperature environment (such as <0°C), the large screen controller 1 can control the current of the micro thermoelectric cooler TEC to reverse through the power control circuit, making the cold end become the hot end, and heating the substrate of the LED chip 3 to prevent circuit damage caused by condensation or low temperature; It should be noted that by changing the current direction, the cold end and hot end of the micro thermoelectric cooler TEC can be switched; when the current is in the forward direction, the micro thermoelectric cooler TEC cools; when the current is in the reverse direction, the micro thermoelectric cooler TEC heats; the refrigeration or heating intensity of the micro thermoelectric cooler TEC is proportional to the current magnitude. Therefore, it is necessary to control the power of the micro thermoelectric cooler TEC by adjusting the current magnitude, specifically achieved by the large screen controller 1 using PWM signals to control the current magnitude.

[0049] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for heat dissipation of LED large screens to achieve intelligent zone temperature control, characterized in that: The steps include: S1. Divide the LED screen into pixel units, and set a micro thermoelectric cooling sheet and a microchannel heat dissipation layer between the LED chip and the substrate of each pixel unit; S2. Divide the LED screen into zones, collect the ambient temperature and humidity values, the brightness values ​​of each zone, and the real-time temperature values ​​of each zone, and use the LSTM model to obtain the temperature prediction value of each zone; S3. Calculate the target temperature of each partition by combining the real-time temperature value and the temperature prediction value of each partition, and control the power of the micro thermoelectric cooling chip according to the target temperature of each partition.

2. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 1 is characterized in that: The specific steps of step S1 are as follows: S11. Determine the size of the pixel unit and divide the LED screen into several pixel units; S12. A micro thermoelectric cooling sheet and a microchannel heat dissipation layer are configured for each pixel unit.

3. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 2 is characterized in that: The specific steps of step S2 are as follows: S21. Determine the partition size of the LED screen, divide the pixel unit of the LED screen according to the partition size, and obtain several partitions; S22. Setting a temperature sensor for each partition on the back of the substrate; S23. Collect historical working parameters of the LED screen, unify the historical working parameters on the time axis, and construct a data set; the historical working parameters include historical ambient temperature and humidity values, historical brightness values ​​of each partition, and historical temperature values ​​of each partition; S24. Use the data set to train the LSTM model to obtain a temperature prediction model for the LED large screen partitions; S25. Collect the real-time working parameters of the LED large screen, and input the real-time working parameters into the LED large screen partition temperature prediction model to obtain the temperature prediction value of each partition within the set time period; the real-time working parameters include the real-time ambient temperature and humidity values, the real-time brightness values ​​of each partition, and the real-time temperature values ​​of each partition.

4. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 3 is characterized in that: The specific steps of step S24 are as follows: S241. Determine input features from the data set; S242. Determine an output label from the data set; S243. Divide the historical time series in the data set into a number of sliding windows, and set each window to contain n time steps of input data and k time steps of output labels; S244. Build an LSTM model, set the input layer of the LSTM model to receive historical environmental temperature and humidity values, historical brightness values, and historical temperature values, set the LSTM layer to capture long-term dependencies in the time series, and set the fully connected layer to map the output of the LSTM layer to the temperature values ​​at k moments in the future; S245. Use mean square error to construct a loss function to measure the temperature value predicted by the LSTM model and the actual temperature value; S246. Use the data of each partition in the data set as shared data to iteratively train the LSTM model, and use the constructed loss function to adjust the LSTM model during the training process until the loss function is less than the set threshold or reaches the maximum number of iterations, complete the training, and obtain the LED large screen partition temperature prediction model.

5. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 3 is characterized in that: The specific steps of step S3 are as follows: S31. Based on the real-time temperature value and the temperature prediction value, a dynamic weight coefficient is set, and the target temperature of each partition is calculated using the dynamic weight coefficient and the real-time temperature value and temperature prediction value of each partition; S32. Prioritize the LED screen according to the target temperature of each partition; If the target temperature of a partition is higher than the upper temperature threshold, the corresponding partition will be marked as high priority; If the target temperature of a partition is less than or equal to the upper temperature threshold and greater than or equal to the lower temperature threshold, the corresponding partition will be marked with medium priority; If the target temperature of a partition is less than the lower temperature threshold, the corresponding partition will be marked as low priority; S33. Controlling the micro-thermoelectric cooling sheet in the high priority partition to work in the forward direction and increasing the power of the micro-thermoelectric cooling sheet to a first set ratio of the maximum power; The micro thermoelectric cooling chip of the medium priority partition is controlled to work in the forward direction, and the power of the micro thermoelectric cooling chip is gradually increased according to the set proportion range of the maximum power; For the low priority partition, determine whether the target temperature is less than the upper limit of the temperature; If not, controlling the corresponding micro-thermoelectric cooling chip to work in the forward direction, and increasing the power of the micro-thermoelectric cooling chip to a second set ratio of the maximum power; If so, the corresponding micro thermoelectric cooling chip is controlled to work in reverse.

6. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 5 is characterized in that: The following steps are also included: S34 monitors the working status of each micro thermoelectric cooling chip of the LED screen; If all micro thermoelectric cooling chips are normal, return to step S25; If there is a faulty micro-thermoelectric cooling chip, go to step S35; S35. Determine the partition to which the faulty micro-thermoelectric cooling sheet belongs, and increase the power of normal micro-thermoelectric cooling sheets in the partition.

7. The LED large screen heat dissipation method for realizing intelligent zone temperature control according to claim 5, characterized in that: In step S33, the specific steps of controlling the micro thermoelectric cooling sheet of the medium priority partition to work in the forward direction and gradually increasing the power of the micro thermoelectric cooling sheet according to the set proportion range of the maximum power are as follows: SS1. Get the set temperature range [T1, T2] of the medium priority partition; SS2. Get the target power of the micro thermoelectric cooling chip of the medium priority partition as the set ratio range of the maximum power [P1, P2]; SS3. Map the set temperature range [T1, T2] to the set ratio range of the maximum power [P1, P2] by linear difference method; SS4. Calculate the ratio P of the target power to the maximum power by using the following formula using the temperature value T of the medium priority partition; 。 8. A large LED screen cooling system that realizes intelligent zone temperature control, characterized in that: Including LED large screen and large screen controller; The LED large screen includes a substrate and LED chips; The LED chip is divided into a number of pixel units, and a micro-thermoelectric cooling sheet and a micro-channel heat dissipation layer are arranged between each pixel unit and the front of the substrate; A temperature sensor is arranged on the back of the substrate corresponding to each pixel unit; The large screen controller collects the real-time temperature value of each pixel unit through the temperature sensor, and predicts the temperature of each pixel unit through the pre-configured LED large screen partition temperature prediction model, and then combines the real-time temperature value of each pixel unit with the predicted temperature value to calculate the target temperature value; The large-screen controller controls the power of the micro thermoelectric cooling chip of each pixel unit through the power control circuit according to the target temperature value of each pixel unit.

9. The LED large screen heat dissipation system for realizing intelligent zone temperature control according to claim 8, characterized in that: The micro thermoelectric cooling sheet is provided with a cold end and a hot end; The cold end is bonded to the LED chip of the corresponding pixel unit, the hot end is bonded to the microchannel heat dissipation layer, and the other end of the microchannel heat dissipation layer is bonded to the front side of the substrate.

10. The LED large screen heat dissipation system realizing intelligent zone temperature control according to claim 9, characterized in that: The power control circuit includes a MOS tube Q1, a MOS tube Q2, a MOS tube Q3 and a MOS tube Q4; The gate of the MOS tube Q1, the gate of the MOS tube Q2, the gate of the MOS tube Q3 and the gate of the MOS tube Q4 are connected to the large screen controller M1; The drain of the MOS tube Q1 is connected to the power supply VCC, the source of the MOS tube Q1 is connected to the hot end of the micro thermoelectric cooling sheet, the cold end of the micro thermoelectric cooling sheet is connected to the drain of the MOS tube Q4, and the source of the MOS tube Q4 is grounded; The drain of MOS tube Q3 is connected to the power supply VCC, the source of MOS tube Q3 is connected to the cold end of the micro thermoelectric cooling chip, the hot end of the micro thermoelectric cooling chip is also connected to the leakage detection of MOS tube Q2, and the source of MOS tube Q2 is grounded.

Images