An OLED module gamma automatic adjustment system and method

Abstract

The present application relates to photoelectric display module calibration and display device manufacturing technical field, disclose a kind of OLED module Gamma automatic calibration system and method.Method acquires the luminance response data of OLED module, temperature data and driving electrical response data, form module state characteristic data;Edge computing module generates standardization luminance curve data, aging state quantity and initial compensation parameter;Establish temperature aging coupling compensation model, through adaptive global optimization algorithm forms register compensation parameter;Write in OLED module register after optical verification, according to optical verification deviation closed loop iteration.System includes data acquisition module, edge computing module, compensation modeling module, parameter optimization module, automatic calibration module, optical verification module and calibration result database.The present application can improve OLED module Gamma calibration consistency, reduce low gray scale deviation and reduce iteration number.

Description

Technical Field

[0001] This invention relates to the field of optoelectronic display module calibration and display device manufacturing technology, specifically to an automatic Gamma calibration system and method for OLED modules. Background Technology

[0002] OLED modules typically include an OLED display panel, display driving circuitry, flexible circuit board, packaging structure, and related connection structures. Before OLED modules enter final assembly or factory testing, their grayscale brightness response needs to be calibrated using Gamma adjustment to ensure that the non-linear relationship between input grayscale and output brightness conforms to the target Gamma curve. Deviations in the Gamma curve affect dark area detail, brightness transition, and color consistency; therefore, Gamma calibration is a crucial step in OLED module production.

[0003] Existing OLED gamma calibration solutions mostly focus on the panel stage or three-channel register parameter calibration. By measuring the brightness values ​​of several grayscale points and adjusting the register parameters, the OLED display panel is brought closer to the target gamma curve. Such solutions have a certain degree of automation on the production line, but the calibration target is often biased towards the OLED display panel itself, and does not fully cover the combined influence of the display driving circuit, packaging structure, electrical connection status, and module-level heat dissipation status on the brightness response within the OLED module.

[0004] OLED modules exhibit brightness response drift under different temperature conditions. The carrier mobility, luminous efficiency, and drive current stability of the same OLED module differ at low, room, and high temperatures, resulting in different Gamma curves for the same register parameters at different temperatures. OLED modules also experience aging effects with increasing usage time, manifesting as threshold voltage shifts, leakage current changes, and brightness decay. If the Gamma calibration process only writes registers based on the initial brightness response, the Gamma curve is prone to shift during subsequent use.

[0005] In mass production scenarios, OLED modules from different production batches exhibit process discreteness. OLED modules within the same batch can share some process characteristics, but initial brightness response, threshold voltage offset, and leakage current variations differ between different batches. If each OLED module starts searching from completely random compensation parameters, it will increase the number of calibration iterations and reduce production cycle time. If only the register values ​​of the previous OLED module are used as the initial values ​​for the next OLED module, incorrect initial values ​​can easily be introduced when switching batches or when there are significant differences in aging states.

[0006] Furthermore, the driving physical mechanisms corresponding to the low grayscale region and the high grayscale region of an OLED module are not the same. The low grayscale region is more susceptible to changes in leakage current, while the high grayscale region is more susceptible to threshold voltage shifts. If a single aging state quantity is used for compensation across the entire grayscale stage, it can easily mask the driving differences between different grayscale regions, leading to calibration deviations such as overcompensation in the low grayscale region and undercompensation in the high grayscale region.

[0007] Therefore, there is a need for an automatic gamma calibration system and method for OLED module production scenarios, which can integrate brightness response data, temperature data, driving electrical response data, production batch characteristics and historical calibration results data to form a register compensation parameter that can be iterated in a closed loop, and can introduce an aging compensation mechanism that matches the physical state of the OLED module driving in different grayscale regions. Summary of the Invention

[0008] The purpose of this invention is to provide an automatic Gamma calibration system and method for OLED modules, which solves the technical problems of insufficient module-level state coverage, insufficient compensation for temperature and aging factors, insufficient utilization of batch differences, confusion of aging mechanisms between low and high grayscale areas, and excessive calibration iterations in the existing OLED module Gamma calibration process.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: An automatic gamma calibration method for an OLED module, the automatic gamma calibration method for an OLED module includes: Collect brightness response data, temperature data, and driving electrical response data of the OLED module at multiple grayscale nodes to form module state characteristic data; Noise suppression, grayscale interpolation, and electrical feature decoupling are performed on the module state characteristic data to obtain standardized brightness curve data, aging state quantities, and initial compensation parameters. A temperature aging coupled compensation model is established based on standardized brightness curve data, aging state quantities, and initial compensation parameters. The temperature aging coupling compensation model is searched for parameters using an adaptive global optimization algorithm to generate register compensation parameters. The register compensation parameters are written into the OLED module register and the optical verification deviation is obtained. The candidate parameter set is adjusted according to the feedback of the optical verification deviation until the optical verification deviation meets the target Gamma curve constraint.

[0010] As a preferred technical solution of the present invention, the module state feature data includes the brightness value, ambient temperature value, threshold voltage offset, leakage current change and production batch identifier corresponding to the grayscale node. The threshold voltage offset is extracted from the forward current-voltage curve of the OLED module, and the leakage current change is extracted from the leakage current measurement value of the OLED module under reverse bias conditions.

[0011] As a preferred embodiment of the present invention, the noise suppression includes performing consistency screening on multiple brightness measurements of the same grayscale node, and performing Kalman filtering on the brightness measurements after consistency screening. The grayscale interpolation dynamically selects the interpolation density based on the brightness change rate of adjacent grayscale nodes, where the brightness change rate in the low grayscale region exceeds the brightness change rate in the adjacent middle grayscale region. When the grayscale value is doubled, interpolation nodes in the low grayscale area are added; the brightness change rate of adjacent grayscale values ​​is the unit grayscale brightness difference, and the reference brightness change rate of the medium grayscale area is the average grayscale brightness change rate of the entire area.

[0012] As a preferred technical solution of the present invention, the temperature aging coupling compensation model generates a reference grayscale brightness based on standardized brightness curve data, generates a temperature compensation coefficient based on temperature data, generates an aging attenuation coefficient based on aging state quantity, and maps the reference grayscale brightness, temperature compensation coefficient and aging attenuation coefficient together to the compensation brightness value of the target grayscale node.

[0013] As a preferred embodiment of the present invention, the initial compensation parameters are generated by the edge computing module based on the production batch identifier, threshold voltage offset, leakage current change and historical register compensation parameters in the calibration result database. The initial compensation parameters are used to limit the initial search range of the candidate parameter set of the adaptive global optimization algorithm.

[0014] As a preferred embodiment of the present invention, the adaptive global optimization algorithm uses the target Gamma curve deviation, the maximum grayscale deviation, and the register write change as optimization objectives, and generates multiple candidate groups of register compensation parameters based on the candidate parameter set. In each iteration, the parameter optimization module filters the candidate groups of register compensation parameters according to the optimization objectives and feeds back the filtering results to the candidate parameter set.

[0015] As a preferred embodiment of the present invention, the optical verification deviation includes root mean square deviation, maximum grayscale deviation and average grayscale deviation; when the optical verification deviation does not meet the target Gamma curve constraint, the parameter optimization module determines the deviation grayscale range according to the optical verification deviation, takes the register compensation parameter corresponding to the deviation grayscale range as the priority adjustment object, and narrows the search boundary corresponding to the non-deviation grayscale range.

[0016] As a preferred embodiment of the present invention, before the register compensation parameters are written into the OLED module register, the automatic calibration module establishes a register address mapping relationship according to the display driver interface protocol and performs double-byte verification on the register compensation parameters; after the writing is completed, the automatic calibration module reads the readback value of the OLED module register and triggers the optical verification module when the readback value is consistent with the register compensation parameters.

[0017] As a preferred technical solution of the present invention, the calibration result database stores the production batch identifier, standardized brightness curve data, temperature data, threshold voltage offset, leakage current change, register compensation parameters and optical verification deviation corresponding to each automatic Gamma calibration. The calibration result database provides the edge computing module with the initial compensation parameter query results for the same production batch or adjacent production batches.

[0018] As a preferred technical solution of the present invention, the edge computing module forms batch classification features based on production batch identifier, threshold voltage offset, leakage current change and historical register compensation parameters, and outputs batch category through batch classification model. The batch category is used to query the initial compensation parameters from the calibration result database.

[0019] As a preferred technical solution of the present invention, the historical register compensation parameters, historical temperature data, historical aging state quantities and historical optical verification deviations in the calibration result database are input into the drift prediction model. The drift prediction model outputs the compensation parameter adjustment amount for future working periods. The compensation parameter adjustment amount is used to update the initial compensation parameters. The input dimension of the drift prediction model is historical four-dimensional features, and the output dimension corresponds one-to-one with the dimension of the single batch module register compensation parameters.

[0020] As a preferred embodiment of the present invention, the drift prediction model includes an input normalization layer, a first long short-term memory network layer, a second long short-term memory network layer, and a fully connected output layer. The training samples of the drift prediction model consist of continuous... The training data consists of historical register compensation parameters, historical temperature data, historical aging state values, and historical optical verification deviations. The training labels are the compensation parameter adjustments of the corresponding OLED module after verification during subsequent working periods.

[0021] As a preferred embodiment of the present invention, the aging state variables in the temperature aging coupling compensation model are replaced by gray-level dynamic aging state variables. These gray-level dynamic aging state variables are obtained by dynamically decoupling and allocating the threshold voltage offset and leakage current change based on the driving physical region where the gray-level node is located, satisfying the following:

[0022] in, This represents the dynamic aging state of the grayscale at the current grayscale node. Assign a function value to the grayscale weight corresponding to the current grayscale node; The threshold voltage aging sub-state quantity is mapped from the threshold voltage offset; The leakage current aging sub-state quantity is mapped by the leakage current change. In the low gray level region, the gray level dynamic aging state quantity is dominated by the leakage current aging sub-state quantity. In the high gray level region, the gray level dynamic aging state quantity is dominated by the threshold voltage aging sub-state quantity.

[0023] As a preferred embodiment of the present invention, when the optical verification deviation does not satisfy the target Gamma curve constraint, the search boundary of the candidate parameter set of the adaptive global optimization algorithm is dynamically adjusted according to the gradient momentum of the optical verification deviation, satisfying:

[0024] in, For the first The search boundary adjustment amount for the next iteration; The iteration number; For the first The search boundary adjustment amount for the next iteration; This is the boundary contraction step size coefficient, with a value range of [value range missing]. to ; For gradient operators; For the first The optimization objective function corresponding to the deviation of the target Gamma curve in the next iteration is composed of the weighted sum of the root mean square deviation of the Gamma curve, the maximum gray level deviation, and the change in register writing. The change in register writing is normalized to a dimensionless value, the normalization reference is the full range of the register, the input is the set of register compensation parameters for the current iteration, and the output is the normalized optimization deviation value. For the first The set of register compensation parameters for the next iteration; This is the momentum compensation coefficient, with a value range of [value range missing]. to ; For the first The gradient momentum of the nth iteration; The gradient momentum of the next iteration is initialized by the gradient of the current parameters. The gradient momentum of subsequent iterations is formed by weighting the gradient momentum of the previous iteration with the gradient of the current parameters according to a momentum forgetting factor. The value range of the momentum forgetting factor is... to By introducing gradient momentum, the search boundary of the corresponding candidate parameter set is expanded when the optical verification deviation direction is consistent in multiple consecutive times, and the search boundary of the corresponding candidate parameter set is shrunk when the optical verification deviation direction oscillates.

[0025] Furthermore, this invention also discloses an OLED module gamma automatic calibration system. This system is used to execute the OLED module gamma automatic calibration method described above. The OLED module gamma automatic calibration system includes a data acquisition module, an edge computing module, a compensation modeling module, a parameter optimization module, an automatic calibration module, an optical verification module, and a calibration result database. The data acquisition module collects brightness response data, temperature data, and driving electrical response data of the OLED module at multiple grayscale nodes and outputs the module state characteristic data to the edge computing module. The edge computing module performs noise suppression, grayscale interpolation, electrical feature decoupling, and batch classification on the module state characteristic data and outputs standardized brightness curve data, aging state quantities, and initial compensation parameters to the compensation modeling module and parameter optimization module. The system comprises the following modules: an optimization module and a compensation modeling module. The former establishes a temperature-aging coupled compensation model based on standardized brightness curve data, aging state parameters, and initial compensation parameters, and outputs this model to the parameter optimization module. The latter generates register compensation parameters based on the temperature-aging coupled compensation model and optical verification deviation, and outputs these parameters to the automatic calibration module. The former writes the register compensation parameters into the OLED module registers via the display driver interface and outputs the write confirmation result to the optical verification module. The latter measures the OLED module's Gamma curve after writing and generates an optical verification deviation, which is then fed back to the parameter optimization module. The calibration result database stores module state characteristic data, register compensation parameters, and optical verification deviation, and provides initial compensation parameter query results to the edge computing module.

[0026] Compared with the prior art, the present invention has the following beneficial effects: This invention collects brightness response data, temperature data, and drive electrical response data to form module state characteristic data, expanding the automatic Gamma calibration process from single brightness correction to module-level state compensation, effectively improving calibration consistency. By establishing a temperature aging coupled compensation model, temperature changes, threshold voltage shifts, and leakage current changes are jointly introduced into the register compensation parameter formation process, improving the stability of the calibrated Gamma curve and reducing Gamma curve shifts caused by ambient temperature and aging conditions. Initial compensation parameters are provided through a calibration result database and batch classification model, transforming the search range of the adaptive global optimization algorithm from a completely random search to a batch-related local constraint search, reducing the number of iterations and improving batch production cycle time.

[0027] This invention distinguishes the dominant aging factors in low-grayscale and high-grayscale regions by using dynamic grayscale aging state quantities. This ensures that compensation in low-grayscale regions is primarily affected by leakage current changes, while compensation in high-grayscale regions is mainly affected by threshold voltage shifts. This reduces system deviations caused by uniform compensation across the entire grayscale stage and improves the consistency of dark area gradation and brightness transitions. By adjusting the search boundary through optical verification deviation and gradient momentum, the adaptive global optimization algorithm can switch between global exploration and local fine-tuning, further improving calibration efficiency and stability. The overall solution integrates data acquisition, model compensation, parameter optimization, register writing, optical verification, and historical data updates in a closed loop, enhancing the feasibility and adaptability of automatic Gamma calibration for OLED modules. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0029] Figure 1 This is a flowchart of the calibration method of the present invention.

[0030] Figure 2 This is a flowchart of the sub-process for forming grayscale dynamic aging state quantities in this invention.

[0031] Figure 3 This is a flowchart of the collaborative process between the batch classification model and the drift prediction model of this invention. Detailed Implementation

[0032] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0033] The following is in conjunction with the appendix Figures 1-3 The embodiments of the present invention will be described in detail below.

[0034] Example 1: This example provides an automatic gamma calibration method for OLED modules. After the OLED module enters the calibration station, the data acquisition module collects brightness response data, temperature data, and driving electrical response data of the OLED module at multiple grayscale nodes. The grayscale nodes can cover... to Grayscale range.

[0035] For example, the number of grayscale nodes is There are [number] grayscale nodes, with an interval of [number] between adjacent grayscale nodes. For low grayscale areas, the brightness change rate between adjacent grayscale levels is the unit grayscale brightness difference; for mid-grayscale areas, the reference brightness change rate is the average of the brightness change rates across the entire grayscale area; when the brightness change rate of any adjacent node in a low grayscale area exceeds the reference brightness change rate in a mid-grayscale area... When the grayscale node spacing in the corresponding low grayscale range is increased by 100%, the edge computing module will adjust the grayscale node spacing of the corresponding low grayscale range from 10 ... Adjusted to This is to increase the sampling density of the brightness response in dark areas.

[0036] The data acquisition module performs data acquisition on each grayscale node. Secondary brightness measurement. Edge computing module for... Consistency screening is performed on any brightness measurement value; when any brightness measurement value is consistent with... The average deviation of the secondary brightness measurement values ​​exceeds At that time, the edge computing module removes the brightness measurement value with the largest deviation and averages the remaining brightness measurements. The edge computing module then applies a Kalman filter to the averaged brightness data to generate standardized brightness curve data. The standardized brightness curve data is normalized using the value corresponding to the maximum brightness as the reference, and the standardized brightness value corresponding to the grayscale node is located at... to between.

[0037] The data acquisition module synchronously collects temperature data from the OLED module. This temperature data includes the ambient temperature of the calibration station and the surface temperature of the OLED module.

[0038] In some preferred embodiments, the temperature sensor is located on the back of the OLED module near the display driving circuit, and the temperature sampling period is [missing information]. The edge computing module performs a moving average processing on the temperature data to generate an effective temperature value within the calibration cycle. For example, the ambient temperature at the calibration workstation is... The surface temperature of the OLED module is The edge computing module uses the surface temperature of the OLED module as the temperature input for the temperature aging coupling compensation model.

[0039] The data acquisition module also collects the driving electrical response data of the OLED module. This data includes forward current-voltage curves and leakage current measurements under reverse bias conditions. The edge computing module extracts the threshold voltage offset from the forward current-voltage curves and the leakage current variation from the leakage current measurements. The threshold voltage offset characterizes the aging degree of the OLED module's driving transistors in the high grayscale region. The leakage current variation characterizes the impact of leakage current on the OLED module's driving transistors in the low grayscale region.

[0040] The edge computing module combines standardized brightness curve data, temperature data, threshold voltage offset, leakage current variation, and production batch identifiers into module status characteristic data. This module status characteristic data serves as input for subsequent temperature aging coupling compensation models and adaptive global optimization algorithms. At this point, the front-end acquired data, temperature data, and drive electrical response data have all been transformed into technical objects used in subsequent steps; no data remains isolated.

[0041] The compensation modeling module generates a baseline grayscale brightness based on standardized brightness curve data, a temperature compensation coefficient based on temperature data, and an aging attenuation coefficient based on aging state parameters. The module then maps the baseline grayscale brightness, temperature compensation coefficient, and aging attenuation coefficient to the compensated brightness value of the target grayscale node. The temperature compensation coefficient is determined by... The aging degradation coefficient is formed by adding a linear temperature term coefficient multiplied by the difference between the actual and reference temperatures, and then adding a nonlinear temperature term coefficient multiplied by the square of that temperature difference. The aging degradation parameter is multiplied by the aging state quantity. Through the aforementioned processing, the effects of temperature change on brightness response and the effects of aging state on brightness response are coupled and expressed in the same model.

[0042] For example, grayscale The standardized brightness value at that location is The reference temperature is ℃, actual temperature is ℃, aging state quantity is The linear temperature coefficient is The coefficient of the nonlinear temperature term is The aging degradation parameter is The compensation modeling module is based on the temperature difference. Obtain the temperature compensation coefficient And obtain the aging decay coefficient based on the aging state quantity. The compensation modeling module will , and Multiply to obtain grayscale. Compensation brightness value at the location .

[0043] The parameter optimization module receives the temperature aging coupled compensation model and initial compensation parameters. The initial compensation parameters are generated by the edge computing module based on the production batch identifier, threshold voltage offset, leakage current variation, and historical register compensation parameters from the calibration result database. The parameter optimization module uses the target Gamma curve deviation, maximum grayscale deviation, and register write change as optimization objectives. The target Gamma curve deviation measures the overall curve consistency. The maximum grayscale deviation limits excessive deviation at a single grayscale node. The register write change prevents abrupt changes in register compensation parameters between adjacent iterations.

[0044] For example, the weight of the deviation from the target Gamma curve is The weight of the maximum grayscale deviation is The weight of the change written to the register is .

[0045] The deviation of the target Gamma curve corresponding to a certain candidate register compensation parameter is The maximum grayscale deviation is The normalized value of the register write change is The parameter optimization module obtains the comprehensive target value. The parameter optimization module compares the combined target values ​​of multiple candidate register compensation parameters and selects the candidate register compensation parameter with the smaller combined target value to proceed to the next iteration.

[0046] The adaptive global optimization algorithm can be executed using a combination of genetic search and local contraction. The parameter optimization module generates a candidate parameter set based on the initial compensation parameters, with each candidate corresponding to a set of register compensation parameters. The module inputs these candidates into a temperature-aging coupled compensation model to obtain candidate compensation brightness curves, which are then compared with the target Gamma curve. Based on the comparison results, the module performs selection, crossover, and mutation operations to obtain the next round of candidate parameter sets. The parameter optimization module continuously... The change in the overall objective value during the round of iterations is less than The iteration can be stopped when the number of iterations reaches a preset limit, such as 1. Second-rate.

[0047] After receiving the register compensation parameters, the automatic calibration module establishes a register address mapping relationship. This mapping relationship is formed based on the OLED module's display driver interface protocol and register table. The display driver interface can be a mobile industrial processor interface or an embedded display port. The automatic calibration module converts the register compensation parameters into register write data and performs double-byte verification. After writing the register compensation parameters, the automatic calibration module reads the OLED module's register readback value. If the readback value matches the register compensation parameters, the automatic calibration module outputs a write confirmation result to the optical verification module.

[0048] The optical verification module performs Gamma curve verification on the OLED module after writing. The module collects brightness response data after writing, generates an actual Gamma curve, and compares it with a target Gamma curve to obtain the optical verification deviation. The optical verification deviation includes root mean square deviation, maximum grayscale deviation, and average grayscale deviation.

[0049] The entire document adopts a unified standard for judging the pass rate of optical verification: simultaneously satisfying the condition that the root mean square deviation of the Gamma curve is no greater than [value missing]. The maximum grayscale deviation is no greater than The average grayscale deviation is no greater than If the target Gamma curve constraint is met, the iterative calibration stops. If any deviation fails to meet the target Gamma curve constraint, the optical verification module feeds back the optical verification deviation to the parameter optimization module.

[0050] The parameter optimization module determines the deviation grayscale range based on the optical verification deviation. For example, if the optical verification deviation in the low grayscale region is greater than that in the medium and high grayscale regions, the parameter optimization module prioritizes adjusting the register compensation parameters corresponding to the low grayscale region and narrows the search boundaries for the medium and high grayscale regions. The parameter optimization module then generates register compensation parameters again based on the new set of candidate parameters. The automatic calibration module writes the register compensation parameters back into the system. The optical verification module performs verification again. This closed-loop iteration continues until the optical verification deviation meets the target Gamma curve constraint.

[0051] Example 2: This example, based on Example 1, further illustrates the formation process of the grayscale dynamic aging state quantity. The pixel driving circuit of the OLED module operates within different semiconductor physical operating ranges under different luminous brightness levels. The thin-film transistors corresponding to the low grayscale region are closer to the subthreshold driving region, and the brightness non-uniformity is mainly affected by leakage current changes. The thin-film transistors corresponding to the high grayscale region are closer to the saturation driving region, and the output current is mainly affected by threshold voltage shift. If a single aging state quantity is used for compensation across the entire grayscale stage, the dominant aging factors in the low and high grayscale regions will be confused, easily leading to overcompensation in the low grayscale region or undercompensation in the high grayscale region.

[0052] The edge computing module receives the driving electrical response data. It maps the threshold voltage offset to a dimensionless threshold voltage aging sub-state variable. It also maps the leakage current change to a dimensionless leakage current aging sub-state variable. The compensation modeling module independently calculates the gray-level dynamic aging state variable for each gray-level node. The gray-level dynamic aging state variable satisfies the following formula:

[0053] in, This represents the dynamic aging state of the grayscale at the current grayscale node. Assign a function value to the grayscale weight corresponding to the current grayscale node; The threshold voltage aging sub-state quantity is mapped from the threshold voltage offset; The leakage current aging sub-state quantity is mapped from the leakage current change.

[0054] The grayscale weight allocation function value increases smoothly from low grayscale areas to high grayscale areas as the grayscale nodes move. The grayscale weight allocation function value can adopt an sigmoid function with natural constants. The base is the negative exponential weight transition steepness coefficient, multiplied by the difference between the current grayscale node and the critical grayscale threshold, and the output is... Divide by The value is added to the power of this exponent. The critical grayscale threshold is used to distinguish between the subthreshold driving region and the saturation driving region of the OLED module. The weight transition steepness coefficient is used to control the transition speed of the grayscale weight allocation function value near the critical grayscale threshold.

[0055] For example, for OLED modules with low-temperature polycrystalline silicon oxide backplanes, the critical grayscale threshold is set to... The weight transition steepness coefficient is set to The threshold voltage aging substate variable is: The leakage current aging substate variable is When the current grayscale node is At that time, the grayscale weight allocation function value is The grayscale dynamic aging state quantity is calculated as follows: The calculation results are close to the leakage current aging sub-state quantity, indicating that compensation in the low grayscale region is mainly affected by changes in leakage current. When the current grayscale node is At that time, the grayscale weight allocation function value is The grayscale dynamic aging state quantity is calculated as follows: The calculation results are close to the threshold voltage aging sub-state variables, indicating that the compensation in the high grayscale region is mainly affected by the threshold voltage offset.

[0056] The compensation modeling module inputs the grayscale dynamic aging state variables into the temperature aging coupled compensation model, replacing the single aging state variable. Thus, the temperature aging coupled compensation model primarily performs aging compensation based on leakage current changes in the low grayscale region, and primarily based on threshold voltage offsets in the high grayscale region. The output of the grayscale dynamic aging state variables is then used to generate the compensated brightness values ​​for the target grayscale nodes. These compensated brightness values ​​are further used by the parameter optimization module to form register compensation parameters. Therefore, the grayscale dynamic aging state variables are not isolated intermediate results, but rather intermediate technical objects in the temperature aging coupled compensation model and register compensation parameter generation process.

[0057] Example 3: This example, based on Example 1, further illustrates the collaborative method of the batch classification model, drift prediction model, and calibration result database. The calibration result database stores the production batch identifier, standardized brightness curve data, temperature data, threshold voltage offset, leakage current change, register compensation parameters, and optical verification deviation corresponding to each automatic Gamma calibration. Each record in the calibration result database is bound to the OLED module serial number and production batch identifier for querying by the edge computing module.

[0058] The edge computing module generates batch classification features based on production batch identifiers, threshold voltage offsets, leakage current variations, and historical register compensation parameters. The batch classification model includes an input normalization layer, a feature extraction layer, and a classification output layer. The input normalization layer converts production batch identifier codes, threshold voltage offsets, leakage current variations, and historical register compensation parameters of different dimensions into feature vectors of a uniform scale. The feature extraction layer can include two fully connected layers; the first fully connected layer outputs a dimension of... The output dimension of the second fully connected layer is The classification output layer outputs the batch category. When training the batch classification model, the edge computing module extracts historical calibration records from the calibration result database, using production batch identifiers and optical verification pass status as supervision information. The cross-entropy loss function is used to train the batch classification model, and the Adam optimizer can be used. The learning rate can be set to... The number of training rounds can be set to The batch size can be set to... The learning rate, training rounds, and batch size mentioned above are the optimal parameters for batch calibration of OLED modules, and can be fine-tuned according to module size, backplane process, and production conditions.

[0059] After the batch classification model outputs the batch category, the edge computing module queries the calibration result database for historical register compensation parameters from the same or adjacent batches based on the batch category. The edge computing module performs consistency screening on the retrieved historical register compensation parameters, removing records where the optical verification deviation exceeds the target Gamma curve constraint, and takes the median value of the remaining historical register compensation parameters to form the initial compensation parameters. These initial compensation parameters are then used by the parameter optimization module to limit the initial search range of the candidate parameter set.

[0060] The drift prediction model is used to predict the adjustment amount of compensation parameters for future working periods based on historical data. The drift prediction model consists of an input normalization layer, a first long short-term memory (LSM) network layer, a second LSM network layer, and a fully connected output layer. The input normalization layer receives historical register compensation parameters, historical temperature data, historical aging state values, and historical optical verification bias. The number of hidden units in the first LSM network layer can be set to... The number of hidden units in the second Long Short-Term Memory (LSTM) layer can be set to... The output dimension of the fully connected output layer is consistent with the dimension of the register compensation parameters. The input dimension of the drift prediction model is historical four-dimensional features, namely historical temperature data, historical aging state quantities, historical register compensation parameters, and historical optical verification bias; the output dimension corresponds one-to-one with the dimension of the register compensation parameters of a single batch of modules. For example, each set of register compensation parameters includes... The compensation value corresponding to each gray level. The training samples of the drift prediction model consist of continuous... The data consists of historical register compensation parameters, historical temperature data, historical aging state parameters, and historical optical verification deviations. The sampling interval can be... The training labels are the adjusted compensation parameters for the corresponding OLED module after verification during subsequent operating periods. The drift prediction model is trained using the mean squared error loss function, and the optimizer can be the Adam optimizer, with the learning rate set to... The batch size can be set to The number of training rounds can be set to The wheel can also be used to verify continuous loss. Training stops when the number of training rounds does not decrease. The learning rate, batch size, and number of training rounds mentioned above are the optimal parameters for batch calibration of OLED modules and can be fine-tuned according to module size, backplane process, and production conditions.

[0061] After the drift prediction model outputs the adjustment amount of the compensation parameters, the edge computing module adds the adjustment amount to the initial compensation parameters to form the updated initial compensation parameters. The updated initial compensation parameters are not directly written to the OLED module register, but instead serve as the center of the candidate parameter set for the parameter optimization module. The parameter optimization module still needs to search and verify based on the temperature aging coupled compensation model and optical verification deviation. Therefore, the output of the drift prediction model is used for optimization starting point selection, rather than replacing closed-loop optical verification.

[0062] In some preferred embodiments, the calibration result database stores no less than Each calibration record includes the OLED module serial number, production batch identifier, grayscale node brightness data, temperature data, threshold voltage offset, leakage current change, register compensation parameters, and optical verification deviation. The edge computing module writes the new calibration record to the calibration result database after each calibration. As the number of records in the calibration result database increases, the batch classification model and drift prediction model can be updated periodically to improve the matching accuracy of the initial compensation parameters.

[0063] For example, in a production batch, the edge computing module queries... One historical calibration record, among which The records satisfy the target Gamma curve constraint. Edge computing module removed. Records that do not meet the target Gamma curve constraint, and... The median values ​​of the register compensation parameters in each record are statistically analyzed to form initial compensation parameters. The parameter optimization module generates a set of candidate parameters centered on these initial compensation parameters, with the initial search range set to the allowable adjustment range of the entire register scale. The full-range adjustable range of the register is to The range of parameter values ​​corresponding to grayscale. Using this method, the average number of iterations is reduced from... The next reduction to Second-rate.

[0064] In some preferred embodiments, the search boundary of the candidate parameter set of the adaptive global optimization algorithm is dynamically adjusted based on the gradient momentum of the optical verification bias. The search boundary adjustment amount satisfies the following formula:

[0065] in, For the first The search boundary adjustment amount for the next iteration; The iteration number; For the first The search boundary adjustment amount for the next iteration; This is the boundary contraction step size coefficient, with a value range of [value range missing]. to ; For gradient operators; For the first The optimization objective function corresponding to the deviation of the target Gamma curve in the next iteration is composed of the weighted sum of the root mean square deviation of the Gamma curve, the maximum gray level deviation, and the change in register writing. The change in register writing is normalized to a dimensionless value, the normalization reference is the full range of the register, the input is the set of register compensation parameters for the current iteration, and the output is the normalized optimization deviation value. For the first The set of register compensation parameters for the next iteration; This is the momentum compensation coefficient, with a value range of [value range missing]. to ; For the first Gradient momentum of the next iteration.

[0066] Edge computing module in the After the second iteration, the optical verification bias is obtained. If the parameter gradient direction points to increasing the value written to the register, the edge computing module records the parameter gradient and initializes the gradient momentum. The optical verification data after the second iteration shows that when the actual brightness is still lower than the target Gamma curve, the parameter gradient direction is similar to that of the third iteration. The iterations remain consistent, with the gradient momentum formed by weighting the previous gradient momentum and the current parameter gradient according to a momentum forgetting factor. The momentum forgetting factor takes values ​​ranging from [value missing]. to The gradient momentum increases during the accumulation process. In updating the... When searching the boundary in the next iteration, the momentum compensation term, which includes the gradient momentum, moves the search boundary of the candidate parameter set toward the direction of increasing register write values ​​in order to cross the smooth deviation interval.

[0067] If the first The actual brightness after the next iteration is higher than the target Gamma curve, causing the parameter gradient direction to reverse, and the new parameter gradient cancels out the accumulated gradient momentum. Based on this, the parameter optimization module determines that the current register compensation parameter is close to the region corresponding to the target Gamma curve, and shrinks the search boundary of the candidate parameter set near the current register compensation parameter. Thus, the adaptive global optimization algorithm shifts from a large-scale search to local fine-tuning.

[0068] Comparative Example 1: The difference between this comparative example and Example 1 is that the compensation modeling module does not establish a temperature aging coupled compensation model, but only generates register compensation parameters based on standardized brightness curve data. The data acquisition module still collects brightness response data, temperature data, and drive electrical response data, but the temperature data, threshold voltage offset, and leakage current change do not participate in the register compensation parameter generation process.

[0069] Under the same OLED module, the same grayscale node, and the same target Gamma curve constraints, the optical verification module verifies the calibrated OLED module. For example, in At ℃, the root mean square deviation of Comparative Example 1 is The maximum grayscale deviation is This allows it to approach the target Gamma curve constraint. At ℃, the root mean square deviation of Comparative Example 1 increased to The maximum grayscale deviation increased to Since temperature data and aging state parameters are not included in the formation process of register compensation parameters, the parameter optimization module cannot adjust the candidate parameter set according to temperature changes and aging state, resulting in increased deviation of the Gamma curve under high temperature conditions.

[0070] Compared to Example 1, Comparative Example 1 lacks a temperature aging coupling compensation model. Example 1 in The root mean square deviation at ℃ is The maximum grayscale deviation is Therefore, it can be seen that the temperature aging coupled compensation model, by inputting temperature data and aging state variables into the register compensation parameter generation process, can reduce the Gamma curve offset caused by ambient temperature and aging state.

[0071] Comparative Example 2: The difference between this comparative example and Example 2 is that the compensation modeling module uses a single aging state quantity and does not dynamically allocate the threshold voltage aging sub-state quantity and leakage current aging sub-state quantity according to the grayscale nodes. The data acquisition module still collects the threshold voltage offset and leakage current change, and the edge computing module still forms the aging state quantity, but the compensation modeling module applies the aging state quantity uniformly to the entire grayscale stage.

[0072] For example, for the same production batch Comparative tests were conducted on OLED modules. The optical verification module recorded the optical verification deviations in the low grayscale and high grayscale regions, and calculated the average number of iterations required for a single OLED module to reach the target Gamma curve constraint. Using global aging and conventional boundary shrinkage, the optical verification deviation in the low grayscale region was... The optical verification deviation in the high grayscale region is The average number of iterations is Next, the adjustment time is / piece. When using grayscale dynamic aging state parameters and conventional boundary shrinkage, the optical verification deviation in the low grayscale region is: The optical verification deviation in the high grayscale region is The average number of iterations is Next, the adjustment time is / piece. When using global aging amount and gradient momentum search boundary adjustment, the optical verification deviation in the low grayscale region is: The optical verification deviation in the high grayscale region is The average number of iterations is Next, the adjustment time is / piece. When adjusting the grayscale dynamic aging state quantity and gradient momentum search boundary, the optical verification deviation in the low grayscale region is: The optical verification deviation in the high grayscale region is The average number of iterations is Next, the adjustment time is / piece.

[0073] The above comparison illustrates that a single aging state variable cannot distinguish the driving differences between low-grayscale and high-grayscale regions. The dynamic grayscale aging state variable allows compensation in low-grayscale regions to be primarily influenced by the leakage current aging sub-state variable, while compensation in high-grayscale regions is mainly influenced by the threshold voltage aging sub-state variable, thereby reducing optical verification deviations in different grayscale regions. Gradient momentum search boundary adjustment can control the search boundary based on changes in the optical verification deviation direction, enabling the adaptive global optimization algorithm to expand the search range when the deviation direction is consistent and shrink the search range when the deviation direction oscillates, thus reducing the number of iterations.

[0074] The OLED module gamma automatic calibration system of this invention can be integrated into an automated OLED module production line. An automatic loading and unloading mechanism delivers the OLED module to the calibration station. The data acquisition module acquires brightness response data, temperature data, and drive electrical response data. The edge computing module performs noise suppression, grayscale interpolation, electrical feature decoupling, batch classification, and initial compensation parameter generation. The compensation modeling module establishes a temperature aging coupled compensation model. The parameter optimization module generates register compensation parameters. The automatic calibration module writes the register compensation parameters. The optical verification module outputs the optical verification deviation. The calibration result database records the complete calibration process and provides initial compensation parameter query results for subsequent OLED modules. These modules constitute a closed-loop process from data acquisition, model compensation, parameter optimization, register writing, optical verification to historical data updates, enabling the OLED module gamma automatic calibration to be applicable to mass production environments.

[0075] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for automatic Gamma adjustment of an OLED module, characterized in that, The automatic Gamma adjustment method for the OLED module includes: Collect brightness response data, temperature data, and driving electrical response data of the OLED module at multiple grayscale nodes to form module state characteristic data; Noise suppression, grayscale interpolation, and electrical feature decoupling are performed on the module state characteristic data to obtain standardized brightness curve data, aging state quantities, and initial compensation parameters. A temperature aging coupled compensation model is established based on standardized brightness curve data, aging state quantities, and initial compensation parameters. The temperature aging coupling compensation model is searched for parameters using an adaptive global optimization algorithm to generate register compensation parameters. The register compensation parameters are written into the OLED module register and the optical verification deviation is obtained. The candidate parameter set is adjusted according to the feedback of the optical verification deviation until the optical verification deviation meets the target Gamma curve constraint.

2. The automatic Gamma adjustment method for an OLED module according to claim 1, characterized in that, The module status feature data includes the brightness value corresponding to the grayscale node, the ambient temperature value, the threshold voltage offset, the leakage current change, and the production batch identifier. The threshold voltage offset is extracted from the forward current-voltage curve of the OLED module, and the leakage current change is extracted from the leakage current measurement value of the OLED module under reverse bias conditions.

3. The automatic Gamma adjustment method for an OLED module according to claim 2, characterized in that, The noise suppression includes consistency screening of multiple brightness measurements at the same grayscale node, followed by Kalman filtering of the consistency-screened brightness measurements. The grayscale interpolation dynamically selects the interpolation density based on the brightness change rate of adjacent grayscale nodes, ensuring that the brightness change rate in the low grayscale region exceeds the brightness change rate in the adjacent middle grayscale region. When the grayscale value is doubled, interpolation nodes in the low grayscale area are added; the brightness change rate of adjacent grayscale values ​​is the unit grayscale brightness difference, and the reference brightness change rate of the medium grayscale area is the average grayscale brightness change rate of the entire area.

4. The automatic Gamma adjustment method for an OLED module according to claim 3, characterized in that, The temperature aging coupled compensation model generates a reference grayscale brightness based on standardized brightness curve data, generates a temperature compensation coefficient based on temperature data, generates an aging attenuation coefficient based on aging state quantities, and maps the reference grayscale brightness, temperature compensation coefficient, and aging attenuation coefficient together to the compensation brightness value of the target grayscale node.

5. The automatic Gamma adjustment method for an OLED module according to claim 4, characterized in that, The initial compensation parameters are generated by the edge computing module based on the production batch identifier, threshold voltage offset, leakage current change, and historical register compensation parameters in the calibration result database. The initial compensation parameters are used to limit the initial search range of the candidate parameter set of the adaptive global optimization algorithm.

6. The automatic Gamma adjustment method for an OLED module according to claim 5, characterized in that, The adaptive global optimization algorithm uses the target Gamma curve deviation, the maximum grayscale deviation, and the register write change as optimization objectives, and generates multiple candidate groups of register compensation parameters based on the candidate parameter set. In each iteration, the parameter optimization module filters the candidate groups of register compensation parameters according to the optimization objectives and feeds back the filtering results to the candidate parameter set.

7. The automatic Gamma adjustment method for an OLED module according to claim 6, characterized in that, The optical verification deviation includes root mean square deviation, maximum grayscale deviation, and average grayscale deviation. When the optical verification deviation does not meet the target Gamma curve constraint, the parameter optimization module determines the deviation grayscale range based on the optical verification deviation, prioritizes the adjustment of the register compensation parameters corresponding to the deviation grayscale range, and narrows the search boundary corresponding to the non-deviation grayscale range.

8. The automatic Gamma adjustment method for an OLED module according to claim 7, characterized in that, Before the register compensation parameters are written into the OLED module register, the automatic calibration module establishes a register address mapping relationship according to the display driver interface protocol and performs double-byte verification on the register compensation parameters. After the writing is completed, the automatic calibration module reads the readback value of the OLED module register and triggers the optical verification module when the readback value matches the register compensation parameter.

9. The automatic Gamma adjustment method for an OLED module according to claim 4, characterized in that, The aging state variables in the temperature aging coupling compensation model are replaced by gray-level dynamic aging state variables. These gray-level dynamic aging state variables are obtained by dynamically decoupling and allocating the threshold voltage offset and leakage current change based on the driving physical region where the gray-level node is located, satisfying the following: in, This represents the dynamic aging state of the grayscale at the current grayscale node. Assign a function value to the grayscale weight corresponding to the current grayscale node; The threshold voltage aging sub-state quantity is mapped from the threshold voltage offset; The leakage current aging sub-state quantity is mapped by the leakage current change. In the low gray level region, the gray level dynamic aging state quantity is dominated by the leakage current aging sub-state quantity. In the high gray level region, the gray level dynamic aging state quantity is dominated by the threshold voltage aging sub-state quantity.

10. An automatic gamma calibration system for an OLED module, characterized in that, The OLED module Gamma automatic calibration system is used to execute the OLED module Gamma automatic calibration method according to any one of claims 1 to 9. The system includes a data acquisition module, an edge computing module, a compensation modeling module, a parameter optimization module, an automatic calibration module, an optical verification module, and a calibration result database. The data acquisition module is used to acquire brightness response data, temperature data, and driving electrical response data of the OLED module at multiple grayscale nodes, and output the module state characteristic data to the edge computing module. The edge computing module is used to perform noise suppression, grayscale interpolation, electrical feature decoupling, and batch classification on the module state characteristic data, and output the standardized brightness curve data, aging state quantity, and initial compensation parameters to the compensation modeling module and the parameter optimization module. The compensation modeling module is used to establish a temperature aging coupled compensation model based on standardized brightness curve data, aging state quantities and initial compensation parameters, and output the temperature aging coupled compensation model to the parameter optimization module. The parameter optimization module generates register compensation parameters based on the temperature aging coupling compensation model and optical verification deviation, and outputs the register compensation parameters to the automatic calibration module; the automatic calibration module writes the register compensation parameters into the OLED module register through the display driver interface, and outputs the write confirmation result to the optical verification module. The optical verification module is used to measure the Gamma curve of the OLED module after writing and form the optical verification deviation, which is fed back to the parameter optimization module. The calibration result database is used to store module state characteristic data, register compensation parameters and optical verification deviation, and to provide the edge computing module with the initial compensation parameter query results.

Images