A health state prediction method of an 8K display device

Abstract

The application provides a health state prediction method of an 8K display device, and relates to the technical field of display health prediction.The application captures subtle changes in the collaborative work between different physical properties in each pixel node of a multi-modal spectral image by extracting a multi-modal feature vector from each pixel node and calculating a mutual information feature vector of the multi-modal feature vector.On this basis, each pixel node is further regarded as a node in a pixel node grid, the similarity between adjacent nodes is obtained by calculating the norm of the mutual information feature vector and quantifying the connection weight, then the decay rate of the connection weight in a continuous time window is calculated and the distribution entropy value is analyzed, a stability degradation entropy index is obtained, trivial connection change information is condensed into a quantitative index representing the degree of decline of the overall order of the system, then the expected time for reaching the upper limit of dynamic control is calculated by performing time series prediction on the historical sequence of the entropy index, and the jump from current state monitoring to future trend prediction is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of display health prediction, and particularly relates to a health state prediction method of an 8K display device. BACKGROUND

[0002] With the rapid development of ultra-high resolution display technology, 8K display devices have been widely used in various fields, and the reliability and stability of long-term operation are crucial. However, the 8K display contains tens of millions or even hundreds of millions of pixels, and the complexity of its structure and the huge pixel scale make it difficult for traditional health monitoring methods based on single performance parameter threshold alarm or simple image processing to cope. The existing technology usually evaluates the health state of the display by monitoring the overall brightness uniformity, chroma coordinates or detecting isolated bad points. These methods have significant limitations. They can only issue an alarm when the performance degradation has reached a certain level and the failure has already appeared, which is a passive reaction after the event and cannot achieve early prediction and early warning of failure. More importantly, the failure of the display device often begins with internal material aging, driving circuit performance degradation or microscopic connection failure, etc. These changes initially manifest as instability in the coordinated work of different physical properties, and eventually manifest as visible display defects. These macroscopic performance indicators cannot reveal early and local performance degradation in the display, nor can they quantify the decline trend of the health state.

[0003] Therefore, it is necessary to provide a health state prediction method of an 8K display device to solve the above technical problems. SUMMARY

[0004] To solve the above technical problems, the present application provides a health state prediction method of an 8K display device, which achieves the beneficial effect of efficiently and accurately predicting the health state of an 8K display.

[0005] The present application provides a health state prediction method of an 8K display device, comprising:

[0006] S1: Based on a preset detection period, periodically synchronously collecting multi-modal spectral images of the display device under a uniform test picture, and aligning the multi-modal spectral images to a unified pixel node grid;

[0007] S2: Extracting a multi-modal feature vector of the multi-modal spectral image corresponding to each pixel node in the pixel node grid, and calculating a mutual information feature vector of each pixel node based on a plurality of consecutive preset detection periods within a preset sliding time window;

[0008] S3: Based on the mutual information feature vector of each pixel node, calculating the connection weight between any two adjacent pixel nodes in the pixel node grid to obtain a connection weight matrix;

[0009] S4: calculating a connection weight decay rate based on the connection weight matrices obtained in the two adjacent preset sliding time windows, and obtaining a weight decay rate matrix;

[0010] S5: calculating a stability degradation entropy index based on the weight decay rate matrix, combining a plurality of continuous stability degradation entropy indexes into a degradation entropy index sequence, and calculating a current dynamic control upper limit, a current degradation rate and a current degradation acceleration based on the degradation entropy index sequence, and generating an entropy index prediction value sequence;

[0011] S6: calculating an expected time window for the entropy index prediction value sequence to reach the current dynamic control upper limit, and calculating a predicted remaining stability time based on the expected time window, and generating a health prediction report in combination with the current degradation rate and the current degradation acceleration.

[0012] Preferably, in step S2, the calculation step of the mutual information feature vector comprises:

[0013] For each pixel node, the instantaneous mutual information between the time series data sequences of any two different modal features in the multi-modal feature vector of the pixel node within the preset sliding time window is calculated, and the values of all the instantaneous mutual information are combined into the mutual information feature vector of the pixel node.

[0014] Preferably, in step S3, the calculation step of the connection weight comprises:

[0015] For each pixel node, the L2 norm value of its mutual information feature vector is calculated, and the L2 norm value is taken as the comprehensive synergy strength of the pixel node.

[0016] The Gaussian similarity between the comprehensive synergy strength values of each pair of adjacent pixel nodes is calculated, and the value of the Gaussian similarity is taken as the connection weight between the pair of adjacent pixel nodes.

[0017] Preferably, the bandwidth parameter used when calculating the Gaussian similarity is determined according to the standard deviation of the comprehensive synergy strength values of all the pixel nodes within a plurality of continuous preset detection periods within the preset sliding time window.

[0018] Preferably, in step S5, the probability distribution of the numerical values of all the elements in the weight decay rate matrix is calculated, and the information entropy of the probability distribution is calculated as the stability degradation entropy index.

[0019] Preferably, in step S5, the calculation step of the current dynamic control upper limit comprises:

[0020] The average value and the standard deviation of all the stability degradation entropy index values in the degradation entropy index sequence are calculated.

[0021] The average value is added to the standard deviation, multiplied by a preset proportion coefficient, and the current dynamic control upper limit is obtained.

[0022] Preferably, in step S5, the current degradation rate and the current degradation acceleration calculation step comprises:

[0023] Applying a polynomial regression algorithm, the degradation entropy index sequence is independently fitted with a function not less than the second order, to obtain an entropy index trend fitting function of the degradation entropy index sequence.

[0024] The first derivative value and the second derivative value of the entropy index trend fitting function are calculated as the current degradation rate and the current degradation acceleration, respectively.

[0025] Preferably, step S6 further comprises triggering a trend deterioration early warning signal when the average of the continuous multiple entropy index prediction values in the entropy index prediction value sequence exceeds the current dynamic control upper limit.

[0026] Preferably, step S6 further comprises the following steps:

[0027] From the weight decay rate matrix corresponding to the trend deterioration early warning signal, the pixel nodes corresponding to the connection weights whose weight decay rate values are higher than the preset high decay threshold are screened out to form an initial abnormal node set;

[0028] Applying a spatial density-based clustering algorithm to the initial abnormal node set, the abnormal node subsets that are closely adjacent in the two-dimensional spatial coordinates of the pixel node grid are identified, and each abnormal node subset is taken as an abnormal node cluster.

[0029] The abnormal nodes contained in the abnormal node cluster are mapped to the physical display area of the display device according to their two-dimensional spatial coordinates in the pixel node grid to form a potential failure area.

[0030] Preferably, the spatial density-based clustering algorithm is the DBSCAN algorithm, and the neighborhood radius parameter in the DBSCAN algorithm is pre-set based on the physical resolution of the display corresponding to the pixel node grid and the typical size of the historical potential failure area.

[0031] Compared with the related art, the health state prediction method of the 8K display device provided by the present application has the following beneficial effects:

[0032] The present application firstly lays a solid foundation for multi-dimensional physical information fusion analysis by synchronously collecting multi-modal spectral images such as visible light, short-wave infrared and ultraviolet fluorescence and performing high-precision registration, and then extracts a multi-modal feature vector from each pixel node and calculates the mutual information feature vector thereof, which can sensitively capture subtle changes in the collaborative work between different physical properties in the node, thereby revealing potential early performance degradation signs at the microscopic level, on the basis of which each pixel node is regarded as a node in the pixel node grid, the connection weight is quantified by calculating the norm of the mutual information feature vector thereof and obtaining the Gaussian similarity between adjacent nodes, so that the object of health assessment is upgraded from an isolated pixel point to a function system related to each other, then the decay rate of the connection weight in the continuous time window is calculated and the distribution entropy value thereof is analyzed to obtain a stability degradation entropy index, which condenses the trivial connection change information into a single quantitative index representing the degree of overall order degradation of the system, thereby improving the efficiency and intuitiveness of state evaluation, and then the history sequence of the entropy index is subjected to time series prediction and the expected time for reaching the upper limit of dynamic control is calculated, thereby realizing the leap from current state monitoring to future trend prediction, and the predicted remaining stable time provided by the present application provides a key time basis for formulating a precise predictive maintenance plan. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 A flowchart of a health state prediction method of an 8K display device. DETAILED DESCRIPTION

[0034] The present application will be further described below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, but not to limit the present application. In addition, it should be noted that only the parts related to the present application are shown in the drawings for the convenience of description, but not all the structures. Furthermore, the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0035] In addition, it should be noted that only the parts related to the present application are shown in the drawings for the convenience of description, but not all the contents. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The processes can be terminated when the operations are completed, but can also have additional steps not included in the drawings. The processes can correspond to methods, functions, procedures, subroutines, subprograms, etc.

[0036] Embodiment One

[0037] The application discloses a health state prediction method of an 8K display device. Figure 1 As shown in the figure,

[0038] Step S1: based on a preset detection period, periodically and synchronously collect multi-modal spectral images of the display device under a uniform test picture, and align the multi-modal spectral images to a unified pixel node grid.

[0039] In the specific implementation process, the multi-modal spectral images presented by the display under the uniform test picture are periodically and synchronously collected through the preset detection period, and the multi-modal spectral images are accurately aligned to a unified pixel node grid, wherein the preset detection period is set according to the actual use frequency and the expected service life of the display device, for example, the collection is triggered every 1 hour to ensure the continuity and representativeness of the data collection, the uniform test picture is selected as a standard test pattern, for example, including but not limited to a gray card recommended by the International Commission on Illumination, to ensure that the display device is in a stable and reproducible working state, thereby eliminating the interference of content changes on the collected data, the multi-modal spectral images include but are not limited to a visible light spectrum image, a short-wave infrared image and an ultraviolet fluorescence image, the visible light spectrum image is collected by a high-resolution visible light camera and is used for capturing the brightness distribution and color uniformity of the display device, the short-wave infrared image is collected by a short-wave infrared camera and is used for detecting the temperature distribution and thermal radiation characteristics of the surface of the display device, thereby indirectly reflecting the working state and heat dissipation efficiency of the driving circuit, and the ultraviolet fluorescence image is collected by an ultraviolet excitation camera and is used for detecting the fluorescence effect caused by the aging degree or potential defects of the optical material of the display device, the synchronous collection of the images synchronizes all the cameras through a hardware trigger signal, ensures that the multi-modal images collected at the same time are completely aligned in time, avoids the registration error caused by the time difference of the collection, and after the collection is completed, the multi-modal images need to be aligned to the unified pixel node grid, the pixel node grid is a digital coordinate system corresponding to the physical pixel array of the 8K display device, each pixel node represents the center position of a physical pixel, the registration process first extracts key feature points in the images of different modalities through an image processing algorithm including but not limited to a scale-invariant feature transform and a speeded up robust features algorithm, then calculates the spatial transformation parameters between the images by using the feature points, thereby twisting and correcting the images of different modalities to the same coordinate system, realizing the accurate alignment at the pixel level, and the registered image data is mapped to the pixel node grid, so that each grid node contains the spectral information from the multi-modal images, laying a data foundation for subsequent feature extraction and health analysis.

[0040] Step S2: Extracting the multi-modal feature vector of the multi-modal spectral image corresponding to each pixel node in the pixel node grid, and calculating the mutual information feature vector of each pixel node based on the multi-modal feature vectors of a plurality of continuous preset detection periods within a preset sliding time window.

[0041] Specifically, in step S2, the calculation step of the mutual information feature vector includes:

[0042] For each pixel node, the instantaneous mutual information between the time series data of any two different modal features in the multi-modal feature vector within the preset sliding time window is calculated, and all the instantaneous mutual information values are combined into the mutual information feature vector of the pixel node.

[0043] In the specific implementation process, the multi-modal feature vector corresponding to each pixel node is extracted from the multi-modal spectral images that have been registered and aligned to the unified pixel node grid, including but not limited to the brightness value and chrominance coordinates extracted from the visible light spectral image, the temperature value obtained by inversion from the short-wave infrared image, and the fluorescence intensity value extracted from the ultraviolet fluorescence image. These specific physical quantities constitute the original feature representation of each node under multi-modal spectrum, and then enter the mutual information calculation stage. For each specific pixel node, the time series data of the multi-modal feature vector within a preset sliding time window is taken, for example, thirty consecutive periods are taken as the length of a preset sliding time window, and the sliding step is the period interval of the preset detection period. Data is collected once per period. The instantaneous mutual information between the time series data of any two different modal features in the multi-modal feature vector of the pixel node is calculated, for example, the mutual information between each pixel node and each pair of modal features within thirty periods is calculated, including but not limited to the instantaneous mutual information between the brightness sequence and the temperature sequence, the instantaneous mutual information between the brightness sequence and the fluorescence intensity sequence, and the instantaneous mutual information between the temperature sequence and the fluorescence intensity sequence. For each pixel node, the instantaneous mutual information values calculated by each pair of modal features are combined into a new vector in a preset order, which is the mutual information feature vector of the pixel node. The mutual information feature vector comprehensively represents the dynamic correlation strength of the optical-thermal-electric material properties and other physical processes of the pixel node in the time dimension. Finally, the mutual information feature vector of each node in the entire pixel node grid is generated, which provides a basis for subsequent analysis of the functional connection between nodes.

[0044] Step S3: Based on the mutual information feature vector of each pixel node, the connection weight between any two adjacent pixel nodes in the pixel node grid is calculated to obtain the connection weight matrix.

[0045] Specifically, in step S3, the calculation step of the connection weight includes:

[0046] For each pixel node, calculate the L2 norm value of its mutual information feature vector, and take the L2 norm value as the comprehensive synergy strength of the pixel node;

[0047] Calculate the Gaussian similarity between the comprehensive synergy strength values of each pair of adjacent pixel nodes, and take the value of the Gaussian similarity as the connection weight between the pair of adjacent pixel nodes.

[0048] Specifically, the bandwidth parameter used when calculating the Gaussian similarity is determined according to the standard deviation of the comprehensive synergy strength values of all pixel nodes in a plurality of continuous preset detection periods within a preset sliding time window.

[0049] In the specific implementation process, based on the mutual information feature vector of each pixel node, the L2 norm of the mutual information feature vector of each pixel node in the pixel node grid is calculated, that is, the square root of the sum of squares of all elements of the mutual information feature vector is calculated, and the L2 norm value is a scalar, which is defined as the comprehensive synergy strength of the pixel node. Next, the connection weight between any pair of directly adjacent pixel nodes in the pixel node grid is calculated, which is calculated using a Gaussian similarity function. Specifically, for a pair of adjacent pixel nodes, their respective comprehensive synergy strength values are substituted into the Gaussian similarity function to obtain the connection weight between the pair of adjacent pixel nodes. The formula of the Gaussian similarity function is wherein, represents the connection weight between the pixel node and the pixel node and represent the comprehensive synergy strength values of the pixel node and the pixel node is the bandwidth parameter of the Gaussian function, which affects the scale sensitivity of the similarity calculation. The bandwidth parameter is not a fixed value but is dynamically determined according to the standard deviation of the comprehensive synergy strength values of all pixel nodes in a plurality of continuous detection periods within the current preset sliding time window. This makes the bandwidth parameter adaptive to the overall dispersion degree of the current network node synergy strength, thereby making the connection weight quantization more robust. For each pair of adjacent nodes, the above Gaussian similarity calculation is repeated and the result is taken as the connection weight of the node pair. Finally, the connection weights of all node pairs are systematically organized into a connection weight matrix, which completely represents the strength of the functional connection between spatially adjacent nodes in the entire display pixel network under a specific time window, thereby laying a foundation for subsequent analysis of the dynamic decay of network connection strength.

[0050] Step S4: Based on the connection weight matrices obtained in the two adjacent preset sliding time windows, calculate the connection weight decay rate to obtain a weight decay rate matrix. ​​

[0051] In the implementation process, first, two connection weight matrices respectively calculated in two adjacent preset sliding time windows are obtained, the two connection weight matrices respectively represent the functional connection strength between all adjacent node pairs in the entire pixel node grid under the previous preset time window and the current preset sliding time window, the current preset sliding time window refers to the preset sliding time window containing the latest collected multi-modal spectral image data, then for the connection weight values of each element at the same position in the connection weight matrix under two consecutive preset sliding time windows, the normalized decay rate thereof is calculated, the specific calculation method is to subtract the connection weight value under the current preset sliding time window from the connection weight value under the previous preset sliding time window and then divide the connection weight value under the previous preset sliding time window, to obtain the weight decay rate of the connection weight matrix at the element position, the positive weight decay rate indicates that the connection weight decays, the greater the weight decay rate, the more intense the decay degree, if the weight decay rate is negative, it indicates that the connection strength is enhanced instead, the reasons include system fluctuation and measurement noise, the weight decay rate of each element in the connection weight matrix is calculated, and finally a new matrix, i.e. a weight decay rate matrix, is formed according to the corresponding positions of the weight decay rates in the connection weight matrix, each element of the weight decay rate matrix accurately records the performance change rate of the connection weight with time, and provides a direct data basis for the next step of statistically analyzing the overall degradation degree of network connection stability from a global perspective.

[0052] Step S5: based on the weight decay rate matrix, calculating the stability degradation entropy index, combining a plurality of consecutive stability degradation entropy indexes into a degradation entropy index sequence, and calculating the current dynamic control upper limit, the current degradation rate and the current degradation acceleration based on the degradation entropy index sequence, and generating an entropy index prediction value sequence.

[0053] Specifically, in step S5, the probability distribution of the values of all elements in the weight decay rate matrix is counted, and the information entropy of the probability distribution is calculated as the stability degradation entropy index.

[0054] Specifically, in step S5, the calculation steps of the current dynamic control upper limit include:

[0055] The average value and the standard deviation of all stability degradation entropy index values in the degradation entropy index sequence are calculated.

[0056] After the average value is added to the standard deviation and multiplied by a preset proportion coefficient, the current dynamic control upper limit is obtained.

[0057] Specifically, in step S5, the calculation steps of the current degradation rate and the current degradation acceleration include:

[0058] A polynomial regression algorithm is applied to independently fit a function not lower than the second order to the degradation entropy index sequence, to obtain an entropy index trend fitting function of the degradation entropy index sequence.

[0059] The first derivative value and the second derivative value of the entropy index trend fitting function are calculated, which are respectively the current degradation rate and the current degradation acceleration.

[0060] In the implementation process, first, the probability distribution of all connection weight decay rate values in the weight decay rate matrix is counted, and the stability degradation entropy index is obtained by calculating the information entropy of the probability distribution, which reflects the overall disorder of the connection decay between pixel nodes; then, the stability degradation entropy indexes calculated continuously are combined into a degradation entropy index sequence in chronological order; based on the degradation entropy index sequence, the current dynamic control upper limit is calculated, and the specific method is to calculate the average value and the standard deviation of all entropy index values in the degradation entropy index sequence, and then multiply the average value plus the standard deviation by a preset proportion coefficient to obtain the dynamic control upper limit, the preset proportion coefficient is an empirical parameter for adjusting the early warning sensitivity, for example, it is set in the range of two to three according to historical data, and its role is to define a statistical tolerance boundary relative to the historical normal fluctuation range; at the same time, a time series prediction algorithm including but not limited to the autoregressive integrated moving average model is applied to fit the degradation entropy index sequence to generate an entropy index prediction value sequence to predict future trends; in addition, in order to quantify the degradation dynamics, a polynomial regression algorithm is applied to fit the degradation entropy index sequence into a function not lower than the second order to obtain an entropy index trend fitting function, and the first derivative value and the second derivative value of the entropy index trend fitting function at the current time point are calculated as the current degradation rate and the current degradation acceleration respectively to represent the speed of change and the acceleration situation.

[0061] Step S6: Calculate the expected time window when the entropy index prediction value sequence reaches the current dynamic control upper limit, and calculate the predicted remaining stable time based on the expected time window, and generate a health prediction report in combination with the current degradation rate and the current degradation acceleration.

[0062] Specifically, step S6 further includes triggering a trend deterioration early warning signal when the average value of the continuous multiple entropy index prediction values in the entropy index prediction value sequence exceeds the current dynamic control upper limit.

[0063] Specifically, step S6 further includes the following steps:

[0064] From the weight decay rate matrix corresponding to the triggering of the trend deterioration early warning signal, the pixel nodes corresponding to the connection weights whose weight decay rate values are higher than the preset high decay threshold are screened out to form an initial abnormal node set;

[0065] The spatial density-based clustering algorithm is applied to the initial abnormal node set to identify abnormal node subsets that are closely adjacent in the two-dimensional spatial coordinates of the pixel node grid, and each abnormal node subset is taken as an abnormal node cluster.

[0066] The abnormal node clusters are mapped to the physical display area of the display device according to the two-dimensional spatial coordinates of the abnormal nodes in the pixel node grid to form the potential failure area.

[0067] Specifically, the spatial density-based clustering algorithm is DBSCAN algorithm, and the neighborhood radius parameter in the DBSCAN algorithm is pre-set based on the physical resolution of the display corresponding to the pixel node grid and the typical size of the historical potential failure area.

[0068] In the implementation process, the sequence of entropy index prediction values is sequentially traversed, and in the traversal process, the first single entropy index prediction value in the sequence of entropy index prediction values that is greater than or equal to the current dynamic control upper limit is recorded as an initial data point, and a plurality of continuous data points after the initial data point are continuously traversed, for example, two data points after the initial data point are continuously traversed, and the entropy index prediction values of the two data points are added to the entropy index prediction value of the initial data point to obtain an arithmetic mean value, if the arithmetic mean value exceeds the current dynamic control upper limit, a trend deterioration early warning signal is immediately triggered, indicating that the system health state presents an accelerated degradation statistical trend, and the future time corresponding to the initial data point is determined as the expected time window; if no initial data point is found in the sequence of entropy index prediction values, the expected time window is extrapolated and calculated based on the sequence end trend using a linear interpolation method, and then the predicted remaining stable time is calculated based on the expected time window. In addition, from the weight decay rate matrix corresponding to the time when the trend deterioration early warning is triggered, the connection weights with weight decay rate values higher than a pre-set high decay threshold are screened out, the pixel nodes corresponding to these abnormal connection weights are found to form an initial abnormal node set, and then the spatial density-based clustering algorithm is applied to the node set, and the DBSCAN algorithm is used, wherein the setting step of the key parameter neighborhood radius is as follows: first, the physical size of a single pixel node is calculated by obtaining the physical size of the display device and the resolution of the pixel node grid, then the typical physical size of the potential failure area is obtained by statistical analysis from the historical failure database, the typical physical size is divided by the physical size of a single pixel node to convert it into a typical size in units of pixel nodes, and the typical size is multiplied by a pre-set scaling coefficient to obtain the final neighborhood radius parameter. The initial abnormal node set is clustered using the DBSCAN algorithm with this parameter, and the abnormal node subsets that are closely adjacent in the two-dimensional spatial coordinates of the pixel node grid are identified, each subset is defined as an abnormal node cluster, and finally these clusters are mapped to the physical display area of the display to form the potential failure area. Finally, the system integrates all the information to generate a health prediction report, the report content includes early warning state flag, predicted remaining stable time, numerical value and direction of current degradation rate and degradation acceleration, detailed location and range of potential failure area, risk level division based on remaining time and degradation acceleration, and corresponding maintenance decision suggestion, so as to comprehensively reflect the health state and future risk of the display device.

[0069] The working principle of the health state prediction method of the 8K display device provided by the application is as follows:

[0070] The application synchronously collects multi-modal spectral images of the display device under a uniform test picture based on a preset detection cycle, such as visible light spectral images, short-wave infrared images and ultraviolet fluorescence images, and accurately registers these images to a unified pixel node grid, establishes a multi-dimensional physical quantity feature vector for each physical pixel, then, based on continuous multiple detection cycle data in a preset sliding time window, calculates the instantaneous mutual information between different modal features in each pixel node, thereby generating a mutual information feature vector reflecting the cooperative working state of the physical processes such as light, heat and electricity in the pixel, next, the comprehensive cooperative strength of the mutual information feature vector of each pixel node is calculated, and the Gaussian similarity of the comprehensive cooperative strength between adjacent pixel nodes is further calculated to quantify the functional connection weight therebetween, thereby forming a connection weight matrix, which depicts the functional correlation network between the pixels of the display device, then the system calculates the normalized decay rate of each connection weight by comparing the connection weight matrices of two adjacent time windows, obtains a weight decay rate matrix, thereby capturing the dynamic degradation behavior of the functional network, calculates the stability degradation entropy index based on the information entropy of the numerical value distribution of all elements of the weight decay rate matrix, quantifies the overall disorder of the network connection decay, and after a plurality of continuous entropy indexes form a sequence, on the one hand, the average value and the standard deviation of the sequence are calculated, and a preset proportion coefficient is combined to determine the dynamic control upper limit, on the other hand, a time series prediction algorithm is applied to generate an entropy index prediction value sequence to predict the future trend, and the current degradation rate and acceleration are calculated by polynomial regression fitting on the historical entropy index sequence to represent the deterioration trend and urgency of the health state, finally, the expected time for the entropy index prediction value sequence to reach the dynamic control upper limit is calculated to obtain the predicted remaining stable time, and a comprehensive health prediction report is generated by fusing the degradation rate acceleration, thereby realizing a complete closed loop from micro physical signal perception to macro system health trend prediction.

[0071] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one flow or multiple flows and / or blocks Figure 1 The device that implements the functions specified in one flow or multiple flows and / or blocks

[0072] Those skilled in the art can understand that all or part of the steps of various methods in the above embodiments can be completed by instructing the relevant hardware by means of a program, and the program can be stored in a computer readable storage medium, including Read-Only Memory (ROM), Random Access Memory (RAM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), One-time Programmable Read-Only Memory (OTPROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM) or other optical disk storage, magnetic disk storage, magnetic tape storage, or any other medium that can be used to carry or store data in a computer readable manner.

[0073] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements, but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the statement "comprising a" does not exclude the existence of additional identical elements in the process, method, article, or apparatus that includes the element.

Claims

1. A method for predicting the health status of an 8K display device, characterized in that, The health state prediction method comprises: S1: periodically synchronously collecting multi-modal spectral images of a display device under a uniform test picture based on a preset detection period, and registering and aligning the multi-modal spectral images to a unified pixel node grid; S2: extracting a multi-modal feature vector of each pixel node in the pixel node grid from the multi-modal spectral images corresponding to the pixel node, and calculating a mutual information feature vector of each pixel node based on a plurality of continuous preset detection periods within a preset sliding time window; S3: calculating a connection weight between any two adjacent pixel nodes in the pixel node grid based on the mutual information feature vector of each pixel node, and obtaining a connection weight matrix; S4: calculating a connection weight decay rate based on the connection weight matrices obtained within two adjacent preset sliding time windows, and obtaining a weight decay rate matrix; S5: calculating a stability degradation entropy index based on the weight decay rate matrix, combining a plurality of continuous stability degradation entropy indexes into a degradation entropy index sequence, calculating a current dynamic control upper limit, a current degradation rate and a current degradation acceleration based on the degradation entropy index sequence, and generating an entropy index prediction value sequence; S6: calculating an expected time window for the entropy index prediction value sequence to reach the current dynamic control upper limit, calculating a predicted remaining stable time based on the expected time window, and generating a health prediction report in combination with the current degradation rate and the current degradation acceleration. 2.The health state prediction method of an 8K display device according to claim 1, wherein, In step S2, the calculation step of the mutual information feature vector comprises: For each pixel node, the instantaneous mutual information between the time series of any two different modal features in the multi-modal feature vector of the pixel node within the preset sliding time window is calculated, and the values of all the instantaneous mutual information are combined into the mutual information feature vector of the pixel node. 3.The health state prediction method of an 8K display device according to claim 2, wherein, In step S3, the calculation step of the connection weight comprises: For each pixel node, the L2 norm value of its mutual information feature vector is calculated as the comprehensive cooperation strength of the pixel node; The Gaussian similarity between the comprehensive cooperation strength values of each pair of adjacent pixel nodes is calculated, and the value of the Gaussian similarity is taken as the connection weight between the pair of adjacent pixel nodes. 4.The health state prediction method of an 8K display device according to claim 3, wherein, The bandwidth parameter used when calculating the Gaussian similarity is determined according to the standard deviation of the comprehensive cooperation strength values of all pixel nodes within a plurality of continuous preset detection periods within the preset sliding time window. 5.The health state prediction method of an 8K display device according to claim 4, wherein, In step S5, the probability distribution of the numerical values of all elements in the weight decay rate matrix is calculated, and the information entropy of the probability distribution is calculated as the stability degradation entropy index. 6.The health state prediction method of an 8K display device according to claim 5, wherein, In step S5, the calculation step of the current dynamic control upper limit comprises: The average value and the standard deviation of all stability degradation entropy index values in the degradation entropy index sequence are calculated; The average value is added to the standard deviation, multiplied by a preset proportion coefficient, and the current dynamic control upper limit is obtained. 7.The health state prediction method of an 8K display device according to claim 6, wherein, In step S5, the current degradation rate and the current degradation acceleration calculation step comprises: A polynomial regression algorithm is applied to independently fit a function of not less than the second order to the degradation entropy index sequence, to obtain an entropy index trend fitting function of the degradation entropy index sequence; The first derivative value and the second derivative value of the entropy index trend fitting function are calculated as the current degradation rate and the current degradation acceleration, respectively. 8.The health state prediction method of an 8K display device according to claim 7, wherein, Step S6 further comprises triggering a trend deterioration early warning signal when the mean of the consecutive plurality of entropy index prediction values in the sequence of entropy index prediction values exceeds the current dynamic control upper limit. 9.The health state prediction method of an 8K display device according to claim 8, wherein, Step S6 further comprises the following steps: From the weight decay rate matrix corresponding to the trend deterioration early warning signal, screening out the pixel nodes corresponding to the connection weights whose weight decay rate values are higher than the preset high decay threshold value, to form an initial abnormal node set; Applying a spatial density-based clustering algorithm to the initial abnormal node set, identifying abnormal node subsets that are closely adjacent in the two-dimensional spatial coordinates of the pixel node grid, and taking each abnormal node subset as an abnormal node cluster; Mapping the abnormal nodes contained in the abnormal node cluster to the physical display area of the display device according to their two-dimensional spatial coordinates in the pixel node grid to form a potential fault area. 10.The health state prediction method of an 8K display device according to claim 8, wherein, The spatial density-based clustering algorithm is a DBSCAN algorithm, and the neighborhood radius parameter in the DBSCAN algorithm is pre-set based on the physical resolution of the corresponding display of the pixel node grid and the typical size of the historical potential fault area.

Images