Display panel operation state detection method and system

Abstract

The invention discloses a display panel operation state detection method and system, and relates to the technical field of display panel operation state detection, and the method comprises the steps: collecting historical electric signals and adjustment data through a sensor, carrying out the preprocessing, calculating the fluctuation amplitude of the electric signal data and the response time of the adjustment data, constructing a decision matrix, and obtaining the operation state of a display panel; obtaining a weight coefficient of the fluctuation amplitude and the response time, defining a target function fusing the weight coefficient, outputting an optimal solution, carrying out optimization according to the optimal solution, and adjusting the sampling frequency to carry out real-time sampling and denoising to obtain denoised data; according to the method, the fluctuation amplitude of the voltage and the current is calculated, the decision matrix is constructed in combination with the response time, the state set is endowed with relative importance by using the analytic hierarchy process, and the fluctuation amplitude and the response time are minimized and optimized by further using the genetic algorithm, so that the precision and the real-time performance of a subsequent detection result are effectively improved; the accuracy and robustness of the operation state detection of the display panel are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of display panel operation state detection, and particularly to a display panel operation state detection method and system. Background Art

[0002] With the rapid development of modern display technologies, display panels have been widely used in various electronic devices, including televisions, computer monitors, smartphones, and various industrial and medical devices. The operation state of a display panel directly affects the performance of the device and the user experience. To ensure the stable operation of the display panel, especially in high-load or long-term use environments, it is particularly important to monitor its operation state in real time. With the continuous progress of display technologies, traditional manual inspection methods are difficult to meet the requirements for efficient and accurate monitoring. Therefore, how to achieve precise detection of the operation state of a display panel through automated means has become a hot topic in the research and industrial communities.

[0003] For traditional display panel operation state detection methods, when removing noise from data, they mostly rely on simple filtering techniques and cannot effectively eliminate high-frequency noise or extremely abnormal data, thus affecting the accuracy of the final state evaluation. Moreover, the accuracy and real-time performance of the detection results are difficult to meet the requirements of high-standard application scenarios. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a display panel operation state detection method and system, which solve the problems that traditional display panel operation state detection methods mostly rely on simple filtering techniques when removing noise from data, cannot effectively eliminate high-frequency noise or extremely abnormal data, thus affecting the accuracy of the final state evaluation, and the accuracy and real-time performance of the detection results are difficult to meet the requirements of high-standard application scenarios.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a display panel operation state detection method, which includes:

[0008] Collect historical electrical signals and adjustment data through sensors for preprocessing, calculate the fluctuation amplitude of the electrical signal data and the response time of the adjustment data respectively, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, define an objective function for fusing the weight coefficients, output the optimal solution, and after optimization according to the optimal solution, adjust the sampling frequency for real-time sampling and denoising to obtain denoised data;

[0009] The electrical signal refers to voltage and current signals, and the adjustment data refers to the initial brightness and the adjusted target brightness of the display panel;

[0010] After decomposing the denoised data by variational mode decomposition technology, use Hilbert transform to extract instantaneous energy and frequency features, and construct feature vectors;

[0011] Use a support vector machine as a classification model, take the feature vectors as input, obtain classification results and confidence levels, sort the confidence levels, and then store the panel states.

[0012] As a preferred solution of the method for detecting the operating state of the display panel according to the present invention, wherein: respectively calculating the fluctuation amplitude of the electrical signal data and the response time of the adjustment data, constructing a decision matrix, obtaining the weight coefficients of the fluctuation amplitude and the response time, defining an objective function for fusing the weight coefficients, and the steps for outputting the optimal solution include the following:

[0013] According to the standardized voltage and current data, calculate the mean values of voltage and current, and then calculate the fluctuation amplitude by taking the standard deviation;

[0014] After taking the difference between the timestamps in the initial state and the target state as the response time, integrate it with the fluctuation amplitude to obtain a state set at different time points;

[0015] According to the response time and the fluctuation amplitude in all state sets, after constructing a decision matrix, use the ratio scale method to obtain the relative importance between the response time and the fluctuation amplitude in the decision matrix;

[0016] Use the analytic hierarchy process, assign values to the response time and the fluctuation amplitude respectively according to the relative importance, and then calculate the average value of each row in the decision matrix to obtain the weight coefficients of the fluctuation amplitude and the response time respectively:

[0017] Take the fluctuation amplitude and the response time within each state set as individuals, and generate a population for initialization;

[0018] Define an objective function to minimize the fluctuation amplitude and the response time;

[0019] Take the objective function value as the fitness value of each individual, calculate the fitness value of each individual, and sort them in ascending order;

[0020] Set the screening threshold as ω, compare the fitness value of each individual with the threshold ω respectively to obtain the retained individuals;

[0021] According to the retained individuals, perform two rounds of competition selection, select the individual with the largest fitness value in the two rounds of competition selection, generate individual pairs, and then perform the operations repeatedly in sequence. After obtaining ω individual pairs, calculate the crossover probability of the individuals in the individual pairs;

[0022] Further, according to the fitness value of each individual, calculate the selection probability of each individual in the population;

[0023] According to the selection probability of each individual, calculate the information entropy of each individual;

[0024] Use the maximization operation to select the maximum information entropy among the individuals, and use the ratio of the information entropy of the i-th individual S i to the maximum information entropy as the adjustment parameter After that, use the Sigmoid function to calculate the mutation probability of each individual;

[0025] Use a pseudo-random number generator to randomly generate the random numbers for crossover and mutation corresponding to each individual;

[0026] Compare the mutation probability of each individual with the corresponding mutation random number to obtain the result of whether each individual performs the mutation operation;

[0027] Then compare the crossover probability of the individuals in the individual pairs selected by tournament selection with the corresponding crossover random numbers to obtain the result of whether the individual pairs perform the crossover operation;

[0028] During the iteration process, according to the fitness value of the individuals, calculate the crossover probability and the mutation probability, and perform crossover and mutation operations according to the crossover probability and the mutation probability to generate new individuals. When the maximum number of iterations is reached, output the minimized fluctuation amplitude and response time.

[0029] As a preferred solution of the method for detecting the operating state of the display panel according to the present invention, wherein: after optimizing according to the optimal solution, adjust the sampling frequency for real-time sampling and denoising, and the steps for obtaining the denoised data are as follows:

[0030] Set the initial adjustment value and the proportional constant;

[0031] Take the minimized fluctuation amplitude and response time as the target state, and use the difference between the target state and the corresponding data in the state set as the deviation value;

[0032] Use the proportional control formula to obtain the new adjustment value of the fluctuation and the new adjustment value of the response;

[0033] Use the exponential decay function combined with the new adjustment value to map the minimized fluctuation amplitude and response time to obtain the optimized fluctuation amplitude and response time;

[0034] According to the optimized response time, use the Nyquist theorem to adjust the sampling frequencies of the voltage and current sensors;

[0035] Use the adjusted sampling frequency to collect the real-time voltage and current data of the display panel;

[0036] According to the optimized fluctuation amplitude, a high or low-pass filter is selected to remove noise from the voltage and current data, and the denoised data is obtained.

[0037] As a preferred solution of the display panel operating state detection method described in the present invention, wherein: after decomposing the denoised data by variational mode decomposition technology, the instantaneous energy and frequency characteristics are extracted by Hilbert transform, and constructing the feature vector includes the following steps:

[0038] According to the denoised data, perform a conversion using the fast Fourier transform to obtain a frequency-domain signal;

[0039] Use the FFT function in Matlab to transform the frequency-domain signal, integrate after obtaining the spectral peak value, and construct a spectral peak value map;

[0040] After counting the number of peaks in the spectrogram, use the number of peaks as the total number of modes;

[0041] Using maximum and minimization operations, select the highest and lowest frequencies in the spectrum, take the difference as the frequency bandwidth, use the highest frequency as the central frequency, and then use the ratio of the frequency bandwidth to the number of modes as the bandwidth constraint value;

[0042] Use modal decomposition technology to decompose the frequency-domain signal to obtain π modal functions;

[0043] Construct an objective function to minimize the reconstruction error and the frequency bandwidth;

[0044] During the iteration process, when the number of iterations reaches the maximum number, output each modal function;

[0045] Each of the modal functions refers to a signal under different frequency responses;

[0046] Using the sliding window technique, divide each modal function, calculate the local mean within each window using the mean formula, arrange them in chronological order, and take the locally adjacent local means in the arrangement result for comparison to obtain local mean retention points;

[0047] According to the retained points, draw a local mean curve, obtain the positive envelope as the actual value of the real part, and then perform a transformation to obtain the expected value of the imaginary part;

[0048] Take the ratio between the expected value of the imaginary part and the actual value of the real part as the imaginary part coefficient;

[0049] Combine the expected value of the imaginary part, the actual value of the real part, and the imaginary part coefficient to obtain an analytic signal;

[0050] Calculate the instantaneous amplitude according to the analytic signal;

[0051] Take the square of the absolute value of the instantaneous amplitude as the instantaneous energy;

[0052] Further, according to the analytical signal, the atan2 function is used to calculate the phase of the analytical signal;

[0053] According to the phase of the analytical signal, the time derivative of the phase is calculated as the instantaneous frequency;

[0054] Using the feature splicing technique, all instantaneous energies and instantaneous frequencies are spliced to obtain a feature vector.

[0055] As a preferred solution of the display panel operating state detection method described in the present invention, wherein: using a support vector machine as a classification model, and taking the feature vector as an input to obtain a classification result and a confidence level. After sorting the confidence levels, the panel state refers to using a support vector machine as a classification model. After inputting the feature vector into the classification model, a classification result and a corresponding confidence level value are output. After sorting the confidence level values in ascending order, the classification result corresponding to the maximum confidence level value is selected as the detection basis.

[0056] As a preferred solution of the display panel operating state detection method described in the present invention, wherein: the storage refers to storing the classification result, the feature vector, the minimized fluctuation amplitude, and the response time in the CSV file format, adding a corresponding timestamp to the CSV file, and then storing the CSV file using a database.

[0057] As a preferred solution of the display panel operating state detection method described in the present invention, wherein: the preprocessing of collecting historical electrical signals and adjustment data through sensors includes the following steps:

[0058] Obtain historical voltage, current data, and adjustment data at different time points from the display panel through the API interface;

[0059] According to the adjustment data, mark the timestamps for the initial state and the target state of the adjustment data;

[0060] The adjustment data refers to the initial brightness and the adjusted target brightness of the display panel;

[0061] After removing outliers from all data using median filtering, a normalization operation is performed.

[0062] In a second aspect, the present invention provides a display panel operating state detection system, including,

[0063] An acquisition and processing module for collecting electrical signals and adjustment data for preprocessing;

[0064] A calculation and output module, which is used to calculate the fluctuation amplitude of the electrical signal data and adjust the response time of the data, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, define an objective function for fusing the weight coefficients, and output the optimal solution;

[0065] An optimization and feedback module, which is used to optimize the optimal solution, readjust the sampling frequency for sampling and denoising, and obtain the denoised data;

[0066] A decomposition and extraction module, which is used to decompose the denoised data, extract the instantaneous energy and frequency characteristics, and construct a feature vector;

[0067] A classification and storage module, which is used to use the support vector machine as a classification model, take the feature vector as the input, obtain the classification result and the confidence level, sort the confidence levels, and store the panel state.

[0068] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the display panel operating state detection method described in the first aspect of the present invention is implemented.

[0069] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the display panel operating state detection method described in the first aspect of the present invention is implemented.

[0070] The beneficial effects of the present invention are as follows: by calculating the fluctuation amplitudes of the voltage and current, constructing a decision matrix in combination with the response time, using the analytic hierarchy process to assign relative importance to the state set, and further using the genetic algorithm to minimize and optimize the fluctuation amplitude and the response time, the accuracy and real-time performance of the subsequent detection results are effectively improved. Secondly, by combining the variational mode decomposition and the Hilbert transform technology to extract the instantaneous energy and frequency characteristics, and using the support vector machine classification model to achieve accurate classification of the display panel state, the accuracy and robustness of the display panel operating state detection are significantly improved. Description of the Drawings

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0072] Figure 1 It is a flowchart of the display panel operating state detection method in Embodiment 1.

[0073] Figure 2 It is the structural diagram of the display panel operation state detection system in Embodiment 1.

[0074] Figure 3 It is the flowchart of parameter optimization in Embodiment 1.

[0075] Figure 4 It is the flowchart of feature extraction in Embodiment 1. Specific implementation manners

[0076] To make the above objects, features and advantages of the present invention more obvious and understandable, the following detailed description of the specific implementation manners of the present invention will be given in conjunction with the accompanying drawings of the specification.

[0077] Many specific details are set forth in the following description in order to fully understand the present invention, but the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0078] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0079] Embodiment 1, referring to Figures 1 to 4 , which is the first embodiment of the present invention. This embodiment provides a method for detecting the operation state of a display panel, including the following steps:

[0080] S1. Collect historical electrical signals and adjustment data through sensors for preprocessing, calculate the fluctuation amplitude of the electrical signal data and the response time of the adjustment data respectively, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, define the objective function of the fusion weight coefficients, output the optimal solution, and after optimization according to the optimal solution, adjust the sampling frequency for real-time sampling and denoising to obtain denoised data;

[0081] Specifically, the preprocessing of collecting historical electrical signals and adjustment data through sensors includes the following steps:

[0082] Obtain historical voltage, current data and adjustment data at different time points from the display panel through the API interface;

[0083] According to the adjustment data, mark the timestamps of the initial state and the target state of the adjustment data;

[0084] The adjustment data refers to the initial brightness and the adjusted target brightness of the display panel;

[0085] After removing outliers from all data using median filtering, perform normalization operations.

[0086] By obtaining the historical voltage, current, and adjustment data of the display panel in real time, combining the timestamp and median filtering technology, the accuracy and stability of data processing are significantly improved. The timestamp marking ensures the chronological order of each group of data, making the adjustment process more precisely traceable. Moreover, median filtering effectively removes abnormal data, avoiding adjustment errors caused by noise or interference, thus ensuring the smoothness and reliability of the data. Secondly, the normalization operation ensures the consistency of different data sets, enabling precise comparison and effective processing of each data, and further improving the accuracy of brightness adjustment.

[0087] Furthermore, calculate the fluctuation amplitude of the electrical signal data and the response time of the adjustment data respectively, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, and define the objective function of the fusion weight coefficient. The steps for outputting the optimal solution are as follows:

[0088] According to the normalized voltage and current data, use the mean formula to calculate the mean values of voltage and current respectively, and then take the standard deviation to obtain the standard deviations of voltage and current respectively. After that, calculate the fluctuation amplitude:

[0089]

[0090] In the formula, Z(e) represents the fluctuation amplitude at time e, σ(C(e)) represents the standard deviation of voltage data at time e, and σ(V(e)) represents the standard deviation of current data at time e;

[0091] After taking the difference between the timestamps in the initial state and the target state as the response time, integrate it with the fluctuation amplitude to obtain the state sets at different time points;

[0092] According to the response time and fluctuation amplitude in all state sets, construct a decision matrix, and then use the ratio scale method to obtain the relative importance between the response time and the fluctuation amplitude in the decision matrix;

[0093] Using the analytic hierarchy process, assign values to the response time and the fluctuation amplitude respectively according to the relative importance, and then perform further normalization operations on each column of the decision matrix;

[0094] Use the mean formula to calculate the average value of each row in the decision matrix to obtain the weight coefficients of the fluctuation amplitude and the response time respectively:

[0095] Take the fluctuation amplitude and the response time within each state set as individuals and generate a population for initialization;

[0096] Define the objective function to minimize the fluctuation amplitude and the response time:

[0097] B(S i ) = δ·Z(S i ) + β·D(S i )

[0098] In the formula, B(S i ) represents the objective function value of the i-th individual S i , δ represents the weight coefficient of the fluctuation amplitude, Z(S i ) represents the fluctuation amplitude of the i-th individual S i , β represents the weight coefficient of the response time, D(S i ) represents the response time of the i-th individual S i ;

[0099] Take the objective function value as the fitness value of each individual, calculate the fitness value of each individual, and sort them in ascending order;

[0100] Set the screening threshold as ω according to the accuracy requirement and relevant domain knowledge, compare the fitness value of each individual with the threshold ω respectively. When the fitness value of an individual is less than or equal to the threshold ω, then exclude this individual, otherwise retain the individual;

[0101] Conduct two rounds of tournament selection based on the retained individuals, select the individual with the largest fitness value in the two rounds of tournament selection. After generating individual pairs, perform the operations repeatedly in sequence. After obtaining ω individual pairs, calculate the crossover probability of the individuals in the individual pairs:

[0102]

[0103] In the formula, G(S i ) represents the crossover probability of the i-th individual S i , B(S i ) represents the fitness value of the i-th individual S i , B(S j ) represents the fitness value of the j-th individual S j ;

[0104] Further calculate the selection probability of each individual in the population according to the fitness value of the individual:

[0105]

[0106] In the formula, p(S i ) represents the selection probability of the i-th individual S i , N represents the total number of individuals;

[0107] Calculate the information entropy of each individual according to the selection probability of each individual:

[0108]

[0109] In the formula, H(S i ) represents the information entropy of the i-th individual S i , and log represents the logarithmic operation;

[0110] Select the maximum information entropy among individuals using the maximization operation, and use the ratio of the information entropy of the i-th individual S i to the maximum information entropy as the adjustment parameter

[0111] Combined with the adjustment parameter, use the Sigmoid function to calculate the mutation probability of each individual:

[0112]

[0113] In the formula, a(S i ) represents the mutation probability of the i-th individual S i , and exp(·) represents the indicator function;

[0114] Use a pseudo-random number generator to randomly generate the random numbers for crossover and mutation corresponding to each individual;

[0115] Compare the mutation probability of each individual with the corresponding mutation random number. When the mutation probability of an individual is greater than or equal to the corresponding mutation random number, mutate the individual, otherwise keep the individual unchanged;

[0116] Then compare the crossover probability of the individuals in the pair selected by tournament selection with the corresponding crossover random number;

[0117] When the crossover probabilities of the corresponding individuals are both greater than or equal to the crossover random number, perform crossover operation on this pair of individuals and generate new individuals. When only one of the pair of individuals has a crossover probability greater than or equal to the crossover random number, keep the individual with a crossover probability greater than or equal to the crossover random number as the next generation and participate in subsequent crossover operations, while the individual with a crossover probability less than the crossover random number is directly used as the next generation and does not participate in subsequent crossover operations. When the crossover probabilities of the pair of individuals are both less than the corresponding crossover random number, do not participate in the crossover operation and directly use them as the next generation;

[0118] During the iteration process, calculate the crossover probability and mutation probability according to the fitness value of the individual, and perform crossover and mutation operations according to the crossover probability and mutation probability to generate new individuals. When the maximum iteration number is reached, output the minimized fluctuation amplitude and response time.

[0119] Calculate the fluctuation amplitude through the standardized voltage and current data, enabling the present invention to effectively eliminate data interference and improve calculation accuracy. The calculation method of the standard deviation not only accurately reflects the fluctuation of the signal but also ensures the robustness of data processing, thus laying a solid foundation for subsequent optimization. The analytic hierarchy process is used to assign weights to the fluctuation amplitude and response time, and the decision matrix is optimized using the weight values, improving the scientificity and rationality of the decision-making, ensuring the weight balance of each objective, and avoiding excessive interference from subjective factors. Moreover, the competition selection and fitness value screening mechanism ensure the transmission of high-quality individuals during the optimization process, screening out unfit individuals, accelerating the convergence speed of the algorithm, and improving the quality of the overall population. And the competition selection gives priority to the individuals with higher fitness for crossover, effectively avoiding the inefficiency that may be caused by completely random selection. Secondly, the calculation of the crossover probability is based on the individual fitness value, ensuring that individuals with high fitness are more likely to participate in the crossover operation, further promoting the transmission of excellent genes and optimizing the effect of genetic operations. The introduction of information entropy further enhances the diversity of the population, preventing the algorithm from falling into local optima during the search process. By calculating the information entropy, the mutation probability can be dynamically adjusted, introducing new gene mutations in a timely manner while ensuring individuals with excellent fitness, avoiding premature convergence, and enabling the present invention to flexibly adjust the search strategy according to the actual situation at different stages, enhancing the global search ability. By dynamically adjusting the operation probabilities of crossover and mutation, the balance between exploration and exploitation during the search process is ensured. Finally, the use of the Sigmoid function adjusts the mutation probability, effectively linking the intensity of mutation with the fitness of individuals and the diversity of the population, improving the adaptability and robustness of the algorithm.

[0120] Furthermore, after optimization according to the optimal solution, adjust the sampling frequency for real-time sampling and denoising. The steps to obtain the denoised data are as follows:

[0121] Set the initial adjustment value and the proportional constant based on empirical knowledge and actual requirements;

[0122] Take the minimized fluctuation amplitude and response time as the target state, and use the difference between the target state and the corresponding data in the state set as the deviation value;

[0123] Using the proportional control formula, take the difference between the product of the initial adjustment value and the proportional constant and the corresponding deviation value as the new adjustment value for the fluctuation amplitude and the response time respectively, obtaining the new adjustment value for the fluctuation and the new adjustment value for the response;

[0124] Use the exponential decay function combined with the new adjustment value to map the minimized fluctuation amplitude and response time, obtaining the optimized fluctuation amplitude and response time:

[0125]

[0126] In the formula, F represents the optimized fluctuation amplitude and response time, e represents the base of the natural logarithm, θ1 represents the new adjustment value of the fluctuation amplitude, and θ2 represents the new adjustment value of the response time;

[0127] According to the optimized response time, the sampling frequencies of the voltage and current sensors are adjusted using the Nyquist theorem;

[0128] Using the adjusted sampling frequencies, the real-time voltage and current data of the display panel are collected;

[0129] According to the optimized fluctuation amplitude, a high- or low-pass filter is selected to remove noise from the voltage and current data, obtaining denoised data.

[0130] Through the calculation of deviation values and the application of the proportional control formula, the new adjustment values of the fluctuation amplitude and response time are adjusted in real time, enabling the present invention to remain near the target state, thereby improving the optimization accuracy. Moreover, through the introduction of the exponential decay function, the optimization process is smoothed, avoiding the instability caused by over-adjustment, ensuring the reasonable range of the fluctuation amplitude and response time. Furthermore, for the adjustment of the sampling frequency, the present invention dynamically adjusts the sampling frequency based on the optimized response time according to the Nyquist theorem, ensuring the accuracy of data acquisition and avoiding signal distortion. Secondly, by selecting an appropriate high- or low-pass filter to remove noise from the data, the denoising effect is significantly improved, making the final data clearer and more reliable. The present invention combines the feedback mechanism, dynamic adjustment, and flexible denoising to achieve precise optimization and stable operation.

[0131] S2. After decomposing the denoised data through the variational mode decomposition technique, the instantaneous energy and frequency characteristics are extracted using the Hilbert transform, and a feature vector is constructed;

[0132] Specifically, after decomposing the denoised data through the variational mode decomposition technique, the steps of extracting the instantaneous energy and frequency characteristics using the Hilbert transform and constructing a feature vector include the following:

[0133] According to the denoised data, it is transformed using the fast Fourier transform to obtain a frequency-domain signal;

[0134] The frequency-domain signal is transformed using the FFT function in Matlab to obtain the spectral peak;

[0135] According to the spectral peak, integration is performed to construct a spectral peak diagram;

[0136] Using the statistical method, after counting the number of peaks in the spectrogram, the number of peaks is used as the total number of modes;

[0137] After selecting the highest and lowest frequencies in the spectrum using the maximum and minimization operations, the difference is taken as the frequency bandwidth, and the highest frequency is used as the central frequency;

[0138] Furthermore, according to the frequency bandwidth, the ratio of the frequency bandwidth to the number of modes is used as the bandwidth constraint value;

[0139] The frequency-domain signal is decomposed using the mode decomposition technique to obtain π mode functions;

[0140] Construct an objective function to minimize the reconstruction error and the frequency bandwidth:

[0141]

[0142] where u k (t) represents the k-th mode function at time t, w k represents the central frequency associated with the k-th mode function, K represents the total number of modes, represents the derivative with respect to time t, O(t) represents the frequency-domain signal at time t, ‖·‖ 2 represents the square of the absolute value, represents the bandwidth constraint value, l k (t) represents the frequency bandwidth associated with the k-th mode function at time t;

[0143] During the iteration process, when the number of iterations reaches the maximum number, each mode function is output;

[0144] Each mode function refers to the signal under different frequency responses;

[0145] Using the sliding window technique, after dividing each mode function, the local mean within each window is calculated using the mean formula;

[0146] Arrange all the local means in chronological order, and take the locally adjacent local means in the arrangement result for comparison. When the taken local mean is greater than the locally adjacent local means before and after, then retain this local mean; otherwise, eliminate this local mean;

[0147] Furthermore, according to the arrangement result, compare the head and tail local means in the arrangement result with their respective adjacent local means. When the head and tail local means are greater than their respective adjacent local means, then retain the head and tail points; otherwise, eliminate the head and tail points;

[0148] Draw a local mean curve based on the retained points to obtain the positive envelope as the actual real part value;

[0149] Use the hilbert(.) function in MATLAB to perform the Hilbert transform on the actual real part value to obtain the expected imaginary part;

[0150] Take the ratio between the expected value of the imaginary part and the actual value of the real part as the imaginary part coefficient;

[0151] Combine the expected value of the imaginary part, the actual value of the real part, and the imaginary part coefficient to obtain the analytic signal:

[0152] A k (t) = M k (t) + mE{M k (t)}

[0153] In the formula, A k (t) represents the analytic signal of the k-th mode function at time t, M k (t) represents the actual value of the real part of the k-th mode function at time t, E{M k (t)} represents the expected value of the imaginary part of the k-th mode function at time t, and m represents the imaginary part coefficient;

[0154] Calculate the instantaneous amplitude according to the analytic signal:

[0155]

[0156] In the formula, o k (t) represents the instantaneous amplitude of the k-th mode function at time t;

[0157] Take the square of the absolute value of the instantaneous amplitude as the instantaneous energy;

[0158] Further, according to the analytic signal, use the atan2 function to calculate the phase of the analytic signal:

[0159]

[0160] In the formula, arg(A k (t)) represents the phase of the analytic signal of the k-th mode function at time t, represents the real part, represents the imaginary part;

[0161] Calculate the time derivative of the phase as the instantaneous frequency according to the phase of the analytic signal:

[0162]

[0163] In the formula, y k (t) represents the instantaneous frequency of the k-th mode function at time t, represents the time derivative;

[0164] Use the feature splicing technology to splice all the instantaneous energies and instantaneous frequencies to obtain the feature vector.

[0165] The complex signal is decomposed into multiple modal components through variational mode decomposition technology, removing high-frequency noise and retaining important frequency characteristics, significantly improving the signal quality. The Hilbert transform is used to extract the instantaneous amplitude and instantaneous frequency of the signal. The instantaneous frequency reveals the changing characteristics of the signal frequency, while the instantaneous amplitude provides the change of the signal intensity over time. This enables the precise capture of the dynamic characteristics of the signal during the feature extraction process. The sliding window technique is used to calculate the local mean of the signal, effectively removing insignificant local features. By comparing the adjacent relationship of the local means before and after, more representative features can be retained, improving the accuracy of feature extraction. Secondly, the introduction of spectrum analysis and bandwidth constraint further optimizes the definition of the signal frequency range, ensuring the efficiency and accuracy of signal reconstruction. Finally, the instantaneous energy and instantaneous frequency are spliced into a feature vector for subsequent analysis.

[0166] S3. Use the support vector machine as a classification model, take the feature vector as the input, obtain the classification result and confidence level, sort the confidence levels, and store the panel state.

[0167] Specifically, using the support vector machine as a classification model and taking the feature vector as the input to obtain the classification result and confidence level, and sorting the confidence levels to obtain the panel state means using the support vector machine as a classification model. After inputting the feature vector into the classification model, the classification result and the corresponding confidence level value are output. After sorting the confidence level values in ascending order, the classification result corresponding to the maximum confidence level value is selected as the detection basis.

[0168] Among them, the construction and training process of the classification model is as follows:

[0169] Set the width parameter and Lagrange multiplier according to personal experience and relevant field knowledge.

[0170] Use the RBF kernel function to map the feature vector in combination with the width parameter to obtain the kernel function value.

[0171] Take the support vector machine as the classification model.

[0172] Combine the kernel function value and the Lagrange multiplier to define the objective function and maximize the objective function:

[0173]

[0174] In the formula, X(α) represents the objective function value related to the Lagrange multiplier α, R represents the total number of samples, α q represents the Lagrange multiplier of the q-th sample, α z represents the Lagrange multiplier of the z-th sample, x q represents the feature vector of the q-th sample, x zDenote the feature vector of the z-th sample, J(s q , s z ) represents the kernel function value, s q denotes the q-th sample, s z denotes the z-th sample;

[0175] Use a quadratic programming solver to update the model parameters. During the iteration process, when the number of iterations reaches the maximum number, stop the iteration and output the final classification model;

[0176] Output the feature vector to the classification model to obtain the classification result and the confidence value.

[0177] Through the input of the feature vector, the support vector machine can output the classification result and the confidence value, improving the accuracy and reliability of the classification. By sorting the confidence values and selecting the classification result corresponding to the maximum confidence, the possibility of misjudgment is further reduced. Secondly, the RBF kernel function solves the non-linear problem by mapping the data to a high-dimensional space, enabling the SVM to handle more complex actual data. The introduction of the Lagrange multiplier optimizes the classification boundary of the model and improves the generalization ability of the classifier by maximizing the objective function, avoiding the overfitting phenomenon. Secondly, the quadratic programming solver can effectively optimize the model parameters to ensure the accurate selection of support vectors, thereby improving the performance of the classification model. Finally, by combining the confidence value sorting with the classification result, the present invention effectively improves the efficiency and accuracy of detecting the operating state of the display panel.

[0178] Furthermore, the storage means storing the classification result, the feature vector, the minimized fluctuation amplitude, and the response time in the CSV file format, adding a corresponding timestamp to the CSV file, and then storing the CSV file using a database.

[0179] By storing the classification result, the feature vector, the fluctuation amplitude, and the response time in the CSV file format, adding a timestamp to each file, and then storing them through the database, the efficient management and access of data are realized. The present invention enhances the traceability of data. The timestamp ensures the timeliness of each record and facilitates the historical traceability of data. Through database storage, large-scale data can be quickly queried and analyzed.

[0180] This embodiment also provides a display panel operating state detection system, including:

[0181] An acquisition and processing module for acquiring electrical signals and adjustment data for preprocessing;

[0182] A calculation and output module for calculating the fluctuation amplitude of the electrical signal data and the response time of the adjustment data, constructing a decision matrix, obtaining the weight coefficients of the fluctuation amplitude and the response time, defining an objective function for fusing the weight coefficients, and outputting the optimal solution;

[0183] An optimization and feedback module, which is used to optimize the optimal solution, readjust the sampling frequency for sampling and denoising, and obtain denoised data;

[0184] A decomposition and extraction module, which is used to decompose the denoised data, extract instantaneous energy and frequency features, and construct a feature vector;

[0185] A classification and storage module, which is used to use a support vector machine as a classification model, take the feature vector as input, obtain a classification result and a confidence level, sort the confidence levels, and then store the panel state.

[0186] This embodiment also provides a computer device, which is applicable to the case of a display panel operation state detection method, and includes: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the display panel operation state detection method proposed in the above embodiment.

[0187] This computer device may be a terminal. This computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication) or other technologies. The display screen of this computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device may be a touch layer covered on the display screen, or a button, a trackball or a touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0188] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for detecting the operating state of a display panel as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM for short), electrically erasable programmable read-only memory (EEPROM for short), erasable programmable read-only memory (EPROM for short), programmable read-only memory (PROM for short), read-only memory (ROM for short), magnetic memory, flash memory, a magnetic disk or an optical disc.

[0189] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for detecting the operating state of a display panel, characterized in that: including Preprocess the historical electrical signals and adjustment data collected by sensors, calculate the fluctuation amplitude of the electrical signal data and the response time of the adjustment data respectively, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, define the objective function of the fusion weight coefficient, output the optimal solution, and after optimization according to the optimal solution, adjust the sampling frequency for real-time sampling and denoising to obtain denoised data; The electrical signals refer to voltage and current signals, and the adjustment data refers to the initial brightness and the adjusted target brightness of the display panel; After decomposing the denoised data by variational mode decomposition technology, use Hilbert transform to extract the instantaneous energy and frequency characteristics, and construct a feature vector; Use the support vector machine as a classification model, take the feature vector as the input, obtain the classification result and confidence level, sort the confidence levels, and then store the panel state.

2. The display panel operation state detection method according to claim 1, wherein: The steps of calculating the fluctuation amplitude of the electrical signal data and the response time of the adjustment data respectively, constructing a decision matrix, obtaining the weight coefficients of the fluctuation amplitude and the response time, and defining the objective function of the fusion weight coefficient and outputting the optimal solution include the following: According to the standardized voltage and current data, calculate the mean values of voltage and current, and then calculate the fluctuation amplitude by taking the standard deviation; After taking the difference between the timestamps in the initial state and the target state as the response time, integrate it with the fluctuation amplitude to obtain the state sets at different time points; According to the response time and fluctuation amplitude in all state sets, construct a decision matrix, and then use the ratio scale method to obtain the relative importance between the response time and the fluctuation amplitude in the decision matrix; Use the analytic hierarchy process, assign values to the response time and the fluctuation amplitude respectively according to the relative importance, and then calculate the average value of each row in the decision matrix to obtain the weight coefficients of the fluctuation amplitude and the response time respectively: Take the fluctuation amplitude and the response time in each state set as individuals, and generate a population for initialization; Define the objective function to minimize the fluctuation amplitude and the response time; Take the objective function value as the fitness value of each individual, calculate the fitness value of each individual, and sort them in ascending order; Set the screening threshold as ω, compare the fitness value of each individual with the threshold ω respectively to obtain the retained individuals; According to the retained individuals, perform two rounds of tournament selection, select the individual with the largest fitness value in the two rounds of tournament selection, generate individual pairs, and then operate repeatedly in turn. After obtaining ω individual pairs, calculate the crossover probability of the individuals in the individual pairs; Further calculate the selection probability of each individual in the population according to the fitness value of the individual; Calculate the information entropy of each individual according to the selection probability of each individual; Select the maximum information entropy among individuals using the maximization operation, and use the ratio of the information entropy of the $i$-th individual $S$ i to the maximum information entropy as the adjustment parameter After that, use the Sigmoid function to calculate the mutation probability of each individual; Use a pseudo-random number generator to randomly generate the random numbers of crossover and mutation corresponding to each individual; Compare the mutation probability of each individual with the corresponding mutation random number to obtain the result of whether each individual performs mutation operation; Then compare the crossover probability of the individuals in the individual pairs selected by the tournament with the corresponding crossover random number to obtain the result of whether the individual pairs perform crossover operation; During the iteration process, according to the fitness values of individuals, the crossover probability and mutation probability are calculated, and crossover and mutation operations are performed according to the crossover probability and mutation probability to generate new individuals. When the maximum number of iterations is reached, the minimized fluctuation amplitude and response time are output.

3. The display panel operation state detection method according to claim 2, wherein: After optimization according to the optimal solution, the sampling frequency is adjusted for real-time sampling and denoising, and the steps for obtaining the denoised data are as follows: Set the initial adjustment value and proportional constant; Take the minimized fluctuation amplitude and response time as the target state, and use the difference between the target state and the corresponding data in the state set as the deviation value; Using the proportional control formula, obtain the new adjustment value for fluctuation and the new adjustment value for response; Using the exponential decay function combined with the new adjustment value, map the minimized fluctuation amplitude and response time to obtain the optimized fluctuation amplitude and response time; According to the optimized response time, adjust the sampling frequency of the voltage and current sensors using the Nyquist theorem; Using the adjusted sampling frequency, collect the real-time voltage and current data of the display panel; According to the optimized fluctuation amplitude, select a high- or low-pass filter to remove noise from the voltage and current data to obtain the denoised data.

4. The display panel operation state detection method according to claim 3, characterized in that: After decomposing the denoised data through the variational mode decomposition technique, the instantaneous energy and frequency characteristics are extracted using the Hilbert transform, and the steps for constructing the feature vector are as follows: According to the denoised data, perform conversion using the fast Fourier transform to obtain the frequency-domain signal; Use the FFT function in Matlab to transform the frequency-domain signal, integrate after obtaining the spectral peaks, and construct a spectral peak diagram; After counting the number of peaks in the spectrogram, take the number of peaks as the total number of modes; Using the maximum and minimization operations, select the highest and lowest frequencies in the spectrum, take the difference as the frequency bandwidth, take the highest frequency as the center frequency, and then take the ratio of the frequency bandwidth to the number of modes as the bandwidth constraint value; Use the mode decomposition technique to decompose the frequency-domain signal to obtain π mode functions; Construct an objective function to minimize the reconstruction error and frequency bandwidth; During the iteration process, when the number of iterations reaches the maximum number, output each mode function; Each of the mode functions refers to a signal under different frequency responses; Using the sliding window technique, divide each mode function, calculate the local mean within each window using the mean formula, arrange them in chronological order, and compare the locally adjacent local means in the arrangement result to obtain the local mean retention points; Draw a local mean curve based on the retained points, obtain the positive envelope as the actual value of the real part, and perform transformation to obtain the expected value of the imaginary part; Take the ratio between the expected value of the imaginary part and the actual value of the real part as the imaginary part coefficient; Combine the expected value of the imaginary part, the actual value of the real part, and the imaginary part coefficient to obtain the analytic signal; Calculate the instantaneous amplitude according to the analytic signal; Take the square of the absolute value of the instantaneous amplitude as the instantaneous energy; Further, according to the analytic signal, use the atan2 function to calculate the phase of the analytic signal; Calculate the time derivative of the phase as the instantaneous frequency according to the phase of the analytic signal; Using the feature splicing technique, splice all the instantaneous energy and instantaneous frequency to obtain the feature vector.

5. The display panel operating state detection method according to claim 4, characterized in that: Using the support vector machine as a classification model, taking the feature vector as the input, obtaining the classification result and the confidence level, sorting the confidence levels, and obtaining the panel state. That is, using the support vector machine as a classification model, after inputting the feature vector into the classification model, outputting the classification result and the corresponding confidence level value, sorting in ascending order according to the confidence level value, and selecting the classification result corresponding to the maximum confidence level value as the detection basis.

6. The display panel operation state detection method according to claim 5, wherein: The so-called storage refers to storing the classification result, the feature vector, the minimized fluctuation amplitude, and the response time in the CSV file format, adding the corresponding timestamp to the CSV file, and then using the database to store the CSV file.

7. The display panel operating state detection method according to claim 6, wherein: The preprocessing of the historical electrical signal and the adjustment data collected by the sensor includes the following steps: Obtaining the historical voltage, current data, and adjustment data at different time points from the display panel through the API interface; According to the adjustment data, marking the timestamps for the initial state and the target state of the adjustment data; The adjustment data refers to the initial brightness of the display panel and the adjusted target brightness; After removing outliers from all data using median filtering, performing a normalization operation.

8. A display panel operation state detection system, based on the display panel operation state detection method according to any one of claims 1 to 7, characterized in that: Including, The acquisition and processing module is used to acquire the electrical signal and the adjustment data for preprocessing; The calculation and output module is used to calculate the fluctuation amplitude of the electrical signal data and the response time of the adjustment data, construct a decision matrix, obtain the weight coefficients of the fluctuation amplitude and the response time, define the objective function of the fusion weight coefficient, and output the optimal solution; The optimization and feedback module is used to optimize the optimal solution, readjust the sampling frequency for sampling and denoising, and obtain the denoised data; The decomposition and extraction module is used to decompose the denoised data, extract the instantaneous energy and frequency features, and construct a feature vector; The classification and storage module is used to use the support vector machine as a classification model, take the feature vector as the input, obtain the classification result and the confidence level, sort the confidence levels, and store the panel state.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the display panel operating state detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the display panel operating state detection method according to any one of claims 1 to 7.

Images