Intelligent LED display screen pixel abnormality monitoring method and system

Abstract

The application discloses an intelligent LED display screen pixel abnormality monitoring method and system, and relates to the technical field of computer data processing.The method collects the pixel point driving current waveform output by the display screen driving chip in a single row scanning period, and obtains an initial current signal containing a rising edge, a platform period and a falling edge by intercepting in a time window triggered after the rising edge; the amplitude spectrum and the phase spectrum are obtained through Hamming window weighting and fast Fourier transform, the high-frequency and low-frequency amplitude components are separated, and the high-frequency amplitude component is normalized; the reference amplitude vector is called from the reference characteristic library by the row and column position index code and is subjected to dimension matching, the characteristic difference value is obtained after calculating the Euclidean distance and performing temperature drift compensation according to the temperature of the driving chip; when the characteristic difference value exceeds the difference threshold value, the aging attenuation and the open circuit and short circuit fault are distinguished according to the ratio of the fundamental frequency and the double-frequency amplitude, and the positioning result is converted into physical coordinates to output a diagnosis report, thereby improving the accuracy of fault diagnosis and the maintenance efficiency.

Description

Technical Field

[0001] This application belongs to the field of computer data processing technology, specifically a method and system for monitoring pixel abnormalities in an intelligent LED display screen. Background Technology

[0002] Intelligent LED displays play a vital role in modern information interaction and visual presentation, and the health of their pixels directly determines the final image quality. In the manufacturing and operation maintenance of display devices, the decline in pixel driving capability often precedes the appearance of visible defects in the image; therefore, early monitoring of pixel driving status is of great significance.

[0003] In related technologies, pixel monitoring methods mostly rely on image acquisition from outside the screen and comparison with the overall image. This method is easily affected by ambient light and is often only passively detected after obvious display errors appear on the screen surface, making it difficult to access the actual physical driving mechanism inside the screen to discover potential problems. From the perspective of the image display process inside the screen, the image presentation depends on the high-speed scanning of the pixel array by the driving current, row by row or column by column. The process of the driving current illuminating each pixel array occurs in a mere microsecond, and the electrical signal feedback in the circuit where the pixel is located is also fleeting. If there is a time difference between the acquisition rhythm of the external monitoring device and the scanning rhythm of the internal driving chip, it is impossible to accurately capture the instantaneous change in electrical signal, resulting in chaotic and disorganized electrical signals that cannot be analyzed to determine the row and column positions corresponding to the electrical signals, and thus making it impossible to locate pixels that failed to respond normally within a specific scanning cycle.

[0004] In addition, the aging and degradation of the driver chip during long-term operation can lead to a decrease in pixel driving capability and may cause open circuit or short circuit faults. Existing methods are difficult to distinguish and accurately locate these two types of abnormalities.

[0005] Therefore, how to ensure that the monitoring process is consistent with the dot matrix scanning timing of the display driver chip, and accurately collect electrical signals and identify and locate abnormal pixels at the microsecond-level illumination moment, is a technical problem that needs to be solved to achieve bottom-level dot matrix pixel abnormality monitoring. Summary of the Invention

[0006] To address the above issues, this application provides a method and system for monitoring pixel anomalies in intelligent LED displays, which solves the problem of monitoring pixel anomalies at the underlying dot matrix level.

[0007] To achieve the above objectives, the technical solution adopted in this application is as follows: According to a first aspect of this application, a method for monitoring pixel anomalies in an intelligent LED display screen is provided, comprising: The pixel drive current waveform output by the display driver chip during a single line scan cycle is acquired, and the initial current signal, including the rising edge, plateau period, and falling edge, is captured within the time window after the rising edge of the drive current is triggered. After applying a Hamming window function to the initial current signal for weighting, a fast Fourier transform is performed to obtain a frequency domain complex sequence. The amplitude spectrum and phase spectrum are then calculated from the frequency domain complex sequence. The frequency bands above the third harmonic are extracted from the amplitude spectrum as high-frequency amplitude components, and the frequency bands from the fundamental frequency to the third harmonic are taken as low-frequency amplitude components. Normalization and correction are then performed on the high-frequency amplitude components. A row and column position index code is generated from the row and column numbers obtained from the current scan. A reference amplitude vector is retrieved from a pre-established reference feature library using the row and column position index code, and the high-frequency amplitude component is matched with the reference amplitude vector in terms of dimension. After dimension matching, the Euclidean distance between the high-frequency amplitude component and the reference amplitude vector is calculated, and temperature drift compensation is performed on the Euclidean distance value according to the current driving chip temperature to obtain the feature difference value; If the characteristic difference value exceeds a preset difference threshold, the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component is extracted. Based on the comparison result of the ratio with a preset attenuation judgment threshold, aging attenuation anomalies and open-circuit / short-circuit faults are distinguished to obtain the anomaly type. An abnormal pixel location result is generated based on the abnormality type and the row and column position index code, and then converted into physical coordinates of the display screen before outputting a pixel abnormality diagnosis report.

[0008] Preferably, the acquisition of the pixel driving current waveform output by the display screen driver chip within a single line scan cycle, and the interception of the initial current signal including the rising edge, plateau period, and falling edge within the time window after the rising edge of the driving current is triggered, includes: The number of sampling points is set to be an integer multiple of the driving pulse width, and the driving current waveform of the pixel is discretized to obtain a discrete current sequence; A first-order difference operation is performed on the discrete current sequence to obtain a difference result. If the difference result is greater than a preset slope threshold, the trigger point of the rising edge of the driving current in the discrete current sequence is determined. A time window is opened based on the trigger point, and the discrete current sequence is extracted within the time window to obtain an initial current signal containing the rising edge, plateau period, and falling edge.

[0009] Preferably, after obtaining the initial current signal, the method further includes: smoothing the initial current signal using a moving average filter to obtain a smoothed current signal with high-frequency noise removed; Calculate the local variance of the smoothed current signal. If the local variance is less than a preset fluctuation threshold, determine the plateau period data segment in the smoothed current signal and take the mean of the plateau period data segment as the reference current value. The pixel driving end time is obtained based on the intersection of the reference current value and the falling edge of the smooth current signal. The actual driving pulse width is determined based on the time difference between the pixel driving end time and the trigger point. The actual driving pulse width is used as the setting of the number of sampling points in the next cycle.

[0010] Preferably, the step of applying a Hamming window function to the initial current signal, performing a fast Fourier transform to obtain a frequency domain complex sequence, and then separating and calculating the amplitude spectrum and phase spectrum from the frequency domain complex sequence includes: A weighted time-domain signal is obtained by applying a Hamming window function to the initial current signal; the driving frequency is acquired, and a frequency resolution parameter is set to one-tenth of the driving frequency. A fast Fourier transform is then performed on the weighted time-domain signal according to the frequency resolution parameter to obtain a frequency-domain complex sequence. The real part sequence and the imaginary part sequence are separated from the frequency domain complex sequence, and the amplitude spectrum and the phase spectrum are calculated based on the real part sequence and the imaginary part sequence.

[0011] Preferably, the step of extracting frequency bands above the third harmonic from the amplitude spectrum as high-frequency amplitude components and frequency bands from the fundamental frequency to the third harmonic as low-frequency amplitude components, and performing normalization processing on the high-frequency amplitude components, includes: The driving fundamental frequency is determined from the amplitude spectrum, and the frequency bands above the third harmonic are separated from the amplitude spectrum as high-frequency amplitude components, and the frequency bands from the fundamental frequency to the third harmonic are separated as low-frequency amplitude components; The high-frequency amplitude components are normalized using range standardization to obtain normalized high-frequency amplitude components with values ​​in the range of zero to one. The deviation value of the adjacent phase difference in the phase spectrum is calculated. If the deviation value is greater than the preset jump determination condition, it is determined that there is a jump point in the phase spectrum, and phase expansion continuity correction is performed on the phase spectrum with the jump point.

[0012] Preferably, the step of generating a row and column position index code from the current scan row and column number, retrieving a reference amplitude vector from a pre-established reference feature library using the row and column position index code, and performing dimensional matching between the high-frequency amplitude component and the reference amplitude vector includes: The binary data of the current scanned row number is shifted left by a preset number of bits and then bitwise ORed with the current scanned column number to obtain the row and column position index code; The corresponding benchmark amplitude vector is retrieved by querying the benchmark feature library using the row and column position index codes. The dimension difference value is obtained by comparing the dimension of the reference amplitude vector with the dimension of the high-frequency amplitude component. If the dimension difference value is greater than a preset reference threshold, the longer vector is truncated. If the dimension difference value is less than the preset reference threshold, the shorter vector is zero-filled, thus completing the dimension matching.

[0013] Preferably, after the calculation dimension is matched, the Euclidean distance value between the two is calculated, and temperature drift compensation is performed on the Euclidean distance value according to the current driving chip temperature to obtain the feature difference value, including: The difference between the high-frequency amplitude component after dimension matching and the corresponding frequency point value of the reference amplitude vector is calculated to obtain a frequency point difference sequence. The arithmetic square root of the sum of squares of the values ​​in the frequency point difference sequence is then used to obtain the Euclidean distance value. The current driver chip temperature is obtained from the temperature sensor, and the temperature deviation value is obtained by calculating the difference between the current driver chip temperature and the preset standard temperature. The temperature drift compensation coefficient is calculated based on the temperature deviation value and the preset drift rate. The Euclidean distance value is then corrected using the temperature drift compensation coefficient to obtain the characteristic difference value after temperature compensation.

[0014] Preferably, if the feature difference value exceeds a preset difference threshold, the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component is extracted. Based on the comparison result of the ratio with a preset attenuation judgment threshold, aging attenuation anomalies and open-circuit / short-circuit faults are distinguished to obtain the anomaly type, including: If the characteristic difference value exceeds a preset difference threshold, then the fundamental frequency amplitude and the second harmonic amplitude are extracted from the low-frequency amplitude component, and the amplitude ratio is obtained by dividing the fundamental frequency amplitude by the second harmonic amplitude. If the amplitude ratio is less than the preset attenuation judgment threshold, then the corresponding pixel is determined to have an abnormal driving capability attenuation, and the abnormality type is recorded as aging attenuation; If the amplitude ratio is greater than the preset attenuation judgment threshold, then the corresponding pixel is determined to have an open circuit or short circuit fault.

[0015] Preferably, the step of generating abnormal pixel location results based on the abnormality type and the row and column position index codes, converting them into physical coordinates on the display screen, and then outputting a pixel abnormality diagnosis report includes: Generate abnormal pixel location results based on the abnormality type and the row and column position index codes, and extract the row and column numbers from the abnormal pixel location results; Multiply the row number and column number by the pixel pitch parameter of the display screen and add the panel border offset to obtain the millimeter-level position value in the physical coordinate system of the display screen; The standard pixel brightness value corresponding to the target pixel is obtained based on the millimeter-level position value. The difference between the actual brightness value of the target pixel and the standard pixel brightness value is used as the feature difference value. The structured fields of the millimeter-level position value, the anomaly type and the feature difference value are filled, and a pixel anomaly diagnosis report is output.

[0016] According to a second aspect of this application, a smart LED display pixel anomaly monitoring system is provided, comprising: The signal acquisition module is configured to acquire the pixel drive current waveform output by the display driver chip in a single line scan cycle, and to capture the initial current signal containing the rising edge, plateau period, and falling edge within the time window after the rising edge of the drive current is triggered; The frequency domain transformation module is configured to apply Hamming window function weighting to the initial current signal and perform a fast Fourier transform to obtain the amplitude spectrum and phase spectrum; The component separation module is configured to extract high-frequency amplitude components and low-frequency amplitude components from the amplitude spectrum, and to perform normalization processing on the high-frequency amplitude components; The feature matching module is configured to retrieve a reference amplitude vector from a reference feature library based on the row and column position index code, perform dimensional matching between the high-frequency amplitude component and the reference amplitude vector, calculate the Euclidean distance value, and perform temperature drift compensation based on the current driver chip temperature to obtain the feature difference value; The anomaly detection module is configured to, when the feature difference value exceeds a preset difference threshold, distinguish between aging attenuation anomalies and open / short circuit faults based on the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component; The positioning output module is configured to generate abnormal pixel positioning results based on the abnormality type and row and column position index codes, convert them into physical coordinates on the display screen, and then output a pixel abnormality diagnostic report.

[0017] The embodiments of this application have the following advantages: By acquiring the pixel drive current waveform output by the display driver chip in a single line scan cycle, and capturing the complete waveform segment within a microsecond-level time window after the rising edge of the drive current is triggered, the acquisition process is kept consistent with the dot matrix scanning timing of the driver chip. This allows for the accurate acquisition of the initial current signal, including the rising edge, plateau period, and falling edge, at the moment of illumination, overcoming the problem that external image acquisition is difficult to capture instantaneous electrical signals.

[0018] By using Hamming window function weighting and fast Fourier transform, the time-domain current signal is converted into frequency-domain amplitude and phase spectra, and the high-frequency amplitude components and low-frequency amplitude components are separated. This allows the physical states of pixel driving circuits, such as aging, attenuation, open circuits, and short circuits, to be clearly represented in the frequency domain, thus improving the accuracy of anomaly identification.

[0019] The high-frequency amplitude components are matched with a pre-established benchmark feature library by Euclidean distance using row and column position index codes, and drift compensation correction is performed according to the temperature of the driving chip, which reduces the interference of temperature changes on the acquired signal and improves the stability of feature difference determination.

[0020] By comparing the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component with the attenuation judgment threshold, two abnormal modes, namely aging attenuation and open / short circuit, are distinguished. The abnormal pixel location results are converted into millimeter-level physical coordinates to generate a diagnostic report, which realizes accurate identification and location of display screen pixel abnormalities, effectively improving the accuracy of fault diagnosis and maintenance efficiency. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the steps of the intelligent LED display pixel anomaly monitoring method in this application embodiment; Figure 2 This is a structural block diagram of the intelligent LED display pixel anomaly monitoring system in this application embodiment. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solution, the present application will be described in detail below with reference to the embodiments. The description in this section is only exemplary and explanatory, and should not be used to limit the scope of protection of the present application in any way.

[0023] like Figure 1 As shown, the intelligent LED display pixel anomaly monitoring method provided in this embodiment is applied to pixel drive status monitoring in display device manufacturing and operation maintenance, and specifically includes the following steps S101 to S107.

[0024] S101 acquires the pixel drive current waveform output by the display driver chip in a single line scan cycle, and captures the initial current signal containing the rising edge, plateau period and falling edge within the time window after the rising edge of the drive current is triggered.

[0025] Specifically, a high-frequency current sensor is connected to the pixel driving circuit to collect the pixel driving current waveform output by the display driver chip within a single line scan cycle. In this embodiment, the line scan cycle is set to 32 microseconds. This value is based on the illumination duration of a single line when the LED display driver chip scans line by line, which is determined by the screen resolution and refresh rate of the monitored display. It is read from the timing configuration register of the driver chip before monitoring to ensure that the acquisition rhythm is consistent with the dot matrix scanning timing. The analog-to-digital conversion module continuously acquires the current waveform at a sampling rate of 200 MHz. The value of this sampling rate is based on the Nyquist sampling theorem, that is, the sampling rate should not be less than twice the highest frequency component of the acquired signal. To ensure the complete restoration of the microsecond-level transient current, this embodiment uses 200 MHz, which is much higher than this lower limit.

[0026] The number of sampling points is set to an integer multiple of the driving pulse width. The tuning method is as follows: first, the driving pulse width is read as two microseconds from the driving chip configuration parameters, and then the number of basic sampling points within a single pulse width is calculated by multiplying the sampling rate by the driving pulse width, that is, two hundred megahertz multiplied by two microseconds to obtain four hundred basic sampling points; then, the total number of sampling points in the interception window is configured as an integer multiple of the number of basic sampling points according to a preset integer multiple coefficient. In this embodiment, the integer multiple coefficient is five, that is, two thousand sampling points are obtained. The value of this integer multiple coefficient is based on the need to cover the complete fluctuation range of rising edge, plateau period and falling edge. If the coefficient is too small, it will truncate the waveform, and if it is too large, it will introduce invalid noise. Therefore, it is five times to balance integrity and signal-to-noise ratio.

[0027] The collected pixel driving current waveforms are discretized according to the aforementioned number of sampling points to obtain a discrete current sequence. A first-order difference operation is then performed on the discrete current sequence to obtain the difference result. The specific calculation method for the first-order difference operation is as follows: the current value of each sampling point in the discrete current sequence is subtracted from the current value of its preceding adjacent sampling point. The difference is then divided by the sampling time interval between the two sampling points to obtain the current change rate at that sampling point. If the difference results of multiple consecutive sampling points are all greater than a preset slope threshold, then that point is determined as the trigger point for the rising edge of the driving current. The slope threshold is tuned as follows: based on the typical rise rate of the current at the instant a normal pixel lights up, with a margin, this embodiment uses 50 microamps per nanosecond. That is, when the current change rate of three consecutive sampling points exceeds 50 microamps per nanosecond, it is determined to be a rising edge trigger. This value can distinguish between normal lighting transitions and small fluctuations during plateau periods.

[0028] A time window is opened based on the trigger point. The time window is in the microsecond range and is tuned to be greater than the drive pulse width to accommodate the complete waveform without significantly exceeding the single line scan cycle. In this embodiment, 10 microseconds is used. Within the time window, a discrete current sequence consisting of the above two thousand sampling points is extracted, and morphological feature recognition is performed on the sequence. The part of the sequence in which the current rises from zero is marked as the rising edge. The part of the middle section where the current fluctuation amplitude is within ±2% is divided into the plateau period. The part of the sequence in which the current falls back to zero is identified as the falling edge, thus obtaining an initial current signal containing the rising edge, plateau period, and falling edge. The plateau period is divided based on the local variance of the smoothed current signal, that is, the local variance of the current sequence within the sliding window is calculated. If the local variance is less than a preset fluctuation threshold, it is determined to be a plateau period data segment. The fluctuation threshold is tuned to take the upper limit of the variance corresponding to 2% of the mean current of the plateau period. This value is based on the premise that the current during normal lighting should be basically stable and the fluctuation should not exceed 2% of the mean.

[0029] S102, apply Hamming window function weighting to the initial current signal to suppress spectral leakage. By setting the frequency resolution parameter to 1 / 10 of the driving frequency, perform fast Fourier transform on the weighted time-domain signal to obtain a frequency-domain complex sequence. Separate the real and imaginary parts from the complex sequence and calculate the amplitude spectrum and phase spectrum.

[0030] Specifically, following the initial current signal obtained in step S101, a Hamming window sequence of the same length as the initial current signal is first generated and multiplied point by point with the initial current signal to obtain a weighted time-domain signal. The specific calculation method of the Hamming window function is as follows: for the nth sampling point in the sequence, its window coefficient is equal to 0.54 minus 0.46 multiplied by the cosine term. The phase of the cosine term is twice pi multiplied by n and then divided by the total length of the sequence minus one. Among them, 0.54 and 0.46 are the standard coefficients of the Hamming window. The basis for this value is to make the two ends of the window function smoothly decay to near zero level, thereby effectively eliminating the abrupt change at the signal truncation edge and suppressing spectral leakage.

[0031] The driving frequency is then read from the timing configuration of the driver chip. In this embodiment, the driving frequency is 50 Hz, and the frequency resolution parameter is set to 1 / 10 of the driving frequency, i.e., 5 Hz. The method and basis for setting the frequency resolution parameter to 1 / 10 of the driving frequency is as follows: the smaller the frequency resolution, the more finely adjacent spectral lines are distinguished. Setting it to 1 / 10 of the driving frequency ensures sufficient spectral line spacing between the fundamental frequency and its harmonics, while avoiding an excessively large number of transformation points that would increase the computational load. From the mathematical relationship that frequency resolution equals sampling rate divided by the number of transformation points, it can be deduced that the required number of transformation points equals sampling rate divided by frequency resolution. The number of transformation points determined in this way matches the total number of sampling points intercepted in step S101.

[0032] A fast Fourier transform (FFT) is performed on the weighted time-domain signal according to the aforementioned number of transform points to obtain a frequency-domain complex sequence. The FFT employs a radix-2 decimation-time algorithm. The real and imaginary parts of the frequency-domain complex sequence are then separated, and the amplitude and phase spectra are calculated.

[0033] The amplitude spectrum is calculated as follows: for each frequency point, the arithmetic square root of the sum of the squares of its real and imaginary parts is taken, and then the number of points is normalized by dividing by the number of transformation points to obtain the amplitude of that frequency point. The phase spectrum is calculated as follows: for each frequency point, the arctangent function value of the ratio of its imaginary to its real parts is taken to obtain the phase of that frequency point. Thus, amplitude and phase spectra corresponding one-to-one with each frequency point are obtained for use in step S103 to extract frequency band components.

[0034] S103: Extract the frequency band above the third harmonic of the driving fundamental frequency from the amplitude spectrum as the high-frequency amplitude component, extract the frequency band between the fundamental frequency and the third harmonic as the low-frequency amplitude component, perform normalization processing on the high-frequency amplitude component to map its numerical range to the zero to one interval, and perform phase expansion continuity correction on the phase spectrum to eliminate jump points.

[0035] Specifically, following the amplitude spectrum and phase spectrum obtained in step S102, the position of the spectral line corresponding to the driving fundamental frequency is first determined from the amplitude spectrum. In this embodiment, the driving fundamental frequency is 50Hz. Using the driving fundamental frequency as a reference, the frequency bands in the amplitude spectrum with frequencies higher than three harmonics of the fundamental frequency are extracted as high-frequency amplitude components, and the frequency bands with frequencies between the fundamental frequency and its third harmonic are extracted as low-frequency amplitude components. The tuning basis for using the third harmonic as the high-low frequency boundary is as follows: weak abnormal characteristics such as aging attenuation and axial friction of the pixel driving circuit are mainly reflected in the higher harmonic frequency band, while the basic operating state is mainly reflected in the low-frequency band between the fundamental frequency and its third harmonic. Therefore, using the third harmonic as the boundary can separate the two types of information, which can be used for anomaly identification and state characterization respectively.

[0036] The high-frequency amplitude components are normalized using range standardization. The specific calculation method is as follows: first, the maximum and minimum amplitude values ​​within the high-frequency band are statistically determined, and the difference between them is taken as the range. Then, for each amplitude data point within the band, the minimum amplitude value is subtracted, and the result is divided by the range, thus accurately mapping all high-frequency amplitude values ​​to the 0-1 value range. For example, when the maximum amplitude value within the band is 10.5 and the minimum amplitude value is 0.2, the range is 10.3. Therefore, for each amplitude value, 0.2 is subtracted, and the result is divided by 10.3. The basis for using range standardization is to eliminate the interference of differences in the dimensions of different pixels or different sensors on subsequent distance calculation and classification.

[0037] The phase spectrum is subjected to phase expansion continuity correction. The specific calculation method is as follows: Since the arctangent function will restrict the phase to between negative and positive pi, the phase sequence is traversed and the absolute value of the phase difference between two adjacent frequency points is calculated. If the absolute value of the phase difference between adjacent points is greater than the preset jump judgment condition, it is determined that there is a jump point, and the subsequent phase data is compensated by an integer multiple of twice pi.

[0038] The jump determination condition is tuned by taking a value close to pi. In this embodiment, it is set to 3.14. The basis for this is that the change in true phase between adjacent frequency points is usually much smaller than pi. Once the difference between adjacent phases exceeds approximately one pi, it can be identified as a false jump caused by arctangent operation. For example, when the originally continuous phase suddenly changes from 2.5 to -2.8, the absolute value of the difference between adjacent phases exceeds 3.14. Then, twice pi, i.e., 6.28, is added to the jump point and the subsequent data to correct it to a continuous true phase value of 3.4.

[0039] S104. Generate row and column position index codes based on the combination of the current scan row and column numbers. Retrieve the reference amplitude vector at the corresponding position from the pre-established reference feature library using the index codes. If the dimension of the reference amplitude vector is inconsistent with the dimension of the current high-frequency amplitude component, truncate the longer vector or zero-padded the shorter vector to complete the feature vector dimension matching.

[0040] Specifically, the current scan row number and the current scan column number are obtained and combined to generate a row and column position index code. The specific calculation method for the row and column position index code is as follows: the binary data of the current scan row number is shifted left by a preset number of bits, and then a bitwise OR operation is performed with the current scan column number to obtain a unique integer index code.

[0041] The preset left shift bit is set by taking a value not less than the maximum binary width of the column number. In this embodiment, 16 bits are used. This is based on the fact that the number of pixels in a single row of the screen is within the range that 16 bits can represent. Shifting left by 16 bits ensures that the row number and column number do not overlap in the integer index code, thus giving each pixel a unique index. For example, when the current scan row number is 128 and the column number is 256, shifting the row number left by 16 bits and performing a bitwise OR operation with the column number yields the integer index code 8388864.

[0042] A hash lookup algorithm is used to locate the storage address corresponding to the row and column position index code in the reference feature library, and the pre-stored reference amplitude vector is retrieved. The reference feature library is a key-value mapping table pre-established and stored in non-volatile memory, with the row and column position index code as the key and the high-frequency amplitude component obtained by collecting and processing the corresponding pixel in a healthy state as the value.

[0043] The benchmark feature library is configured as follows: before the display device is manufactured and shipped, or during the first power-on self-test, the high-frequency amplitude components of each pixel on the screen are collected and processed according to steps S101 to S103, and stored one by one using the corresponding row and column position index code as the key, thereby establishing a benchmark for the health status of each pixel. In this embodiment, the benchmark amplitude vector is set as an array containing 64 floating-point elements. The value of this dimension is based on the number of spectral lines of the high-frequency amplitude component in the selected high-frequency band. In the example, the first element of the benchmark amplitude vector corresponding to a certain index code is 0.15 and the last element is 0.89.

[0044] Following the high-frequency amplitude component obtained in step S103, the dimension difference value is obtained by comparing the dimension of the reference amplitude vector with that of the high-frequency amplitude component. If the dimension difference value is greater than a preset reference threshold, a truncation operation is performed on the longer vector, that is, redundant elements are removed from the end of the longer vector; if the dimension difference value is less than the preset reference threshold, a zero-padding operation is performed on the shorter vector, that is, zero-valued elements are appended to the end of the shorter vector.

[0045] The benchmark threshold is set to zero, meaning that dimension matching is triggered whenever the dimensions of the two vectors are not equal. This is based on the requirement that the dimensions of the two vectors be strictly consistent for subsequent Euclidean distance calculations. For example, if the current high-frequency amplitude component has a dimension of 72 and the benchmark vector has a dimension of 64, the dimension difference is 8. The last eight redundant high-frequency edge elements of the longer high-frequency amplitude component are truncated to reduce it to 64 dimensions. If the current high-frequency amplitude component is only 60 dimensions, four zero-valued elements are appended to its end to expand and align it to 64 dimensions, thereby ensuring that the two feature vectors participating in the subsequent distance calculation are completely consistent in data structure.

[0046] S105 calculates the sum of squares of the differences between the high-frequency amplitude components and the corresponding frequency points of the reference amplitude vector, takes the square root of the sum of squares to obtain the Euclidean distance value, obtains the current driving chip temperature from the temperature sensor, calculates the temperature drift compensation coefficient based on the difference between the temperature and the standard temperature, and corrects the Euclidean distance value with the compensation coefficient to obtain the characteristic difference value after temperature compensation.

[0047] Specifically, following the high-frequency amplitude component and the reference amplitude vector after dimensional matching in step S104, the difference between their corresponding frequency points is calculated to obtain a frequency difference sequence. The square root of the sum of the squares of the frequency difference sequence is then used to obtain the Euclidean distance. The specific calculation method for the Euclidean distance is shown in the following formula: Where d represents the Euclidean distance value, N represents the dimension of the vector after dimension matching, and a i b represents the value of the i-th frequency point of the high-frequency amplitude component. iThis represents the value at the i-th frequency point of the reference amplitude vector. For example, if we extract three frequency points: 1 kHz, 2 kHz, and 3 kHz, the high-frequency amplitude component values ​​are 2.5, 1.8, and 0.9, respectively. The corresponding values ​​of the reference amplitude vector are 2.1, 1.5, and 0.8. The corresponding differences are 0.4, 0.3, and 0.1, which square to 0.16, 0.09, and 0.01. The sum of the squares is 0.26, and taking the square root yields an Euclidean distance of approximately 0.5099.

[0048] Because the driver chip generates heat under prolonged high-load operation, it alters the impedance characteristics of its internal analog-to-digital conversion circuit, causing temperature drift errors in the acquired high-frequency signals. Therefore, the current driver chip temperature is obtained from a temperature sensor, and temperature drift compensation is performed. Specifically, the current driver chip temperature is obtained by reading the register data of the built-in temperature sensor in real time through the internal integrated circuit bus, and the difference between the current driver chip temperature and the preset standard temperature is calculated to obtain the temperature deviation value.

[0049] The standard temperature is set by taking the normal operating temperature reference of the driver chip. In this embodiment, 25 degrees Celsius is used, based on the fact that the reference amplitude vector in the reference feature library was collected and established at this normal temperature. Therefore, this temperature is used as the zero point for temperature compensation. If the temperature deviation is greater than the preset deviation threshold, the temperature drift compensation coefficient is calculated. The deviation threshold is set by taking a small temperature tolerance to avoid unnecessary compensation near the normal temperature. In this embodiment, 5 degrees Celsius is used.

[0050] The specific calculation method for the temperature drift compensation coefficient is shown in the following formula: Where k represents the temperature drift compensation coefficient, T represents the current driver chip temperature, T0 represents the standard temperature, and β represents the preset linear drift rate. The linear drift rate is tuned by calibrating and measuring high-frequency signals collected from the same driver chip at different temperatures, and determining it by the relative offset of the signal amplitude caused by a unit temperature change. In this embodiment, it is set to 0.0015 per degree Celsius. Multiplying the temperature drift compensation coefficient by the Euclidean distance value yields the characteristic difference value after temperature compensation. For example, if the current driver chip temperature is 85.0 degrees Celsius, the difference from the standard temperature of 25.0 degrees Celsius is 60.0 degrees Celsius. Multiplying this by the linear drift rate of 0.0015 gives 0.09. Subtracting 0.09 from 1 gives the temperature drift compensation coefficient of 0.91. Multiplying the initial Euclidean distance of 0.5099 by 0.91 finally yields a characteristic difference value of approximately 0.4640 after temperature compensation.

[0051] S106, if the feature difference value exceeds the preset difference threshold, the ratio of the fundamental frequency amplitude to the second harmonic amplitude is extracted from the low frequency amplitude component. If the ratio is less than the preset attenuation judgment threshold, the pixel is determined to have abnormal driving capability attenuation, and the abnormality type is recorded as aging attenuation. If the ratio is greater than the attenuation judgment threshold, it is determined to be an open circuit or short circuit fault.

[0052] Specifically, following the temperature-compensated feature difference value obtained in step S105, it is compared with a preset difference threshold. The difference threshold is determined by taking the upper bound of the distribution of feature difference values ​​for healthy pixels and allowing for a margin. It is determined by the statistical results from the baseline feature library establishment phase, based on the assumption that the feature difference values ​​of healthy pixels should be stably at a low level; exceeding this threshold indicates that the pixel deviates from a healthy state. In this embodiment, when the calculated feature difference value of a pixel is 45 and the preset difference threshold is 30, since 45 exceeds 30, the pixel is determined to be abnormal, and the subsequent differentiation process is triggered.

[0053] When the feature difference value exceeds the difference threshold, the fundamental frequency amplitude and the second harmonic amplitude are extracted from the low-frequency amplitude component obtained in step S103, and the amplitude ratio is obtained by dividing the fundamental frequency amplitude by the second harmonic amplitude. The amplitude ratio is compared with a preset attenuation judgment threshold: if the amplitude ratio is less than the attenuation judgment threshold, it is determined that the pixel has an abnormal drive capability attenuation, and the abnormality type is recorded as aging attenuation; if the amplitude ratio is greater than the attenuation judgment threshold, it is determined that the pixel has an open circuit or short circuit fault.

[0054] The attenuation judgment threshold is set as follows: by statistically analyzing the fundamental frequency second harmonic amplitude ratio of aging attenuation samples and open-circuit / short-circuit samples, the boundary value of the amplitude ratio distribution of the two types of samples is taken. In this embodiment, it is set to 200%. The basis is that the fundamental frequency energy decreases relatively during aging attenuation, resulting in a lower amplitude ratio, while the waveform distortion during open-circuit / short-circuit causes a significant change in the second harmonic energy relative to the fundamental frequency, resulting in a higher amplitude ratio.

[0055] For example, if a pixel has a fundamental frequency amplitude of 120 microvolts and a second harmonic amplitude of 80 microvolts, with an amplitude ratio of 150%, which is less than the attenuation judgment threshold of 200%, it is judged as an aging attenuation anomaly and recorded in the database; while if an adjacent pixel has a fundamental frequency amplitude of 300 microvolts and a second harmonic amplitude of 100 microvolts, with an amplitude ratio of 300%, which is greater than the attenuation judgment threshold of 200%, then it is judged as an open circuit or short circuit fault, and the corresponding anomaly type is obtained.

[0056] S107, Generate abnormal pixel location results based on the abnormality type and row and column position index codes, convert the row and column numbers in the location results into physical coordinates on the display screen, calculate the millimeter-level position of the pixel on the panel using the physical coordinates, and output a pixel abnormality diagnosis report containing position coordinates, abnormality type, and feature difference values.

[0057] Specifically, based on the anomaly type obtained in step S106 and the row and column position index codes obtained in step S104, anomaly pixel location results are generated, and row and column numbers are extracted from these results. The row and column numbers are then multiplied by the pixel pitch parameter of the display screen to obtain the physical coordinate values ​​in the display screen's physical coordinate system. The pixel pitch parameter is determined by reading it from the display screen's factory specifications, representing the center-to-center distance between adjacent pixels on the panel; in this embodiment, it is set to 150 micrometers. The display screen's size parameters also include screen resolution; in this embodiment, it is set to 3840 x 2160, and these values ​​are all based on the factory specifications of the monitored display screen.

[0058] The specific calculation method for the millimeter-level position is as follows: multiply the column number by the pixel pitch to obtain the horizontal physical coordinate, multiply the row number by the pixel pitch to obtain the vertical physical coordinate, then divide the horizontal and vertical coordinates by 1000 to convert from micrometers to millimeters, and add the panel border offset to obtain the absolute millimeter-level position value of the pixel on the panel. The panel border offset is determined by taking the fixed edge distance of the upper left corner of the effective display area of ​​the panel relative to the physical origin of the panel, which is determined by the panel structure. In this embodiment, both the horizontal and vertical offsets are 5 millimeters. For example, when the anomaly type is a dark spot and the row and column position index codes correspond to row number 1080 and column number 1920, column number 1920 multiplied by the pixel pitch of 150 micrometers gives a horizontal offset of 288,000 micrometers, and row number 1080 multiplied by 150 micrometers gives a vertical offset of 162,000 micrometers. Dividing these by 1000 converts the position to a horizontal offset of 288 millimeters and a vertical offset of 162 millimeters. Adding the horizontal and vertical border offsets of 5 millimeters gives an absolute position of 293 millimeters horizontally and 167 millimeters vertically.

[0059] The standard pixel brightness value corresponding to the target pixel is obtained based on the millimeter-level position value. The difference between the actual brightness value of the target pixel and the standard pixel brightness value is used as the feature difference value. The millimeter-level position value, anomaly type, and feature difference value are filled with structured fields, encapsulated into a standard data format, written to the underlying database, and a pixel anomaly diagnosis report is output. The standard pixel brightness value is adjusted by taking the rated grayscale reference value of the pixel when it is displayed normally. In this embodiment, it is taken as 255. For example, if the actual grayscale value of the target pixel is 12, the feature difference value 243 is obtained by subtracting it from the standard grayscale value 255. The diagnosis report is output along with the position coordinates of 293 mm horizontally and 167 mm vertically and the dark spot anomaly type.

[0060] In another embodiment of this application, the horizontal scanning period is set to 16 microseconds, the driving pulse width is 1 microsecond, and the sampling rate is still 200 MHz. Therefore, the number of basic sampling points within a single pulse width is 200 MHz multiplied by 1 microsecond, which is 200 basic sampling points. The integer multiple is still 5, resulting in 1000 sampling points, and the time window is correspondingly set to 5 microseconds. The processing flow of the remaining steps S102 to S107 is the same as in Embodiment 1. This embodiment shows that the number of sampling points, time window, and other parameters of this application are adaptively determined according to the driving pulse width and the horizontal scanning period, thereby adapting to different display screen specifications.

[0061] In another embodiment of this application, the current high-frequency amplitude component has a dimension of 60, the reference amplitude vector has a dimension of 64, the dimension difference is 4, and the current component is relatively short. Therefore, a zero-padding operation is performed on the current high-frequency amplitude component, appending four elements with values ​​of 0 to the end to expand it to 64 dimensions. In Embodiment 1, when the current component is longer, a truncation operation is performed. The remaining steps are the same as in Embodiment 1. This embodiment shows that this application can adaptively select truncation or zero-padding to complete dimension matching based on the dimensional relationship between the current high-frequency amplitude component and the reference amplitude vector.

[0062] like Figure 2 As shown, this embodiment provides an intelligent LED display pixel anomaly monitoring system, including a signal acquisition module, a frequency domain transformation module, a component separation module, a feature matching module, an anomaly determination module, and a positioning output module. Wherein: The signal acquisition module is configured to execute step S101, acquiring the pixel drive current waveform output by the display driver chip within a single line scan cycle and capturing the initial current signal; The frequency domain transformation module is configured to execute step S102, which involves windowing the initial current signal and performing a fast Fourier transform to obtain the amplitude spectrum and phase spectrum. The component separation module is configured to perform step S103, extracting high-frequency amplitude components and low-frequency amplitude components and normalizing the high-frequency amplitude components; The feature matching module is configured to execute steps S104 and S105, which involve retrieving the reference amplitude vector, performing dimensional matching, calculating the Euclidean distance, and performing temperature drift compensation to obtain feature difference values. The anomaly detection module is configured to execute step S106 to distinguish between aging degradation anomalies and open / short circuit faults; The positioning output module is configured to execute step S107, generate abnormal pixel positioning results, convert them into physical coordinates, and then output a diagnostic report.

[0063] Each of the above modules and method steps corresponds to one another. The output of the signal acquisition module serves as the input of the frequency domain transformation module, and they are passed in sequence to form a complete monitoring system.

[0064] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the technical solutions of this application; the above examples are only for the purpose of helping to understand the methods and core ideas of this application.

[0065] The above description is merely a preferred embodiment of this application. It should be noted that, due to the limitations of written expression, there are objectively infinite specific structures. For those skilled in the art, without departing from the principles of this application, several improvements, modifications, or variations can be made, and the above technical features can be combined in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of this application to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A method for monitoring pixel abnormalities in an intelligent LED display screen, characterized in that, include: The pixel drive current waveform output by the display driver chip during a single line scan cycle is acquired, and the initial current signal, including the rising edge, plateau period, and falling edge, is captured within the time window after the rising edge of the drive current is triggered. After applying a Hamming window function to the initial current signal for weighting, a fast Fourier transform is performed to obtain a frequency domain complex sequence. The amplitude spectrum and phase spectrum are then calculated from the frequency domain complex sequence. The frequency bands above the third harmonic are extracted from the amplitude spectrum as high-frequency amplitude components, and the frequency bands from the fundamental frequency to the third harmonic are taken as low-frequency amplitude components. Normalization and correction are then performed on the high-frequency amplitude components. A row and column position index code is generated from the row and column numbers obtained from the current scan. A reference amplitude vector is retrieved from a pre-established reference feature library using the row and column position index code, and the high-frequency amplitude component is matched with the reference amplitude vector in terms of dimension. After dimension matching, the Euclidean distance between the high-frequency amplitude component and the reference amplitude vector is calculated, and temperature drift compensation is performed on the Euclidean distance value according to the current driving chip temperature to obtain the feature difference value; If the characteristic difference value exceeds a preset difference threshold, the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component is extracted. Based on the comparison result of the ratio with a preset attenuation judgment threshold, aging attenuation anomalies and open-circuit / short-circuit faults are distinguished to obtain the anomaly type. An abnormal pixel location result is generated based on the abnormality type and the row and column position index code, and then converted into physical coordinates of the display screen before outputting a pixel abnormality diagnosis report.

2. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The acquisition of the pixel drive current waveform output by the display screen driver chip within a single line scan cycle, and the interception of the initial current signal including the rising edge, plateau period, and falling edge within the time window after the rising edge of the drive current triggers it, includes: The number of sampling points is set to be an integer multiple of the driving pulse width, and the driving current waveform of the pixel is discretized to obtain a discrete current sequence; A first-order difference operation is performed on the discrete current sequence to obtain a difference result. If the difference result is greater than a preset slope threshold, the trigger point of the rising edge of the driving current in the discrete current sequence is determined. A time window is opened based on the trigger point, and the discrete current sequence is extracted within the time window to obtain an initial current signal containing the rising edge, plateau period, and falling edge.

3. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 2, characterized in that, After obtaining the initial current signal, the method further includes: smoothing the initial current signal using a moving average filter to obtain a smoothed current signal with high-frequency noise removed; Calculate the local variance of the smoothed current signal. If the local variance is less than a preset fluctuation threshold, determine the plateau period data segment in the smoothed current signal and take the mean of the plateau period data segment as the reference current value. The pixel driving end time is obtained based on the intersection of the reference current value and the falling edge of the smooth current signal. The actual driving pulse width is determined based on the time difference between the pixel driving end time and the trigger point. The actual driving pulse width is used as the setting of the number of sampling points in the next cycle.

4. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The process of applying a Hamming window function to weight the initial current signal and then performing a Fast Fourier Transform to obtain a complex sequence in the frequency domain, followed by separating and calculating the amplitude spectrum and phase spectrum from the complex sequence in the frequency domain, includes: A weighted time-domain signal is obtained by applying a Hamming window function to the initial current signal; the driving frequency is acquired, and a frequency resolution parameter is set to one-tenth of the driving frequency. A fast Fourier transform is then performed on the weighted time-domain signal according to the frequency resolution parameter to obtain a frequency-domain complex sequence. The real part sequence and the imaginary part sequence are separated from the frequency domain complex sequence, and the amplitude spectrum and the phase spectrum are calculated based on the real part sequence and the imaginary part sequence.

5. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The step of extracting frequency bands above the third harmonic from the amplitude spectrum as high-frequency amplitude components and frequency bands from the fundamental frequency to the third harmonic as low-frequency amplitude components, and performing normalization processing on the high-frequency amplitude components, includes: The driving fundamental frequency is determined from the amplitude spectrum, and the frequency bands above the third harmonic are separated from the amplitude spectrum as high-frequency amplitude components, and the frequency bands from the fundamental frequency to the third harmonic are separated as low-frequency amplitude components; The high-frequency amplitude components are normalized using range standardization to obtain normalized high-frequency amplitude components with values ​​in the range of zero to one. The deviation value of the adjacent phase difference in the phase spectrum is calculated. If the deviation value is greater than the preset jump determination condition, it is determined that there is a jump point in the phase spectrum, and phase expansion continuity correction is performed on the phase spectrum with the jump point.

6. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The process of generating row and column position index codes from the row and column numbers obtained from the current scan, retrieving a reference amplitude vector from a pre-established reference feature library using the row and column position index codes, and performing dimensional matching between the high-frequency amplitude components and the reference amplitude vector includes: The binary data of the current scanned row number is shifted left by a preset number of bits and then bitwise ORed with the current scanned column number to obtain the row and column position index code; The corresponding benchmark amplitude vector is retrieved by querying the benchmark feature library using the row and column position index codes. The dimension difference value is obtained by comparing the dimension of the reference amplitude vector with the dimension of the high-frequency amplitude component. If the dimension difference value is greater than a preset reference threshold, the longer vector is truncated. If the dimension difference value is less than the preset reference threshold, the shorter vector is zero-filled, thus completing the dimension matching.

7. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The calculated dimensionality is matched, and the Euclidean distance between the two is calculated. Temperature drift compensation is then performed on the Euclidean distance based on the current driver chip temperature to obtain feature difference values, including: The difference between the high-frequency amplitude component after dimension matching and the corresponding frequency point value of the reference amplitude vector is calculated to obtain a frequency point difference sequence. The square root of the sum of the squares of the values ​​in the frequency point difference sequence is then taken to obtain the Euclidean distance value. The current driver chip temperature is obtained from the temperature sensor, and the temperature deviation value is obtained by calculating the difference between the current driver chip temperature and the preset standard temperature. The temperature drift compensation coefficient is calculated based on the temperature deviation value and the preset drift rate. The Euclidean distance value is then corrected using the temperature drift compensation coefficient to obtain the characteristic difference value after temperature compensation.

8. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, If the characteristic difference value exceeds a preset difference threshold, the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component is extracted. Based on the comparison result of the ratio with a preset attenuation judgment threshold, aging attenuation anomalies and open-circuit / short-circuit faults are distinguished to obtain the anomaly type, including: If the characteristic difference value exceeds a preset difference threshold, then the fundamental frequency amplitude and the second harmonic amplitude are extracted from the low-frequency amplitude component, and the amplitude ratio is obtained by dividing the fundamental frequency amplitude by the second harmonic amplitude. If the amplitude ratio is less than the preset attenuation judgment threshold, then the corresponding pixel is determined to have an abnormal driving capability attenuation, and the abnormality type is recorded as aging attenuation; If the amplitude ratio is greater than the preset attenuation judgment threshold, then the corresponding pixel is determined to have an open circuit or short circuit fault.

9. The method for monitoring pixel abnormalities in an intelligent LED display screen according to claim 1, characterized in that, The step of generating abnormal pixel location results based on the abnormality type and the row and column position index codes, converting them into physical coordinates on the display screen, and then outputting a pixel abnormality diagnosis report includes: Generate abnormal pixel location results based on the abnormality type and the row and column position index codes, and extract the row and column numbers from the abnormal pixel location results; Multiply the row number and column number by the pixel pitch parameter of the display screen and add the panel border offset to obtain the millimeter-level position value in the physical coordinate system of the display screen; The standard pixel brightness value corresponding to the target pixel is obtained based on the millimeter-level position value. The difference between the actual brightness value of the target pixel and the standard pixel brightness value is used as the feature difference value. The structured fields of the millimeter-level position value, the anomaly type and the feature difference value are filled, and a pixel anomaly diagnosis report is output.

10. A smart LED display pixel anomaly monitoring system, characterized in that, include: The signal acquisition module is configured to acquire the pixel drive current waveform output by the display driver chip in a single line scan cycle, and to capture the initial current signal containing the rising edge, plateau period, and falling edge within the time window after the rising edge of the drive current is triggered; The frequency domain transformation module is configured to apply Hamming window function weighting to the initial current signal and perform a fast Fourier transform to obtain the amplitude spectrum and phase spectrum; The component separation module is configured to extract high-frequency amplitude components and low-frequency amplitude components from the amplitude spectrum, and to perform normalization processing on the high-frequency amplitude components; The feature matching module is configured to retrieve a reference amplitude vector from a reference feature library based on the row and column position index code, perform dimensional matching between the high-frequency amplitude component and the reference amplitude vector, calculate the Euclidean distance value, and perform temperature drift compensation based on the current driver chip temperature to obtain the feature difference value; The anomaly detection module is configured to, when the feature difference value exceeds a preset difference threshold, distinguish between aging attenuation anomalies and open / short circuit faults based on the ratio of the fundamental frequency amplitude to the second harmonic amplitude in the low-frequency amplitude component; The positioning output module is configured to generate abnormal pixel positioning results based on the abnormality type and row and column position index codes, convert them into physical coordinates on the display screen, and then output a pixel abnormality diagnostic report.

Images