A pressure-resistant detection method for an OLED metal mesh touch screen

Abstract

This invention discloses a compressive strength testing method for OLED metal mesh touchscreens, relating to the field of touchscreen performance testing technology. The method includes: acquiring node coordinate perturbation data, angle perturbation data, and spacing perturbation data of an irregularly oriented metal mesh touch layer to generate a randomized mesh reference map; applying an initial electrical excitation signal to the touch layer and acquiring a first electrical response map; applying a dynamically increasing surface pressure load to the cover glass and simultaneously acquiring a second electrical response map; calculating the difference between the two maps to generate a perturbation deviation matrix; performing eigenvalue decomposition on the perturbation deviation matrix and extracting abnormal perturbation feature vectors; and matching the abnormal perturbation feature vectors with pre-stored standard failure mode vectors to determine the compressive strength failure type and failure location.

Description

Technical Field

[0001] This invention relates to the field of touch screen performance testing technology, specifically to a compressive strength testing method for OLED metal mesh touch screens. Background Technology

[0002] OLED metal mesh touchscreens use an irregular, randomized metal mesh as the touch layer, employing randomized node coordinates, mesh spacing, and angles to eliminate moiré fringes and improve optical performance. This structure results in random perturbations within the design tolerances for each touch layer, leading to significant individual differences and posing challenges to pressure resistance testing. Conventional pressure resistance testing methods are based on regular mesh models, using an ideal periodic conductive network as a reference, and determining failure by comparing the overall change in capacitance or resistance before and after pressure application. When this regular model-based method is directly applied to an irregular, randomized metal mesh, the mesh's inherent randomness is superimposed on the pressure-induced changes in electrical signals, easily leading to misjudgments of pressure damage and numerous false positives. Furthermore, traditional testing methods often only provide qualitative conclusions regarding overall pressure resistance or failure status, failing to differentiate the microscopic types of failure. For example, they cannot identify whether it's a broken metal mesh line, a short circuit between adjacent meshes, or interface peeling caused by interlayer stress, nor can they output the specific coordinates of the failure location. In manufacturing process control and defect tracing scenarios, it is not only necessary to know whether the touchscreen has experienced compressive failure, but also to identify the failure mode and its spatial location. Existing detection technologies are insufficient to meet this diagnostic requirement. To effectively perform compressive testing on irregularly shaped metal mesh touch layers, it is urgent to address how to construct individualized detection benchmarks based on their random disturbance characteristics, and how to separate and accurately classify and locate abnormal disturbance modes characterizing local failures from the complex electrical response changes under dynamic pressure. Summary of the Invention

[0003] The present invention aims to provide a compressive strength testing method for OLED metal mesh touch screens, which adapts to the random disturbance characteristics of irregular metal meshes, and achieves accurate identification of the compressive strength failure type of the touch layer and precise determination of the failure location.

[0004] The objective of this invention can be achieved through the following technical solutions: This invention provides a compressive strength testing method for an OLED metal mesh touchscreen, comprising the following steps: acquiring node coordinate perturbation data, angle perturbation data, and spacing perturbation data of an irregular random metal mesh touch layer of the OLED metal mesh touchscreen to be tested, and generating a random mesh reference map; applying an initial electrical excitation signal to the irregular random metal mesh touch layer, and acquiring a first electrical response map of the irregular random metal mesh touch layer; applying a dynamically increasing surface pressure load to the cover glass of the OLED metal mesh touchscreen to be tested, and simultaneously acquiring a second electrical response map of the irregular random metal mesh touch layer; calculating the difference between the first electrical response map and the second electrical response map, and generating a perturbation deviation matrix; performing eigenvalue decomposition on the perturbation deviation matrix to extract at least one abnormal perturbation feature vector; matching the abnormal perturbation feature vector with multiple pre-stored standard failure mode vectors, and determining the compressive strength failure type and failure location of the irregular random metal mesh touch layer based on the matching result.

[0005] This method establishes a baseline spectrum based on the irregular randomness characteristics of the touch layer itself. Combined with the differences in electrical response under pressure and no-pressure conditions, it can accurately identify abnormal disturbances caused by pressure, rather than simply attributing all signal changes to anomalies. This adapts to the intrinsic discreteness of irregular metal meshes and significantly reduces misjudgments. By matching the abnormal disturbance feature vector with the standard failure mode vector, it can not only determine whether compressive failure has occurred, but also distinguish different failure types such as mesh breakage, mesh bridging short circuits, and interface peeling, and pinpoint the specific physical location of the failure. This provides a precise basis for evaluating the compressive performance and analyzing the failure of OLED metal mesh touchscreens.

[0006] As a preferred embodiment of the present invention, the step of obtaining node coordinate perturbation data, angle perturbation data, and spacing perturbation data of the irregular random metal mesh touch layer of the OLED metal mesh touch screen to be tested, and generating a random mesh reference map, specifically includes: retrieving the design file of the irregular random metal mesh touch layer from the production database, and parsing the regular reference node coordinates, reference spacing, and reference angle in the design file; calling a pseudo-random number generator to apply node coordinate perturbation coefficients, spacing perturbation coefficients, and angle perturbation coefficients to the regular reference node coordinates, the reference spacing, and the reference angle, respectively, to generate actual node coordinates, actual spacing, and actual angles; mapping the actual node coordinates, the actual spacing, and the actual angles to a two-dimensional plane coordinate system to construct the random mesh reference map. By introducing pseudo-random perturbations to simulate the mesh irregularities caused by the process during mass production, the reference map is made closer to the actual geometry of the real touch layer, avoiding systematic deviations caused by using an ideal regular shape as a reference, and enhancing the basic accuracy of subsequent perturbation analysis.

[0007] Furthermore, when applying perturbation to the reference node coordinates, a first node coordinate perturbation coefficient is assigned to the horizontal coordinate of the regular reference node coordinates, and a second node coordinate perturbation coefficient is assigned to the vertical coordinate of the regular reference node coordinates. The pseudo-random number generator is invoked to generate a first random number between -1 and +1. This first random number is multiplied by the first node coordinate perturbation coefficient and added to the horizontal coordinate to generate the perturbated horizontal coordinate. Similarly, the pseudo-random number generator is invoked to generate a second random number between -1 and +1. This second random number is multiplied by the second node coordinate perturbation coefficient and added to the vertical coordinate to generate the perturbated vertical coordinate. The combination of the perturbated horizontal coordinate and the perturbated vertical coordinate is used as the actual node coordinates. This method of independently applying random perturbations to the horizontal and vertical coordinates can more accurately reproduce anisotropic deviations that may occur in the manufacturing process, giving the generated random grid reference map higher fidelity and reference value.

[0008] As a preferred embodiment of the present invention, the step of applying an initial electrical excitation signal to the irregular metal mesh touch layer and acquiring a first electrical response spectrum of the irregular metal mesh touch layer specifically includes: dividing the initial electrical excitation signal into multiple excitation time slots in chronological order, with each excitation time slot corresponding to a frequency code; sequentially injecting the multiple excitation time slots into all transmitting channels of the irregular metal mesh touch layer through a driving circuit; synchronously sampling the transient induced voltage value at the end of each excitation time slot on all receiving channels of the irregular metal mesh touch layer; and arranging the transient induced voltage values ​​acquired on all receiving channels in channel index order to form the first electrical response spectrum. By employing the excitation method of time slot division and frequency coding, multi-frequency scanning of the touch layer can be performed, and the multi-dimensional response spectrum composed of the acquired transient induced voltage values ​​can sensitively capture the electrical characteristics of the mesh structure, providing a high-resolution initial reference for comparison after pressure loading.

[0009] As a preferred embodiment of the present invention, the step of applying a dynamically increasing surface pressure load to the cover glass of the OLED metal mesh touchscreen under test and simultaneously acquiring the second electrical response spectrum of the irregular metal mesh touch layer specifically involves: controlling a pressure loading head to descend vertically from an initial height until the contact surface of the pressure loading head is completely in contact with the surface of the cover glass; controlling the pressure loading head to gradually increase the pressure value applied to the cover glass according to a preset stepped pressure curve; repeating the step of acquiring the first electrical response spectrum during the holding period of each pressure step to obtain candidate electrical response spectra corresponding to the current pressure value; and stacking the candidate electrical response spectra corresponding to all pressure steps in ascending order of pressure value to generate the second electrical response spectrum. By using stepped loading and acquiring the response during the stable period of each pressure step, transient interference during the dynamic loading process is effectively avoided, ensuring that the electrical response at each pressure level truly reflects the mesh state under that load. The second electrical response spectrum generated by this stacking can accurately characterize the continuous evolution of the electrical properties of the touch layer as the pressure increases.

[0010] Preferably, after the pressure loading head reaches the target pressure value of the current pressure step, a timer is started; when the timer count reaches a preset stabilization period, a data acquisition command is triggered; in response to the data acquisition command, the initial electrical excitation signal is reapplied to the irregular random metal mesh touch layer; the response signal output by the irregular random metal mesh touch layer under the stable state of the target pressure value is acquired, and the response signal is recorded as the candidate electrical response spectrum. Applying excitation and acquiring the response after the pressure stabilizes eliminates the random fluctuations in the signal caused by pressure fluctuations, so that the candidate electrical response spectrum of each pressure step has good repeatability and reliability.

[0011] As a preferred embodiment of the present invention, during the application of the dynamically increasing surface pressure load, the acoustic emission signal of the microcracks in the cover glass is monitored in real time. By synchronously monitoring the acoustic emission signal and the electrical response spectrum, the acoustic characteristics of microcrack initiation and electrical anomalies can be correlated in the time domain, further improving the accuracy of failure determination and early warning capability.

[0012] As a preferred embodiment of the present invention, both the first and second electrical response spectra are bandpass filtered before acquisition to remove power frequency interference and environmental noise. This filtering effectively suppresses the influence of external electromagnetic interference on weak induced voltage signals, improves the signal-to-noise ratio of the response spectra, and ensures the accuracy of subsequent disturbance deviation matrix calculations.

[0013] As a preferred embodiment of the present invention, the step of calculating the difference between the first electrical response spectrum and the second electrical response spectrum to generate a disturbance deviation matrix specifically includes: extracting a second induced voltage value from the second electrical response spectrum at the same receiving channel index and the same excitation time slot position as the first electrical response spectrum; subtracting the second induced voltage value from the first induced voltage value at the corresponding position in the first electrical response spectrum to obtain a voltage difference; associating the voltage difference with its corresponding receiving channel index and excitation time slot position to form a two-dimensional difference-position mapping table; performing normalization processing on the difference-position mapping table, and using the result after normalization processing as the disturbance deviation matrix. Directly taking the difference of the induced voltage values ​​at the corresponding positions and constructing a two-dimensional matrix can completely preserve the spatial distribution characteristics and time slot characteristics of the electrical response changes caused by pressure; the normalization processing eliminates the scale inconsistency problem caused by the difference in signal amplitude between different channels or different time slots, making the abnormal disturbance stand out in the matrix.

[0014] Preferably, before generating the perturbation deviation matrix, the method further includes a step of time alignment calibration of the first electrical response spectrum and the second electrical response spectrum. Specifically, this involves: extracting the start time of the first excitation time slot from the first electrical response spectrum as a first reference time; extracting the start time of the first excitation time slot from the second electrical response spectrum as a second reference time; calculating the time offset of the second reference time relative to the first reference time; and performing a global time-domain shift on the second electrical response spectrum based on the time offset to align the second reference time with the first reference time. Time alignment calibration eliminates time-domain misalignment caused by acquisition trigger delay or clock drift, ensuring that the subtraction process compares the actual response change at the same excitation time slot position, thus avoiding calculation errors introduced by time mismatch.

[0015] As a preferred embodiment of the present invention, the step of performing eigenvalue decomposition on the disturbance deviation matrix to extract at least one abnormal disturbance feature vector specifically includes: calculating the covariance matrix of the disturbance deviation matrix and obtaining all eigenvalues ​​of the covariance matrix and the eigenvector corresponding to each eigenvalue; sorting all eigenvalues ​​and their corresponding eigenvectors in descending order of eigenvalues; selecting the three eigenvectors corresponding to the first three eigenvalues ​​after sorting as principal component vectors; calculating the energy distribution concentration of each principal component vector, and marking the principal component vectors with energy distribution concentration exceeding a preset concentration threshold as the abnormal disturbance feature vectors. By extracting the main change patterns with the most concentrated energy through principal component analysis and selecting feature vectors representing local anomalies based on energy distribution concentration, it is possible to effectively distinguish between globally uniform strain caused by pressure and singular disturbances caused by local defects, thereby improving the pertinence and accuracy of failure feature extraction.

[0016] As a preferred technical solution of the present invention, the step of matching the abnormal disturbance feature vector with multiple pre-stored standard failure mode vectors and determining the compressive failure type and failure location of the irregular metal mesh touch layer based on the matching result specifically includes: sequentially calculating the cosine similarity between the abnormal disturbance feature vector and each standard failure mode vector; selecting the standard failure mode vector with the largest cosine similarity value as the target failure mode vector; reading the failure type label pre-bound to the target failure mode vector and determining the failure type label as the compressive failure type; extracting the peak coordinates corresponding to the peak elements from the abnormal disturbance feature vector, mapping the peak coordinates back from the feature space to the physical space coordinates of the touch screen, and determining the physical space coordinates as the failure location. Using cosine similarity to measure the directional consistency between the abnormal disturbance mode and the known failure mode is unaffected by the vector magnitude and can robustly identify the failure type; simultaneously, using the reverse mapping of the peak coordinates to locate the failure location achieves coordinated output of failure type determination and failure location calibration, significantly improving the practicality of detection.

[0017] Preferably, the pre-stored multiple standard failure mode vectors include at least mesh disconnection failure mode vectors, mesh bridging short-circuit failure mode vectors, and interface peeling failure mode vectors. Vectorizing typical failure modes and pre-setting them in the system enables the detection process to automatically classify and identify the most common and impactful failure modes, enhancing the engineering applicability of the method.

[0018] The beneficial effects of this invention are: By retrieving the design files of the irregular, random-degree metal mesh touch layer from the production database, the coordinates of the regular reference nodes, the reference spacing, and the reference angle are extracted. Then, a pseudo-random number generator is invoked to apply first and second node coordinate perturbation coefficients to the horizontal and vertical coordinates, respectively, and corresponding perturbation coefficients are applied to the spacing and angle. This generates a random-degree mesh reference map carrying random deviations from the actual production process. This reference map establishes a unique, individualized electrical reference model for each touch layer, realistically reflecting its irregular, random degree distribution. After applying a dynamically increasing surface pressure load and acquiring a second electrical response map, the response difference is calculated based on this reference. This allows the inherent random perturbations of the mesh to be separated from the pressure-induced electrical changes, enabling the subtle electrical signal changes caused purely by the pressure load to be presented independently. This fundamentally avoids background interference caused by traditional regular template superposition, eliminates the risk of the mesh's random irregularity being misjudged as pressure damage, and significantly improves the realism and accuracy of pressure resistance testing.

[0019] After obtaining the first and second electrical response spectra, the induced voltage values ​​at the same receiving channel index and excitation time slot location are extracted and their differences are calculated to generate a disturbance deviation matrix, which is then normalized. This transforms the disturbance information of all channels and time slots into a structured numerical matrix. Eigenvalue decomposition is performed on this disturbance deviation matrix to calculate all eigenvalues ​​and eigenvectors of the covariance matrix. The top three principal component vectors are selected, and anomalous disturbance feature vectors are marked from them based on the energy distribution concentration threshold. These vectors collectively characterize the specific disturbance modes caused by local failures. The anomalous disturbance feature vectors are then sequentially compared with pre-stored mesh disconnection failure mode vectors, mesh bridging short-circuit failure mode vectors, and interface peeling failure mode vectors using cosine similarity calculation. The standard failure mode with the highest similarity is automatically matched to determine the failure type. Simultaneously, the coordinates of the peak elements are extracted from the anomalous disturbance feature vectors and mapped back from the feature space to the physical space of the touchscreen to obtain the failure location coordinates. This processing method decouples the macroscopic compressive test response into independently analyzable local anomaly features, enabling automatic differentiation and precise location of various failure modes such as metal mesh fracture, bridging short circuit, and interlayer peeling. It breaks through the limitation of traditional detection methods that can only provide an overall strength judgment, and provides a refined analysis method for failure diagnosis and process optimization of OLED metal mesh touch screens. Attached Figure Description

[0020] The invention will now be further described with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart of the pressure resistance testing method for OLED metal mesh touchscreens; Figure 2 This is a flowchart of the process for generating a random grid baseline map; Figure 3 This is a flowchart of the first electrical response spectrum acquisition process for an irregularly shaped metal mesh touch layer. Figure 4 This is the flowchart for acquiring the second electrical response spectrum; Figure 5 This is a flowchart of the time alignment process for acoustic emission alarm and electrical response spectrum of microcracks in cover glass pressure testing; Figure 6 This is a flowchart of a method for detecting the compressive failure of an irregularly shaped metal mesh touch layer. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] See Figure 1 This invention provides a compressive strength testing method for OLED metal mesh touchscreens. The method acquires node coordinate perturbation data, angle perturbation data, and spacing perturbation data of the irregular, randomized metal mesh touch layer of the OLED metal mesh touchscreen to be tested, generating a randomized mesh reference map. An initial electrical excitation signal is applied to the irregular, randomized metal mesh touch layer, and a first electrical response map of the irregular, randomized metal mesh touch layer is acquired. A dynamically increasing surface pressure load is applied to the cover glass of the OLED metal mesh touchscreen to be tested, and a second electrical response map of the irregular, randomized metal mesh touch layer is acquired simultaneously. The difference between the first and second electrical response maps is calculated to generate a perturbation deviation matrix. Eigenvalue decomposition is performed on the perturbation deviation matrix to extract at least one abnormal perturbation feature vector. The abnormal perturbation feature vector is matched with multiple pre-stored standard failure mode vectors, and the compressive strength failure type and failure location of the irregular, randomized metal mesh touch layer are determined based on the matching results.

[0024] In specific implementation, please refer to Figure 2 The process of acquiring the node coordinate perturbation data, angle perturbation data, and spacing perturbation data of the irregular random metal mesh touch layer of the OLED metal mesh touch screen to be tested, and generating a random mesh reference map is described in detail below: The design file for the irregular, randomized metallic mesh touch layer is retrieved from the production database. The design file stores the baseline parameters for an ideal, regular mesh. Parsing the design file extracts the coordinates of the baseline nodes, the baseline spacing, and the baseline angle. The coordinates of the baseline nodes are the design coordinates of the intersection points of each mesh in the regular mesh within a two-dimensional plane. The baseline spacing is the design distance between two adjacent parallel mesh lines, and the baseline angle is the design angle formed by the intersecting mesh lines.

[0025] A pseudo-random number generator is invoked to apply node coordinate perturbation coefficients, spacing perturbation coefficients, and angle perturbation coefficients to the regular baseline node coordinates, baseline spacing, and baseline angle, respectively, to generate the actual node coordinates, actual spacing, and actual angle. The perturbation application follows the formula below: ; in, This represents the baseline parameter value of the rule to which the perturbation is to be applied; This represents the disturbance coefficient corresponding to the type of rule reference parameter, and its value is determined based on historical process deviation data; This represents a pseudo-random number output by the pseudo-random number generator, with a value range of [-1, 1]. This indicates the actual parameter values ​​generated.

[0026] For the perturbation of the regular reference node coordinates, the regular reference node coordinates are decomposed into horizontal and vertical coordinates. A first node coordinate perturbation coefficient is assigned to the horizontal coordinate, and a second node coordinate perturbation coefficient is assigned to the vertical coordinate. The values ​​of the first and second node coordinate perturbation coefficients are obtained from the process deviation statistics table in the production database, specifically the standard deviations of the horizontal and vertical coordinate deviations of the metal mesh touch layer nodes calculated from historical manufacturing data. Following the Six Sigma criterion in statistical process control, the first node coordinate perturbation coefficient is set to three times the standard deviation of the horizontal coordinate deviation, and the second node coordinate perturbation coefficient is set to three times the standard deviation of the vertical coordinate deviation. A pseudo-random number generator is used to generate a first random number, with a value between -1 and 1. The first random number is multiplied by the first node coordinate perturbation coefficient, and the product is added to the horizontal coordinate in the regular reference node coordinates to generate the perturbated horizontal coordinate. A pseudo-random number generator is invoked to generate a second random number, with a value between -1 and 1. This second random number is multiplied by the second node coordinate perturbation coefficient, and the product is added to the ordinate of the baseline node coordinates to generate the perturbed ordinate. The perturbed abscissa and the perturbed ordinate are combined to form the actual node coordinates.

[0027] For disturbances in the reference spacing, the reference spacing is used as... Spacing perturbation coefficient as The spacing perturbation coefficient is calculated based on historical process data, and the standard deviation of the spacing deviation is set to three times the standard deviation of the spacing deviation. A pseudo-random number generator is called to generate a third random number, with the value range of [-1, 1]. The third random number is multiplied by the spacing perturbation coefficient, and the product is added to the reference spacing to generate the actual spacing.

[0028] For disturbances in the reference angle, the reference angle is used as... Angular perturbation coefficient as The angle disturbance coefficient, calculated based on historical process data, is set to three times the standard deviation of the angle deviation. A pseudo-random number generator is called to generate a fourth random number, with a value range of [-1, 1]. The fourth random number is multiplied by the angle disturbance coefficient, and the product is added to the reference angle to generate the actual angle.

[0029] After generating all actual node coordinates, actual spacing, and actual angles, the above actual data are mapped to a two-dimensional plane coordinate system. The origin is set at the lower left corner of the visible area of ​​the touchscreen, with the horizontal axis extending along the long side of the touchscreen and the vertical axis extending along the short side. Each actual node coordinate is positioned in the coordinate system according to its perturbed horizontal and vertical coordinates. For adjacent actual node coordinates, the length of the line connecting them is determined based on the actual spacing, and the deflection of the line at the node intersection is determined based on the actual angle. This constructs a pixel-level mesh topology map of the irregular random metal mesh touch layer, serving as the baseline map for the random mesh.

[0030] The random grid reference map is stored in the form of a two-dimensional matrix. The row index and column index of the matrix correspond to the discrete coordinate position in the two-dimensional plane coordinate system. The matrix element value indicates whether the position is covered by the metal grid line and the grid line number to which it belongs.

[0031] In specific implementation, please refer to Figure 3 The specific process of applying an initial electrical excitation signal to an irregularly shaped metal mesh touch layer and acquiring the first electrical response spectrum of the irregularly shaped metal mesh touch layer is described as follows: The signal generation module generates an initial electrical excitation signal, which is then divided into multiple excitation time slots by the timing control unit in chronological order. The duration of each excitation time slot is determined based on the RC time constant of the irregular metal mesh touch layer, and is set to 5 times the maximum time constant in the irregular metal mesh touch layer. Each excitation time slot is assigned a frequency code, which is taken from a preset frequency set, where the interval between adjacent frequencies is greater than twice the bandwidth of the receiving channel of the irregular metal mesh touch layer.

[0032] The driving circuit receives the excitation time slot sequence and corresponding frequency code output by the timing control unit. Internally, the driving circuit contains a number of transmit amplifiers equal to the number of transmit channels. At the beginning of each excitation time slot, the driving circuit generates an AC drive voltage of the corresponding frequency based on the frequency code of the current excitation time slot, and synchronously push-pull outputs the AC drive voltage to all transmit channels of the irregularly shaped metal mesh touch layer. A transmit channel refers to a set of parallel electrode lines on the irregularly shaped metal mesh touch layer used for injecting excitation signals.

[0033] All receiving channels of the irregularly shaped metal mesh touch layer are connected to a multi-channel synchronous acquisition circuit. A receiving channel refers to another set of electrode lines on the irregularly shaped metal mesh touch layer that intersect with the transmitting channels. In the signal link of each receiving channel, a transimpedance amplifier and an anti-aliasing filter are connected in series. At the end of each excitation time slot, the timing control unit sends a sampling trigger pulse. When the rising edge of the sampling trigger pulse arrives, the analog-to-digital converters of all channels in the multi-channel synchronous acquisition circuit simultaneously sample the induced voltage after processing by the anti-aliasing filter. The sampling result is quantized and output as a transient induced voltage value. The transient induced voltage value is the digital quantity of the induced voltage of that receiving channel at the end of the current excitation time slot.

[0034] The transient induced voltage values ​​acquired by all receiving channels across all excitation time slots are organized using a unified data structure. The receiving channel index is used as the first dimension, numbered sequentially from 1 to the total number of receiving channels. The excitation time slot number is used as the second dimension, numbered sequentially from 1 to the total number of excitation time slots. The transient induced voltage value corresponding to each receiving channel index and each excitation time slot number is filled into the corresponding position in a two-dimensional matrix. The resulting two-dimensional matrix is ​​the first electrical response spectrum.

[0035] In specific implementation, please refer to Figure 4 The process of applying a dynamically increasing surface pressure load to the cover glass of the OLED metal mesh touchscreen under test and simultaneously acquiring the second electrical response spectrum of the irregular metal mesh touch layer is described in detail below: A pressure loading head is controlled to descend vertically from an initial height. Driven by a servo motor and ball screw assembly, the pressure loading head has a polyurethane contact surface mounted on its lower end face, with an area larger than the visible area of ​​the cover glass of the OLED metal mesh touchscreen to be tested. The pressure loading head incorporates a high-precision pressure sensor with a measurement range of 0 to 500 Newtons and a resolution of 0.01 Newtons. The pressure loading head descends from its initial height at a speed of 2 millimeters per second, while the pressure sensor reads the current contact force value in real time. When the pressure sensor detects that the contact force jumps from zero to a non-zero value and remains at this non-zero value for more than 100 milliseconds, it is determined that the contact surface of the pressure loading head has achieved complete contact with the surface of the cover glass, and the servo motor immediately stops driving.

[0036] The pressure loading head gradually increases the pressure applied to the cover glass according to a preset stepped pressure curve. The stepped pressure curve is generated by the pressure control module, with preset parameters including the initial pressure value, pressure increment, and number of steps. The initial pressure value is set to 10 Newtons, the pressure increment to 10 Newtons, and the number of steps to 20. The pressure control module uses a PID closed-loop control algorithm with a proportional coefficient of 2.5, an integral coefficient of 0.8, a derivative coefficient of 0.1, and a control cycle of 10 milliseconds. In each control cycle, the pressure control module reads the feedback value from the pressure sensor, compares it with the target pressure value of the current pressure step, calculates the deviation, and then drives the servo motor to perform pressure compensation, stabilizing the actual applied pressure within ±0.5 Newtons of the target pressure value.

[0037] During the holding period of each pressure step, the steps of acquiring the first electrical response spectrum are repeated to obtain candidate electrical response spectra corresponding to the current pressure value. Specifically, the pressure control module starts a timer after the pressure loading head reaches the target pressure value of the current pressure step. The timer is implemented using hardware and has a timing accuracy of 1 microsecond. From the start of the timer, the pressure control module continuously monitors the feedback value from the pressure sensor and performs pressure compensation to keep the applied pressure within the allowable error range of the target pressure value. When the timer count reaches a preset stabilization period, the timer overflow signal triggers an acquisition command. The stabilization period is set to 5 seconds based on the stress relaxation characteristics of the metal mesh touch layer. The acquisition command is sent to the excitation signal generation module. In response to the acquisition command, the excitation signal generation module regenerates the initial electrical excitation signal, whose waveform parameters are exactly the same as the initial electrical excitation signal used when acquiring the first electrical response spectrum. The excitation signal generation module sends the initial electrical excitation signal to the driving circuit. The driving circuit injects the excitation signal into all the transmitting channels of the irregular random metal mesh touch layer. The multi-channel synchronous acquisition circuit synchronously samples the transient induced voltage values ​​of all receiving channels at the end of each excitation time slot. The signals are arranged according to the same channel index order and excitation time slot order as the first electrical response spectrum to form the response signal output by the irregular random metal mesh touch layer under the stable state of the target pressure value. The response signal is recorded as a candidate electrical response spectrum.

[0038] After acquiring candidate electrical response maps during the holding periods of all pressure steps, the candidate electrical response maps corresponding to all pressure steps are stacked in ascending order of pressure value. The stacking operation is performed on the third dimension of the data storage device. The resulting data structure is a three-dimensional array. The first dimension of the three-dimensional array is the receive channel index, the second dimension is the excitation time slot number, and the third dimension is the pressure step number, which increases in ascending order along the third dimension. After stacking, the entire three-dimensional array is output as the second electrical response map.

[0039] In specific implementation, please refer to Figure 5 During the application of dynamically increasing surface pressure loads, the acoustic emission signals of microcracks in the cover glass were monitored in real time. Multiple broadband piezoelectric acoustic emission sensors were attached to the four corners of the lower surface of the cover glass using vacuum coupling agent. Each broadband piezoelectric acoustic emission sensor had a frequency response range of 100 kHz to 1 MHz. The output of each broadband piezoelectric acoustic emission sensor was connected to a preamplifier with a gain set to 40 dB. The preamplifier output was then connected to a multi-channel acoustic emission acquisition card with a sampling rate set to 5 Mbps. The multi-channel acoustic emission acquisition card incorporated a real-time feature extraction module that continuously calculated the rise time, duration, and ring count for each acoustic emission signal. A feature threshold determination module obtained rise time thresholds, duration thresholds, and ring count thresholds. The rise time threshold was set to 50 microseconds, the duration threshold to 200 microseconds, and the ring count threshold to 10 rings. These three thresholds were obtained by looking up tables in a material acoustic emission characteristic database based on the cover glass material properties. When the acquired acoustic emission signal simultaneously satisfies the following conditions: rise time greater than rise time threshold, duration greater than duration threshold, and ring count greater than ring count threshold, the feature threshold determination module outputs a microcrack alarm signal. The alarm signal triggers the pressure control module to record the current pressure value of the pressure sensor and stores the pressure value and the corresponding microcrack acoustic emission event timestamp in the event log file.

[0040] In practice, both the first and second electrical response spectra undergo bandpass filtering before acquisition to remove power frequency interference and environmental noise. An analog bandpass filter is inserted between the output of the transimpedance amplifier and the input of the anti-aliasing filter in each receiving channel. The analog bandpass filter employs a fourth-order Butterworth topology, with its low-end cutoff frequency set to 0.8 times the lowest frequency component of the excitation signal and its high-end cutoff frequency set to 1.2 times the highest frequency component of the excitation signal. The lowest and highest frequency components of the excitation signal are extracted from the frequency encoding set of the initial electrical excitation signal. The low-end cutoff frequency of the analog bandpass filter suppresses the 50 Hz power frequency and its harmonic components, while the high-end cutoff frequency suppresses high-frequency environmental noise caused by external electromagnetic radiation. The signal processed by the analog bandpass filter is then fed into the anti-aliasing filter and the analog-to-digital converter.

[0041] In the specific implementation, before generating the disturbance deviation matrix, time alignment calibration is performed on the first and second electrical response spectra. The timestamp of the start time of the first excitation time slot is read from the data acquisition log of the first electrical response spectra, serving as the first reference time. The data acquisition log of the first electrical response spectra is generated and saved by the timing control unit when sending the initial electrical excitation signal. The timestamp of the start time of the first excitation time slot when the candidate electrical response spectra are first acquired under the first pressure step is read from the data acquisition log of the second electrical response spectra, serving as the second reference time. The time offset is calculated by subtracting the value of the first reference time from the value of the second reference time. Based on the time offset, the second electrical response spectra is shifted in the time domain. The second electrical response spectra is a three-dimensional array. The time domain shift operation specifically involves traversing all two-dimensional data planes corresponding to all pressure step numbers in the second electrical response spectra, uniformly subtracting the time offset from the time labels of all sampling points in each two-dimensional data plane, so that the adjusted time labels of the second reference time are aligned with the time labels of the first reference time. After the translation operation is completed, the adjusted second electrical response spectrum is used for the subsequent calculation of the disturbance deviation matrix.

[0042] In specific implementation, please refer to Figure 6 The specific process of calculating the difference between the first and second electrical response spectra and generating a perturbation deviation matrix is ​​described below: The second induced voltage value at the same receiver channel index and excitation time slot position as the first electrical response spectrum is extracted from the time-aligned and calibrated second electrical response spectrum. The second electrical response spectrum is a three-dimensional array. During extraction, the pressure step number is fixed, and the receiver channel index and excitation time slot number are traversed sequentially to read the voltage value at the corresponding position under each pressure step. Since the perturbation deviation matrix represents the average difference effect under multi-level pressure, the extracted second induced voltage value is the average voltage at the same receiver channel index and excitation time slot position under all pressure steps. The second induced voltage value is subtracted from the first induced voltage value at the same receiver channel index and excitation time slot position in the first electrical response spectrum to obtain a voltage difference. The sign of the voltage difference is retained to reflect the direction of increase or decrease of the induced voltage before and after pressure application.

[0043] A two-dimensional difference-position mapping table is formed by associating the voltage difference with its corresponding receiving channel index and excitation time slot position. The number of rows in the difference-position mapping table is equal to the total number of receiving channels, and the number of columns is equal to the total number of excitation time slots. The element filled in at the i-th row and j-th column position in the table is the voltage difference corresponding to receiving channel index i and excitation time slot number j.

[0044] Normalization is performed on the difference-position mapping table. The normalization process uses the range normalization method, calculated according to the following formula: ; in, The difference and position mapping table represents the first... The index of the receiving channel, the first Voltage difference at each excitation time slot; This represents the minimum voltage difference among all voltage differences in the position mapping table, starting from the set of all voltage differences. Obtained through iterative comparisons; This represents the maximum value among all voltage differences in the difference-location mapping table, starting from the entire set. Obtained through iterative comparisons; Represents the normalized th The index of the receiving channel, the first The matrix element values ​​at each excitation time slot position. All normalized matrix element values ​​constitute the perturbation bias matrix, whose row and column dimensions are consistent with the total number of receiving channels and the total number of excitation time slots.

[0045] In practice, the process of performing eigenvalue decomposition on the disturbance deviation matrix to extract at least one anomalous disturbance eigenvector is described as follows: Calculate the covariance matrix of the perturbation bias matrix. The calculation process is as follows: remove the mean from each column of the perturbation bias matrix to obtain the mean-reduced perturbation bias matrix. Multiply the mean-reduced perturbation bias matrix by its transpose, and divide the product by the total number of excitation slots minus one to obtain the covariance matrix. Solve the characteristic equation of the covariance matrix to obtain all eigenvalues ​​and the corresponding eigenvectors of each eigenvalue. Eigenvalue decomposition is performed numerically using the Jacobi iterative method, with an iterative convergence accuracy set to 1e-6.

[0046] All eigenvalues ​​and their corresponding eigenvectors are sorted in descending order of eigenvalue. The first eigenvalue in the sorted sequence is the largest eigenvalue, and its corresponding eigenvector is the first principal component vector. The second eigenvalue is the second largest eigenvalue, and its corresponding eigenvector is the second principal component vector. The third eigenvalue is the third largest eigenvalue, and its corresponding eigenvector is the third principal component vector. The three eigenvectors corresponding to the first three eigenvalues ​​after sorting are then selected as the principal component vectors.

[0047] Calculate the energy distribution concentration for each principal component vector. The energy distribution concentration is defined as the sum of the squares of the k elements with the largest absolute values ​​in the principal component vector, divided by the sum of the squares of all elements in the principal component vector. The value of k is set to 5% of the dimension of the principal component vector and rounded down. Compare the calculated energy distribution concentration with a preset concentration threshold of 0.75. This value is based on the relatively uniform energy distribution of the principal components in an irregular, randomized metal mesh touch layer under normal pressure, and is derived through statistical analysis of historical normal samples. Principal component vectors with energy distribution concentration exceeding the preset concentration threshold are marked as anomalous perturbation feature vectors.

[0048] In practical implementation, the process of matching the abnormal disturbance feature vector with multiple pre-stored standard failure mode vectors and determining the compressive failure type and failure location of the irregular random metal mesh touch layer based on the matching results is described as follows: The cosine similarity between the anomalous perturbation feature vector and each standard failure mode vector is calculated sequentially. During cosine similarity calculation, the anomalous perturbation feature vector and the stored standard failure mode vector are both L2-norm normalized. The inner product of the two normalized vectors is then calculated to obtain the cosine similarity value. The standard failure mode vector with the largest cosine similarity value is selected as the target failure mode vector. The failure type label pre-bound to the target failure mode vector is read. The failure type label is stored as an integer in the metadata field of the standard failure mode vector, and the failure type label is identified as the compressive strength failure type.

[0049] The peak coordinates corresponding to the peak elements are extracted from the anomalous disturbance feature vector. A peak element is the element with the maximum absolute value in the anomalous disturbance feature vector, and its coordinates are determined by the index position of the peak element within the feature vector. These peak coordinates are then mapped back from the feature space to the physical space coordinates of the touchscreen. This mapping is achieved by querying a correspondence table between the feature vector index and the actual receiving channel and excitation time slot. This table records the association between the index number of each element in the feature vector and the physical layout coordinates of the receiving channel, as well as the touchscreen area corresponding to the excitation time slot. The obtained physical space coordinates are then used to determine the failure location.

[0050] In practical implementation, the pre-stored standard failure mode vectors include at least three types: mesh breakage failure mode vector, mesh bridging short-circuit failure mode vector, and interface peeling failure mode vector. These standard failure mode vectors are generated by pre-collecting OLED metal mesh touchscreen samples with known failure types, performing the aforementioned excitation acquisition and difference matrix generation steps on each sample, extracting abnormal perturbation feature vectors from the perturbation deviation matrix of known failed samples, and then clustering and averaging the abnormal perturbation feature vectors of samples with the same type of failure. The mesh breakage failure mode vector corresponds to the failure mode caused by metal mesh line breakage; the mesh bridging short-circuit failure mode vector corresponds to the failure mode caused by short-circuit bridging formed by extrusion and overlapping of adjacent mesh lines; and the interface peeling failure mode vector corresponds to the failure mode caused by interface detachment between the mesh layer and the substrate layer.

[0051] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for testing the compressive strength of an OLED metal mesh touchscreen, characterized in that, Includes the following steps: Acquire the node coordinate perturbation data, angle perturbation data, and spacing perturbation data of the irregular random metal mesh touch layer of the OLED metal mesh touch screen to be tested, and generate a random mesh reference map; An initial electrical excitation signal is applied to the irregular random metal mesh touch layer, and the first electrical response spectrum of the irregular random metal mesh touch layer is acquired; A dynamically increasing surface pressure load is applied to the cover glass of the OLED metal mesh touch screen to be tested, and the second electrical response spectrum of the irregular metal mesh touch layer is acquired simultaneously. The difference between the first electrical response spectrum and the second electrical response spectrum is calculated to generate a perturbation deviation matrix, specifically: Extract the second induced voltage value from the second electrical response spectrum at the same receiving channel index and the same excitation time slot position as the first electrical response spectrum; Subtract the second induced voltage value from the first induced voltage value at the corresponding position in the first electrical response spectrum to obtain a voltage difference value; The voltage difference is associated with its corresponding receiving channel index and excitation time slot position to form a two-dimensional difference-position mapping table; The difference-position mapping table is normalized, and the result after normalization is used as the disturbance deviation matrix. The disturbance deviation matrix is ​​subjected to eigenvalue decomposition to extract at least one anomalous disturbance feature vector, specifically: Calculate the covariance matrix of the disturbance deviation matrix, and obtain all the eigenvalues ​​of the covariance matrix and the eigenvector corresponding to each eigenvalue; Sort all the eigenvalues ​​and their corresponding eigenvectors in descending order of eigenvalue; The three eigenvectors corresponding to the first three eigenvalues ​​after sorting are selected as principal component vectors. Calculate the energy distribution concentration of each principal component vector, and mark the principal component vectors whose energy distribution concentration exceeds a preset concentration threshold as the abnormal perturbation feature vectors. The abnormal disturbance feature vector is matched with multiple pre-stored standard failure mode vectors, and the compressive failure type and failure location of the irregular random metal mesh touch layer are determined based on the matching results.

2. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, The step of acquiring the node coordinate perturbation data, angle perturbation data, and spacing perturbation data of the irregular random metal mesh touch layer of the OLED metal mesh touch screen to be tested, and generating a random mesh reference map, specifically includes: Retrieve the design file of the irregular random metal mesh touch layer from the production database, and parse the coordinates of the regular reference nodes, the reference spacing, and the reference angle in the design file; A pseudo-random number generator is invoked to apply node coordinate perturbation coefficient, spacing perturbation coefficient, and angle perturbation coefficient to the rule reference node coordinates, the reference spacing, and the reference angle, respectively, to generate actual node coordinates, actual spacing, and actual angle; The actual node coordinates, the actual spacing, and the actual angle are mapped to a two-dimensional plane coordinate system to construct the random degree grid reference map.

3. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, The steps of applying an initial electrical excitation signal to the irregular metal mesh touch layer and acquiring the first electrical response spectrum of the irregular metal mesh touch layer are as follows: The initial electrical excitation signal is divided into multiple excitation time slots in chronological order, and each excitation time slot corresponds to a frequency code. The driving circuit sequentially injects the multiple excitation time slots into all the transmission channels of the irregular random metal mesh touch layer; On all receiving channels of the irregular metal mesh touch layer, the transient induced voltage value at the end of each excitation time slot is sampled synchronously; The transient induced voltage values ​​collected from all receiving channels are arranged in channel index order to form the first electrical response spectrum.

4. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, The steps of applying a dynamically increasing surface pressure load to the cover glass of the OLED metal mesh touchscreen under test, and simultaneously acquiring the second electrical response spectrum of the irregularly shaped metal mesh touch layer, are as follows: Control a pressure loading head to descend vertically from an initial height until the contact surface of the pressure loading head is completely in contact with the surface of the cover glass; The pressure loading head is controlled to gradually increase the pressure applied to the cover glass according to a preset stepped pressure curve; During the holding period of each pressure step, the step of acquiring the first electrical response spectrum is repeated to obtain the candidate electrical response spectrum corresponding to the current pressure value; The candidate electrical response spectra corresponding to all pressure steps are stacked in ascending order of pressure value to generate the second electrical response spectra.

5. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, During the application of the dynamically increasing surface pressure load, the acoustic emission signals of the microcracks in the cover glass are monitored in real time.

6. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, Both the first and second electrical response spectra were bandpass filtered before acquisition to remove power frequency interference and environmental noise.

7. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, The steps of matching the abnormal disturbance feature vector with multiple pre-stored standard failure mode vectors, and determining the compressive failure type and failure location of the irregular random metal mesh touch layer based on the matching results, are as follows: Calculate the cosine similarity between the anomalous perturbation feature vector and each standard failure mode vector in sequence; The standard failure mode vector with the largest cosine similarity value is selected as the target failure mode vector. Read the failure type label pre-bound to the target failure mode vector, and determine the failure type label as the compressive failure type; The peak coordinates corresponding to the peak elements are extracted from the abnormal disturbance feature vector, and the peak coordinates are mapped back from the feature space to the physical space coordinates of the touch screen. The physical space coordinates are then determined as the failure location.

8. The compressive strength testing method for an OLED metal mesh touchscreen according to claim 1, characterized in that, The pre-stored multiple standard failure mode vectors include at least the mesh disconnection failure mode vector, the mesh bridging short-circuit failure mode vector, and the interface peeling failure mode vector.

Images