A method and system for partitioned dimming of a backlit liquid crystal display thermoelectric coupling

Abstract

The present application relates to the technical field of display, and discloses a backlight liquid crystal display thermoelectric coupling partition dimming method and system, comprising the following steps: S1: partition pre-cooling dimming based on heat diffusion prediction: predicting future N frames of heat field by thermal resistance network / GNN, creating a "temperature depression" in the adjacent area in advance, and preventing fire in advance; obtaining the image brightness requirement and real-time thermal state data of each partition, wherein the real-time thermal state data includes a surface temperature sequence collected by a sparsely arranged physical temperature sensor, the present application scheme predicts the full-partition temperature field of N control cycles in the future through a thermal resistance network model, identifies the hot spot partition in advance and generates a dynamic temperature depression requirement diagram, and the reduced brightness of the hot spot partition is compensated to the adjacent area according to visual saliency and thermal crosstalk intensity, so that the generation of local hot spots is effectively avoided.

Description

Technical Field

[0001] This invention belongs to the field of display technology, specifically a method and system for local dimming of backlit liquid crystal displays using thermoelectric coupling. Background Technology

[0002] With the rapid development of new backlight technologies such as Mini-LED and Micro-LED, LCD displays are evolving towards higher brightness, higher contrast, and more local dimming. In local dimming technology, by dividing the backlight module into spatial sections and independently controlling the brightness of each section, the dynamic range of the display image quality can be significantly improved. However, the ultra-high density of the local dimming layout and the extremely high peak brightness requirements result in a huge heat flux density generated by the light-emitting elements per unit area.

[0003] In existing local dimming schemes, most technologies only focus on image quality optimization, such as passively dealing with temperature issues through lookup table compensation or fixed threshold dimming. Traditional solutions usually rely on temperature sensors to detect overheating before triggering dimming protection, which is a "post-event fire suppression" control. It cannot predict the heat accumulation process in the next few frames, resulting in hot spots forming and causing image quality fluctuations or device damage. It ignores the thermal crosstalk effect between local dimming zones. The local dimming zones on the backlight board are physically close together, and heat will be conducted between adjacent zones through printed circuit boards, copper foils, and thermal vias. Existing methods usually treat each zone as an independent hot node and do not establish an asymmetric thermal resistance network model, causing the optimization results to fail due to thermal crosstalk in actual operation. To address this, we propose a local dimming method and system for thermoelectric coupling of backlight liquid crystal displays. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, the present invention provides a local dimming method and system for thermoelectric coupling of backlight liquid crystal displays, which effectively solves the above problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a local dimming method for thermoelectric coupling of a backlight liquid crystal display, comprising the following steps: S1: Partition pre-cooling dimming based on thermal diffusion prediction: Using thermal resistance network / GNN to predict the thermal field of the next N frames, creating "temperature depressions" in neighboring areas in advance to prevent fires in advance; The system acquires image brightness requirements and real-time thermal status data for each partition. The real-time thermal status data includes surface temperature sequences collected by sparsely arranged physical temperature sensors, and electrical power data calculated from the driving voltage, current, and duty cycle of each partition in real time. Based on a thermal resistance network model, the system predicts the temperature field of the entire partition over the next N control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements characterizing the intensity of thermal crosstalk between partitions. Based on the predicted temperature field, hotspot partitions are identified, the heat accumulation rate and remaining safety margin of each partition are calculated, and a dynamic temperature depression demand map of the entire screen is generated, including emergency hotspot partitions. For negative cooling requirements, normal partitions have positive or zero heat dissipation margins. Based on the temperature depression requirement map, brightness redistribution is performed in the neighborhood of the hot spot partition to create a temperature depression. The reduced brightness of the hot spot partition is compensated to the neighboring area according to the visual salience and thermal crosstalk intensity ratio of the neighboring area. A joint cost function is constructed using image quality error, thermal cost, thermal crosstalk penalty, and power consumption. Under the preset brightness range, temperature constraints, image quality constraints, and power consumption constraints, the backlight drive value of each partition that minimizes the joint cost is solved and output. At the same time, the actual output brightness is compared with the temperature-luminous efficacy decay curve, and the weights are updated and optimized in the next control cycle. S2: Establish a thermoelectric synergistic model for predictive control: Actively allocate brightness in constrained optimization with multi-zone junction temperature, thermal crosstalk, and future heat capacity as the state; The system acquires the image brightness requirements and real-time thermal status data for each partition, including the junction temperature of the light-emitting element, the liquid crystal layer temperature, and the shell temperature for each partition. Based on a pre-established thermal resistance network model, it predicts the thermal field distribution of the entire partition for multiple consecutive control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements to characterize the thermal crosstalk intensity between different partitions. A joint cost function is constructed using image quality error, thermal cost, color difference, and power consumption. The thermal cost explicitly includes a thermal crosstalk penalty term calculated from the off-diagonal elements of the thermal resistance network model. Under preset brightness range, thermal safety budget, and color coordinate constraints, the joint cost function is solved in the rolling time domain to obtain the backlight driving value sequence for each partition for multiple future control cycles. The corresponding partition backlight source is driven according to the backlight driving value sequence, and the actual thermal status data after driving is fed back to the thermal resistance network model to update the model parameters and generate a luminous efficacy decay curve. S3: Thermal, Optical, and Color Consistency Feedback Compensation: Based on the junction temperature of each zone, the pre-calibrated luminous efficacy decay curve is consulted to estimate the thermally induced optical decay of each zone. The pre-calibrated peak wavelength drift curve is also consulted to estimate the peak wavelength drift of the red, green, and blue light-emitting elements in each zone. Based on the peak wavelength drift of the red, green, and blue elements, the color coordinate error of each zone is calculated, and the color gamut mapping gain of each zone is corrected accordingly. The net display brightness of each zone is calculated, where the net display brightness is the command backlight value for each zone. The actual output brightness and chromaticity after the combined effect of the light efficiency attenuation coefficient and the color gamut mapping gain; a joint cost function is constructed using image quality error, thermal cost, color difference and power consumption, and the net display brightness is used as an optimization variable to re-enter the dimming optimization solution in the next control cycle to obtain the backlight drive value of each zone and output the drive; the actual output data after output drive is compared with the target data, and when the deviation exceeds the preset tolerance value, at least one of the light efficiency attenuation curve, peak wavelength drift curve and color gamut mapping gain is updated in the subsequent control cycle; S4: Joint optimization of brightness in each backlight zone under thermal crosstalk constraints: Obtain the target image brightness requirements of each backlight zone in the current display frame, and simultaneously obtain the real-time thermal state data of each zone. The acquisition method includes: collecting local surface temperature through physical temperature sensors non-uniformly arranged on the backlight panel, and calculating the real-time electrical power of each zone by inverting the driving voltage, driving current and pulse width modulation duty cycle of each zone in real time; fusing the surface temperature and the electrical power to estimate and output the estimated junction temperature of the light-emitting elements of all zones, including the sensorless zone; A thermal resistance matrix is ​​constructed based on the physical structural parameters of the backlight panel, including the thermal conductivity of the printed circuit board material, copper foil thickness, thermal via density, light-emitting element spacing, and thermal paste thickness. The thermal resistance matrix contains self-thermal resistance elements on the diagonal and mutual thermal resistance elements on the off-diagonal. The off-diagonal elements are used to quantify the intensity and direction of heat conduction between different zones to describe the thermal crosstalk effect between adjacent zones. The thermal resistance matrix is ​​constructed as an asymmetric matrix to reflect the spatial asymmetry of the heat dissipation path caused by frame heat dissipation and structural obstruction of the backlight panel. A joint cost function is constructed using image quality error, thermal cost, and power consumption. The image quality error is used to measure the difference between the optimized backlight brightness and the target image brightness requirement. The thermal cost is used to constrain the peak junction temperature of each partition to not exceed a safe threshold. In the joint cost function, a thermal crosstalk penalty term composed of the off-diagonal elements of the thermal resistance matrix is ​​further introduced. The thermal crosstalk penalty term is configured to generate a penalty amount when there is a significant temperature difference between a partition and its adjacent partitions. This is used to prompt the optimization algorithm to actively transfer the local heat load to the neighboring area with better heat dissipation conditions and lower thermal resistance, thereby balancing the heat distribution across the entire screen. Under preset constraints, the joint cost function is solved to obtain the backlight driving value sequence for each partition. The constraints include: the backlight brightness of each partition is within a preset upper and lower limit range, the predicted junction temperature of each partition does not exceed the temperature safety threshold, the total power consumption of the entire screen does not exceed the preset power consumption budget, and the image quality error caused by brightness adjustment does not exceed the threshold imperceptible to the human eye. The solution process adopts a rolling time-domain optimization method, and outputs the backlight driving value that minimizes the joint cost function in each control cycle. The corresponding zone light source is driven according to the backlight driving value; after the driving is executed, the actual temperature distribution is collected by the temperature sensor, the actual thermal state is compared with the predicted thermal state, and the model error is calculated; based on the model error, the off-diagonal elements of the thermal resistance matrix are corrected and updated online to improve the accuracy of thermal field prediction in subsequent control cycles.

[0006] Preferably, the thermal resistance matrix of the thermal resistance network model is constructed in the following way: the initial thermal resistance is calculated based on the physical structural parameters of the backlight board, including the thermal conductivity of the PCB material, copper thickness, via density, LED spacing, and thermal paste thickness; a known power pattern is applied and the steady-state temperature distribution is obtained through an infrared thermal imager calibration experiment, and the accurate thermal resistance matrix is ​​obtained through inversion correction, wherein the thermal resistance matrix is ​​an asymmetric matrix to reflect the asymmetry of the heat dissipation path, each partition is modeled as a graph node, and the heat conduction path between partitions is modeled as an edge, with the edge weight equal to the element of the thermal resistance matrix; through iterative diffusion calculation, the current temperature vector is multiplied by the normalized thermal resistance adjacency matrix, and the electrical power vector of each partition is superimposed, thereby predicting the temperature field of the entire partition for the next N frames.

[0007] Preferably, the acquisition of real-time thermal state data for each partition specifically includes: non-uniformly arranging physical temperature sensors on the backlight panel to collect local surface temperature sequences; real-time acquisition of the driving voltage, driving current, and pulse width modulation duty cycle of each partition, and calculation of the real-time electrical power of each partition; and using the extended Kalman filter algorithm, fusing the local surface temperature sequence and the real-time electrical power as observation vectors to estimate the real-time junction temperature of all partitions, including the sensorless partition.

[0008] Preferably, when it is predicted that a certain partition will experience heat accumulation in a future control cycle, the optimization algorithm actively reduces the brightness allocation weight of that partition and simultaneously increases the brightness weight of adjacent partitions with better heat dissipation conditions and lower thermal resistance. This allows the algorithm to spatially utilize the heat capacity of neighboring regions to balance the heat load of the entire screen. Based on the partition junction temperature, the algorithm queries the preset temperature-luminous efficacy decay curve and temperature-peak wavelength drift curve to calculate the actual output brightness and color coordinate offset of the current partition. The difference between the actual output brightness and the target brightness, as well as the color coordinate offset, are fed back into the joint cost function of the next control cycle to dynamically adjust the color gamut mapping matrix, so that the heat-induced brightness loss and color difference are evenly distributed across the partitions.

[0009] Preferably, the ratio of the thermally induced brightness loss of each partition to the heat dissipation margin of that partition is calculated; when the ratio of a certain partition exceeds the average value of the whole screen and reaches a preset threshold, the brightness allocation weight of that partition is reduced in the optimization solution, and the brightness weight of neighboring areas with better heat dissipation conditions and better thermal stability is increased at the same time, so that the thermally induced brightness loss is evenly distributed across the whole screen. When the actual output brightness is lower than the target brightness and the difference exceeds the tolerance value, the image quality error weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the brightness loss; when the color coordinate error exceeds the preset color difference threshold, the color difference weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the color shift.

[0010] Preferably, when any partition is predicted to be a hotspot partition, a target partition with better heat dissipation conditions is identified based on the mutual thermal resistance elements of each partition within its neighborhood window, and the backlight brightness of the hotspot partition is reduced. At the same time, the released brightness budget is redistributed according to the heat dissipation margin of the target partition and the visual saliency weight. The visual saliency weight is calculated based on the brightness contrast and gradient features of the image content in the corresponding partition, so that brightness compensation is preferentially allocated to areas with high human eye attention, and the image quality error before and after brightness redistribution does not exceed the threshold imperceptible to the human eye. The feedback heat of the neighboring area due to the increased heat generation caused by receiving brightness compensation is calculated and fed back to the hotspot partition through mutual thermal resistance. When the feedback heat exceeds a preset tolerance value, the weight ratio of brightness allocation is iteratively adjusted until convergence.

[0011] A system for a local dimming method using thermoelectric coupling in a backlit liquid crystal display: This includes: a pre-cooling dimming module, a thermoelectric synergistic predictive control module, a light and color feedback compensation module, and a joint optimization module; The pre-cooling dimming module is used to acquire the image brightness requirements and real-time thermal status data of each zone. The real-time thermal status data includes the surface temperature sequence collected by sparsely arranged physical temperature sensors and the power data calculated by real-time acquisition of the driving voltage, current and duty cycle of each zone. Based on the thermal resistance network model, the temperature field of the entire zone is predicted for the next N control cycles. Based on the predicted temperature field, hot spot zones are identified and a dynamic temperature depression demand map of the whole screen is generated. Brightness redistribution is performed in the neighborhood of the hot spot zone to create temperature depressions. The reduced brightness of the hot spot zone is compensated to the neighboring zone according to the visual salience and thermal crosstalk intensity ratio of the neighboring zone. The thermoelectric co-predictive control module, connected to the pre-cooling dimming module, is used to obtain the junction temperature of the light-emitting element, the liquid crystal layer temperature and the shell temperature of each zone. Based on the thermal resistance network model, it predicts the thermal field distribution of the entire zone for multiple consecutive control cycles in the future. A joint cost function is constructed with image quality error, thermal cost, color difference and power consumption. The thermal cost explicitly includes a thermal crosstalk penalty term calculated from the off-diagonal elements of the thermal resistance network model. Under preset constraints, the backlight driving value sequence of each zone for multiple control cycles in the future is obtained by rolling time domain solution. The light and color feedback compensation module is connected to the thermoelectric co-predictive control module. It is used to query the pre-calibrated light efficiency attenuation curve and peak wavelength drift curve according to the junction temperature of each zone, estimate the thermal light decay and three-color wavelength drift respectively, calculate the color coordinate error and correct the color gamut mapping gain, and after obtaining the net display brightness, it is re-entered into the dimming optimization solution of the next control cycle. The joint optimization module, connected to the light and color feedback compensation module, is used to fuse surface temperature and power data to estimate the junction temperature of the entire zone, construct an asymmetric thermal resistance matrix based on the physical structure parameters of the backlight panel, introduce a thermal crosstalk penalty term composed of off-diagonal elements of the thermal resistance matrix in the joint cost function to promote the transfer of local heat load to neighboring areas with better heat dissipation conditions, solve the backlight driving value of each zone and drive the corresponding zone light source, and compare the actual temperature distribution after driving with the predicted value to correct the off-diagonal elements of the thermal resistance matrix online.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention predicts the temperature field of the entire zone for the next N control cycles using a thermal resistance network model, identifies hot spot zones in advance and generates a dynamic temperature depression demand map, and compensates the reduced brightness of hot spot zones to neighboring zones according to visual salience and thermal crosstalk intensity, effectively avoiding the generation of local hot spots. 2. This invention constructs a thermoelectric collaborative model predictive control architecture based on multi-zone junction temperature, thermal crosstalk, and future heat capacity. The system can plan the brightness sequence for multiple control cycles in advance and actively use the future heat capacity of neighboring zones to allocate brightness, which significantly reduces the heat accumulation rate and ensures the long-term convergence of image quality error and thermal safety budget in dynamic scenes. 3. This invention estimates the thermal luminous decay and three-color wavelength drift in each zone in real time by querying the pre-calibrated luminous efficacy decay curve and peak wavelength drift curve, and corrects the color gamut mapping gain accordingly. The net display brightness is then used as an optimization variable and re-input into the solution of the next control cycle, forming a closed-loop feedback of thermo-optical-color, which effectively offsets the brightness and color decay caused by high temperature and ensures that the display can maintain the factory-grade color accuracy after long-term operation. 4. This invention accurately quantifies the intensity and direction of heat conduction between partitions by constructing an asymmetric thermal resistance matrix containing off-diagonal elements. A thermal crosstalk penalty term is introduced into the joint cost function, forcing the optimization algorithm to actively transfer local heat loads to neighboring regions with better heat dissipation and lower thermal resistance. This strategy balances heat distribution across the entire screen, avoiding localized overheating and safety hazards caused by heat accumulation in traditional methods. Attached Figure Description

[0013] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0014] In the attached diagram: Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0016] Please refer to Figure X, a local dimming method for thermoelectric coupling of a backlit liquid crystal display, S1: Local pre-cooling dimming based on thermal diffusion prediction: using a thermal resistance network / GNN to predict the thermal field of the next N frames, creating "temperature depressions" in neighboring areas in advance, thus preventing fires in advance; The system acquires image brightness requirements and real-time thermal status data for each partition. Real-time thermal status data includes surface temperature sequences collected by sparsely arranged physical temperature sensors, and electrical power data calculated from the driving voltage, current, and duty cycle of each partition in real time. Based on a thermal resistance network model, the system predicts the temperature field of the entire partition over the next N control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements characterizing the intensity of thermal crosstalk between partitions. Based on the predicted temperature field, hotspot partitions are identified, the heat accumulation rate and remaining safety margin of each partition are calculated, and a dynamic temperature depression demand map of the entire screen is generated, including emergency hotspot partitions. For negative cooling requirements, normal zones have positive or zero heat dissipation margins. Based on the temperature depression requirement map, brightness redistribution is performed in the neighborhood of hot zone to create temperature depressions. The reduced brightness of hot zone is compensated to neighboring zones according to the visual salience and thermal crosstalk intensity ratio of neighboring zones. A joint cost function is constructed using image quality error, thermal cost, thermal crosstalk penalty, and power consumption. Under preset brightness range, temperature constraints, image quality constraints, and power consumption constraints, the backlight drive values ​​of each zone that minimize the joint cost are solved and output. At the same time, the actual output brightness is compared with the temperature-luminous efficacy decay curve, and the weights are updated and optimized in the next control cycle. The thermal resistance matrix of the thermal resistance network model is constructed as follows: Initial thermal resistance is calculated based on the physical structural parameters of the backlight board, including the thermal conductivity of the PCB material, copper thickness, via density, LED spacing, and thermal paste thickness. A known power pattern is applied, and the steady-state temperature distribution is obtained through an infrared thermal imager calibration experiment. The accurate thermal resistance matrix is ​​then obtained through inversion correction. This thermal resistance matrix is ​​asymmetric to reflect the asymmetry of the heat dissipation path. Each partition is modeled as a graph node, and the heat conduction path between partitions is modeled as an edge, with the edge weight equal to the element of the thermal resistance matrix. Through iterative diffusion calculation, the current temperature vector is multiplied by the normalized thermal resistance adjacency matrix, and the electrical power vectors of each partition are superimposed to predict the temperature field of the entire partition for the next N frames. The thermal resistance network model is constructed as follows: the initial thermal resistance is calculated based on the physical structural parameters of the backlight, and the accurate thermal resistance matrix is ​​obtained by applying a known power pattern and inverting through infrared thermal imager calibration experiments. The thermal resistance matrix is ​​an asymmetric matrix. Brightness redistribution is performed in the neighborhood of the hot spot partition. Specifically, this includes: calculating the visual saliency map of the image content, and distributing the reduced brightness of the hot spot partition according to the proportion of the product of the visual saliency score of the neighboring area and the thermal crosstalk intensity, so that the neighboring areas with high visual saliency and strong thermal crosstalk receive more brightness compensation. The process involves acquiring real-time thermal state data for each partition, specifically including: non-uniformly arranging physical temperature sensors on the backlight panel to collect local surface temperature sequences; real-time acquisition of the driving voltage, driving current, and pulse width modulation duty cycle for each partition, and calculation of the real-time electrical power for each partition; and using an extended Kalman filter algorithm, fusing the local surface temperature sequences and real-time electrical power as observation vectors to estimate the real-time junction temperature of all partitions, including the sensorless partition. S2: Establish a thermoelectric co-model for predictive control, with multi-zone junction temperature, thermal crosstalk, and future heat capacity as states, and actively allocate brightness in the constrained optimization process; The system acquires the image brightness requirements and real-time thermal status data for each partition, including the junction temperature of the light-emitting elements, the liquid crystal layer temperature, and the shell temperature for each partition. Based on a pre-established thermal resistance network model, it predicts the thermal field distribution of the entire partition over multiple consecutive control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements to characterize the thermal crosstalk intensity between different partitions. A joint cost function is constructed using image quality error, thermal cost, color difference, and power consumption, where the thermal cost explicitly includes a thermal crosstalk penalty term calculated from the off-diagonal elements of the thermal resistance network model. Under preset brightness range, thermal safety budget, and color coordinate constraints, the joint cost function is solved in the rolling time domain to obtain the backlight driving value sequence for each partition over multiple future control cycles. The corresponding partition backlight source is driven according to the backlight driving value sequence, and the actual thermal status data after driving is fed back to the thermal resistance network model to update the model parameters and generate a luminous efficacy decay curve. The thermal crosstalk penalty term is constructed by weighting and summing the off-diagonal elements of the thermal resistance matrix between each partition with the temperature difference between each partition. It is used to encourage the optimization process to actively suppress the thermal diffusion of local hot spots to adjacent areas. When it is predicted that a certain partition will experience heat accumulation in a future control cycle, the optimization algorithm actively reduces the brightness allocation weight of that partition and simultaneously increases the brightness weight of adjacent partitions with better heat dissipation conditions and lower thermal resistance. This allows the algorithm to spatially borrow the heat capacity of neighboring regions to balance the heat load of the entire screen. Based on the partition junction temperature, the algorithm queries the preset temperature-luminous efficacy decay curve and temperature-peak wavelength drift curve to calculate the actual output brightness and color coordinate offset of the current partition. The difference between the actual output brightness and the target brightness, as well as the color coordinate offset, are fed back into the joint cost function of the next control cycle to dynamically adjust the color gamut mapping matrix, so that the heat-induced brightness loss and color difference are evenly distributed in the partitions. Calculate the ratio of thermally induced brightness loss to the heat dissipation margin of each partition; when the ratio of a certain partition exceeds the average value of the whole screen and reaches the preset threshold, reduce the brightness allocation weight of that partition in the optimization solution, and simultaneously increase the brightness weight of neighboring areas with better heat dissipation conditions and better thermal stability, so that the thermally induced brightness loss is evenly distributed across the whole screen. When the actual output brightness is lower than the target brightness and the difference exceeds the tolerance value, the image quality error weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the brightness loss; when the color coordinate error exceeds the preset color difference threshold, the color difference weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the chromaticity shift. S3: Thermal, Light, and Color Consistency Feedback Compensation: Based on the junction temperature of each zone, the pre-calibrated luminous efficacy decay curve is consulted to estimate the thermally induced light decay of each zone, and the pre-calibrated peak wavelength drift curve is consulted to estimate the peak wavelength drift of the red, green, and blue light-emitting elements in each zone; based on the peak wavelength drift of red, green, and blue, the color coordinate error of each zone is calculated, and the color gamut mapping gain of each zone is corrected accordingly; the net display brightness of each zone is calculated, which is the actual output brightness and chromaticity after the combined effect of the command backlight value, luminous efficacy decay coefficient, and color gamut mapping gain of each zone; a joint cost function is constructed using image quality error, thermal cost, color difference, and power consumption, and the net display brightness is used as an optimization variable to re-enter the dimming optimization solution in the next control cycle to obtain the backlight drive value of each zone and output the drive; the actual output data after output drive is compared with the target data, and when the deviation exceeds the preset tolerance value, at least one of the luminous efficacy decay curve, peak wavelength drift curve, and color gamut mapping gain is updated in the subsequent control cycle; When re-entering the dimming optimization solution for the next control cycle, the net display brightness also includes: using the balanced distribution of thermally induced brightness loss across all screen zones as a constraint, ensuring that the brightness compensation obtained by zones with severe thermal decay is lower than that of zones with good thermal stability. S4: Joint optimization of brightness in each backlight zone under thermal crosstalk constraints: Obtain the target image brightness requirements of each backlight zone in the current display frame. At the same time, obtain the real-time thermal state data of each zone. The acquisition methods include: collecting local surface temperature through physical temperature sensors that are not uniformly arranged on the backlight panel, and calculating the real-time electric power of each zone by inverting the driving voltage, driving current and pulse width modulation duty cycle of each zone in real time; fusing the surface temperature and electric power data, estimating and outputting the estimated junction temperature of the light-emitting elements of all zones, including the sensorless zone; When any partition is predicted to be a hotspot partition, a target partition with better heat dissipation conditions is identified based on the mutual thermal resistance elements of each partition within its neighborhood window, and the backlight brightness of the hotspot partition is reduced. At the same time, the released brightness budget is redistributed according to the heat dissipation margin of the target partition and the visual saliency weight. The visual saliency weight is calculated based on the brightness contrast and gradient features of the image content in the corresponding partition, so that brightness compensation is preferentially allocated to areas with high human eye attention, and the image quality error before and after brightness redistribution does not exceed the threshold imperceptible to the human eye. The heat generated by the neighboring area due to receiving brightness compensation is calculated and fed back to the hotspot partition through mutual thermal resistance. When the feedback heat exceeds the preset tolerance value, the weight ratio of brightness allocation is iteratively adjusted until convergence. A thermal resistance matrix is ​​constructed based on the physical structural parameters of the backlight panel, including the thermal conductivity of the printed circuit board material, copper foil thickness, thermal via density, light-emitting element spacing, and thermal paste thickness. The thermal resistance matrix contains self-thermal resistance elements on the diagonal and mutual thermal resistance elements on the off-diagonal. The off-diagonal elements are used to quantify the intensity and direction of heat conduction between different zones to describe the thermal crosstalk effect between adjacent zones. The thermal resistance matrix is ​​constructed as an asymmetric matrix to reflect the spatial asymmetry of the heat dissipation path caused by frame heat dissipation and structural obstruction of the backlight panel. The local heat load is spatially transferred to neighboring areas with better heat dissipation conditions. Specifically, this includes: identifying target areas with lower thermal resistance and larger heat dissipation margins in the neighborhood of hotspot areas; redistributing the released brightness budget to the target areas according to the heat dissipation margin ratio and visual salience weight of neighboring areas; and iteratively adjusting the allocation weights until the thermal crosstalk feedback heat converges. A joint cost function is constructed using image quality error, thermal cost, and power consumption. Image quality error is used to measure the difference between the optimized backlight brightness and the target image brightness requirement. Thermal cost is used to constrain the peak junction temperature of each partition to not exceed the safety threshold. In the joint cost function, a thermal crosstalk penalty term composed of off-diagonal elements of the thermal resistance matrix is ​​further introduced. The thermal crosstalk penalty term is configured to generate a penalty amount when there is a significant temperature difference between a partition and its neighboring partitions. This is used to prompt the optimization algorithm to actively transfer the local heat load to the neighboring area with better heat dissipation conditions and lower thermal resistance, thereby balancing the heat distribution across the entire screen. Under preset constraints, the joint cost function is solved to obtain the backlight drive value sequence for each partition. The constraints include: the backlight brightness of each partition is within the preset upper and lower limits, the predicted junction temperature of each partition does not exceed the temperature safety threshold, the total power consumption of the whole screen does not exceed the preset power consumption budget, and the image quality error caused by brightness adjustment does not exceed the threshold imperceptible to the human eye. The solution process adopts a rolling time-domain optimization method, and outputs the backlight drive value that minimizes the joint cost function in each control cycle. Local surface temperatures are collected using non-uniformly arranged physical temperature sensors, and real-time power is calculated by inverting the driving voltage, driving current, and pulse width modulation duty cycle. Surface temperature and power data are fused to estimate and output junction temperatures of all light-emitting elements in all zones, including sensorless zones. An asymmetric thermal resistance matrix is ​​constructed based on the physical structural parameters of the backlight panel, containing off-diagonal elements characterizing the intensity and direction of heat conduction between zones. A thermal crosstalk penalty term composed of off-diagonal elements is introduced into the joint cost function to encourage the optimization algorithm to actively transfer local heat load to neighboring zones with better heat dissipation conditions. Under preset brightness range, temperature safety threshold, power consumption budget, and image quality constraints, a rolling time-domain optimization method is used to solve for the backlight driving values ​​of each zone and drive the light source. After driving execution, the actual temperature distribution is collected, and the off-diagonal elements of the thermal resistance matrix are corrected and updated online based on model errors. The corresponding zone light source is driven according to the backlight driving value; after the driving is executed, the actual temperature distribution is collected by the temperature sensor, the actual thermal state is compared with the predicted thermal state, and the model error is calculated; based on the model error, the off-diagonal elements of the thermal resistance matrix are corrected and updated online to improve the accuracy of thermal field prediction in subsequent control cycles.

[0017] A system for a thermoelectrically coupled local dimming method for a backlit liquid crystal display includes: a pre-cooling dimming module, a thermoelectric co-predictive control module, a light and color feedback compensation module, and a joint optimization module; The pre-cooled dimming module is used to acquire the image brightness requirements and real-time thermal status data of each zone. The real-time thermal status data includes the surface temperature sequence collected by sparsely arranged physical temperature sensors and the power data calculated by real-time acquisition of the driving voltage, current and duty cycle of each zone. Based on the thermal resistance network model, the temperature field of the entire zone is predicted for the next N control cycles. Based on the predicted temperature field, hot spot zones are identified and a dynamic temperature depression demand map of the whole screen is generated. Brightness redistribution is performed in the neighborhood of the hot spot zone to create temperature depressions. The reduced brightness of the hot spot zone is compensated to the neighboring zone according to the visual salience of the neighboring zone and the ratio of thermal crosstalk intensity. The thermoelectric co-predictive control module, connected to the pre-cooling dimming module, is used to obtain the junction temperature of the light-emitting element, the liquid crystal layer temperature and the shell temperature of each zone. Based on the thermal resistance network model, it predicts the thermal field distribution of the entire zone for multiple consecutive control cycles in the future. A joint cost function is constructed with image quality error, thermal cost, color difference and power consumption. The thermal cost explicitly includes the thermal crosstalk penalty term calculated by the off-diagonal elements of the thermal resistance network model. Under preset constraints, the backlight driving value sequence of each zone for multiple control cycles in the future is obtained by rolling time domain solution. The light and color feedback compensation module is connected to the thermoelectric co-predictive control module. It is used to query the pre-calibrated light efficiency attenuation curve and peak wavelength drift curve according to the junction temperature of each zone, estimate the thermal light decay and three-color wavelength drift respectively, calculate the color coordinate error and correct the color gamut mapping gain, and then re-enter the dimming optimization solution for the next control cycle after obtaining the net display brightness. The joint optimization module, connected to the light and color feedback compensation module, is used to fuse surface temperature and power data to estimate the junction temperature of the entire zone. It constructs an asymmetric thermal resistance matrix based on the physical structure parameters of the backlight panel. In the joint cost function, a thermal crosstalk penalty term composed of off-diagonal elements of the thermal resistance matrix is ​​introduced to promote the transfer of local heat load to neighboring zones with better heat dissipation conditions. It solves the backlight driving value of each zone and drives the corresponding zone light source. It also compares the actual temperature distribution after driving with the predicted value to correct the off-diagonal elements of the thermal resistance matrix online.

Claims

1. A method for local dimming of a backlit liquid crystal display using thermoelectric coupling, characterized in that: Includes the following steps: S1: Partition pre-cooling dimming based on thermal diffusion prediction: Using thermal resistance network / GNN to predict the thermal field of the next N frames, creating "temperature depressions" in neighboring areas in advance to prevent fires in advance; The image brightness requirements and real-time thermal status data of each partition are obtained. The real-time thermal status data includes the surface temperature sequence collected by sparsely arranged physical temperature sensors, and the power data calculated by real-time acquisition of the driving voltage, current and duty cycle of each partition. The temperature field of the entire partition is predicted based on a thermal resistance network model for the next N control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements representing the intensity of thermal crosstalk between partitions. Hotspot partitions are identified based on the predicted temperature field, and the heat accumulation rate and remaining safety margin of each partition are calculated. A dynamic temperature depression demand map of the entire screen is generated, where emergency hotspot partitions correspond to negative cooling demand, and normal partitions correspond to positive heat dissipation margin or zero. Based on the temperature depression demand map, brightness redistribution is performed in the neighborhood of the hotspot partition to create temperature depressions. The reduced brightness of the hotspot partition is compensated to the neighboring area according to the visual salience and thermal crosstalk intensity ratio of the neighboring area. A joint cost function is constructed based on image quality error, thermal cost, thermal crosstalk penalty, and power consumption. Under preset brightness range, temperature constraints, image quality constraints, and power consumption constraints, the backlight drive value of each partition that minimizes the joint cost is solved and output. At the same time, the actual output brightness is compared with the temperature-luminous efficiency decay curve, and the weights are updated and optimized in the next control cycle. S2: Establish a thermoelectric synergistic model for predictive control: Actively allocate brightness in constrained optimization with multi-zone junction temperature, thermal crosstalk, and future heat capacity as the state; The system acquires the image brightness requirements and real-time thermal status data for each partition, including the junction temperature of the light-emitting element, the liquid crystal layer temperature, and the shell temperature for each partition. Based on a pre-established thermal resistance network model, it predicts the thermal field distribution of the entire partition for multiple consecutive control cycles. The state transition matrix of the thermal resistance network model contains off-diagonal elements to characterize the thermal crosstalk intensity between different partitions. A joint cost function is constructed using image quality error, thermal cost, color difference, and power consumption. The thermal cost explicitly includes a thermal crosstalk penalty term calculated from the off-diagonal elements of the thermal resistance network model. Under preset brightness range, thermal safety budget, and color coordinate constraints, the joint cost function is solved in the rolling time domain to obtain the backlight drive value sequence for each partition for multiple future control cycles. Drive the corresponding partitioned backlight source according to the backlight driving value sequence, and feed back the actual thermal state data after driving back to the thermal resistance network model to update the model parameters and generate the light efficiency decay curve. S3: Thermal, optical, and color consistency feedback compensation: Based on the junction temperature of each partition, the pre-calibrated luminous efficacy decay curve is consulted to estimate the thermally induced optical decay of each partition, and the pre-calibrated peak wavelength drift curve is consulted to estimate the peak wavelength drift of the red, green, and blue light-emitting elements in each partition respectively; based on the peak wavelength drift of the red, green, and blue elements, the color coordinate error of each partition is calculated, and the color gamut mapping gain of each partition is corrected accordingly. Calculate the net display brightness of each zone, where the net display brightness is the actual output brightness and chromaticity after the combined effect of the command backlight value of each zone, the light efficiency attenuation coefficient, and the color gamut mapping gain. A joint cost function is constructed using image quality error, thermal cost, color difference, and power consumption. The net display brightness is then used as an optimization variable and re-input into the dimming optimization solution of the next control cycle to obtain the backlight drive value of each zone and output the drive. The actual light output data after output drive is compared with the target data. When the deviation exceeds the preset tolerance value, at least one of the light effect attenuation curve, peak wavelength drift curve, and color gamut mapping gain is updated in subsequent control cycles. S4: Joint optimization of brightness in each backlight zone under thermal crosstalk constraints: Obtain the target image brightness requirements of each backlight zone in the current display frame, and simultaneously obtain the real-time thermal state data of each zone. The acquisition method includes: collecting local surface temperature through physical temperature sensors non-uniformly arranged on the backlight panel, and calculating the real-time electrical power of each zone by inverting the driving voltage, driving current and pulse width modulation duty cycle of each zone in real time; fusing the surface temperature and the electrical power to estimate and output the estimated junction temperature of the light-emitting elements of all zones, including the sensorless zone; A thermal resistance matrix is ​​constructed based on the physical structural parameters of the backlight panel, including the thermal conductivity of the printed circuit board material, copper foil thickness, thermal via density, light-emitting element spacing, and thermal paste thickness. The thermal resistance matrix contains self-thermal resistance elements on the diagonal and mutual thermal resistance elements on the off-diagonal. The off-diagonal elements are used to quantify the intensity and direction of heat conduction between different zones to describe the thermal crosstalk effect between adjacent zones. The thermal resistance matrix is ​​constructed as an asymmetric matrix to reflect the spatial asymmetry of the heat dissipation path caused by frame heat dissipation and structural obstruction of the backlight panel. A joint cost function is constructed using image quality error, thermal cost, and power consumption. The image quality error is used to measure the difference between the optimized backlight brightness and the target image brightness requirement. The thermal cost is used to constrain the peak junction temperature of each partition to not exceed a safe threshold. In the joint cost function, a thermal crosstalk penalty term composed of the off-diagonal elements of the thermal resistance matrix is ​​further introduced. The thermal crosstalk penalty term is configured to generate a penalty amount when there is a significant temperature difference between a partition and its adjacent partitions. This is used to prompt the optimization algorithm to actively transfer the local heat load to the neighboring area with better heat dissipation conditions and lower thermal resistance, thereby balancing the heat distribution across the entire screen. Under preset constraints, the joint cost function is solved to obtain the backlight driving value sequence for each partition. The constraints include: the backlight brightness of each partition is within a preset upper and lower limit range, the predicted junction temperature of each partition does not exceed the temperature safety threshold, the total power consumption of the entire screen does not exceed the preset power consumption budget, and the image quality error caused by brightness adjustment does not exceed the threshold imperceptible to the human eye. The solution process adopts a rolling time-domain optimization method, and outputs the backlight driving value that minimizes the joint cost function in each control cycle. The corresponding zone light source is driven according to the backlight driving value; after the driving is executed, the actual temperature distribution is collected by the temperature sensor, the actual thermal state is compared with the predicted thermal state, and the model error is calculated; based on the model error, the off-diagonal elements of the thermal resistance matrix are corrected and updated online to improve the accuracy of thermal field prediction in subsequent control cycles.

2. The method for local dimming of a backlit liquid crystal display with thermoelectric coupling according to claim 1, characterized in that: The thermal resistance matrix of the thermal resistance network model is constructed as follows: the initial thermal resistance is calculated based on the physical structural parameters of the backlight board, including the thermal conductivity of the PCB material, copper thickness, via density, LED spacing, and thermal paste thickness; a known power pattern is applied and the steady-state temperature distribution is obtained through an infrared thermal imager calibration experiment; the accurate thermal resistance matrix is ​​obtained through inversion correction, wherein the thermal resistance matrix is ​​an asymmetric matrix to reflect the asymmetry of the heat dissipation path; each partition is modeled as a graph node, and the heat conduction path between partitions is modeled as an edge, with the edge weight equal to the element of the thermal resistance matrix; By iterative diffusion calculation, the current temperature vector is multiplied by the normalized thermal resistance adjacency matrix and superimposed with the electrical power vectors of each partition, thereby predicting the temperature field of the entire partition for the next N frames.

3. The method for local dimming of a backlit liquid crystal display with thermoelectric coupling according to claim 1, characterized in that: The acquisition of real-time thermal state data for each partition specifically includes: non-uniformly arranging physical temperature sensors on the backlight panel to collect local surface temperature sequences; real-time acquisition of the driving voltage, driving current, and pulse width modulation duty cycle of each partition, and calculation of the real-time electrical power of each partition; and using the extended Kalman filter algorithm, fusing the local surface temperature sequence and the real-time electrical power as observation vectors to estimate the real-time junction temperature of all partitions, including the sensorless partition.

4. The local dimming method for thermoelectric coupling of a backlight liquid crystal display according to claim 1, characterized in that: When it is predicted that a certain partition will experience heat accumulation in the future control cycle, the optimization algorithm actively reduces the brightness allocation weight of that partition and simultaneously increases the brightness weight of the adjacent partitions with better heat dissipation conditions and lower thermal resistance. In this way, the heat capacity of the neighboring areas is used to balance the heat load of the entire screen. Based on the junction temperature of the partition, the preset temperature-luminous efficacy decay curve and temperature-peak wavelength drift curve are queried to calculate the actual output brightness and color coordinate offset of the current partition. The difference between the actual output brightness and the target brightness, as well as the color coordinate offset, are fed back into the joint cost function of the next control cycle to dynamically adjust the color gamut mapping matrix, so that the thermally induced brightness loss and color difference are evenly distributed in the intervals.

5. The method for local dimming of a backlit liquid crystal display with thermoelectric coupling according to claim 1, characterized in that: Calculate the ratio of thermally induced brightness loss to the heat dissipation margin of each partition; when the ratio of a certain partition exceeds the average value of the whole screen and reaches a preset threshold, reduce the brightness allocation weight of that partition in the optimization solution, and simultaneously increase the brightness weight of neighboring areas with better heat dissipation conditions and better thermal stability, so that the thermally induced brightness loss is evenly distributed across the whole screen. When the actual output brightness is lower than the target brightness and the difference exceeds the tolerance value, the image quality error weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the brightness loss; when the color coordinate error exceeds the preset color difference threshold, the color difference weight of that partition is increased in the joint cost function of the next control cycle to prompt subsequent control cycles to compensate for the color shift.

6. The method for local dimming of a backlit liquid crystal display with thermoelectric coupling according to claim 1, characterized in that: When any partition is predicted to be a hotspot partition, a target partition with better heat dissipation conditions is identified based on the mutual thermal resistance elements of each partition within its neighborhood window, and the backlight brightness of the hotspot partition is reduced. At the same time, the released brightness budget is redistributed according to the heat dissipation margin and visual saliency weight of the target partition. The visual saliency weight is calculated based on the brightness contrast and gradient features of the image content in the corresponding partition, so that brightness compensation is preferentially allocated to the area with high human eye attention, and the image quality error before and after brightness redistribution does not exceed the threshold imperceptible to the human eye. The heat generated by the neighboring cell due to brightness compensation is calculated and fed back to the hot spot partition via mutual thermal resistance. When the feedback heat exceeds a preset tolerance value, the weight ratio of brightness allocation is iteratively adjusted until convergence.

7. A system for a local dimming method of thermoelectric coupling in a backlight liquid crystal display according to any one of claims 1-6, characterized in that: include: Pre-cooling dimming module, thermoelectric synergistic predictive control module, light and color feedback compensation module, and joint optimization module; The pre-cooled dimming module is used to acquire the image brightness requirements and real-time thermal status data of each zone. The real-time thermal status data includes the surface temperature sequence collected by sparsely arranged physical temperature sensors and the power data calculated by real-time acquisition of the driving voltage, current and duty cycle of each zone. Based on the thermal resistance network model, the full-zone temperature field of the next N control cycles is predicted. Based on the predicted temperature field, hot spots are identified and a dynamic temperature depression demand map of the whole screen is generated. Brightness redistribution is performed in the neighborhood of the hot spots to create temperature depressions. The reduced brightness of the hot spots is compensated to the neighboring areas according to the visual salience and thermal crosstalk intensity ratio of the neighboring areas. The thermoelectric co-predictive control module, connected to the pre-cooling dimming module, is used to obtain the junction temperature of the light-emitting element, the liquid crystal layer temperature and the shell temperature of each zone. Based on the thermal resistance network model, it predicts the thermal field distribution of the entire zone for multiple consecutive control cycles in the future. A joint cost function is constructed with image quality error, thermal cost, color difference and power consumption. The thermal cost explicitly includes a thermal crosstalk penalty term calculated from the off-diagonal elements of the thermal resistance network model. Under preset constraints, the backlight driving value sequence of each zone for multiple control cycles in the future is obtained by rolling time domain solution. The light and color feedback compensation module is connected to the thermoelectric co-predictive control module. It is used to query the pre-calibrated light efficiency attenuation curve and peak wavelength drift curve according to the junction temperature of each zone, estimate the thermal light decay and three-color wavelength drift respectively, calculate the color coordinate error and correct the color gamut mapping gain, and after obtaining the net display brightness, it is re-entered into the dimming optimization solution of the next control cycle. The joint optimization module, connected to the light and color feedback compensation module, is used to fuse surface temperature and power data to estimate the junction temperature of the entire zone, construct an asymmetric thermal resistance matrix based on the physical structure parameters of the backlight panel, introduce a thermal crosstalk penalty term composed of off-diagonal elements of the thermal resistance matrix in the joint cost function to promote the transfer of local heat load to neighboring areas with better heat dissipation conditions, solve the backlight driving value of each zone and drive the corresponding zone light source, and compare the actual temperature distribution after driving with the predicted value to correct the off-diagonal elements of the thermal resistance matrix online.

Images