A method for correcting test data of a liquid crystal display module

Abstract

This application discloses a test data correction method for a liquid crystal display module, comprising: collecting touch pressure values ​​to construct a touch response curve, calculating curvature and torsion, mapping sampling points to phase space to form a phase space trajectory of a historical curve, collecting real-time touch pressure values ​​to obtain the phase space trajectory of the measured curve and cutting it into a target segment, cutting historical segments of the same length from the phase space trajectory of the historical curve, calculating manifold distance and taking the historical segment with the smallest distance as the best matching reference segment, comparing the measured pressure value of the target segment with the corresponding matching historical reference pressure value to obtain the pressure deviation, and finally outputting the corrected touch test data. This avoids cross-stage mismatch, and the matching process does not depend on time axis alignment, eliminating the matching misalignment problem caused by the randomness of the touch-off moment.

Description

Technical Field

[0001] This invention relates to the field of liquid crystal data correction technology, and in particular to a test data correction method for liquid crystal display modules. Background Technology

[0002] As a core display component of modern electronic devices, the reliability and accuracy of the touch function of a Liquid Crystal Display Module (LCM) directly impacts the user experience. Touch performance testing is an essential step in the research, development, production, and quality inspection of LCMs. By applying touch operations to the module and collecting the corresponding pressure response signals, key indicators such as touch sensitivity, response consistency, and anti-interference capabilities can be evaluated. However, in actual testing, factors such as environmental noise, sensor nonlinearity, individual module differences, and the randomness of testing operations often result in varying degrees of deviation in the collected touch pressure test data, leading to test results that fail to accurately reflect the module's actual performance. Therefore, how to effectively calibrate the touch test data of LCD modules has become a pressing technical problem to be solved in this field.

[0003] Currently, calibration methods for touch test data mainly focus on two aspects: signal filtering and time-domain alignment. Signal filtering methods smooth the acquired pressure signal to suppress high-frequency noise, such as using moving average filtering or low-pass filtering. However, these methods can only eliminate noise interference and cannot correct systematic deviations introduced by the touch operation itself. Time-domain alignment methods attempt to achieve calibration by matching the measured curve with the historical standard curve on the time axis. However, the pressing, holding, and releasing attenuation stages in touch operation have different physical characteristics, and simple time-domain alignment cannot distinguish the morphological differences between different stages, easily leading to cross-stage mismatches. In addition, the touch-off moment has an inherent randomness, and matching methods relying on time-axis alignment are prone to calibration errors due to time misalignment.

[0004] Furthermore, the touch pressure response signal of a liquid crystal display module is essentially a composite signal composed of multiple physical modes superimposed. When the touch operation is released, the pressure decay process simultaneously incorporates contributions from two physical mechanisms: mechanical elastic recovery and liquid crystal orientation rearrangement. On one hand, the touchscreen's glass cover and optically transparent adhesive layer (OCA) undergo elastic recovery after being deformed by pressure, generating a mechanical mode decay signal. On the other hand, during the touch process, liquid crystal molecules are compressed and undergo orientation rearrangement, gradually returning to their initial equilibrium state after release, generating a liquid crystal mode decay signal. These two physical modes have significantly different decay characteristics—the mechanical mode has a smaller time constant and decays faster, while the liquid crystal mode has a larger time constant and decays slower. However, existing correction methods typically treat the acquired pressure signal as the output of a single physical process, directly correcting the mixed signal without separating the contributions of the two physical modes. Due to the significant differences in the decay characteristics of the two modes, applying a uniform correction amount to the mixed signal inevitably leads to insufficient correction accuracy, and in some cases, even over-correction or under-correction.

[0005] Furthermore, during the touch-release attenuation phase, the attenuation rate of the pressure signal is not constant but exhibits natural inflection points and transitions. Existing technologies often process the attenuation segment by dividing it into equal parts of fixed length. This method ignores the physical evolution of the attenuation process itself, resulting in inconsistent attenuation behavior within the divided segments and confusion regarding their physical meaning. Simultaneously, when the signal amplitude is low at the end of the attenuation phase, traditional unidirectional prediction or extrapolation correction methods tend to amplify small deviations, leading to unreliable correction results. Existing methods also lack effective quantitative indicators and graded output mechanisms for assessing the reliability of the correction results, making it difficult for testers to determine the reliability of the corrected data. Summary of the Invention

[0006] This application provides a test data correction method for liquid crystal display modules. By using curvature and torsion phase space, different physical stages of touch response can be distinguished, ensuring that matching always occurs between stages of the same type, avoiding cross-stage mismatch. The matching process does not depend on time axis alignment, eliminating the matching misalignment problem caused by the randomness of the touch-off moment.

[0007] This application provides a test data calibration method for a liquid crystal display module, including: S101: Collect touch pressure values, construct touch response curves based on the collected pressure values, calculate curvature and deflection based on the touch response curves, and map the sampling points of the touch response curves to phase space to form the phase space trajectory of the historical curves. S102: Collect real-time touch pressure values, obtain the phase space trajectory of the measured curve, cut the phase space trajectory of the measured curve to a fixed length to form a target segment, cut the historical segment to the same length as the target segment based on the phase space trajectory of the historical curve, calculate the manifold distance between the target segment and the historical segment, and take the historical segment with the smallest manifold distance as the best matching reference segment. S103, compare the measured pressure value of the target segment with the corresponding historical reference pressure value of the target segment to obtain the pressure deviation, subtract the corresponding deviation from the measured pressure value of the sampling point to obtain the corrected pressure value, and splice the correction results of all target segments according to the original time sequence to output the corrected touch test data.

[0008] Preferably, the correction method further includes: S201, based on the touch response curve, identifies the attenuation response segment, performs double exponential curve fitting on the attenuation response segment, and separates the mechanical mode sub-curve and the liquid crystal mode sub-curve; S202, extract the morphological features of the mechanical mode sub-curve and the liquid crystal mode sub-curve respectively, and perform phase space mapping based on the morphological features, wherein the morphological features include discrete decay rate and decay acceleration coefficient; S203, the mechanical modal sub-curve and the liquid crystal modal sub-curve generate their respective best matching reference segments based on step S102, obtain the mechanical modal correction amount and the liquid crystal modal correction amount, and weight the mechanical modal correction amount and the liquid crystal modal correction amount to obtain the correction result.

[0009] Preferably, the weighting method for the mechanical modal correction and the liquid crystal modal correction is as follows: obtain the current module's resting time before this touch test, and determine the weight of the liquid crystal modal correction based on the resting time; when the resting time is greater than the time constant of the liquid crystal modality, the weight approaches 1; when the resting time is less than the time constant of the liquid crystal modality, the weight approaches 0; when the resting time is between the two, the weight monotonically increases with the increase of the resting time.

[0010] Preferably, the correction method further includes: S301, calculates the first and second derivatives based on the pressure sequence of the decay response segment in the touch response curve, identifies the zero-crossing point through the second derivative, and divides the segment based on the zero-crossing point; S302, perform single exponential decay fitting based on sub-segments, calculate forward prediction error and backward prediction error based on the single exponential decay fitting results, and construct a two-way cross-correction matrix based on the forward prediction error and backward prediction error. S303, calculate the signal-to-noise ratio weight of the segment, and obtain the intrinsic decay curve based on the single exponential decay fitting and the signal-to-noise ratio weight; S304 calculates the quantity to be corrected based on the intrinsic decay curve and the original data, calculates the consistency index using a two-way cross-calibration matrix, corrects the sampling points of the sub-segment using the consistency index and the quantity to be corrected, calculates the self-calibration confidence score based on the consistency index, and assigns a calibration strategy according to the calibration confidence score.

[0011] Preferably, the method for identifying zero-crossing points is as follows: check whether the second derivatives of two adjacent sampling points have opposite signs. If the second derivative of the previous point is positive and the second point is negative, or vice versa, it indicates that there is a zero-crossing point between the two points. Only when the absolute value of the second derivative is greater than a preset noise threshold is the zero-crossing point considered valid. The noise threshold is set to 1% of the maximum absolute value of the entire second derivative sequence. Both of the above conditions must be met simultaneously for the position to be determined as a zero-crossing point.

[0012] Preferably, the steps for assigning a correction strategy based on the correction confidence score are as follows: calculate the arithmetic mean of the segment consistency index as the self-correction confidence score; when the confidence score is greater than the high confidence threshold, directly output the corrected complete stress sequence and add a high confidence label; when the confidence score is between the medium confidence threshold and the high confidence threshold, output the corrected stress sequence, and add a medium confidence label and a suggestion to weightedly fuse it with the results from the historical database; when the confidence score is less than or equal to the medium confidence threshold, discard the correction result, directly output the original uncorrected stress sequence, and add a low confidence label.

[0013] One or more technical solutions provided in this application have at least the following technical effects or advantages: different physical stages of touch response can be distinguished through curvature and torsion phase space, ensuring that matching is always carried out between the same type of stages, avoiding cross-stage mismatch, the matching process does not depend on time axis alignment, and eliminating the matching misalignment problem caused by the randomness of the touch off screen time. By using double exponential decomposition, the mixed signal is decomposed into a single physical mechanical mode and a liquid crystal mode. The calibration object is refined from the mixed signal to a single physical process signal, fundamentally eliminating cross-modal mismatch. The synthesis weight of the liquid crystal mode calibration is dynamically modulated with the module's resting time, fully considering the nonlinear influence of liquid crystal orientation stress accumulation on the calibration requirements, and avoiding overshoot or undershoot of the calibration amount when the test rhythm changes. By identifying the natural inflection point of the decay rate within the release segment and performing dynamic segmentation, the decay behavior within each sub-segment remains uniform, thereby eliminating the confusion of physical meaning caused by equal division. By constructing a consistency matrix through bidirectional cross-prediction, the sub-segments can mutually verify each other rather than being unidirectionally dependent, effectively suppressing the amplification of extrapolation errors at the end of decay. Attached Figure Description

[0014] Figure 1This is a flowchart illustrating a test data correction method for a liquid crystal display module according to the present invention. Figure 2 This is a schematic diagram of the process for identifying mechanical modal sub-curves and liquid crystal modal sub-curves according to the present invention; Figure 3 This is a schematic diagram illustrating the process of constructing a bidirectional mutual calibration matrix according to the present invention. Detailed Implementation

[0015] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0016] It should be noted that the terms "vertical," "horizontal," "up," "down," "left," "right," and similar expressions used in this article are for illustrative purposes only and do not represent the only possible implementation.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0018] Example 1: Figure 1 This is a flowchart illustrating a test data correction method for a liquid crystal display module according to an embodiment of the present invention, including: S101: Collect touch pressure values, construct touch response curves based on the collected pressure values, calculate curvature and deflection based on the touch response curves, and map the sampling points of the touch response curves to phase space to form the phase space trajectory of the historical curves. Specifically, touch operations are performed on the target LCD module. A test probe is used to press vertically onto the center area of ​​the touchscreen with a fixed force, then quickly released at a natural speed. The touch duration is recorded. The sampling rate is set to at least 500Hz to ensure the capture of high-frequency details of residual sliding. Touch pressure values ​​are collected using the above sampling configuration. The collected touch pressure data is stored as a time-pressure binary sequence. A valid interval is extracted from the collected data: from the first detected pressure increase to the point where the pressure returns to the floor noise level, forming a time series. The average pressure value within the first 100ms of the test is calculated as the floor noise baseline. The net pressure is obtained by subtracting the floor noise baseline from all touch pressure values. The signal is smoothed by convolution using a Savitzky-Golay smoothing filter to obtain a smoothed sequence, i.e., the touch response curve. The touch response curve includes a pressing phase, a holding phase, and a release decay phase. The pressing phase starts when the pressure value first exceeds three standard deviations of the base noise and ends when the pressure reaches a local maximum value. During this phase, the pressure increases monotonically. The holding phase starts from the peak value and ends when the pressure begins to continuously decrease by more than 5% of the peak value. During this phase, the pressure fluctuates around the peak value. The release decay phase starts from the end of the holding phase and ends when the pressure value falls back to within three standard deviations of the base noise and remains stable. During this phase, the pressure exhibits a decay trend.

[0019] The formula for calculating curvature based on the touch response curve is: ,in, Let be the discrete curvature at the i-th point. Let be the change in pressure value within the i-th interval. A positive value indicates increasing pressure, while a negative value indicates decreasing pressure. This represents the change in pressure value within the (i-1)th interval. The formula for calculating torsion is: (where the time interval is specified). ,in, A positive value for the discrete torsion at the i-th point indicates that the decay is accelerating, while a negative value indicates that the decay is decelerating. Let be the change in pressure value within the i-th interval. Let Δt be the change in pressure value within the (i-1)th interval, and let Δt be the time interval. The curvature and torsion calculated above constitute two-dimensional phase space coordinates. Replace the two-dimensional time domain index of each sampling point in the touch response curve with two-dimensional spatial coordinates, and connect all the two-dimensional spatial coordinates in chronological order to form the phase space trajectory of the historical curve.

[0020] S102: Collect real-time touch pressure values, obtain the phase space trajectory of the measured curve, cut the phase space trajectory of the measured curve to a fixed length to form a target segment, cut the historical segment to the same length as the target segment based on the phase space trajectory of the historical curve, calculate the manifold distance between the target segment and the historical segment, and take the historical segment with the smallest manifold distance as the best matching reference segment. Furthermore, real-time touch pressure values ​​are collected, and the phase space trajectory is calculated in step S101. The phase space trajectory of the measured curve is obtained, and a fixed length is set according to the characteristic shape of residual sliding that the window can capture. The phase space trajectory of the measured curve is then segmented according to the set fixed length to form a target segment. A historical segment of fixed length is also segmented from the starting position of the phase space trajectory of the historical curve. The manifold distance between the target segment and the historical segment is calculated based on the formula: Where D is the manifold distance value, representing the weighted Euclidean distance between the target segment and the historical segment in phase space. The smaller the distance value, the closer the two segments are in phase space, meaning their curvature-torsion morphology combinations are more similar. For the current target sub-segment, As a candidate segment for history, This is the index of the point position within the sub-segment. For a fixed length, α is a weighting coefficient for curvature differences, satisfying 0 < α < 1 and α + β = 1, used to adjust the contribution ratio of curvature differences to the total distance. α is the weighting coefficient for the torsion difference, satisfying 0 < β < 1 and α + β = 1. It is used to adjust the contribution ratio of the torsion difference to the total distance. Let j be the discrete curvature value of the j-th point in the target sub-segment. Let j be the discrete curvature value of the j-th point in the historical sub-segment. Let j be the discrete torsion value at the j-th point in the target sub-segment. Given the discrete torsion value of the j-th point in the historical sub-segment, based on the manifold distance calculated above, find the historical sub-segment with the smallest manifold distance as the best matching reference segment. Repeat the above steps for the starting position of each target sub-segment of the measured trajectory to obtain the best matching reference segment corresponding to each target sub-segment, thus forming a matching result list.

[0021] S103, compare the measured pressure value of the target segment with the corresponding historical reference pressure value of the target segment to obtain the pressure deviation, subtract the corresponding deviation from the measured pressure value of the sampling point to obtain the corrected pressure value, and splice the correction results of all target segments according to the original time sequence to output the corrected touch test data.

[0022] Specifically, the first sampling point in the target sub-segment is paired with the first sampling point in the historical reference sub-segment, the second sampling point in the target sub-segment is paired with the second sampling point in the historical reference sub-segment, and so on, until all sampling points in the sub-segment are paired. For each paired sampling point, the difference between the current measured pressure value and the historical reference pressure value is calculated. For each sampling point in the target sub-segment, the measured pressure value is subtracted from the difference to obtain the corrected pressure value of the sampling point. For adjacent target sub-segments with overlapping areas, the distance from the center of the previous target sub-segment is considered. The closer the sampling point, the greater the weight of the correction value given by the previous sub-segment; the closer the sampling point is to the center of the next target sub-segment, the greater the weight of the correction value given by the next sub-segment. The weights gradually transition linearly from front to back. In the latter half of the previous sub-segment, the weight gradually decreases, while the weight of the next sub-segment gradually increases. At the midpoint of the overlapping area, the weights of the two are equal. Each target sub-segment obtains a corrected pressure value. These target sub-segments are arranged according to their original order on the measured curve and spliced ​​together to form a complete sequence of corrected pressure values.

[0023] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: the curvature and torsion phase space can distinguish different physical stages of touch response, ensuring that matching is always performed between stages of the same type, avoiding cross-stage mismatch, and the matching process does not depend on time axis alignment, thus eliminating the matching misalignment problem caused by the randomness of the touch off-screen moment.

[0024] Example 2: Example 1 avoids misalignment in the time axis by mapping the touch response curve to the curvature-torsion phase space. However, the pressure signal involved in the matching simultaneously includes the superimposed contributions of two physical modes: mechanical elasticity and liquid crystal alignment rearrangement. Significant deviations exist in the attenuation characteristics of each physical mode, resulting in a lack of accuracy at the physical mode level in the subsequent application of correction amounts. Figure 2 As shown.

[0025] S201, based on the touch response curve, identifies the attenuation response segment, performs double exponential curve fitting on the attenuation response segment, and separates the mechanical mode sub-curve and the liquid crystal mode sub-curve; Furthermore, the timing data of the touch response curve is scanned point by point. Starting from the touch start time, the pressure values ​​of adjacent sampling points are compared sequentially. When the pressure value of a certain point is greater than the pressure values ​​of the previous and next points at the same time, that point is the local peak point. In the touch response curve, the maximum value reached after the press rise segment is the global pressure peak point. The amplitude of the peak point and the corresponding time coordinate are recorded. The time coordinate marks the turning point from the end of the press phase to the beginning of the release phase. When the pressure value decays to less than 3 times the floor noise level, the signal is considered to have basically returned to a steady state. Starting from the pressure peak point, the time axis is scanned backward. When the pressure values ​​of several consecutive sampling points (such as 5 consecutive points) are all less than 3 times the floor noise level, the starting point of this continuous segment is recorded as the end time of the decay response segment. From the complete touch response timing data, the time from the pressure peak point to the end time of the decay response segment is extracted to form the decay response segment. The decay response segment contains the complete decay process from the start of the pressure peak to the complete decay to the floor noise level, and records all the effective information of the two physical modes of mechanical elastic recovery and liquid crystal alignment rearrangement.

[0026] Based on the attenuation response segment fitted with a bi-exponential curve, the mixed attenuation signal in the attenuation response segment is separated into mechanical mode sub-curves and liquid crystal mode sub-curves. The attenuation response segment is described by a decomposition model, and the expression is: ,in, For the attenuation response segment, The initial amplitude coefficient of the mechanical mode reflects the proportion of the contribution of glass / OCA deformation release to the total signal. The mechanical modal time constant, The initial amplitude coefficient of the liquid crystal mode reflects the proportion of the contribution of liquid crystal orientation recovery to the total signal. The time constant of the liquid crystal mode. The residual term refers to the system measurement noise and high-order small perturbations not captured by the double exponential model. The decay response segment is fitted using the Levenberg-Marquardt algorithm according to the convergence condition and initial values. The convergence condition is that the rate of change of the sum of squared residuals between two consecutive iterations is less than... When this condition is met, the algorithm is considered to have converged to a stable solution, and the iteration stops. In each iteration, the algorithm calculates the residual between the model output and the measured data under the current parameter combination, adjusts the values ​​of the four parameters according to the gradient direction of the residual, and gradually approaches the optimal solution. After the algorithm converges, the final convergence values ​​of the mechanical modal amplitude coefficient, mechanical modal time constant, liquid crystal modal amplitude coefficient, and liquid crystal modal time constant are recorded. At the same time, the energy level of the final fitted residual is recorded. Modal sub-curves are constructed based on the final convergence values ​​obtained from the fitting. The modal sub-curves include mechanical modal sub-curves and liquid crystal modal sub-curves. The formula for the mechanical modal sub-curve is: ,in, The pressure contribution value of the mechanical modal sub-curve at time t. The initial amplitude coefficients for the mechanical modes, Let be the time constant of the mechanical mode, and the formula for the liquid crystal modal sub-curve is: ,in, This represents the pressure contribution value of the liquid crystal modal sub-curve at time t, where t is the time variable. is the base of the natural logarithm. The initial amplitude coefficient of the liquid crystal mode. is the time constant of the liquid crystal mode.

[0027] S202, extract the morphological features of the mechanical mode sub-curve and the liquid crystal mode sub-curve respectively, and perform phase space mapping based on the morphological features, wherein the morphological features include discrete decay rate and decay acceleration coefficient; Specifically, the mechanical mode sub-curve and the liquid crystal mode sub-curve are denoted as a discrete sampling point sequence. Each sampling point includes a time coordinate and an amplitude. For each sampling point in the sub-curve, the discrete attenuation rate is calculated based on the adjacent points before and after it, using the following formula: ,in, Let i be the discrete decay rate of the i-th sampling point. The amplitude of the (i+1)th sampling point The amplitude of the (i-1)th sampling point Let be the amplitude of the i-th sampling point. Let i be the time coordinate of the (i+1)th sampling point. Let i be the time coordinate of the (i-1)th sampling point. When the signal decays rapidly, the amplitude difference between the preceding and following points is large, resulting in a larger numerator. The value is relatively high; when the signal decays slowly, the amplitude difference between the preceding and following points is small, and the numerator is small. The mechanical mode has a lower time constant and decays faster. Therefore, the decay rate of the mechanical mode sub-curve is significantly higher than that of the liquid crystal mode sub-curve. Based on the calculated decay rate, the decay acceleration coefficient is calculated using the following formula: ,in, The decay acceleration coefficient, The discrete decay rate of the (i+1)th sampling point. The discrete decay rate of the (i-1)th sampling point. Let i be the time coordinate of the (i+1)th sampling point. Let i be the time coordinate of the (i-1)th sampling point, when When the value is greater than 0, it indicates that the attenuation rate is increasing and the signal attenuation is accelerating, which is reflected in the curve as a change from a flat to a steep one; when... When <0, it indicates that the attenuation rate is decreasing, and the signal attenuation is slowing down. This is reflected on the curve as a change from a steep to a flat curve. Each effective sampling point is converted from the original time-amplitude two-dimensional coordinate system. , ) is mapped to a point in phase space ( , This can be represented as a phase space where the decay rate is the horizontal axis and the decay acceleration coefficient is the vertical axis. The temporal morphology of the sub-curve is re-expressed in the phase space. All valid sampling points are connected sequentially in the original time order to form a discrete phase space trajectory. The geometry of this trajectory is completely determined by the dynamic characteristics of the decay process. The decay rate determines the position of the trajectory on the horizontal axis (decay rate), and the acceleration or deceleration trend of the decay determines the position of the trajectory on the vertical axis. Since the time information is implicitly retained in the arrangement order of the trajectory points, the shape of the phase space trajectory is not related to the absolute time of the touch, but only to the evolution law of the decay process itself.

[0028] S203, the mechanical modal sub-curve and the liquid crystal modal sub-curve generate their respective best matching reference segments based on step S102, obtain the mechanical modal correction amount and the liquid crystal modal correction amount, and weight the mechanical modal correction amount and the liquid crystal modal correction amount to obtain the correction result.

[0029] Furthermore, based on the phase space trajectories of the mechanical modal sub-curve and the liquid crystal modal sub-curve, the optimal matching reference segment for the mechanical mode and the optimal matching reference segment for the liquid crystal mode are generated respectively using the method in step S102. The current mechanical modal pressure value of each sampling point in the target sub-segment is compared point by point with the historical mechanical modal pressure value at the corresponding position in the optimal matching reference segment. The difference between the current measured value and the historical reference value is calculated. This difference is the deviation that the mechanical mode at that sampling point needs to be corrected, forming a mechanical modal correction quantity sequence. The liquid crystal modal correction quantity sequence is obtained using the same method. The static waiting time of the current module before this touch test is obtained. The time interval from the end of the last touch test of the module to the start of this test is the static waiting time. The weight of the liquid crystal modal correction quantity is determined based on the static waiting time. When the static waiting time is greater than the time constant of the liquid crystal mode, This indicates that the module has undergone sufficient rest, and the liquid crystal molecules have fully recovered to their initial equilibrium state from the orientation disturbance caused by the previous touch, with the weight approaching 1. When the rest time is less than the time constant of the liquid crystal mode, it indicates that the module has just undergone frequent testing, and the orientation stress accumulated from the previous touch in the liquid crystal layer has not yet fully relaxed, so the weight is set to approach 0. When the rest time is between the two, the weight increases monotonically with the increase of the rest time, and the mechanical mode correction amount is a fixed weight of 1. The liquid crystal mode correction amount is multiplied by the weight of the liquid crystal mode correction amount, and the result of the multiplication is added to the mechanical mode correction amount to obtain the correction result. If the correction result is positive, it means that the original value is lower than the expected value and needs to be adjusted upward; if the correction result is negative, it means that the original value is higher than the expected value and needs to be adjusted downward. The corrected attenuation segment is spliced ​​with the press segment in the touch response curve that does not need to be corrected to obtain the corrected touch response curve.

[0030] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: by decomposing the mixed signal into a single physical mechanical mode and liquid crystal mode through double exponential decomposition, the calibration object is refined from the mixed signal into a single physical process signal, fundamentally eliminating cross-modal mismatch. The synthesis weight of the liquid crystal mode calibration amount is dynamically modulated with the module's resting waiting time, which fully considers the nonlinear influence of the accumulation of liquid crystal orientation stress on the calibration requirements and avoids overshoot or undershoot of the calibration amount when the test rhythm changes.

[0031] Example 3: By performing a double exponential decomposition on the pressure signal of the touch release segment, the mixed attenuation signal is separated into a mechanoelastic recovery mode and a liquid crystal orientation rearrangement mode. In this example, the curvature feature is dynamically segmented to ensure that the attenuation behavior within each segment is relatively uniform. A bidirectional cross-calibration consistency matrix is ​​constructed for the segmented segments, achieving dual protection for physical model fitting and data verification, such as... Figure 3 As shown.

[0032] S301, calculates the first and second derivatives based on the pressure sequence of the decay response segment in the touch response curve, identifies the zero-crossing point through the second derivative, and divides the segment based on the zero-crossing point; Specifically, the pressure sequence of the decay response segment in the touch response curve is calculated using a central difference scheme. The first derivative reflects the rate of pressure change over time. In the release segment, the first derivative is negative, and its absolute value indicates the rate of decay—the larger the absolute value, the faster the decay; the smaller the absolute value, the smoother the decay. When the second derivative is positive, the absolute value of the first derivative (negative) is decreasing, indicating that the decay is decelerating. When the second derivative is negative, the absolute value of the first derivative (negative) is increasing, indicating that the decay is accelerating. When the second derivative is zero, the decay rate reaches a local extremum, which is the critical point where the decay behavior changes. The zero-crossing point refers to the critical point where the second derivative changes from positive to negative or from negative to positive. The position corresponds to the moment when the decay rate reaches a local extremum (i.e., the decay behavior changes direction). The method for identifying zero-crossing points is as follows: check whether the second derivatives of two adjacent sampling points have opposite signs. If the second derivative of the previous point is positive and the second point is negative, or vice versa, it indicates that there is a zero-crossing point between the two points. Only when the absolute value of the second derivative is greater than the preset noise threshold is the zero-crossing point considered valid. The noise threshold is 1% of the maximum absolute value of the entire second derivative sequence. Both of the above conditions must be met simultaneously for the position to be determined as a zero-crossing point.

[0033] Using all identified zero-crossing points as dividing boundaries, the release segment is divided into several continuous sub-segments along the time axis. These sub-segments are arranged in chronological order, and the second derivative sign is consistent within each sub-segment, meaning that the direction of the decaying acceleration remains unchanged within the time interval covered by that sub-segment.

[0034] S302, perform single exponential decay fitting based on sub-segments, calculate forward prediction error and backward prediction error based on the single exponential decay fitting results, and construct a two-way cross-correction matrix based on the forward prediction error and backward prediction error. Furthermore, the Levenberg-Marquardt nonlinear least squares algorithm is used to fit the sub-segments. The four parameters are adjusted iteratively to minimize the sum of squared residuals between the model output and the measured data. After fitting, the time constant fitting result for each sub-segment is recorded. The time constant reflects the characteristic rate of signal attenuation within the time interval covered by the sub-segment—the smaller the value, the faster the attenuation; the larger the value, the slower the attenuation. If the fitted time constant of a certain sub-segment deviates significantly from that of other sub-segments, it indicates that the data in that sub-segment may have been interfered with by abnormal factors. For any two different sub-segments, where one sub-segment is earlier than the other, forward prediction is performed on the two different sub-segments. The attenuation characteristics of the earlier sub-segment are used to predict the attenuation pattern that the later sub-segment should exhibit. Specifically... The method involves substituting the initial amplitude and initial time of the later segment into an exponential function with the time constant of the earlier segment as the decay rate to construct a forward prediction curve. The forward prediction curve is then compared point-by-point with the actual measurement data of the later segment. The difference between the actual and predicted values ​​is divided by the actual value to obtain the forward prediction error. This forward prediction error represents the deviation between the predicted and actual values ​​when using the decay rate of the earlier segment to predict the later segment. A smaller value indicates that the decay behavior of the two segments is consistent in the time direction; a larger value indicates that the decay characteristics of the earlier segment cannot describe the actual decay behavior of the later segment. For backward prediction, the same method is used to construct a backward prediction curve to obtain the backward prediction error. This backward prediction error refers to the deviation between the predicted and actual values ​​when using the decay rate of the later segment to infer the earlier segment.

[0035] For any two distinct sub-segments, the absolute difference between their forward and backward prediction errors is taken as the element value of the bidirectional cross-correction matrix. The diagonal elements of the matrix are set to 0. The numerical meaning of the elements of the bidirectional cross-correction matrix is: the degree of asymmetry between the two sub-segments in bidirectional cross-correction. When the element value is very small, it indicates that the forward and backward prediction errors are approximately equal—the error of predicting a later sub-segment using an earlier sub-segment is not significantly different from the error of predicting an earlier sub-segment using a later sub-segment, indicating that the two sub-segments point to the same decay time constant. When the element value is large, it indicates that there is significant asymmetry in the bidirectional prediction of the two sub-segments, with the prediction deviation in one direction being significantly greater than that in the other direction, indicating that at least one sub-segment has anomalies in its data, and its decay behavior deviates from the overall consistency pattern.

[0036] S303, calculate the signal-to-noise ratio weight of the segment, and obtain the intrinsic decay curve based on the single exponential decay fitting and the signal-to-noise ratio weight; Specifically, the signal-to-noise ratio (SNR) weights of each segment are calculated. The SNR weight for each segment is defined as the ratio of the signal amplitude range within that segment to the standard deviation of the system's floor noise. The specific calculation method is as follows: subtract the minimum pressure amplitude from the maximum pressure amplitude within the segment to obtain the amplitude range of that segment, then divide this by the standard deviation of the system's floor noise. The larger this ratio, the stronger the signal strength relative to the noise in that segment. In the release phase, the signal amplitude of the earlier segments is higher and the amplitude range is larger, so the SNR weight value is usually larger; the signal of the later segments has attenuated to a lower level and the amplitude range is smaller, so the SNR weight value is usually smaller. A loss function is constructed based on the single exponential decay fitting and the SNR weights. Specifically, for each segment... The loss function is obtained by summing the squared fitting errors of all sampling points in a segment and multiplying them by the signal-to-noise ratio weight of that segment. The weighted summation of all segments yields the loss function. The Levenberg-Marquardt nonlinear least squares algorithm is used to iteratively solve the loss function. Starting from a set of initial estimates, the loss function value under the current parameters and its gradient with respect to each parameter are calculated. Then, the parameter values ​​are adjusted according to the gradient direction to gradually decrease the loss function value. This process is repeated until convergence. The global parameters obtained from convergence are substituted into the expression of the fitting function to calculate the corresponding fitting pressure value for each sampling time point in the release segment, forming the intrinsic decay curve.

[0037] S304 calculates the quantity to be corrected based on the intrinsic decay curve and the original data, calculates the consistency index using a two-way cross-calibration matrix, corrects the sampling points of the sub-segment using the consistency index and the quantity to be corrected, calculates the self-calibration confidence score based on the consistency index, and assigns a calibration strategy according to the calibration confidence score.

[0038] Furthermore, the original release segment pressure sequence was compared point by point with the intrinsic decay curve. For each sampling point within the release segment, the original pressure value is subtracted from the fitted value of the intrinsic decay curve at that point to obtain the calibration value for that sampling point. The calibration value represents the degree of deviation of the measured pressure value at the sampling point from the ideal decay response. When the calibration value is positive, it indicates that the measured value is higher than the ideal value and needs to be adjusted downward; when the calibration value is negative, it indicates that the measured value is lower than the ideal value and needs to be adjusted upward; when the calibration value is zero, it indicates that the measured value at that point is consistent with the ideal value and no calibration is required. The consistency index is calculated as follows: for the k-th field, based on all off-diagonal elements in the k-th row of the bidirectional cross-calibration matrix, take the negative exponent with the natural logarithm as the base for each element, then sum these exponent values ​​and divide by the total number of sub-segments minus one. The consistency index ranges from 0 to 1. The closer the value is to 1, the more reliable the data quality of the sub-segment, and its calibration value should be trusted and fully applied; the closer the value is to 0, the more abnormal the sub-segment is in cross-calibration, and its calibration value should be suppressed to avoid amplifying the abnormality into a larger error.

[0039] For each sampling point within a sub-segment, the square of the consistency index of that sub-segment is used as the correction application coefficient, multiplied by the amount to be corrected for that sampling point, to obtain the final applied correction amount. Then, the original pressure value is subtracted from the final correction amount to obtain the corrected pressure value for that sampling point. The arithmetic mean of the sub-segment consistency indices is calculated as the self-correction confidence score. Based on the range of the self-correction confidence score, a three-level graded output strategy is implemented. When the confidence score is greater than the high confidence threshold, it indicates that the self-correction result is highly reliable and no significant abnormal intervals are detected. At this point, the direct... The system outputs the corrected complete stress sequence and adds a high-confidence marker to the output data. When the confidence score is between the medium and high confidence thresholds, it indicates that the self-calibration result has medium reliability. In this case, the corrected stress sequence is output, along with a medium-confidence marker and a suggestion to weightedly fuse it with the results from the historical database. When the confidence score is less than or equal to the medium confidence threshold, it indicates that the self-calibration result is unreliable. The calibration result is discarded, and the original uncorrected stress sequence is output directly with a low-confidence marker, suggesting manual review or retesting.

[0040] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: by identifying the natural inflection point of the decay rate within the release segment and performing dynamic segmentation, the decay behavior within each sub-segment remains uniform, thereby eliminating the confusion of physical meaning caused by equal division; by constructing a consistency matrix through bidirectional cross-prediction, the sub-segments mutually verify each other rather than being unidirectionally dependent, effectively suppressing the amplification of extrapolation error at the end of decay. The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for calibrating test data of a liquid crystal display module, characterized in that, include: S101: Collect touch pressure values, construct touch response curves based on the collected pressure values, calculate curvature and deflection based on the touch response curves, and map the sampling points of the touch response curves to phase space to form the phase space trajectory of the historical curves. S102: Collect real-time touch pressure values, obtain the phase space trajectory of the measured curve, cut the phase space trajectory of the measured curve to a fixed length to form a target segment, cut the historical segment to the same length as the target segment based on the phase space trajectory of the historical curve, calculate the manifold distance between the target segment and the historical segment, and take the historical segment with the smallest manifold distance as the best matching reference segment. S103, compare the measured pressure value of the target segment with the corresponding historical reference pressure value of the target segment to obtain the pressure deviation, subtract the corresponding deviation from the measured pressure value of the sampling point to obtain the corrected pressure value, and splice the correction results of all target segments according to the original time sequence to output the corrected touch test data.

2. The test data correction method for a liquid crystal display module according to claim 1, characterized in that, The formula for calculating curvature based on the touch response curve is: ,in, Let be the discrete curvature at the i-th point. Let be the change in pressure value within the i-th interval. A positive value indicates increasing pressure, while a negative value indicates decreasing pressure. This represents the change in pressure value within the (i-1)th interval. The time interval is given; the formula for calculating the torsion is: ,in, A positive value for the discrete torsion at the i-th point indicates that the decay is accelerating, while a negative value indicates that the decay is decelerating.

3. The test data correction method for a liquid crystal display module according to claim 2, characterized in that, The manifold distance between the target segment and the historical segment is calculated using the following formula: Where D is the manifold distance value, representing the weighted Euclidean distance between the target segment and the historical segment in phase space. For the current target sub-segment, As a candidate segment for history, This is the index of the point position within the sub-segment. For a fixed length, α is a weighting coefficient for curvature differences, satisfying 0 < α < 1 and α + β = 1, used to adjust the contribution ratio of curvature differences to the total distance. α is the weighting coefficient for the torsion difference, satisfying 0 < β < 1 and α + β = 1, used to adjust the contribution ratio of the torsion difference to the total distance. Let j be the discrete curvature value of the j-th point in the target sub-segment. Let j be the discrete curvature value of the j-th point in the historical sub-segment. Let j be the discrete torsion value at the j-th point in the target sub-segment. It represents the discrete torsion value of the j-th point in the historical segment.

4. The test data correction method for a liquid crystal display module according to claim 1, characterized in that, The correction methods also include: S201, based on the touch response curve, identifies the attenuation response segment, performs double exponential curve fitting on the attenuation response segment, and separates the mechanical mode sub-curve and the liquid crystal mode sub-curve; S202, extract the morphological features of the mechanical mode sub-curve and the liquid crystal mode sub-curve respectively, and perform phase space mapping based on the morphological features, wherein the morphological features include discrete decay rate and decay acceleration coefficient; S203, the mechanical modal sub-curve and the liquid crystal modal sub-curve generate their respective best matching reference segments based on step S102, obtain the mechanical modal correction amount and the liquid crystal modal correction amount, and weight the mechanical modal correction amount and the liquid crystal modal correction amount to obtain the correction result.

5. The test data correction method for a liquid crystal display module according to claim 4, characterized in that, The formula for the mechanical modal sub-curve is: ,in, The pressure contribution value of the mechanical modal sub-curve at time t. The initial amplitude coefficients for the mechanical modes, is the time constant of the mechanical mode.

6. The test data correction method for a liquid crystal display module according to claim 5, characterized in that, The formula for the liquid crystal modal sub-curve is: ,in, This represents the pressure contribution value of the liquid crystal modal sub-curve at time t, where t is the time variable. is the base of the natural logarithm. The initial amplitude coefficient of the liquid crystal mode. is the time constant of the liquid crystal mode.

7. The test data correction method for a liquid crystal display module according to claim 6, characterized in that, The weighting method for mechanical modal correction and liquid crystal modal correction is as follows: obtain the current module's resting time before this touch test, and determine the weight of the liquid crystal modal correction based on the resting time; when the resting time is greater than the time constant of the liquid crystal modality, the weight approaches 1; when the resting time is less than the time constant of the liquid crystal modality, the weight approaches 0; when the resting time is between the two, the weight monotonically increases with the increase of the resting time.

8. The test data correction method for a liquid crystal display module according to claim 4, characterized in that, The correction method further includes: S301, calculates the first and second derivatives based on the pressure sequence of the decay response segment in the touch response curve, identifies the zero-crossing point through the second derivative, and divides the segment based on the zero-crossing point; S302, perform single exponential decay fitting based on sub-segments, calculate forward prediction error and backward prediction error based on the single exponential decay fitting results, and construct a two-way cross-correction matrix based on the forward prediction error and backward prediction error. S303, calculate the signal-to-noise ratio weight of the segment, and obtain the intrinsic decay curve based on the single exponential decay fitting and the signal-to-noise ratio weight; S304 calculates the quantity to be corrected based on the intrinsic decay curve and the original data, calculates the consistency index using a two-way cross-calibration matrix, corrects the sampling points of the sub-segment using the consistency index and the quantity to be corrected, calculates the self-calibration confidence score based on the consistency index, and assigns a calibration strategy according to the calibration confidence score.

9. The test data correction method for a liquid crystal display module according to claim 8, characterized in that, The method for identifying zero-crossing points is as follows: check whether the second derivatives of two adjacent sampling points have opposite signs. If the second derivative of the previous point is positive and the second point is negative, or vice versa, it indicates that there is a zero-crossing point between the two points. Only when the absolute value of the second derivative is greater than the preset noise threshold is the zero-crossing point considered valid. The noise threshold is 1% of the maximum absolute value of the entire second derivative sequence. Both of the above conditions must be met simultaneously for the position to be determined as a zero-crossing point.

10. The test data correction method for a liquid crystal display module according to claim 9, characterized in that, The steps for assigning a correction strategy based on the correction confidence score are as follows: Calculate the arithmetic mean of the segment consistency index as the self-correction confidence score; when the confidence score is greater than the high confidence threshold, directly output the corrected complete stress sequence and add a high confidence label; when the confidence score is between the medium and high confidence thresholds, output the corrected stress sequence, add a medium confidence label, and suggest weighted fusion with historical database results; when the confidence score is less than or equal to the medium confidence threshold, discard the correction result, directly output the original uncorrected stress sequence, and add a low confidence label.

Images