Display fault prediction system based on big data

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a big data-based display fault prediction system, which comprises a gray-scale synchronization module, a fluctuation identification module, a stage judgment module, a trajectory comparison module and a trend early warning module, analyzes gray-scale time sequence data output by a gray-scale optical acquisition instrument based on a display device, and determines whether a fault occurs or not by checking data integrity. And comparing the gray scale content of each frame with a driving chip synchronizing signal. According to the invention, by checking the synchronism and sequence integrity of the collected data, accurate arrangement of information on a multi-time sequence level, continuity analysis based on a change track and feature node extraction are realized, the difference expression of a key moment and a turning point of a focusing stage in a display process is clarified, and the brightness and a gray scale correlation trend are subjected to multi-dimensional comparison, so that the accuracy of the display process is improved. The dynamic state of the risk area is comprehensively assessed in combination with the trend continuity between the nodes, the capability of capturing display abnormal symptoms is improved, early sensing and timely early warning are realized, and the stability and risk prevention capability of fault prediction of the display device are effectively enhanced.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to a display fault prediction system based on big data. Background Technology

[0002] Fault diagnosis primarily involves identifying and judging abnormalities or potential problems that occur in various equipment, systems, or components during operation. It covers multiple aspects, including hardware circuit faults, electrical system failures, mechanical component wear, and electronic equipment malfunctions, and is widely used in industrial manufacturing, transportation, aerospace, information systems, and consumer electronics. Traditional display fault prediction systems refer to technical solutions used to identify potential display anomalies in advance during the operation of display devices through specific methods. They focus on how to predict faults based on the operating status data of display devices to achieve preventative maintenance.

[0003] Traditional display fault prediction systems suffer from problems such as coarse-grained analysis of display data, insufficient control over temporal changes, limited data synchronization and identification of abnormal nodes, and frequent neglect of multi-stage change trends within the operating cycle. This makes it difficult to capture subtle signs of display anomalies in a timely manner. In practical applications, they are prone to missing initial risk signals under the interference of multiple factors, and the judgment of abnormal trends is lagging, failing to efficiently support accurate early warning of risk areas and affecting the reliability and prevention and control effectiveness of display device fault prediction. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a big data-based display fault prediction system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a big data-based display fault prediction system, the system comprising: The grayscale synchronization module is based on the display device. By analyzing the timing data of the grayscale optical acquisition instrument, it performs synchronization verification on the content of each frame and the driving signal, checks the time tags one by one, and filters out unaligned frames to obtain grayscale timing reconstruction data. Based on the grayscale time-series reconstruction data, the fluctuation recognition module performs a continuity analysis on the grayscale change trajectory of adjacent frames, judges the stability of the change direction of each segment, identifies the key time points that appear in the continuous change, and obtains a set of grayscale direction change nodes. The stage determination module, based on the set of grayscale direction change nodes and combined with the temporal fluctuation pattern of the display operation cycle, compares the grayscale change direction segment by segment for each stage, identifies the trend inflection point nodes, and obtains the stage inflection point associated node group. The trajectory comparison module, based on the stage inflection point associated node group, corresponds the brightness record of the optical sensor with the grayscale data at the same time point, creates a continuously changing trajectory, analyzes the morphological differences at the curve turning points, compares the morphological changes with key nodes point by point, filters consistent nodes, and obtains the offset turning consistency features. Based on the aforementioned offset turning point consistency characteristics, the trend warning module sorts out the order of node appearance throughout the entire cycle, ranks the gray-scale downward trend corresponding to the nodes, compares the continuity of the downward magnitude between nodes, evaluates the trend performance of risk areas, and obtains a displayed risk warning indicator.

[0006] The present invention is improved in that the grayscale time-series reconstruction data includes a set of synchronization frames, a time mapping table, and a valid data segment; the grayscale direction change node set includes a change node index, a fluctuation trend label, and distribution characteristic parameters; the stage inflection point associated node group includes an inflection point time series, a stage identifier, and an association relationship parameter; the offset turning point consistency feature includes a turning point consistency quantity, key node information, and a trajectory difference factor; and the display risk warning indicator includes a risk level, a trend partition, and a warning node.

[0007] The present invention is improved in that the grayscale synchronization module includes: The data stream receiving submodule is based on the display device and analyzes the time-series data stream acquired from the display device and grayscale optical acquisition instrument. By checking the frame number and acquisition order, it determines whether there are discontinuous or repetitive problems in the frame sequence, identifies abnormal frames, and obtains frame integrity characteristics. The time tag comparison submodule, based on the frame integrity feature, calls the data frame sequence without anomalies, extracts the time tag of each frame content, compares it with the time sequence of the synchronization signal output by the driver chip, compares the time correspondence between each frame tag and the synchronization signal, screens out data frames with mismatched time pairing status, and obtains the synchronization offset frame index. The temporal sequence reshaping submodule, based on the synchronization offset frame index, arranges the corresponding grayscale frame content in chronological order according to the synchronization time and frame number order, removes frames with time conflicts and order abnormalities, integrates and summarizes continuous frame data with consistent sequence structure, and obtains grayscale temporal reconstruction data.

[0008] The present invention is improved in that the fluctuation recognition module includes: The grayscale trajectory extraction submodule analyzes the grayscale changes in adjacent frames based on the grayscale temporal reconstruction data, calculates the directional information of grayscale differences between each pair of frames, determines the continuation of the differences in the continuous frame interval, identifies change segments with consistent directions, and integrates the continuous change segments in chronological order to obtain a set of grayscale change trajectories. The direction of change determination submodule determines the directional trend of each trajectory sequence based on the grayscale change trajectory set, analyzes the duration of the direction within the trajectory in time, compares the duration ratio of the direction change of each trajectory segment, identifies data segments whose direction maintenance time exceeds a preset duration threshold, and obtains grayscale direction stable segments. The time node identification submodule, based on the grayscale direction stable segment, compares the direction changes of adjacent trajectory segments, determines the time position of the direction switch, calculates the start and end time of the same direction change segment, identifies the time node that appears when the direction changes, and obtains the grayscale direction change node set.

[0009] The present invention is improved in that the stage determination module includes: The trend segmentation submodule calculates the gray-scale direction change identifier between adjacent nodes based on the gray-scale direction change node set. According to the fluctuation trend label and distribution characteristic parameters of the nodes, it performs continuity detection on the gray-scale direction change identifier, determines the continuous interval of direction change, and obtains the stage gray-scale trend sequence. The inflection point node identification submodule identifies interval nodes whose direction changes from positive to negative or from negative to positive based on the grayscale trend sequence of the stage, extracts the time label sequence of each node, calculates the time interval, obtains the trend turning point amplitude, and filters out nodes whose trend turning point amplitude is greater than the change gradient benchmark amount to obtain the trend inflection point index set. The overlapping node filtering submodule compares the time tags of each node with those of the preceding abnormal nodes based on the trend inflection point index set, determines the time correspondence between the two groups of nodes, identifies the node combinations with time differences less than a preset time difference threshold, and obtains the stage inflection point associated node group.

[0010] The present invention is improved in that the trajectory comparison module includes: The trajectory drawing submodule compares the brightness change sequence with the corresponding inflection point time based on the stage inflection point associated node group, calculates the grayscale acquisition results at the same time point, determines whether the time labels match one by one, filters the continuously matching grayscale brightness data, and obtains the grayscale brightness trajectory sequence. The transition analysis submodule, based on the grayscale brightness trajectory sequence, determines the brightness and grayscale change trend of consecutive frames, analyzes the time when the change direction reverses, compares the time distribution of each transition point with the inflection point, obtains the grayscale brightness transition difference, and filters nodes whose difference is within the range and coincides with the inflection point time to obtain the set of overlapping morphological change nodes. The node filtering submodule analyzes the grayscale and brightness change trajectory of each node in the preceding and following multiple frames based on the set of overlapping morphological change nodes, determines whether the change converges at the node, compares the differences between nodes, identifies nodes with consistent temporal and trajectory change characteristics, and obtains the offset turning consistency feature.

[0011] The present invention is improved in that the trend early warning module includes: The grayscale trend processing submodule analyzes the temporal distribution of each feature node based on the offset turning consistency feature, determines the order of nodes within the running cycle, compares the direction of grayscale change trends between adjacent nodes, identifies the node sequence with continuously decreasing grayscale, and obtains the grayscale decreasing trend sequence. The trend stability judgment submodule, based on the grayscale downward trend sequence, judges the continuity of grayscale changes in each segment, analyzes the consistency of grayscale trend changes in each segment within the time interval, compares the trend stability of continuously decreasing segments, and filters segments where grayscale trends remain stable to obtain stable trend segments. The risk segment summary submodule adjusts the time boundaries of each segment based on the stable trend segment, analyzes the performance between grayscale trend segments, classifies time segments with similar trend changes, judges the risk characteristics of each segment and organizes them into an index to obtain the displayed risk warning indicators.

[0012] The present invention is improved in that the grayscale content refers to the set of all grayscale values ​​corresponding to a certain moment of the display device, which is collected in a single frame, and the fluctuation pattern refers to the fluctuation pattern or periodic and staged characteristics of the grayscale data as time goes by during the entire working cycle of the display device.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by verifying the synchronization and sequence integrity of the collected data, the information is accurately organized at multiple time-series levels. Based on the continuous analysis of the change trajectory and the extraction of feature nodes, the key moments in the display process are identified, the differences in the turning points of each stage are focused on, the correlation trends between brightness and grayscale are compared in multiple dimensions, and the dynamics of the risk zone are comprehensively evaluated by combining the trend continuity between nodes. This improves the ability to capture abnormal signs of display, enables early perception and timely warning, and effectively enhances the stability of display device fault prediction and risk prevention capabilities. Attached Figure Description

[0014] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the grayscale synchronization module in this invention; Figure 3 This is a flowchart of the fluctuation recognition module in this invention; Figure 4 This is a flowchart of the stage determination module in this invention; Figure 5 This is a flowchart of the trajectory comparison module in this invention; Figure 6 This is a flowchart of the trend warning module in this invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0017] All user-related information involved in this invention (including but not limited to biometric information, identity verification information, behavioral data, device information, and other data that can be used for identity verification and personalized services) is collected and processed with the user's full knowledge and voluntary consent. The collection, storage, and use of all information strictly comply with applicable national and regional laws and regulations, and meet relevant data protection standards and policy requirements. The use of data is limited to purposes necessary for providing the technical services of this invention, and reasonable technical and management measures will be taken to ensure the security and confidentiality of users' personal information in terms of information protection and privacy.

[0018] Example: Please see Figure 1 This invention provides a technical solution: a big data-based display fault prediction system comprising: The grayscale synchronization module is based on the display device. It analyzes the grayscale timing data output by the grayscale optical acquisition instrument. By checking the integrity of the data, it compares the grayscale content of each frame with the synchronization signal of the driver chip, compares the time tags of each frame one by one, removes misaligned frames, and reorganizes the sequence to obtain grayscale timing reconstruction data. The fluctuation recognition module analyzes the continuous trajectory of grayscale changes in adjacent frames based on grayscale time-series reconstruction data, judges the stability of the change direction segment by segment, identifies the time point of change by comparing the extension trend of continuous change segments, identifies the relationship between the time points in the sequence, and records the concentrated time points of each change direction as feature nodes to obtain the grayscale direction change node set. The stage determination module analyzes the temporal fluctuation pattern of the display device's operating cycle based on the set of grayscale direction change nodes, compares the grayscale change direction of each stage segment by segment, determines the nodes where the trend turns into an inflection point, identifies the parts where the time overlaps with the previous abnormal nodes, organizes the overlapping nodes as the basis for stage determination, and obtains the stage inflection point associated node group. The trajectory comparison module is based on the stage inflection point associated node group, compares the brightness changes recorded by the optical sensor, retrieves the grayscale records at the same time point, draws the continuous trajectory corresponding to the brightness and grayscale, analyzes the curve turning position, judges whether the morphological differences at the turning point are concentrated in the key node, compares the morphological changes with the key nodes point by point, filters the consistent nodes, and obtains the offset turning consistency characteristics. The trend warning module analyzes the order of node appearance throughout the entire operation cycle based on the consistency characteristics of offset turning points. It arranges the gray-scale downward trend corresponding to the node in the time direction, judges the stability of trend changes by comparing the continuity of the downward amplitude between nodes, and uniformly organizes the trend performance of risk areas to obtain the displayed risk warning indicators.

[0019] The grayscale time-series reconstruction data includes a set of synchronization frames, a time mapping table, and valid data segments. The grayscale direction change node set includes change node indexes, fluctuation trend labels, and distribution characteristic parameters. The stage inflection point associated node group includes inflection point time series, stage identifiers, and association parameters. The offset turning point consistency characteristics include turning point consistency quantity, key node information, and trajectory difference factors. The risk warning indicators include risk level, trend partitions, and warning nodes.

[0020] In the grayscale synchronization module, grayscale timing data refers to the grayscale output data stream with a clear time sequence label acquired by the grayscale optical acquisition instrument during the continuous operation of the display device (such as LCD, OLED, etc.), reflecting the grayscale (brightness) state changes of screen pixels or areas at each moment; grayscale content refers to the set of all grayscale values ​​acquired in a single frame corresponding to a certain moment of the display device, reflecting the brightness or grayscale distribution of the display area at that moment; the driver chip refers to the core electronic chip responsible for the working state of the display device (such as refresh, grayscale control), and its output synchronization signal is used to calibrate the physical acquisition time of each frame of grayscale data to achieve alignment between data sources; the time label refers to the acquisition or occurrence time identifier assigned to each frame of grayscale data, usually generated by the drive signal or system clock, used to ensure strict timing consistency of different data streams; the reordering sequence refers to reordering the frame data that was rejected due to anomalies, frame loss, or misalignment according to the time label to form a continuous, complete, and time-misaligned grayscale change sequence, providing high-quality input for subsequent analysis.

[0021] In the fluctuation recognition module, the continuous trajectory of grayscale changes refers to the sequence formed by arranging the grayscale values ​​of all adjacent frames in chronological order, reflecting the path or trend curve of grayscale changes over time; the stability of the direction of change refers to judging whether the grayscale value continuously rises, falls, or fluctuates back and forth over a period of time, reflecting whether the grayscale trend of this segment is singular, regular, or has abnormal disturbances; the extension trend of continuous change segments refers to observing whether the direction of change of grayscale data remains consistent over a continuous period of time (such as continuously darkening or brightening), and inferring its subsequent development direction to help identify abnormal inflection points; the time point of change refers to the position in the grayscale sequence where there is a significant rise, fall, or abrupt change, and the time point usually indicates that the device may experience performance degradation or abnormality; the sequential relationship refers to focusing on the order and interval of the above-mentioned time points of change on the time axis, providing a basis for subsequent fluctuation segmentation and node classification; the feature node refers to the important time point that has been screened and judged as representative and can reflect the grayscale fluctuation trend, used as an "anchor point" to characterize key stages or abnormal changes.

[0022] In the stage determination module, the temporal fluctuation pattern refers to the fluctuation pattern or periodic and staged characteristics of grayscale data over time in the entire working cycle of the display device; the grayscale change direction refers to the overall change trend of grayscale data in an analysis stage (such as overall rise, fall, or reversal), providing a judgment standard for distinguishing different working states or degradation stages; the inflection point refers to the key time point in the sequence where one change trend changes to another (such as from rise to fall), which is an important indicator for judging state switching or accelerated degradation; the overlapping part refers to comparing the newly detected inflection point with the abnormal node identified in the previous step in time, finding the node whose time is consistent or close to the other, which is regarded as the "key attention moment" under multiple criteria; the stage judgment basis refers to the time set composed of overlapping nodes, which serves as a reference standard for subsequent stage division and key node tracking.

[0023] In the trajectory comparison module, the curve turning point refers to the location in the brightness-grayscale joint change trajectory where the curve shows a trend change such as turning back or bending, which usually corresponds to the moment when the performance of the display device changes significantly; morphological difference refers to the significant change in the correspondence between brightness and grayscale at the turning point, which reflects the microscopic differences or abnormal signs of the internal characteristics of the display device; key node refers to the time point that has been screened in the previous stage and is considered to be significantly representative, and further comparison is needed to see if the actual brightness-grayscale trajectory performance matches; screening consistent nodes refers to selecting those nodes that overlap in time and match in phenomenon by comparing the morphological change position and key node, which are regarded as the signal points that "truly" reflect the risk of degradation.

[0024] In the trend warning module, the grayscale decline trend is determined during the risk analysis phase by arranging all key node grayscale values ​​by time to form the overall trajectory of the equipment's grayscale output decreasing over time. The continuity of the decline range focuses on whether the grayscale changes between key nodes are continuous and smooth, or exhibit jumps or discontinuities, which helps to determine whether the risk trend is gradual or sudden. The trend performance of the risk area refers to the time period with the most abnormal characteristics identified through the above analysis in the overall grayscale change curve, and the area corresponds to the potential or existing fault risk in the equipment operation.

[0025] Please see Figure 2 The grayscale synchronization module includes: The data stream receiving submodule is based on the display device and analyzes the time-series data stream acquired from the display device and grayscale optical acquisition instrument. By checking the frame number and acquisition order, it determines whether there are discontinuous or repetitive problems in the frame sequence, identifies abnormal frames, and obtains frame integrity characteristics. First, the refresh rate, scan cycle, and signal interface timing parameters of the display device are extracted. For example, for an LCD screen with a refresh rate of 60Hz, the single frame refresh time is approximately 16.67 milliseconds. Based on this, the raw timing data stream acquired by the grayscale optical acquisition device is called. Each frame in this data stream contains a frame number, acquisition time information, and the corresponding grayscale image content. When processing the data, the receiving module arranges all frame data in ascending order of acquisition time, sequentially reads the difference in the number between every two frames, and determines whether the values ​​are consecutive. For example, from number 208 to 209 is a normal sequence, but if it jumps from number 208 to 211, then frames 209 and 210 are missing, indicating that two frames are missing and are recorded as discontinuous frames. Then, it is determined whether the number is repeated. For example, if the number appears in three consecutive frames... In cases 221, 221, and 222, it is determined that number 221 is repeated once. For all data frames, the number and percentage of the above discontinuous or repeated frames are counted. If the total number of frames collected in 10 seconds is 600, and 7 discontinuous frames and 5 repeated frames are found, then there are a total of 12 abnormal frames, accounting for 2% of the total. These 12 frames are recorded as abnormal frames and an abnormal frame index table is established. The table records the number of each abnormal frame, the abnormal type (such as "number jump" or "number repetition"), the time interval between the preceding and following frames, etc. The remaining frame data is further segmented. Based on the continuity of the number and the stability of the collection time interval, it is divided into several normal segments. Each segment is assigned a number, and the start and end frames and the corresponding time range are recorded so that the sorted normal frame segments can be used for time series analysis to obtain frame integrity characteristics and complete data processing.

[0026] The time tag comparison submodule, based on frame integrity features, calls the data frame sequence without anomalies, extracts the time tag of each frame content, compares it with the time sequence of the synchronization signal output by the driver chip, compares the time correspondence between each frame tag and the synchronization signal, screens out data frames with mismatched time pairing status, and obtains the synchronization offset frame index. The module extracts time stamp data from all frames within the normal frame segment. Each time stamp identifies the absolute time of frame acquisition in milliseconds. For example, the data acquisition time for frame number 150 is 2523.71 milliseconds. Simultaneously, it calls the synchronization signal sequence output by the driver chip. Each rising edge in this sequence corresponds to the theoretical refresh time of one frame of data. For example, the 150th signal is 2523.68 milliseconds. The module compares the two time points and determines whether the difference is within the allowable range. This range is set based on a combination of the acquisition device and display response delay. For example, the maximum allowable error is set to ±2 milliseconds. If the time difference between a frame and its corresponding synchronization signal is 2.6 milliseconds, it exceeds the threshold and is considered mismatched with the synchronization signal. If the time difference is 1.2 milliseconds, it is considered a successful match. This process is repeated for all frames in the normal frame segment, and the numbers of all mismatched frames are recorded to form a synchronization offset frame index. For example, if there are 400 frames in the normal frame segment and 23 frames are found to have a time difference exceeding the set threshold, these 23 frames are identified as synchronization offset frames. The frame number, actual timestamp, corresponding synchronization signal time value, difference between the two, and whether it exceeds the limit are written as entries into the index table for use in subsequent timing resetting steps.

[0027] The temporal sequence reshaping submodule is based on the synchronization offset frame index. According to the synchronization time and frame number order, the corresponding grayscale frame content is arranged in chronological order, frames with time conflicts and order abnormalities are removed, and continuous frame data with consistent sequence structure are integrated and summarized to obtain grayscale temporal reconstruction data. Based on the generated synchronous offset frame index, these offset frames are removed from all frame data, retaining the remaining normally synchronized frame sequence. Then, the remaining frame data is reordered using the time label as the key field to ensure logical consistency in time sequence. For each pair of consecutive frames, the difference in acquisition time between the two frames is calculated, and it is checked whether the difference falls within a predefined time window. For example, if the display device refresh cycle is 16.67 milliseconds, the acceptable minimum interval between frames is set to 8 milliseconds and the maximum interval to 25 milliseconds. If the acquisition time interval between a frame and the next frame is 5 milliseconds, it indicates that the acquisition is too dense and may be a mis-acquisition, so it is removed. If the interval is 30 milliseconds, it indicates that the frame is lost and is removed. Through this time difference check, after removing all frames with abnormal acquisition intervals, the remaining frame data is numbered and reordered so that the frame numbers are consecutive integers and the acquisition time strictly increases. For example, after sorting, 372 frames are obtained and renumbered from 0 to 371. The time label and grayscale image content corresponding to each frame are recorded for use by the grayscale trajectory recognition module in the next stage, forming a grayscale time-series reconstruction data structure with stable numbering order and time consistency, ensuring the accuracy of the basic data for analysis.

[0028] Please see Figure 3 The fluctuation recognition module includes: The grayscale trajectory extraction submodule analyzes grayscale changes in adjacent frames based on grayscale temporal reconstruction data, calculates the directional information of grayscale differences between each pair of frames, determines the continuation of differences in continuous frame intervals, identifies change segments with consistent directions, and integrates continuous change segments in chronological order to obtain a set of grayscale change trajectories. The process reads the frame number and grayscale image content of each frame in the reconstructed sequence, extracts the average grayscale value of all pixels in the image matrix frame by frame as a representative value, and constructs a time-series grayscale data linked list in sequence. During this process, data processing is performed. Then, for each pair of adjacent frames, a difference operation is performed. If the grayscale value of the later frame is greater than that of the previous frame, it's considered an upward direction; if it's less, it's considered a downward direction; if they are equal, it's marked as a stationary state. For example, if frame number 120 corresponds to a grayscale value of 142, and frame 121 corresponds to 146, then the direction is marked as upward. This process continues to iterate through all inter-frame differences, recording the direction information in frame order to form a direction vector. Finally, the process reads the values ​​that continuously maintain the same direction from the direction vector. For each segment, a consistent direction of change is constructed. The position of the point where the direction change occurs is taken as the end point of the segment. If the continuous grayscale direction between frame numbers 150 and 158 is downward, it is integrated into a downward segment. The start number, end number, and corresponding time label are then combined to form the basic record item of the grayscale change segment. All change segments are arranged in ascending order of time label to obtain a complete set of grayscale change trajectories. During this process, the minimum length of a single change segment needs to be determined. If the number of consecutive consistent direction frames is less than 3, the segment is discarded to avoid misjudgment caused by jitter. The start and end times, grayscale direction, change value range, number of participating frames, and other information of each segment are output in the form of a change segment sequence to form a set of grayscale change trajectories.

[0029] The direction of change judgment submodule is based on the grayscale change trajectory set to determine the directional trend of each trajectory sequence, analyze the duration of the direction within the trajectory in time, compare the duration of the direction change of each trajectory segment, identify the data segment with the longer direction maintenance time, and obtain the grayscale direction stable segment. The system sequentially reads the direction information and time interval length recorded in each trajectory segment, calculates the duration based on the start and end times of the segment, and then summarizes and compares the durations of all segments in the trajectory set. Each segment's directional trend is identified and grouped; for example, all upward trajectory segments are grouped into an ascending group, and downward trajectory segments are grouped into a descending group. The number of trajectory segments and the total duration within each group are counted, and the proportion of each trajectory segment in the group is calculated. Trajectory segments with a proportion greater than 20% are selected as candidates for directionally stable segments. Data processing and further judgment are performed on the candidate segments to determine if there are any interruptions. For example, if there are missing frames or the proportion of interrupted frames exceeds 20% between the start and end times, the segment is excluded. If the duration is continuous and the frame sequence is complete, it is marked as a directionally stable segment. For example, if a segment numbered 300 to 320 continuously descends for 340 milliseconds, accounting for 26% of the total duration of all descending segments, and the frame completeness rate is 98%, it is retained as a descending directionally stable segment. Simultaneously, the direction, time interval, number of frames, and total grayscale change are recorded for each stable segment, and a stable segment index table is constructed for subsequent modules to call, thus obtaining grayscale directionally stable segments.

[0030] The time node recognition submodule is based on the gray-scale direction stable segment, compares the direction changes of adjacent trajectory segments, determines the time position of the direction switch, calculates the start and end time of the same direction change segment, identifies the time node that appears when the direction changes, and obtains the gray-scale direction change node set. The system iterates through the directional information between two adjacent stable segments, extracting the end time of the previous segment and the start time of the next segment. If the direction changes from upward to downward or from downward to upward, the directional switching behavior is marked at the connection point between the two segments. The time label of the connection point is extracted as the directional switching time node. At the same time, the original frame number and grayscale value corresponding to the switching point are checked back, and the switching node information is processed and verified. Then, based on the start and end time intervals recorded within the stable segment, the duration and number of frames within the segment are calculated respectively. It is confirmed whether there is a sudden change in the grayscale change between the location of the directional conversion and the segments before and after. For example, if the grayscale value at the end of the previous segment is 186 and the starting value of the next segment is 176, the grayscale value drops by 10, and the directional change changes from upward to downward, with an inter-frame time of 20 milliseconds, the conditions for the switching node are met. Then, the time of this node is recorded as the directional change point. A grayscale direction change node set is constructed, and the time, direction of the segments before and after, grayscale value difference, frame number, and start and end information of adjacent segments are recorded for each node. This information is used for subsequent stage identification and node comparison analysis.

[0031] Please see Figure 4The stage determination module includes: The trend segmentation submodule is based on the set of gray-scale direction change nodes. It calculates the gray-scale direction change identifier between adjacent nodes. Based on the fluctuation trend label and distribution characteristic parameters of the nodes, it performs continuity detection on the gray-scale direction change identifier, determines the continuous interval of direction change, and obtains the stage gray-scale trend sequence. The grayscale direction change identifiers of all adjacent nodes are extracted from the node set. The acquisition process involves reading the sign value of the grayscale change direction for each pair of adjacent nodes. If the preceding node is in an upward direction and the following node is in a downward direction, it is marked as a reversal; if the directions are the same, it is marked as continuous. All direction change identifiers are then arranged in chronological order to form a preliminary direction change sequence. Based on this, the fluctuation trend label attached to each node is called. This label records the trend stage of the node in the grayscale change, divided into three categories: rapid change segment, stable segment, and disturbance segment. The label value is derived from the continuity of the grayscale difference in the preceding grayscale data segment and the number of direction changes. If a node switches direction no more than once in the first 10 frames and the total grayscale difference is greater than 15 grayscale values, it is classified as a rapid change segment. If the number of switches is less than or equal to 2 and the difference is less than 5 grayscale values, it is marked as a stable segment. This process is combined with the label of each node. The distribution feature parameters of the current node are then extracted. These parameters represent the positional distribution density of the current node in the entire node sequence, calculated using the number of nodes per unit time. For example, if 3 nodes are detected within 1 second, the density is 3. The density continuity judgment threshold is set to 2 nodes per second. If the density exceeds this threshold within a continuous time period, it is considered a continuously changing region. By comparing the direction change indicators and trend labels of adjacent node pairs segment by segment, continuous intervals are grouped and summarized. If 3 or more consecutive pairs of node indicators are in the same direction and the labels are consistent, the segment is considered a continuous direction change segment. The starting and ending node numbers and time range of the segment are further statistically analyzed. All intervals that meet the continuity judgment conditions are integrated to form a stage grayscale trend sequence. For example, between frames 50 and 70, there are 5 nodes that are all in an upward direction, the fluctuation label is a stable segment, and the node density is 2.5. This interval meets the continuity judgment and is identified as a complete trend sequence segment.

[0032] The inflection point node identification submodule, based on the stage grayscale trend sequence, identifies interval nodes where the direction changes from positive to negative or from negative to positive, extracts the time label sequence of each node, and calculates the time interval using the formula: ; Obtain the magnitude of trend reversal By filtering out nodes whose trend inflection amplitude is greater than the baseline change gradient, a set of trend inflection point indices is obtained. Indicates the first The magnitude of a trend reversal, This indicates the number of grayscale direction segments within each cycle. Indicates the first Within the first cycle of change, the first Gray-scale directional encoding of segments, Indicates the first Within the first cycle of change, the first Gray-scale directional encoding of segments, Indicates the first A time tag for a trend inflection point Indicates the first A time tag for a trend inflection point; The magnitude of a trend reversal refers to the magnitude of the change between two adjacent trend inflection points ( and Between these two points, the overall intensity of the change in the direction of grayscale change, combined with the directional change in the grayscale change trend during this period and the duration of this trend change, quantitatively characterizes the "severity" or "amount of change" when each trend makes a significant turn. It is an important metric used to quantitatively describe the "degree of change" of the data trend at a stage of turning point. The larger the value, the more obvious and longer the grayscale change trend switch before and after the turning point is, and the more representative and valuable it is for monitoring. Analyze the grayscale direction change between any two consecutive trend segments, identify the connecting nodes of adjacent trend segments where the grayscale direction changes from positive to negative or from negative to positive, extract the time labels corresponding to the two trend segments at the connection point, and record them as the end time of the previous segment. With the start of the next segment Construct a trend switching timeline and number and organize its sequence, then call the current period. Compared with the previous cycle The grayscale directional encoded data is divided into the same number of segments within each trend segment. The direction segments are calculated, and the grayscale change direction of each segment is encoded as follows: , or The upward direction is , remain unchanged The direction of descent is For example, let's say In the current cycle, the first Segment direction encoding is: ; In the previous period Segment direction encoding is: ; Then, the squared difference of the directional coding for each segment is: ; Next, we perform the summation: ; Taking the average and square root, we obtain the term for the intensity of the directional change: ; Then calculate the time stamp difference, assuming: , ; The time span is: ; Substitute into the formula: ; The original trend reversal magnitude was calculated as follows: ; To ensure comparability of trend amplitude values ​​across different periods, a minimum-maximum normalization method is used. To perform normalization, the minimum and maximum values ​​in the sample data are defined as follows: , ; The normalized value is then calculated as follows: ; Based on the preset trend amplitude segmentation standard, the normalization result is divided into the following three judgment intervals: when When this occurs, it is determined to be a weak trend change segment, indicating that the grayscale direction changes are small and the direction switching is slow; when When the time frame is reached, it is determined to be a medium-term trend change segment, indicating that there is some directional shift but it is not yet significant. when When the grayscale direction changes abruptly over a longer period of time, it is considered a significant trend change segment, indicating a high intensity of trend reversal.

[0033] therefore, satisfy It is classified as a segment of significant trend change. This judgment result is consistent with the threshold condition for the inflection point amplitude in the formula calculation steps, indicating that the trend connection node constitutes a key turning point in the gray-scale fluctuation process and should be included in the trend inflection point index set.

[0034] The overlapping node filtering submodule is based on the trend inflection point index set. It compares the time labels of each node with the preceding abnormal node, determines the time correspondence between the two groups of nodes, identifies the node combination with small time difference, and obtains the stage inflection point associated node group. The time tag of each inflection point node in the set is extracted as a comparison benchmark. The time tags of all abnormal nodes are extracted from the abnormal node set identified in the previous stage. Each time value in both node sets is read sequentially and compared pairwise. The time difference for each pair of time values ​​is calculated. The reference range for time difference judgment is set between 0.016 seconds and 0.050 seconds, which corresponds to a frame time interval of 0.0167 seconds on a conventional display device. If the time difference is less than 0.050 seconds, the two nodes are considered to be within the same logical time frame. This judgment method is used to match and compare all inflection point nodes with abnormal nodes, recording each pair of node combinations with a time difference less than a threshold. To eliminate occasional interference, a combination repetition judgment benchmark is further set. If an abnormal node is combined with three or more inflection point nodes, then... The abnormal node is considered to have high-frequency overlap characteristics and is therefore filtered out, retaining only one-to-one or at most one-to-many matching relationships. After completing the comparison of all combinations, the frame number and time label in the original data are recorded for each matching result, and a node association identifier field is established to indicate whether the combination belongs to valid overlap. All node pairs with time differences within a specified range and repetition within a set range are combined into a stage inflection point associated node group. For example, in practical applications, if the abnormal node time is 12.016 seconds and the inflection point node time is 12.030 seconds, the time difference is 0.014 seconds, which is less than the threshold, and the combination is valid. If the abnormal node can also be combined with the nodes at 12.032 seconds and 12.034 seconds, only the first combination with 12.030 seconds is retained, and the rest are excluded. After the filtering is completed, the stage inflection point associated node group is generated.

[0035] Please see Figure 5 The trajectory comparison module includes: The trajectory drawing submodule is based on the stage inflection point associated node group, compares the brightness change sequence with the inflection point time, calculates the gray level acquisition results at the same time point, determines whether the time labels match one by one, filters out the continuously matching gray level brightness data, and obtains the gray level brightness trajectory sequence. The time label corresponding to each node in the node group is extracted as a comparison benchmark. Then, the recorded brightness change sequence is called, and the matching brightness data point is searched one by one against the time label of each node. If there is a completely consistent sampling time point in the time dimension, the brightness value is recorded as the brightness data value corresponding to the current node. Further, the grayscale value of the corresponding time point is extracted from the synchronized grayscale time series data to construct the "time-grayscale-brightness" triplet structure data corresponding to each node. Then, the continuity detection on the time axis is performed on all nodes. The detection logic is to determine whether the time interval between two consecutive nodes is within the allowable error range. The error tolerance threshold is set to ±0.016 seconds, that is, the time difference between two consecutive nodes is within this range. If the sampling is continuous, it is considered continuous; otherwise, the interrupted frames are removed and the continuous trajectory segment is terminated. Then, the subsequence containing only continuous matching nodes is selected. The subsequence is arranged in chronological order to construct a grayscale brightness trajectory sequence. During the drawing process, time is used as the horizontal axis, and grayscale values ​​and brightness values ​​are used as two sets of vertical axis sequences to draw curves. The curve trend at each node is compared and observed to see if there is a turning structure. In the example, if the grayscale of the node at 12.016 seconds is 140 and the brightness is 145, and the grayscale of the node at 12.033 seconds drops to 135 and the brightness drops to 139, then the curve at this position will bend downwards in the drawing, which is judged as a point of change. All continuous time nodes and their corresponding grayscale and brightness values ​​are integrated and output as a grayscale brightness trajectory sequence.

[0036] The transition analysis submodule, based on the grayscale brightness trajectory sequence, determines the brightness and grayscale change trends of consecutive frames, analyzes the time when the change direction reverses, compares the time distribution of each transition point with the inflection point, and uses the following formula: ; The grayscale brightness transition difference is obtained, and nodes whose difference falls within the range and coincides with the inflection point time are selected to obtain the set of nodes with overlapping morphological changes. Indicates the first The difference in grayscale brightness transition at each transition node Indicates the first The difference in brightness before and after each turning point Indicates the first The difference in grayscale before and after each turning point Indicates the first The time stamps corresponding to each turning point Indicates the relationship with the first The time tag of the inflection point closest to the turning point. Grayscale brightness transition difference is used to measure the joint intensity of brightness and grayscale changes at a given transition point. Combined with a quantitative indicator of the time interval between that point and a reference inflection point, grayscale brightness transition difference reflects the combined change in brightness and grayscale at each detected transition point in the grayscale-brightness trajectory (calculated using Euclidean distance). This change amplitude is normalized to the time distance between the transition point and its nearest inflection point. It describes whether the display device experiences significant changes in both brightness and grayscale at a given point in time, and whether such drastic changes occur near a critical inflection point. A higher value indicates a more concentrated and closer concentration of drastic changes in brightness and grayscale near the node, and closer to a critical inflection point. Determine the relative direction of change between grayscale and brightness in each group of consecutive frames, and calculate the grayscale difference of the current frame. The brightness difference is If the product of two consecutive differences is less than zero, it indicates a direction reversal and is marked as a candidate turning point. The time stamp of that point is also recorded. Further retrieval of inflection point associated node groups and comparison With all inflection point time tags The difference between the two values ​​is taken as the reference inflection point time for that node. To address the inconsistency in the dimensions of grayscale, brightness, and time, the following measures were taken: , , Range normalization is used, and the normalization formula is as follows: The original range of grayscale difference is set to . , Brightness difference set to , The time difference is set to , The data collected at a certain test node is as follows: , , ; The corresponding normalized values ​​are as follows: , , ; Substituting the normalized values ​​into the formula, we get: ; Take the data from the second node again. , , After normalization, they are as follows: , , Substituting into the formula, we get: ; This result indicates that the computational difference of the first node is... Falling into the set filter range Within this range, the node belongs to the "effective change concentration segment," indicating that it exhibits significant changes in both grayscale and brightness dimensions, and its occurrence time is close to the key inflection point, meeting the screening criteria of both difference intensity and temporal aggregation. The calculation result for the second node is... Falling within the interval Within this range, it falls under the "non-significant change segment," indicating that the fluctuation of the node in brightness or grayscale is insufficient, or its changes are not focused around the inflection point, and therefore it is removed; furthermore, for future expansion scenarios, if... Nodes exhibiting abnormal fluctuations are categorized as "abnormal fluctuation segments." While these nodes are high in intensity, they represent noise disturbances or irregular fluctuations and can be further marked or removed based on application requirements. Therefore, The specific range of values ​​can be determined in three categories: when At that time, it is not included in the set of overlapping nodes; when At that time, it was identified as a node of overlapping morphological changes; when At that time, it is recorded as an abnormally drastic node but not directly included in the master node set.

[0037] The node filtering submodule analyzes the grayscale and brightness change trajectory of each node in the preceding and following frames based on the set of overlapping morphological change nodes, determines whether the change converges at the node, compares the differences between nodes, identifies nodes with consistent temporal and trajectory change characteristics, and obtains the offset turning consistency feature. For each node in the node set, the preceding and following frame data are extracted. At least three frames are read forward and three frames backward from the time position of each target node to obtain the grayscale change sequence and brightness change sequence before and after the current node, forming a local trajectory window. Then, the direction consistency of the grayscale and brightness data within this window is judged, and the change directions of grayscale and brightness are synchronously compared. When both grayscale and brightness show a change in direction at the node (e.g., both changing from increasing to decreasing or vice versa), the node is initially marked as a change aggregation node. Further, the change amplitude at this node is compared with the difference between the preceding and following frames. If the change amplitude exceeds a set threshold (grayscale change amplitude threshold set to ±5, brightness change amplitude threshold set to ±6), then the change amplitude at a node is considered a change aggregation node. The above steps mark the node as a strong change node. After performing the above judgment on all nodes, extract the set of all nodes that meet the requirements of consistent direction and amplitude exceeding the threshold. Then, compare the differences between the nodes within the set to determine whether the differences are focused on a specific time interval. If multiple nodes appear in a concentrated manner within a ±0.05 second time window and their grayscale brightness change trends are consistent, then the region is marked as a consistency feature region. From the above screening, identify nodes with concentrated time positions and similar trajectory change directions and amplitudes as offset turning point consistency features. For example, if node A's time is 10.200 seconds, the grayscale decreases from 150 to 142, and the brightness decreases from 152 to 144, with a grayscale difference of 8 and a brightness difference of 8, the direction is the same and the amplitude exceeds the threshold, it is judged as a consistency node and is grouped into the same feature set with similar nodes in adjacent time periods.

[0038] Please see Figure 6 The trend warning module includes: The grayscale trend consolidation submodule analyzes the temporal distribution of each feature node based on the offset turning point consistency feature, determines the order of nodes within the running cycle, compares the direction of grayscale change trends between adjacent nodes, identifies the node sequence with continuously decreasing grayscale, and obtains the grayscale decreasing trend sequence. Extract the time tags and corresponding grayscale values ​​of each key node, and arrange them in ascending order of time tags to establish a node time series table. Read the time difference and grayscale difference between each pair of adjacent nodes, and determine whether the grayscale value of the preceding and following nodes decreases. If the grayscale value of three or more consecutive nodes decreases continuously with each step change not less than 1 grayscale unit, it is marked as the start of a downward trend. For each new node added, continue to determine whether its grayscale value is lower than that of the previous node. If it meets the condition, it is included in the current downward sequence until the grayscale value stops decreasing, at which point the identification of that segment is terminated. Repeat this process to traverse all nodes. Point combination is used to identify all downward trend segments. For example, in the sequence of key nodes of a certain device, the grayscale values ​​are 189, 186, 183, 179, and 178 respectively, and the time interval is within 30 milliseconds. They all meet the requirement of continuous decline and constitute a downward trend sequence. For each downward sequence, the start and end time, start and end grayscale values, number of nodes, and average decline amplitude are recorded. If the average decline amplitude is greater than 1 unit per node and the total decline value is greater than 5 units, it is identified as a continuous downward trend segment of grayscale. Otherwise, the segment is excluded. All the downward sequences that meet the requirements are combined in sequence to form a grayscale downward trend sequence.

[0039] The trend stability judgment submodule is based on the gray-scale downward trend sequence to judge the continuity of gray-scale changes in each segment, analyze the consistency of gray-scale trend changes in each segment within the time interval, compare the trend stability of continuously decreasing segments, and screen segments where the gray-scale trend remains stable to obtain stable trend segments. The node list in each descending segment is read segment by segment. The time interval and grayscale decrease value of adjacent nodes in each segment are calculated to determine whether the continuity requirements are met. For time continuity, the sampling interval between adjacent nodes must not exceed 1.5 times the device refresh cycle. For grayscale continuity, the number of points where the descending change is reversed or unchanged must not exceed 20% of the total number of nodes in the segment. If a segment contains 10 nodes, and 3 of them have a decrease value of 0 or a positive value, it is considered a discontinuous segment and is not retained. If the continuity conditions are met, the segment is further calculated. The average descent rate, calculated by dividing the total grayscale decrease by the total duration, is used to determine the stability of a segment. If the volatility of this rate is less than a set stability threshold, such as 15% (meaning the difference between the maximum and minimum rate does not exceed 15% of the average rate), then the segment is marked as a stable descent segment. For example, if a segment's grayscale decreases from 202 to 190 over 180 milliseconds with 12 nodes and a descent rate between 0.8 and 1.1 units per 20 milliseconds, with a volatility of approximately 14%, then it is considered a stable trend segment. All the stable segments that pass the screening are then merged in chronological order to establish a set of stable trend segments.

[0040] The Risk Segment Summary Submodule is based on stable trend segments, adjusts the time boundaries of each segment, analyzes the performance between grayscale trend segments, classifies time segments with similar trend changes, judges the risk characteristics of each segment and organizes them into an index to obtain displayed risk warning indicators. The start and end times and grayscale change values ​​of each segment are read. Segment merging is performed based on their adjacency. If the interval between two adjacent segments is less than 50 milliseconds, and they are in the same direction with continuous grayscale changes, the two segments are merged, and the time boundaries are adjusted uniformly. After merging, the new start and end times are recorded. Then, the descent rate, average descent amplitude, and duration within each segment are analyzed sequentially to determine if they meet the risk characteristic judgment criteria. For example, if the total grayscale descent value is greater than 12 units, the duration exceeds 300 milliseconds, and the average descent rate is not less than 1 unit every 25 milliseconds, then… The segment is identified as having risk characteristics. All segments that meet the risk criteria are assigned a risk rating, such as Level 3 (mild), Level 2 (moderate), and Level 1 (severe). The rating is determined based on a combination of the rate and magnitude of the decrease. For example, a rate less than 0.5 is Level 3, 0.5-1.2 is Level 2, and greater than 1.2 is Level 1. A grayscale decrease of more than 20 units is directly classified as Level 1. Finally, the number, start and end time, grayscale change value, duration, and risk level of all identified risk segments are compiled into a risk index table, which is output as a risk warning indicator.

[0041] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A display fault prediction system based on big data, characterized in that, The system includes: The grayscale synchronization module is based on the display device. By analyzing the timing data of the grayscale optical acquisition instrument, it performs synchronization verification on the content of each frame and the driving signal, checks the time tags one by one, and filters out unaligned frames to obtain grayscale timing reconstruction data. Based on the grayscale time-series reconstruction data, the fluctuation recognition module performs a continuous analysis on the grayscale change trajectory of adjacent frames, judges the stability of the change direction of each segment, identifies the key time points that appear in the continuous change, and obtains a set of grayscale direction change nodes. The stage determination module, based on the set of grayscale direction change nodes and combined with the temporal fluctuation pattern of the display operation cycle, compares the grayscale change direction segment by segment for each stage, identifies the trend inflection point nodes, and obtains the stage inflection point associated node group. The trajectory comparison module, based on the stage inflection point associated node group, corresponds the brightness record of the optical sensor with the grayscale data at the same time point, creates a continuously changing trajectory, analyzes the morphological differences at the curve turning points, compares the morphological changes with key nodes point by point, filters consistent nodes, and obtains the offset turning consistency features. Based on the aforementioned offset turning point consistency characteristics, the trend warning module sorts out the order of node appearance throughout the entire cycle, ranks the gray-scale downward trend corresponding to the nodes, compares the continuity of the downward magnitude between nodes, evaluates the trend performance of risk areas, and obtains a displayed risk warning indicator.

2. The big data-based display fault prediction system according to claim 1, characterized in that, The grayscale time-series reconstruction data includes a set of synchronization frames, a time mapping table, and a valid data segment. The grayscale direction change node set includes a change node index, a fluctuation trend label, and distribution characteristic parameters. The stage inflection point associated node group includes an inflection point time series, a stage identifier, and an association parameter. The offset turning point consistency feature includes a turning point consistency quantity, key node information, and a trajectory difference factor. The display risk warning indicator includes a risk level, a trend partition, and a warning node.

3. The big data-based display fault prediction system according to claim 1, characterized in that, The grayscale synchronization module includes: The data stream receiving submodule is based on the display device. It analyzes the time-series data streams collected from the display device and the grayscale optical acquisition instrument. By checking the frame number and acquisition order, it determines whether there are discontinuous or repetitive problems in the frame sequence, identifies abnormal frames, and obtains frame integrity characteristics. The time tag comparison submodule, based on the frame integrity feature, calls the data frame sequence without anomalies, extracts the time tag of each frame content, compares it with the time sequence of the synchronization signal output by the driver chip, compares the time correspondence between each frame tag and the synchronization signal, screens out data frames with mismatched time pairing status, and obtains the synchronization offset frame index. The temporal sequence reshaping submodule, based on the synchronization offset frame index, arranges the corresponding grayscale frame content in chronological order according to the synchronization time and frame number order, removes frames with time conflicts and order abnormalities, integrates and summarizes continuous frame data with consistent sequence structure, and obtains grayscale temporal reconstruction data.

4. The display fault prediction system based on big data according to claim 1, characterized in that, The fluctuation recognition module includes: The grayscale trajectory extraction submodule analyzes the grayscale changes in adjacent frames based on the grayscale temporal reconstruction data, calculates the directional information of grayscale differences between each pair of frames, determines the continuation of the differences in the continuous frame interval, identifies change segments with consistent directions, and integrates the continuous change segments in chronological order to obtain a set of grayscale change trajectories. The direction of change determination submodule determines the directional trend of each trajectory sequence based on the grayscale change trajectory set, analyzes the duration of the direction within the trajectory in time, compares the duration ratio of the direction change of each trajectory segment, identifies data segments whose direction maintenance time exceeds a preset duration threshold, and obtains grayscale direction stable segments. The time node identification submodule, based on the grayscale direction stable segment, compares the direction changes of adjacent trajectory segments, determines the time position of the direction switch, calculates the start and end time of the same direction change segment, identifies the time node that appears when the direction changes, and obtains the grayscale direction change node set.

5. The display fault prediction system based on big data according to claim 1, characterized in that, The stage determination module includes: The trend segmentation submodule calculates the gray-scale direction change identifier between adjacent nodes based on the gray-scale direction change node set. According to the fluctuation trend label and distribution characteristic parameters of the nodes, it performs continuity detection on the gray-scale direction change identifier, determines the continuous interval of direction change, and obtains the stage gray-scale trend sequence. The inflection point node identification submodule identifies interval nodes whose direction changes from positive to negative or from negative to positive based on the grayscale trend sequence of the stage, extracts the time label sequence of each node, calculates the time interval, obtains the trend turning point amplitude, and filters out nodes whose trend turning point amplitude is greater than the change gradient benchmark amount to obtain the trend inflection point index set. The overlapping node filtering submodule compares the time tags of each node with those of the preceding abnormal nodes based on the trend inflection point index set, determines the time correspondence between the two groups of nodes, identifies the node combinations with time differences less than a preset time difference threshold, and obtains the stage inflection point associated node group.

6. The display fault prediction system based on big data according to claim 1, characterized in that, The trajectory comparison module includes: The trajectory drawing submodule compares the brightness change sequence with the corresponding inflection point time based on the stage inflection point associated node group, calculates the grayscale acquisition results at the same time point, determines whether the time labels match one by one, filters the continuously matching grayscale brightness data, and obtains the grayscale brightness trajectory sequence. The transition analysis submodule, based on the grayscale brightness trajectory sequence, determines the brightness and grayscale change trend of consecutive frames, analyzes the time when the change direction reverses, compares the time distribution of each transition point with the inflection point, obtains the grayscale brightness transition difference, and filters nodes whose difference is within the range and coincides with the inflection point time to obtain the set of overlapping morphological change nodes. The node filtering submodule analyzes the grayscale and brightness change trajectory of each node in the preceding and following multiple frames based on the set of overlapping morphological change nodes, determines whether the change converges at the node, compares the differences between nodes, identifies nodes with consistent temporal and trajectory change characteristics, and obtains the offset turning consistency feature.

7. The display fault prediction system based on big data according to claim 1, characterized in that, The trend early warning module includes: The grayscale trend processing submodule analyzes the temporal distribution of each feature node based on the offset turning consistency feature, determines the order of nodes within the running cycle, compares the direction of grayscale change trends between adjacent nodes, identifies the node sequence with continuously decreasing grayscale, and obtains the grayscale decreasing trend sequence. The trend stability judgment submodule, based on the grayscale downward trend sequence, judges the continuity of grayscale changes in each segment, analyzes the consistency of grayscale trend changes in each segment within the time interval, compares the trend stability of continuously decreasing segments, and filters segments where grayscale trends remain stable to obtain stable trend segments. The risk segment summary submodule adjusts the time boundaries of each segment based on the stable trend segment, analyzes the performance between grayscale trend segments, classifies time segments with similar trend changes, judges the risk characteristics of each segment and organizes them into an index to obtain the displayed risk warning indicators.

8. The big data-based display fault prediction system according to claim 1, characterized in that, The grayscale content refers to the set of all grayscale values ​​collected in a single frame that correspond to a certain moment of the display device. The fluctuation pattern refers to the fluctuation mode or periodic and staged characteristics of the grayscale data over time during the entire working cycle of the display device.

Images