Multi-source data fusion display module temperature rise abnormal intelligent early warning method and system

Abstract

This invention relates to the field of display module thermal management and intelligent early warning technology, specifically to a method and system for intelligent early warning of abnormal temperature rise in display modules based on multi-source data fusion. The method includes: acquiring electrical excitation data and thermal observation data of the display module to obtain basic parameters; generating temperature deviation values ​​based on the basic parameters; acquiring initial thermal resistance and capacitance parameters of the display module; correcting the initial thermal resistance and capacitance parameters according to the temperature deviation values ​​to generate parameter correction values; generating transient temperature field prediction values ​​by combining the parameter correction values ​​and transient spatial power consumption density data; extracting spatial temperature gradient data and transient temperature rise rate data based on the transient temperature field prediction values, and generating hot spot early warning indicators; generating early warning judgment results by combining the hot spot early warning indicators; executing the early warning judgment results and outputting a hot spot abnormality early warning signal for the display module. This invention reduces the missed detection of local hot spots and early warning lag.

Description

Technical Field

[0001] This invention relates to the field of display module thermal management and intelligent early warning technology, specifically to a method and system for intelligent early warning of abnormal temperature rise in display modules based on multi-source data fusion. Background Technology

[0002] During the operation of the display module, the pixel array continuously generates electrical power consumption under different driving states, which further causes temperature changes inside the module. When a local area is under high load or rapid switching conditions for a long time, it is easy to form concentrated temperature rise and hot spot accumulation, thereby affecting the stability and reliability of the display module. In order to prevent the risk of abnormal temperature from developing into an uncontrollable state, it is usually necessary to continuously monitor the thermal status of the display module and issue abnormal warnings. In traditional methods, the temperature rise of the display module is determined by relying on the temperature measurement results of physical sensors at the edge, or by making a rough estimate based on the driving power consumption. When the frequency of local heat load changes exceeds the preset change threshold, the existing monitoring methods still cannot simultaneously take into account the time correspondence, spatial distribution and prediction accuracy, which leads to problems such as missed detection of local hot spots, delayed warnings and unclear identification of abnormal locations. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent early warning method and system for abnormal temperature rise in display modules based on multi-source data fusion, so as to avoid spatial temperature distribution distortion caused by fixed parameter modeling and sparse observation interpolation, and to solve the problems of missed detection of local hot spots, delayed early warning, and unclear identification of abnormal locations caused by the difficulty of existing monitoring methods in simultaneously taking into account time correspondence, spatial distribution and prediction accuracy.

[0004] The objective of this invention can be achieved through the following technical solutions: A thermal resistance-capacitance network model and a temperature prediction model for the display module are pre-established; electrical excitation data and thermal observation data of the display module are acquired, and the electrical excitation data and thermal observation data are processed by timestamp mapping to obtain basic parameters; Transient spatial power density data is obtained by multiplying the electrical excitation data in the basic parameters, and the initial temperature data is obtained by combining the thermal resistance-capacitance network model. The initial temperature data and thermal observation data are synchronously compared and calculated to generate temperature deviation values. Obtain the initial thermal resistance and capacitance parameters of the display module; correct the initial thermal resistance and capacitance parameters according to the temperature deviation value to generate parameter correction values; combine the parameter correction values ​​and transient spatial power consumption density data to calculate the deviation of the heat conduction equation of the temperature prediction model and generate transient temperature field prediction values. Based on the transient temperature field prediction values, spatial temperature gradient data and transient temperature rise rate data are extracted, and hot spot early warning indicators are generated. The safety limit threshold is obtained by extracting the thermodynamic properties of the display module material. The safety limit threshold is then compared with the hot spot warning index to generate a warning judgment result. The warning judgment result is executed to output a hot spot abnormality warning signal for the display module.

[0005] Preferably, the voltage distribution matrix and current distribution matrix of the pixel array of the display module are obtained at the first sampling frequency to obtain electrical excitation data; Temperature observations from the physical sensors in the non-display area of ​​the display module are acquired using a second sampling frequency to obtain thermal observation data. The electrical excitation data and thermal observation data are then processed by timestamp mapping to obtain basic parameters.

[0006] As a preferred method, transient spatial power density data is obtained by multiplying the voltage distribution matrix and current distribution matrix in the electrical excitation data. Transient heat conduction calculations are performed by combining transient spatial power density data and a pre-set thermal resistance-capacitance network model to obtain initial temperature data; the initial temperature data and thermal observation data are then compared and calculated simultaneously to obtain the temperature deviation value.

[0007] As a preferred method, the equivalent heat capacity parameter and equivalent thermal resistance parameter are obtained by extracting the thermodynamic properties of the display module material, and used as the initial thermal resistance-capacity parameter; The equivalent heat capacity and equivalent thermal resistance parameters are iteratively optimized based on the temperature deviation value to obtain the parameter adjustment result; the trend term of the parameter adjustment result is extracted by a low-pass filtering algorithm to obtain the parameter correction value.

[0008] Preferably, the transient spatial power density data is used as the transient boundary excitation input to the temperature prediction model to obtain the initial temperature field data; The deviation of the heat conduction equation is calculated by combining the parameter correction values ​​and the initial temperature field data to obtain the equation deviation data; the model parameters of the temperature prediction model are updated to obtain the transient temperature field prediction value.

[0009] As a preferred method, the absolute temperature peak value is extracted from the transient temperature field prediction value to obtain the absolute temperature peak value data; the spatial partial derivative and time partial derivative calculations are performed on the transient temperature field prediction value to obtain the spatial temperature gradient data and transient temperature rise rate data, respectively. By combining absolute temperature peak data, spatial temperature gradient data, and transient temperature rise rate data, and through weighted calculation using multiple weighting coefficients, a hot spot early warning index is obtained.

[0010] Preferably, it is determined whether the hot spot warning index is greater than or equal to the safety limit threshold. If so, an abnormal triggering logic judgment result is generated. If not, the step of obtaining the electrical excitation data and thermal observation data of the display module is returned. After generating the abnormal trigger logic judgment result, the hot spot coordinates are located and the predicted peak value is extracted based on the transient temperature field prediction value to obtain the early warning judgment result.

[0011] The intelligent early warning system for abnormal temperature rise of display modules based on multi-source data fusion includes a data acquisition module, a deviation calculation module, a task processing module, an analysis module, an indicator extraction module, a judgment module, and an execution module. The data acquisition module is used to pre-establish the thermal resistance-capacitance network model and temperature prediction model of the display module; and to acquire the electrical excitation data and thermal observation data of the display module, and to perform timestamp mapping processing on the electrical excitation data and thermal observation data to obtain the basic parameters; The deviation calculation module is used to perform product calculations based on the electrical excitation data in the basic parameters to obtain transient spatial power density data, and to calculate the initial temperature data by combining the thermal resistance-capacitance network model. The initial temperature data and thermal observation data are synchronously compared and calculated to generate temperature deviation values. The task processing module is used to obtain the initial thermal resistance and capacitance parameters of the display module, correct the initial thermal resistance and capacitance parameters according to the temperature deviation value, and generate parameter correction values. The analysis module is used to combine parameter correction values ​​and transient spatial power density data to calculate the deviation of the heat conduction equation in the temperature prediction model and generate transient temperature field prediction values. The index extraction module is used to extract spatial temperature gradient data and transient temperature rise rate data based on the transient temperature field prediction value, and generate hot spot early warning indexes. The judgment module is used to obtain the safety limit threshold by extracting the thermodynamic properties of the display module material, and compare the safety limit threshold with the hot spot warning index to generate the warning judgment result. The execution module is used to execute the early warning judgment results and output the hot spot abnormality early warning signal of the display module.

[0012] Compared with the prior art, the present invention has the following advantages: 1. This invention first acquires the electrical excitation data and thermal observation data of the display module within the system cycle, and then uses timestamp mapping to organize the voltage distribution matrix, current distribution matrix and temperature observation values ​​under different sampling rhythms onto a unified time reference. This can effectively avoid the problem of time misalignment caused by inconsistent sampling frequencies, thereby solving the defect of delayed temperature rise judgment in traditional monitoring methods. 2. This invention generates transient spatial power consumption density data based on voltage distribution matrix and current distribution matrix, calculates initial temperature data by combining thermal resistance-capacitance network model, and then compares the initial temperature data with thermal observation data synchronously to obtain temperature deviation value, so as to establish a unified thermal conduction correspondence between electrical heating and thermal response, and ensure that the thermal state during local high load changes can be quantitatively identified. 3. This invention iteratively optimizes, extracts trends, and corrects the equivalent heat capacity and equivalent thermal resistance parameters based on the temperature deviation value. The corrected parameter value is then applied to the temperature prediction model along with transient spatial power density data. Under the constraints of the heat conduction equation, observation point constraints, and boundary conditions, the transient temperature field prediction value is generated, which solves the prediction distortion problem caused by fixed parameter modeling and sparse observation interpolation. 4. This invention further extracts absolute temperature peak data, spatial temperature gradient data, and transient temperature rise rate data from the transient temperature field prediction value, and generates hot spot early warning indicators by weighting the three types of features. Then, it combines the safety limit threshold and spatial continuity constraints to determine the early warning judgment result, which can simultaneously identify abnormal risks such as absolute high temperature, local concentrated heating, and rapid temperature rise. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application and the prior art, the accompanying drawings used in the description of the embodiments and the prior art will be briefly introduced below.

[0014] Figure 1 This diagram illustrates a flowchart of an intelligent early warning method for abnormal temperature rise in a display module based on multi-source data fusion, according to an embodiment of the present disclosure. Figure 2 This diagram illustrates a block diagram of an intelligent early warning system for abnormal temperature rise in a display module based on multi-source data fusion, according to an embodiment of the present disclosure. Detailed Implementation

[0015] To make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments will be described in detail below with reference to the accompanying drawings; Please see Figure 1 A multi-source data fusion-based intelligent early warning method for abnormal temperature rise in display modules includes: A thermal resistance-capacitance network model and a temperature prediction model for the display module are pre-established; electrical excitation data and thermal observation data of the display module are acquired, and the electrical excitation data and thermal observation data are processed by timestamp mapping to obtain basic parameters; Transient spatial power density data is obtained by multiplying the electrical excitation data in the basic parameters, and the initial temperature data is obtained by combining the thermal resistance-capacitance network model. The initial temperature data and thermal observation data are synchronously compared and calculated to generate temperature deviation values. Obtain the initial thermal resistance and capacitance parameters of the display module; The initial thermal resistance and capacitance parameters are corrected based on the temperature deviation value to generate parameter correction values; By combining parameter correction values ​​and transient spatial power density data, the deviation of the heat conduction equation of the temperature prediction model is calculated, and the transient temperature field prediction value is generated. Based on the transient temperature field prediction values, spatial temperature gradient data and transient temperature rise rate data are extracted, and hot spot early warning indicators are generated. The safety limit threshold is obtained by extracting the thermodynamic properties of the display module material. The safety limit threshold is then compared with the hot spot warning index to generate a warning judgment result. The system executes the early warning judgment result and outputs an abnormal hot spot warning signal for the display module. Acquire electrical excitation data and thermal observation data of the display module, perform timestamp mapping processing on the electrical excitation data and thermal observation data to obtain basic parameters, including: The voltage distribution matrix and current distribution matrix of the pixel array of the display module are obtained at the first sampling frequency to obtain electrical excitation data; Temperature observations from the physical sensors in the non-display area of ​​the display module are obtained using a second sampling frequency to acquire thermal observation data. The electrical excitation data and thermal observation data are timestamped to obtain the basic parameters; Transient spatial power density data is obtained by multiplying the electrical excitation data in the basic parameters, and the initial temperature data is calculated by combining it with the thermal resistance-capacitance network model. The initial temperature data and the thermal observation data are then compared and calculated simultaneously to generate temperature deviation values, including: Transient spatial power density data is obtained by multiplying the voltage distribution matrix and current distribution matrix in the electrical excitation data. Transient heat conduction calculations are performed by combining transient spatial power density data and a pre-defined thermal resistance-capacitance network model to obtain initial temperature data; The initial temperature data and thermal observation data are compared and calculated simultaneously to obtain the temperature deviation value.

[0016] This embodiment provides a method for intelligent early warning of abnormal temperature rise in display modules by multi-source data fusion. The method first acquires the electrical excitation data and thermal observation data of the display module within the system cycle, and then organizes the two types of data with different sampling rhythms into a unified time base through timestamp mapping to form basic parameters. The transient spatial power density data is then calculated using the voltage distribution matrix and the current distribution matrix, and the initial temperature data corresponding to the power input is obtained by using the thermal resistance-capacitance network model. Furthermore, the temperature deviation value is generated by synchronously comparing the initial temperature data with the thermal observation data. The temperature deviation value is used as the basis for subsequent thermal resistance-capacitance parameter correction and temperature field prediction, and finally outputs a hot spot abnormality warning signal for the display module. The processing configuration is to integrate electrical excitation data and thermal observation data so that the temperature rise judgment is based on the same time reference, thereby reducing the missed detection of local hot spots and the lag in early warning; Acquire electrical excitation data and thermal observation data of the display module, namely the temperature observation values ​​output by the physical sensors distributed in the non-display area. Based on the pixel array driving state and the physical sensor output, perform timestamp mapping processing on the electrical excitation data and thermal observation data to generate basic parameters. These basic parameters are the combined results of electrical data and thermal data after completing the unified time correspondence, which are used for subsequent power consumption calculation and temperature comparison. In this process, the first sampling frequency is used to characterize the acquisition rhythm of the voltage distribution matrix and the current distribution matrix. Its value should be greater than the second sampling frequency to retain the high-frequency alternating characteristics of the drive power consumption. The second sampling frequency is used to characterize the sampling period of the temperature observation value. Its value can be set according to the response capability of the physical sensor. In practical implementation, the voltage distribution matrix of the display module pixel array is obtained at the first sampling frequency. and current distribution matrix Using spatial location variables and and the time variable corresponding to the system cycle The voltage distribution matrix describes the driving voltage that each pixel experiences in the current cycle, and the current distribution matrix describes the driving current that flows through each pixel in the current cycle. Together, they reflect the electrical excitation intensity of the display module in the current cycle. At the same time, a set of temperature observations from multiple physical sensors in the non-display area is acquired at a second sampling frequency. By sensor serial number Identify the spatial location of different observation points, in order to Indicates the first The installation coordinates of each physical sensor are determined, and the observation period time variable is used. Corresponding to the second sampling frequency; Since the first sampling frequency and the second sampling frequency are usually different, the first sampling frequency corresponds to the preset electrical change rate and the second sampling frequency corresponds to the preset thermal response change rate. If the two types of data are processed independently, time misalignment is likely to occur, causing the subsequent temperature calculation to deviate from the operating condition. This implementation uses timestamp mapping to map each set of thermal observation data to electrical excitation data within adjacent system cycles; The execution logic for timestamp mapping is configured as follows: when at time... When a set of thermal observation data is acquired, the timestamp is extracted from the electrical excitation data stream. closest The voltage distribution matrix and current distribution matrix of a system cycle are used as matching objects, so that the voltage and current information with the high sampling frequency and the temperature observation information with the low sampling frequency are established in a strict correspondence under a unified time reference. Through the above processing steps, the thermal-electric correspondence of the display module during local high load changes can be captured, and the judgment ambiguity caused by inconsistent sampling frequency can be transformed into basic parameter features that can be directly calculated. Obtain basic parameters, and calculate power consumption for electrical excitation data in the basic parameters based on the product relationship between voltage distribution matrix and current distribution matrix, generating transient spatial power consumption density data, which displays the instantaneous heat intensity distribution of each spatial location of the module within the system cycle; Then, based on the preset thermal resistance-capacitance network model, transient heat conduction calculations are performed on the transient spatial power consumption density data to generate the basic temperature result, i.e., the initial temperature data, which is calculated by the model based on the current power consumption input. Further, based on the synchronous comparison results of the initial temperature data and thermal observation data, a temperature deviation value is determined to characterize the degree of deviation between the model calculation results and the observations. This temperature deviation value is the difference between the initial temperature data and the thermal observation data. In the specific implementation, we first use the formula for the transient spatial power density matrix: ; in, This represents the product operator. and These represent individual pixel grids in direction and Physical dimensions of the direction; The formula is used to convert the voltage and current distribution of the pixel array into a heat source input that directly reflects the degree of heat generation, utilizing the driving voltage distribution described earlier. With drive current distribution Performing the product yields the position used to describe the system's period. Transient spatial power density data on the intensity of heat generation under the current driving conditions ; Since the heat source of the display module is determined by the power consumption, the result of this product directly corresponds to the heat input item in the subsequent heat conduction calculation; After obtaining the transient spatial power density data, transient heat conduction calculations are performed using a thermal resistance-capacitance network model. The calculation configuration utilizes the equivalent thermal resistance and equivalent thermal capacity of the multilayer materials in the display module to convert electrical heating into a temperature quantity that can be compared with sensor observations. Discrete state update relationships can be used. ; In this relationship, the thermal resistive-capacitive network model outputs initial temperature data representing the basic thermal state driven by power consumption during the current system cycle. Its update depends on the sampling step size that limits the model update interval. ; The calculation process incorporates the system's equivalent heat capacity parameter, which characterizes the thermal inertia of the temperature change after the display module absorbs heat. And the system equivalent thermal resistance parameter characterizing the heat transfer resistance properties from the module interior to the external environment. ; Meanwhile, the global power consumption, which reflects the total heat generation level of the entire display module in the current cycle, is obtained by summing the transient spatial power density data over the entire spatial domain of the display module. Global power consumption The specific discrete accumulation formula is as follows: ; in, and These represent the pixel array of the display module in... direction and Total number of grid cells in each direction and Represents the physical size of a single pixel grid; And combined with the ambient temperature characterizing the external heat dissipation boundary conditions Together, they complete the iterative update of the temperature state; Based on the system cycle generated by low-frequency temperature observations, the initial temperature data is then synchronously compared with the thermal observation data to obtain the temperature deviation value; the deviation relationship can be used as follows: ; Calculate the difference between the temperature estimated by the quantification model and the observed temperature; by aggregating the number of samples in the non-display area. The physical sensor during the observation period Temperature observation values ​​below Calculate the average observed temperature and subtract the initial temperature data for the same observation period. This allows us to derive the current observation period temperature deviation value, which indicates whether the thermal resistive-capacitive network model underestimates or overestimates the module temperature. ; The spatial heat distribution is obtained by the power consumption product relationship, and the electrical excitation is mapped to the temperature quantity through the thermal resistance-capacitance network model. Synchronous comparison is performed under a unified time base. It can capture the temperature rise trend of the display module during local screen changes and transform the local anomalies that are easily masked by thermal inertia in traditional solutions into quantifiable temperature deviation values. By obtaining the spatial heat distribution through the power consumption product relationship, the initial temperature data is made consistent with the actual heat conduction law. Then, the deviation of the model is evaluated by the temperature deviation value. Therefore, it is more suitable for handling the working conditions where high-frequency excitation and low-frequency observation coexist.

[0017] In this embodiment, the initial thermal resistance and capacitance parameters of the display module are obtained; the initial thermal resistance and capacitance parameters are corrected according to the temperature deviation value to generate parameter correction values, including: The equivalent thermal capacity parameter and equivalent thermal resistance parameter are obtained by extracting the thermodynamic properties of the display module material, and used as the initial thermal resistance-capacity parameter; The equivalent heat capacity and equivalent thermal resistance parameters are iteratively optimized based on the temperature deviation values ​​to obtain the parameter adjustment results; The parameter adjustment results are extracted using a low-pass filtering algorithm to obtain the parameter correction value; By combining parameter correction values ​​and transient spatial power density data, the deviation of the heat conduction equation in the temperature prediction model is calculated to generate transient temperature field predictions, including: The transient spatial power density data is used as the transient boundary excitation input to the temperature prediction model to obtain the initial temperature field data; The deviation of the heat conduction equation is calculated by combining the parameter correction values ​​and the initial temperature field data, and the equation deviation data is obtained. The model parameters of the temperature prediction model are updated to obtain the predicted value of the transient temperature field.

[0018] Obtain the initial thermal resistance and capacitance parameters of the display module, determine the initial thermal resistance and capacitance parameters based on the thermodynamic properties of the display module material, and generate the equivalent thermal capacitance parameters and equivalent thermal resistance parameters. Then, based on the temperature deviation value, the equivalent heat capacity parameter and the equivalent thermal resistance parameter are iteratively optimized and the trend is extracted to generate parameter correction values; further, based on the parameter correction values ​​and transient spatial power consumption density data, the deviation of the heat conduction equation of the temperature prediction model is calculated to generate transient temperature field prediction values. In this process, the basic thermal parameters abstracted from the thermal properties of the multi-layer materials of the display module are used as the initial thermal resistance and capacitance parameters; the thermal parameter adjustment results extracted based on the temperature deviation value are used to form parameter correction values ​​to compensate for thermal response drift caused by material aging, environmental changes or assembly differences. The spatial temperature distribution result output by the temperature prediction model within the current system cycle is the transient temperature field prediction value, which is used to describe the temperature rise status at each location within the display area. The initial thermal resistance and capacitance parameters of the display module are obtained. Based on the thermodynamic properties of the display module material, the thermal parameters of the display module are equivalently processed to generate the initial thermal resistance and capacitance parameters. The thermodynamic properties of these display module materials include at least the equivalent heat capacity related properties, which reflect the rate of temperature rise after the material absorbs heat, and the equivalent thermal resistance related properties, which reflect the degree of resistance to heat transfer to the environment. For multi-layered heterogeneous display modules, the heat absorption capacity under the combined action of multiple materials can be summarized into an equivalent heat capacity parameter. The heat dissipation resistance under the combined action of multiple materials is reduced to the equivalent thermal resistance parameter. ; The equivalent processing configuration is set to characterize the overall thermal response characteristics of the display module with preset quantity parameters, which facilitates subsequent online adjustments. The initial thermal resistance-capacitance parameters are corrected based on the temperature deviation value, generating parameter adjustment results; the calculation configuration is designed to adjust the model parameters in accordance with changes in the thermal behavior of the display module, avoiding prediction deviations caused by long-term use of fixed parameters; an update relationship can be used: ; ; To calculate the partial derivatives, a sensitivity state variable is introduced: ; Its discrete update formula is: ; ; During the update process, the temperature deviation value, which is calculated above and represents the degree of matching between the current model parameters and the thermal response, is used. The learning rate is set to a positive value for the heat capacity parameter. and thermal resistance parameter learning rate The magnitude of each adjustment to the control parameters determines the equivalent heat capacity parameter for the current observation period. and equivalent thermal resistance parameters Iterate each time to the updated equivalent heat capacity parameters and equivalent thermal resistance parameters ; To address the disturbances caused by single observation fluctuations to the parameter adjustment results, the parameter adjustment results can be low-pass filtered to extract the changing trends and obtain the parameter correction values. In practice, low-pass filtering can be performed using a first-order exponential smoothing mechanism. The calculation rule is to weight and fuse the current parameter adjustment result with the smoothing result of the previous observation period according to a preset smoothing coefficient. The smoothing coefficient controls the weight of the latest adjustment result, and its value range is usually set to... By giving greater weight to historical smoothing results, the drastic parameter changes caused by a single surge in temperature deviation can be mitigated. The low-pass filtering algorithm is configured to retain the effective components of thermal parameters as they change over a long period of time, while suppressing the spike effects caused by short-term measurement disturbances. The parameter correction values ​​obtained after trend extraction are more suitable for input into subsequent temperature prediction models to maintain the stability of model parameter updates. Based on the temperature deviation value, the equivalent heat capacity parameter and the equivalent thermal resistance parameter are adaptively adjusted; it can capture the changes in thermal response caused by changes in material properties and environmental boundaries, and transform the long-term accumulated error under the fixed parameter mode into a parameter correction value feature that can be gradually corrected. Obtain parameter correction values ​​and transient spatial power density data, and calculate the deviation of the temperature prediction model based on the constraints of the heat conduction equation to generate transient temperature field prediction values. The temperature prediction model here is specifically designed to output the spatial temperature distribution of the display area. After receiving the current power consumption input, it will generate initial temperature field data as a preliminary spatial temperature estimate and calculate equation deviation data that reflects the degree of deviation between the predicted output and the constraints of the heat conduction equation, thereby constraining the direction of model update. In practical implementation, transient spatial power density data will be used. As a transient boundary excitation input into the temperature prediction model, through model parameters Obtain the temperature prediction model in spatial location and system cycle The output temperature field result, i.e., the initial temperature field data. ; Configure a temperature prediction model to generate a spatial temperature estimate corresponding to the current power consumption distribution, replacing the interpolation inference step based on sensor data; The deviation of the heat conduction equation is calculated by combining the parameter correction value and the initial temperature field data to obtain the equation deviation data. In order to introduce the parameter correction value after the layer adjustment, that is, the updated equivalent heat capacity parameter and equivalent thermal resistance parameter, into the layer equation, this embodiment establishes a spatial conversion constraint between the two. Specifically, the equivalent heat capacity parameter in the parameter correction value is proportionally converted into heat capacity density, i.e., equivalent density, by displaying the equivalent volume of the module. With equivalent specific heat capacity The product of the effective heat transfer cross-sectional area and thickness characteristic ratio of the display module is used to convert the equivalent thermal resistance parameter in the parameter correction value into the equivalent thermal conductivity. ; Through this computational decomposition and transfer rule, the thermal parameters obtained from model optimization can be directly applied to the constraints of the partial differential equations; based on this conversion, the equation deviation data are calculated. Using partial differential equations: ; When the absolute value of the deviation data in this equation approaches zero, it indicates that the current temperature field conforms to the heat conduction law that incorporates the latest parameter correction results; in the formula, This is a simplified form of the aforementioned initial temperature field data. This refers to the transient spatial power density data mentioned earlier; The equation utilizes the equivalent density obtained from the parameter correction values. With equivalent specific heat capacity The time-varying term acting together on the temperature field Simultaneously, the equivalent thermal conductivity calculated based on parameter correction values ​​is utilized. Spatial diffusion term acting on the temperature field and ; And through the module equivalent thickness The planar power density is converted into the volume heat source intensity, thereby completing the equation constraints of the layer; During model updates, model parameters can be adjusted using composite constraint relationships. Adjustments are made to ensure that the obtained transient temperature field prediction simultaneously satisfies the power consumption input constraint, the heat conduction equation constraint, and the sparse observation point constraint; a loss relation can be used: ; In this loss function, a constraint term for the heat conduction equation is introduced to suppress the equation bias data. Observation point data constraints used to ensure that the predicted temperature is consistent with the physical sensor measurements. and boundary condition constraints used to ensure that the heat dissipation state at the module edge conforms to the boundary. And through preset positive number weighting coefficients , and Adjust the proportion of each constraint item separately; update This allows us to obtain a transient temperature field prediction that matches the current power input and corrected thermal parameters; The temperature prediction model employs a deep multilayer perceptron structure based on a physical information neural network, with the network input feature vector being three-dimensional spatiotemporal coordinates. The output is a scalar predicted temperature. The network learnable parameter matrix include Weight matrix of fully connected layer With bias vector Data constraint terms in the loss function and boundary constraint terms Defined by the following mathematical formulas respectively: ; in, Indicates the physical boundaries of the display module. Represents the boundary normal vector. The convective heat transfer coefficient is given during the current system cycle. Below, the Adam optimization algorithm is used to optimize the model parameters. The update is performed using the following formula: ; in, Indicates the number of iterations in the inner loop. The preset learning rate (the range of values ​​is set to...) to ); By guiding the update of thermal resistance and capacitance parameters through parameter correction values ​​and directly writing transient spatial power density data into the heat conduction equation, the model output is driven by the current power distribution and constrained by thermal laws. Therefore, it is easier to maintain the availability of temperature prediction under extreme conditions such as rapid switching of local highlighting.

[0019] In this embodiment, spatial temperature gradient data and transient temperature rise rate data are extracted based on transient temperature field prediction values, and hot spot early warning indicators are generated, including: The absolute temperature peak value is extracted from the transient temperature field prediction value to obtain the absolute temperature peak value data; Spatial and temporal partial derivatives of the transient temperature field predictions were calculated to obtain spatial temperature gradient data and transient temperature rise rate data, respectively. By combining absolute temperature peak data, spatial temperature gradient data, and transient temperature rise rate data, and by weighting calculations using multiple weighting coefficients, a hot spot early warning index is obtained. By combining hot spot early warning indicators with safety limit thresholds, an early warning judgment result is generated, including: Determine whether the hot spot warning index is greater than or equal to the safety limit threshold. If yes, generate an abnormal trigger logic judgment result. If no, return to the step of obtaining the electrical excitation data and thermal observation data of the display module. After generating the abnormal trigger logic judgment result, the hot spot coordinates are located and the predicted peak value is extracted based on the transient temperature field prediction value to obtain the early warning judgment result; Please see Figure 2 The intelligent early warning system for abnormal temperature rise of display modules based on multi-source data fusion includes: a data acquisition module, which is used to pre-establish the thermal resistance-capacitance network model and temperature prediction model of the display module; and to acquire the electrical excitation data and thermal observation data of the display module, and to perform timestamp mapping processing on the electrical excitation data and thermal observation data to obtain basic parameters; The deviation calculation module is used to calculate the transient spatial power density data by multiplying the electrical excitation data in the basic parameters, and to calculate the initial temperature data by combining the thermal resistance-capacitance network model. The initial temperature data and thermal observation data are compared and calculated simultaneously to generate the temperature deviation value. The task processing module is used to obtain the initial thermal resistance and capacitance parameters of the display module, correct the initial thermal resistance and capacitance parameters according to the temperature deviation value, and generate parameter correction values. The analysis module is used to combine parameter correction values ​​and transient spatial power density data to calculate the deviation of the heat conduction equation of the temperature prediction model and generate transient temperature field prediction values. The index extraction module is used to extract spatial temperature gradient data and transient temperature rise rate data based on the transient temperature field prediction value, and generate hot spot early warning indexes. The judgment module is used to obtain the safety limit threshold by extracting the thermodynamic properties of the display module material, and compare the safety limit threshold with the hot spot warning index to generate the warning judgment result. The execution module is used to execute the early warning judgment results and output the hot spot abnormality early warning signal of the display module.

[0020] The transient temperature field prediction value is obtained. Based on the spatial temperature distribution and the temperature change within the system cycle, the transient temperature field prediction value is feature extracted to generate absolute temperature peak data, spatial temperature gradient data, and transient temperature rise rate data. Then, the three types of features are weighted and calculated to generate hot spot early warning indicators. Further, the hot spot warning indicators are compared and processed according to the safety limit threshold to determine the warning judgment result; in this process, the absolute temperature peak data is used to characterize the temperature result corresponding to the highest temperature point in the display area. Spatial temperature gradient data is used to characterize the severity of temperature changes between adjacent regions to reflect whether the heat distribution is concentrated; transient temperature rise rate data is used to describe how fast the temperature changes over time within a system cycle to reflect whether hot spots are in a rapid growth state. The final hot spot warning index is a judgment result that comprehensively quantifies the absolute temperature, the steepness of the temperature distribution, and the rate of temperature rise. The absolute temperature peak value is extracted from the transient temperature field prediction value to obtain the absolute temperature peak value data; the extraction step is configured to determine the highest temperature position of the current display module; the peak value extraction relationship can be used: ; Predicted values ​​of transient temperature fields at various locations Take the maximum value and calculate the current system periodic absolute temperature peak data that reflects the highest temperature level in the display area. ; This feature is suitable for identifying whether the overall temperature has approached the material's allowable upper limit. However, relying solely on this feature can easily overlook situations where local areas have not yet reached their peak temperature but are rapidly deteriorating. Spatial and temporal partial derivatives of the transient temperature field predictions are calculated to obtain spatial temperature gradient data and transient temperature rise rate data, respectively. The calculation configuration quantifies heat concentration and temperature rise rate; the gradient and rate relationship can be used for calculation. ; ; Spatial temperature gradient data were obtained through spatial partial derivative calculations. Its physical meaning lies in reflecting the spatial temperature distribution gradient around a certain location, and this value is positively correlated with the local heat concentration. Transient temperature rise rate data were obtained by time partial derivative calculation. Its physical meaning lies in reflecting how fast the temperature at that location changes with the system's cycle, and this value is positively correlated with the rate of temperature rise; For display modules, a spatial temperature gradient greater than a preset gradient threshold indicates an increased risk of local stress and uneven material heating, while a temperature rise rate greater than a preset rate threshold indicates that hot spots are forming or expanding. By combining absolute temperature peak data, spatial temperature gradient data, and transient temperature rise rate data, a hot spot early warning index is obtained through weighted calculation. This index is used to comprehensively assess the current location's thermal risk level. A weighted relationship can be used: ; An absolute temperature reference derived from material properties is introduced into the calculation. Space temperature gradient reference benchmark and transient temperature rise rate reference And with weighting coefficients that sum to 1 , and To balance the contributions of the three types of risks—absolute high temperature, localized concentrated heating, and rapid temperature rise—under the current operating conditions; To ensure the effective fusion of characteristics from different dimensions, a reference standard is used. , and The module material is pre-calibrated by displaying the upper limit of the rated operating temperature, the empirical value of the maximum allowable local thermal stress gradient, and the theoretical value of the initial heating slope of the material under the limit power consumption. Meanwhile, regarding the weighting coefficients, since the risk focus differs in different application scenarios, they can be calibrated through regression analysis using multiple sets of historical fault sample data. Under conventional operating conditions where the primary concern is preventing high-temperature burnout, the values ​​typically fall within a certain range. , , This makes the normalized values ​​dimensionless and biased towards the prevention of absolute high temperature. This method combines absolute temperature, spatial temperature gradient, and transient temperature rise rate for quantitative judgment; it can capture areas where local heat generation is already concentrated but has not yet reached the highest temperature, as well as areas where the temperature is not yet high but is rising rapidly, transforming the problem of insufficient sensitivity of single indicators into the feature of multi-indicator collaborative judgment. The system acquires hot spot warning indicators, determines the safety limit threshold based on the thermodynamic properties of the display module material, compares the hot spot warning indicators with the safety limit threshold, and generates a warning judgment result. During this comparison process, the safety limit threshold is set as the comprehensive judgment boundary. The underlying properties are determined by the thermodynamic properties of the display module material, which are normalized by the above-mentioned absolute temperature reference, spatial temperature gradient reference, and temperature rise rate reference, corresponding to the comprehensive heat load boundary that the material can accept during long-term operation. When the hot spot warning indicators reach the preset danger conditions, an abnormal triggering logic judgment result will be generated to determine whether to enter the hot spot location and warning output stage. In practical implementation, it is determined whether the hot spot early warning index is greater than or equal to the safety limit threshold; specifically, it is determined whether the following conditions are met. ,For example The value is 1.0; When there is any position When this condition is met, an exception triggering logic judgment result is generated; the judgment condition here... This represents the normalized danger boundary, whose physical meaning is that the overall thermal risk at the current location has reached the upper limit of acceptable material risk. Considering that occasional transient electromagnetic interference may cause abrupt changes in the predicted temperature at a single point, a spatial continuity constraint is added to the judgment, that is, only when... There must be at least multiple consecutive warning indicators that simultaneously reach the preset neighborhood of the center. Only then can the result of the abnormal triggering logic be finally confirmed, thereby avoiding false alarms caused by a single noise point; When the trigger logic determines an anomaly, the transient temperature field prediction value is used. By calculating the two-dimensional matrix index corresponding to the maximum value of the early warning index in the spatial grid, the hot spot coordinates are located and the predicted peak value is extracted. The specific calculation formula is as follows: ; Extracted coordinates With peak Together with the current transient temperature rise rate, they form the early warning judgment result; hot spot coordinate positioning is used to give the spatial location of the high-risk area within the display area, and predicted peak extraction is used to give the temperature peak and temperature rise status of the corresponding area, thereby forming a hot spot abnormality early warning signal that can be read by the host computer or external diagnostic system. By constraining the early warning indicators with multi-dimensional thermal features, and providing hot spot coordinates and predicted peak values ​​after triggering, it is more suitable for handling display conditions where local areas are heated unevenly and change rapidly. During application implementation, new voltage distribution matrices and current distribution matrices continuously flow into the data acquisition module, and new physical sensor temperature observations synchronously flow into the data acquisition module, which are then mapped by timestamps and enter the deviation calculation module. The temperature deviation value output by the deviation calculation module is input to the task processing module, and the parameter correction value output by the task processing module is input to the analysis module. The transient temperature field prediction value output by the analysis module is input to the index extraction module and the judgment module. When the preset logic judgment conditions are met, the judgment module outputs the early warning judgment result, and the execution module generates a hot spot abnormality early warning signal. This abnormality early warning signal is only used as a status output for external systems to read and does not perform adjustment actions on the display drive circuit. This implementation organizes electrical input, thermal observation, parameter correction, temperature field prediction, and early warning output into a unified system according to the data flow sequence; it can reduce the judgment bias caused by the disconnect between modules and make the early warning results directly based on the continuous data transmission relationship.

[0021] The embodiments described above do not constitute a limitation on the scope of protection of this technical solution. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the above embodiments should be included within the scope of protection of this technical solution.

Claims

1. A method for intelligent early warning of abnormal temperature rise in display modules based on multi-source data fusion, characterized in that, include: A thermal resistance-capacitance network model and a temperature prediction model for the display module are pre-established. Acquire electrical excitation data and thermal observation data of the display module, and perform timestamp mapping processing on the electrical excitation data and thermal observation data to obtain basic parameters; Transient spatial power density data is obtained by multiplying the electrical excitation data in the basic parameters and calculating the initial temperature data by combining the thermal resistance-capacitance network model. The initial temperature data and the thermal observation data are then compared and calculated synchronously to generate a temperature deviation value. Obtain the initial thermal resistance and capacitance parameters of the display module; The initial thermal resistance-capacitance parameter is corrected based on the temperature deviation value to generate a parameter correction value; By combining the parameter correction values ​​and the transient spatial power density data, the temperature prediction model is subjected to heat conduction equation deviation calculation to generate transient temperature field prediction values. Based on the predicted transient temperature field values, spatial temperature gradient data and transient temperature rise rate data are extracted, and hot spot early warning indicators are generated. The safety limit threshold is obtained by extracting the thermodynamic properties of the display module material. The safety limit threshold is then compared with the hot spot warning index to generate a warning judgment result. The warning judgment result is executed, and a hot spot abnormality warning signal is output to the display module.

2. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 1, characterized in that, Acquire electrical excitation data and thermal observation data of the display module, perform timestamp mapping processing on the electrical excitation data and the thermal observation data to obtain basic parameters, including: The voltage distribution matrix and current distribution matrix of the pixel array of the display module are obtained at the first sampling frequency to obtain electrical excitation data; Temperature observations from the physical sensors in the non-display area of ​​the display module are obtained using a second sampling frequency to acquire thermal observation data. The electrical excitation data and the thermal observation data are timestamped to obtain the basic parameters.

3. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 2, characterized in that, Transient spatial power density data is obtained by multiplying the electrical excitation data in the basic parameters, and initial temperature data is calculated by combining it with the thermal resistance-capacitance network model. The initial temperature data and the thermal observation data are then compared synchronously to generate a temperature deviation value, including: Based on the product calculation of the voltage distribution matrix and current distribution matrix in the electrical excitation data, the transient spatial power density data is obtained. The transient spatial power density data and the preset thermal resistance-capacitance network model are combined to perform transient heat conduction calculations to obtain the initial temperature data. The initial temperature data and the thermal observation data are synchronously compared and calculated to obtain the temperature deviation value.

4. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 1, characterized in that, Obtain the initial thermal resistance and capacitance parameters of the display module; correct the initial thermal resistance and capacitance parameters according to the temperature deviation value to generate parameter correction values, including: The equivalent thermal capacity parameter and equivalent thermal resistance parameter are obtained by extracting the thermodynamic properties of the display module material, and used as the initial thermal resistance and capacity parameters. The equivalent heat capacity parameter and the equivalent thermal resistance parameter are iteratively optimized based on the temperature deviation value to obtain the parameter adjustment result; The parameter adjustment results are subjected to a low-pass filtering algorithm to extract the trend term, and the parameter correction value is obtained.

5. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 1, characterized in that, Combining the parameter correction values ​​and the transient spatial power density data, the temperature prediction model is subjected to heat conduction equation deviation calculation to generate transient temperature field prediction values, including: The transient spatial power density data is used as a transient boundary excitation input into the temperature prediction model to obtain the initial temperature field data; The deviation of the heat conduction equation is calculated by combining the parameter correction value and the initial temperature field data to obtain the equation deviation data; The model parameters of the temperature prediction model are updated to obtain the transient temperature field prediction value.

6. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 1, characterized in that, Based on the predicted transient temperature field values, spatial temperature gradient data and transient temperature rise rate data are extracted, and hot spot early warning indicators are generated, including: The absolute temperature peak value is extracted from the transient temperature field prediction value to obtain the absolute temperature peak value data; Spatial and temporal partial derivatives are calculated on the predicted transient temperature field values ​​to obtain spatial temperature gradient data and transient temperature rise rate data, respectively. By combining the absolute temperature peak data, the spatial temperature gradient data, and the transient temperature rise rate data, and performing weighted calculations using multiple weighting coefficients, a hot spot early warning index is obtained.

7. The intelligent early warning method for abnormal temperature rise of display modules based on multi-source data fusion according to claim 1, characterized in that, Based on the aforementioned hot spot early warning indicators, the safety limit threshold is compared and processed to generate an early warning judgment result, including: Determine whether the hot spot warning index is greater than or equal to the safety limit threshold. If yes, generate an abnormal trigger logic judgment result. If no, return to the step of obtaining the electrical excitation data and thermal observation data of the display module. After generating the trigger logic judgment result of the anomaly, hot spot coordinates are located and predicted peak values ​​are extracted based on the transient temperature field prediction value to obtain the early warning judgment result.

8. A multi-source data fusion-based intelligent early warning system for abnormal temperature rise in display modules, used to implement the multi-source data fusion-based intelligent early warning method for abnormal temperature rise in display modules as described in claims 1-7, characterized in that, include: The data acquisition module is used to pre-build the thermal resistance-capacitance network model and temperature prediction model of the display module; It also acquires the electrical excitation data and thermal observation data of the display module, performs timestamp mapping processing on the electrical excitation data and the thermal observation data, and obtains the basic parameters; The deviation calculation module is used to perform product calculation based on the electrical excitation data in the basic parameters to obtain transient spatial power density data, and to calculate the initial temperature data in combination with the thermal resistance-capacitance network model. The initial temperature data and the thermal observation data are synchronously compared and calculated to generate a temperature deviation value. The task processing module is used to obtain the initial thermal resistance and capacitance parameters of the display module, correct the initial thermal resistance and capacitance parameters according to the temperature deviation value, and generate parameter correction values. The analysis module is used to combine the parameter correction values ​​and the transient spatial power density data to calculate the deviation of the heat conduction equation of the temperature prediction model and generate the transient temperature field prediction value. The index extraction module is used to extract spatial temperature gradient data and transient temperature rise rate data based on the transient temperature field prediction value, and generate hot spot early warning index. The judgment module is used to obtain the safety limit threshold by extracting the thermodynamic properties of the display module material, and compare the safety limit threshold with the hot spot warning index to generate a warning judgment result. The execution module is used to execute the warning judgment result and output the hot spot abnormality warning signal of the display module.

Images