Dead pixel monitoring method and system for electroluminescent device

Abstract

The invention relates to the technical field of display device monitoring, and discloses a dead pixel monitoring method and system for an electroluminescent device, and the method comprises the steps: dividing the electroluminescent device into a plurality of to-be-monitored regions, and setting monitoring features according to a historical data set of each to-be-monitored region; acquiring real-time monitoring data of each to-be-monitored area according to the monitoring characteristics, generating a real-time correction influence coefficient of each real-time monitoring data, and correcting the real-time monitoring data to obtain real-time correction monitoring data; constructing a plurality of real-time monitoring data sequences according to the plurality of real-time correction monitoring data of the same to-be-monitored area, analyzing each real-time monitoring data sequence, and calculating a sequence coefficient according to an analysis result; and according to the sequence coefficient, setting a region state of the corresponding to-be-monitored region, and if the region state is a defective pixel state, determining defective pixel features and generating an operation and maintenance strategy, thereby improving defective pixel monitoring precision and monitoring efficiency.

Description

Technical Field

[0001] This application relates to the field of display device monitoring technology, and in particular to a method and system for detecting dead pixels in electroluminescent devices. Background Technology

[0002] Electroluminescent devices such as OLEDs, LEDs, and QLEDs are widely used in displays, lighting, and other fields due to their advantages such as self-illumination, fast response speed, and wide viewing angle. However, during the production and use of electroluminescent devices, defects such as dark spots, bright spots, flickering spots, and color deviation spots can easily appear due to material defects, process errors, and aging. These defects can seriously affect the performance and lifespan of electroluminescent devices.

[0003] In existing technologies, a single monitoring method is usually used to identify defective pixels. However, this method is easily affected by ambient light and device characteristics, resulting in inaccurate monitoring data, reduced defective pixel judgment accuracy, and inability to determine defective pixel characteristics. Consequently, it is impossible to formulate reasonable operation and maintenance strategies. Therefore, there is an urgent need for a defective pixel monitoring method and system for electroluminescent devices to improve the accuracy of defective pixel monitoring. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method and system for detecting defects in electroluminescent devices. The method involves dividing the electroluminescent device into multiple monitoring areas, setting monitoring characteristics, acquiring real-time monitoring data, generating a real-time correction influence coefficient to correct the real-time monitoring data, obtaining real-time corrected monitoring data, constructing several real-time monitoring data sequences, calculating sequence coefficients, and determining the area status. If defects are present, the defect characteristics are determined, and an operation and maintenance strategy is generated. By setting reasonable monitoring characteristics, the accuracy and efficiency of defect detection are improved.

[0005] In some embodiments of this application, a method for detecting defective pixels in an electroluminescent device is provided, including: The electroluminescent device is divided into multiple monitoring areas, and monitoring characteristics are set according to the historical data set of each monitoring area. Real-time monitoring data for each area to be monitored is obtained according to the monitoring characteristics, a real-time correction influence coefficient for each real-time monitoring data is generated, and the real-time monitoring data is corrected to obtain real-time corrected monitoring data. Several real-time monitoring data sequences are constructed based on multiple real-time corrected monitoring data of the same area to be monitored. Each real-time monitoring data sequence is analyzed, and the sequence coefficients are calculated based on the analysis results. The region status of the corresponding area to be monitored is set according to the sequence coefficient. If the region status is a state with bad pixels, the bad pixel characteristics are determined and an operation and maintenance strategy is generated.

[0006] In some embodiments of this application, monitoring characteristics are set based on historical data sets for each area to be monitored, including: The monitoring features include the number of monitoring points and the monitoring time interval; Obtain several historical bad spot monitoring logs for each area to be monitored, and extract the historical bad spot monitoring strategy, historical bad spot monitoring data, and historical status data from each historical bad spot monitoring log; A historical data set for each area to be monitored is constructed based on historical defect monitoring data and historical status data. Several defect evaluation indicators and several device evaluation indicators are pre-defined; The historical bad spot monitoring data in the historical data set is evaluated based on several bad spot evaluation indicators to obtain the first monitoring sub-evaluation value; The historical state data in the historical dataset is evaluated based on several device evaluation indicators to obtain the second monitoring sub-evaluation value; Based on the first and second monitoring sub-evaluation values, a monitoring evaluation value is generated for the corresponding area to be monitored. The number of monitoring points and the monitoring time interval are set according to the monitoring evaluation values ​​for the corresponding monitoring area.

[0007] In some embodiments of this application, the monitoring features are set based on the historical data set of each area to be monitored, and the method further includes: The monitoring features also include monitoring strategies; The types of defects in the corresponding monitoring areas are determined based on historical data sets, as well as the attention coefficient for each defect type; Based on the correspondence between the extracted historical defect monitoring strategies and historical defect monitoring data, as well as the correspondence between historical defect monitoring data and defect types, several historical defect monitoring strategies corresponding to each defect type are determined. Based on several preset optimization rules, the optimization analysis is performed on several historical bad point monitoring strategies corresponding to each bad point type, and the historical optimization coefficient of each historical bad point monitoring strategy is determined according to the optimization analysis results. The historical bad point monitoring strategies corresponding to each bad point type are sorted according to the historical optimization coefficient to obtain the sorting results; The number of sorting and filtering steps is set according to the attention coefficient of each bad pixel type, and the sorting results are filtered to obtain the bad pixel monitoring strategy for each bad pixel type. Based on the bad pixel monitoring strategies for all bad pixel types within the same monitoring interval, a monitoring strategy for the corresponding monitoring area is generated.

[0008] In some embodiments of this application, the historical optimization coefficient of each historical bad pixel monitoring strategy is determined based on the optimization analysis results, including: Several preset optimization rules are set, including reliable rules, anti-interference rules, low-cost rules and low-latency rules, and each preset optimization rule includes several preset optimization intervals, and each preset optimization interval is mapped to a preset sub-optimization coefficient. Based on several preset optimization rules, the optimization analysis of several historical bad point monitoring strategies corresponding to each bad point type is carried out to obtain several historical optimization intervals in which the historical bad point monitoring data corresponding to each historical bad point monitoring strategy is located. Compare the historical optimization intervals with the preset optimization intervals under the same preset optimization rule. If they are in the same preset optimization interval, set the preset sub-optimization coefficient of the corresponding preset optimization interval to the historical sub-optimization coefficient of the corresponding historical optimization interval. If they are not in the same preset optimization interval, determine the interval proportion of the preset optimization interval involved, set the weight coefficient according to the interval proportion, and calculate the historical sub-optimization coefficient of the corresponding historical optimization interval in combination with the preset sub-optimization coefficient of the preset optimization interval involved. The historical optimization coefficients of the corresponding historical bad pixel monitoring strategy are generated based on several historical sub-optimization coefficients of each historical bad pixel monitoring strategy corresponding to each bad pixel type and the weight coefficients of the corresponding preset optimization rules.

[0009] In some embodiments of this application, a real-time correction influence coefficient is generated for each real-time monitoring data, and the real-time monitoring data is corrected to obtain real-time corrected monitoring data, including: Obtain historical monitoring logs for each area to be monitored, and calculate the degree of influence of different historical environmental parameters in the historical monitoring logs on the historical monitoring parameters. Historical environmental parameters whose impact exceeds a preset impact threshold are set as the impact factors of the corresponding historical monitoring parameters. Construct an impact factor-historical impact coefficient mapping table for each historical monitoring parameter. The impact factor-historical impact coefficient mapping table includes several impact factors corresponding to the historical monitoring parameter. The historical monitoring parameter includes several historical monitoring data. Each impact factor includes several historical environmental data. Each historical monitoring data corresponds to a historical environmental data for each influencing factor, and each historical environmental data is mapped to a historical influence coefficient. The historical monitoring data of each historical monitoring parameter and the historical environmental data of each corresponding influencing factor are used as training input data, and the historical influence coefficient mapped to each historical environmental data of each corresponding influencing factor is used as training output data. The neural network is trained to obtain the influence model. Input the real-time monitoring data and the real-time environmental data of several influencing factors corresponding to the real-time monitoring data into the influence model to obtain several real-time influence coefficients; A real-time corrected impact coefficient is generated based on several real-time impact coefficients and the degree of influence of the corresponding impact factors. The real-time monitoring data is corrected according to the real-time correction influence coefficient to obtain the real-time corrected monitoring data.

[0010] In some embodiments of this application, each real-time monitoring data sequence is analyzed, and sequence coefficients are calculated based on the analysis results, including: The real-time corrected monitoring data of the same area to be monitored are classified according to the data type, and several real-time monitoring data sequences are constructed according to the location relationship of the monitoring points. Each real-time corrected monitoring data in the same real-time monitoring data sequence is compared with the corresponding standard monitoring data range. If it is within the standard monitoring data range, the corresponding real-time corrected monitoring data is set as normal data, and the normal coefficient is calculated. If the data is not within the standard monitoring data range, set the corresponding real-time correction monitoring data as abnormal data and calculate the abnormality coefficient. The sequence coefficients of the corresponding real-time monitoring data sequence are calculated based on the number of normal data, the corresponding normal coefficients, the number of abnormal data, and the corresponding abnormal coefficients.

[0011] In some embodiments of this application, the formula for calculating the sequence coefficients is as follows: ; X is the sequence coefficient, r1 is the first transformation coefficient, r2 is the second transformation coefficient, n1 is the number of normal data points, n2 is the number of outlier data points, and Z1i is the normal coefficient of the i-th normal data point. Z1i is the normal coefficient threshold, a1i is the weight coefficient of the i-th normal data, Z2s is the abnormal coefficient of the s-th abnormal data, and a2s is the weight coefficient of the s-th abnormal data.

[0012] In some embodiments of this application, the regional state of the corresponding monitored area is set according to the sequence coefficients, including: The state coefficient of the corresponding monitoring area is generated based on the sequence coefficients of several real-time monitoring data sequences of the same monitoring area and the weight coefficients of the corresponding real-time monitoring data. A first preset state coefficient threshold and a second preset state coefficient threshold are preset. If the state coefficient is not greater than the first preset state coefficient threshold, the corresponding area state is set as having bad points, and the bad point characteristics of the corresponding area to be monitored are determined. An operation and maintenance strategy is generated based on the bad point characteristics. If the state coefficient is between the first preset state coefficient threshold and the second preset state coefficient threshold, the corresponding area state is set as a risk state, and the monitoring characteristics of the corresponding area to be monitored are adjusted. If the state coefficient is not less than the second preset state coefficient threshold, the corresponding area state is set to the state of no bad pixels.

[0013] In some embodiments of this application, determining the characteristics of bad pixels corresponding to the area to be monitored includes: Obtain the sequence coefficients of several real-time monitoring data sequences of the monitored area with defective pixels; Real-time monitoring data sequences with sequence coefficients less than a preset sequence coefficient threshold are selected, and abnormal data in the selected real-time monitoring data sequences are extracted and marked to the corresponding monitoring points in the corresponding monitoring area. Construct a bad pixel feature reference library, which includes several preset bad pixel types. Each preset bad pixel type is mapped to several preset bad pixel data intervals, and each preset bad pixel data interval is mapped to a corresponding preset bad pixel feature. The similarity is obtained by performing a similarity analysis between all abnormal data of the same marked monitoring point and several preset bad point data intervals of each preset bad point type in the bad point feature reference library; If the similarity is greater than the preset similarity threshold, the corresponding preset bad point type will be used as the current bad point type of the corresponding marked monitoring point. If all abnormal data of the marked monitoring point falls within a number of preset bad point data intervals of the current bad point type, then the preset bad point feature mapped to the corresponding preset bad point data interval will be used as the current bad point feature of the corresponding marked monitoring point.

[0014] In some embodiments of this application, a defect detection system for electroluminescent devices is also included: The setting module is used to divide the electroluminescent device into multiple monitoring areas and set monitoring characteristics based on the historical data set of each monitoring area. The correction module is used to acquire real-time monitoring data for each area to be monitored according to the monitoring characteristics, generate a real-time correction influence coefficient for each real-time monitoring data, and correct the real-time monitoring data to obtain real-time corrected monitoring data. The analysis module is used to construct several real-time monitoring data sequences based on multiple real-time corrected monitoring data of the same area to be monitored, analyze each real-time monitoring data sequence, and calculate the sequence coefficients based on the analysis results. The determination module is used to set the regional status of the corresponding area to be monitored based on the sequence coefficients. If the regional status is a state with bad pixels, the bad pixel characteristics are determined and an operation and maintenance strategy is generated.

[0015] The defect detection method and system for an electroluminescent device according to embodiments of this application have the following advantages compared with the prior art: By dividing the electroluminescent device into multiple monitoring areas, setting monitoring characteristics and acquiring real-time monitoring data, generating a real-time correction influence coefficient to correct the real-time monitoring data, obtaining real-time corrected monitoring data and constructing several real-time monitoring data sequences, calculating sequence coefficients and determining the area status, if a defective state exists, determining the defective characteristics and generating an operation and maintenance strategy, setting reasonable monitoring characteristics, and improving the accuracy and efficiency of defective monitoring. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for detecting defects in an electroluminescent device according to an embodiment of this application. Figure 2 This is a schematic diagram of a defect detection system for an electroluminescent device according to an embodiment of this application. Detailed Implementation

[0017] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application 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. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] like Figure 1 As shown in the figure, a method for detecting defects in an electroluminescent device according to an embodiment of this application includes: Step S101: Divide the electroluminescent device into multiple monitoring areas, and set monitoring characteristics based on the historical data set of each monitoring area; Step S102: Obtain real-time monitoring data for each area to be monitored according to the monitoring characteristics, generate a real-time correction influence coefficient for each real-time monitoring data, and correct the real-time monitoring data to obtain real-time corrected monitoring data; Step S103: Construct several real-time monitoring data sequences based on multiple real-time corrected monitoring data of the same area to be monitored, analyze each real-time monitoring data sequence, and calculate the sequence coefficients based on the analysis results; Step S104: Set the region status of the corresponding region to be monitored according to the sequence coefficients. If the region status is a state with bad points, determine the bad point characteristics and generate an operation and maintenance strategy.

[0022] In some embodiments of this application, monitoring characteristics are set based on historical data sets for each area to be monitored, including: The monitoring features include the number of monitoring points and the monitoring time interval; Obtain several historical bad spot monitoring logs for each area to be monitored, and extract the historical bad spot monitoring strategy, historical bad spot monitoring data, and historical status data from each historical bad spot monitoring log; A historical data set for each area to be monitored is constructed based on historical defect monitoring data and historical status data. Several defect evaluation indicators and several device evaluation indicators are pre-defined; The historical bad spot monitoring data in the historical data set is evaluated based on several bad spot evaluation indicators to obtain the first monitoring sub-evaluation value; The historical state data in the historical dataset is evaluated based on several device evaluation indicators to obtain the second monitoring sub-evaluation value; Based on the first and second monitoring sub-evaluation values, a monitoring evaluation value is generated for the corresponding area to be monitored. The number of monitoring points and the monitoring time interval are set according to the monitoring evaluation values ​​for the corresponding monitoring area.

[0023] In this embodiment, the historical defect monitoring log refers to the log in which the monitoring result shows the presence of defects. The historical defect monitoring data includes, but is not limited to, the type of historical defect, the location of historical defect, the reliability of historical defect, and the degree of historical defect. The historical status data includes, but is not limited to, the device operating status, performance, and the remaining lifespan of the device.

[0024] In this embodiment, the defect evaluation indicators include, but are not limited to, defect probability, defect frequency, defect impact degree, and whether the defect distribution is regular. The device evaluation indicators include, but are not limited to, different historical defect types in the current monitored area, the impact of historical defect severity on the device during use, the degree of lifespan reduction, and the degree of performance degradation.

[0025] In this embodiment, the higher the monitoring evaluation value, the more monitoring points are set and the shorter the monitoring time interval. The monitoring points are reasonably allocated according to the area to be monitored and the historical distribution of bad points, thereby improving the efficiency and accuracy of bad point monitoring.

[0026] In this embodiment, by calculating the first monitoring sub-evaluation value and the second monitoring sub-evaluation value, the different types of historical defects, the probability of historical defects, the degree of historical defects, and their impact on the operating status, performance, and lifespan of the device in each monitored area are accurately evaluated, thereby formulating reasonable monitoring characteristics and improving the defect monitoring efficiency of electroluminescent devices.

[0027] In some embodiments of this application, the monitoring features are set based on the historical data set of each area to be monitored, and the method further includes: The monitoring features also include monitoring strategies; The types of defects in the corresponding monitoring areas are determined based on historical data sets, as well as the attention coefficient for each defect type; Based on the correspondence between the extracted historical defect monitoring strategies and historical defect monitoring data, as well as the correspondence between historical defect monitoring data and defect types, several historical defect monitoring strategies corresponding to each defect type are determined. Based on several preset optimization rules, the optimization analysis is performed on several historical bad point monitoring strategies corresponding to each bad point type, and the historical optimization coefficient of each historical bad point monitoring strategy is determined according to the optimization analysis results. The historical bad point monitoring strategies corresponding to each bad point type are sorted according to the historical optimization coefficient to obtain the sorting results; The number of sorting and filtering steps is set according to the attention coefficient of each bad pixel type, and the sorting results are filtered to obtain the bad pixel monitoring strategy for each bad pixel type. Based on the bad pixel monitoring strategies for all bad pixel types within the same monitoring interval, a monitoring strategy for the corresponding monitoring area is generated.

[0028] In this embodiment, the attention coefficient is set based on the historical defect probability, historical defect frequency, historical defect severity, and the impact on the device's performance, operating status, and lifespan. When the historical defect probability, historical defect frequency, historical defect severity, and impact are higher, the corresponding attention coefficient is higher, and the number of sorting and filtering options is greater, and vice versa.

[0029] In this embodiment, the number of sorting and filtering is at least one. If there are multiple, multiple historical bad pixel monitoring strategies are filtered out according to the sorting results, and the current bad pixel monitoring strategy for the bad pixel type and the monitoring strategy for each area to be monitored are constructed to improve the bad pixel monitoring accuracy for each monitoring area.

[0030] In some embodiments of this application, the historical optimization coefficient of each historical bad pixel monitoring strategy is determined based on the optimization analysis results, including: Several preset optimization rules are set, including reliable rules, anti-interference rules, low-cost rules and low-latency rules, and each preset optimization rule includes several preset optimization intervals, and each preset optimization interval is mapped to a preset sub-optimization coefficient. Based on several preset optimization rules, the optimization analysis of several historical bad point monitoring strategies corresponding to each bad point type is carried out to obtain several historical optimization intervals in which the historical bad point monitoring data corresponding to each historical bad point monitoring strategy is located. Compare the historical optimization intervals with the preset optimization intervals under the same preset optimization rule. If they are in the same preset optimization interval, set the preset sub-optimization coefficient of the corresponding preset optimization interval to the historical sub-optimization coefficient of the corresponding historical optimization interval. If they are not in the same preset optimization interval, determine the interval proportion of the preset optimization interval involved, set the weight coefficient according to the interval proportion, and calculate the historical sub-optimization coefficient of the corresponding historical optimization interval in combination with the preset sub-optimization coefficient of the preset optimization interval involved. The historical optimization coefficients of the corresponding historical bad pixel monitoring strategy are generated based on several historical sub-optimization coefficients of each historical bad pixel monitoring strategy corresponding to each bad pixel type and the weight coefficients of the corresponding preset optimization rules.

[0031] In this embodiment, the trusted rule is used to evaluate the monitoring accuracy of each historical bad spot monitoring strategy. The corresponding preset optimization interval refers to multiple preset trusted coefficient intervals. When the trusted coefficient interval is larger, the corresponding preset optimization sub-coefficient is larger, and vice versa.

[0032] In this embodiment, the anti-interference rule is used to assess the degree of influence of changes in the surrounding environment on the monitoring process or results of the historical bad spot monitoring strategy. The corresponding preset optimization interval refers to multiple preset anti-interference coefficient intervals. When the degree of influence is smaller, the anti-interference coefficient interval is larger and the corresponding preset optimization sub-coefficient is larger, and vice versa.

[0033] In this embodiment, the low-cost rule is used to evaluate the monitoring cost of each historical bad pixel monitoring strategy. The corresponding preset optimization interval refers to multiple preset cost intervals. When the cost interval is larger, the corresponding preset optimization sub-coefficient is smaller, and vice versa.

[0034] In this embodiment, the low latency rule is used to evaluate the monitoring timeliness of each historical bad point monitoring strategy. The corresponding preset optimization interval refers to multiple preset latency intervals. When the latency interval is larger, the corresponding preset optimization sub-coefficient is smaller, and vice versa.

[0035] In this embodiment, the historical optimization interval is obtained based on the historical bad spot monitoring data corresponding to the historical bad spot monitoring strategy, namely, the historical reliability coefficient interval obtained by multiple historical reliability coefficients corresponding to the same historical bad spot monitoring strategy, the historical anti-interference coefficient interval obtained by multiple historical anti-interference coefficients, the historical cost interval obtained by multiple historical monitoring costs, and the historical delay time interval obtained by multiple historical delay time.

[0036] In this embodiment, the interval percentage refers to the proportion of the historical optimization interval in each of the preset optimization intervals involved. For example, if the historical optimization interval is (50, 120) and the preset optimization intervals involved include (40, 80) and (80, 120), then the interval percentages of each of the preset optimization intervals involved are 30 / 70 and 40 / 70, respectively. When the interval percentage is larger, the corresponding weight coefficient is larger.

[0037] In this embodiment, by calculating the historical optimization coefficient of each historical defect monitoring strategy, the monitoring accuracy, timeliness, cost, and anti-interference capability of each historical defect monitoring strategy are accurately evaluated from multiple dimensions, thereby determining the defect monitoring strategy for each defect type. This lays the foundation for subsequently calculating the sequence coefficient of each monitoring point and improving the defect monitoring accuracy of electroluminescent devices.

[0038] In some embodiments of this application, a real-time correction influence coefficient is generated for each real-time monitoring data, and the real-time monitoring data is corrected to obtain real-time corrected monitoring data, including: Obtain historical monitoring logs for each area to be monitored, and calculate the degree of influence of different historical environmental parameters in the historical monitoring logs on the historical monitoring parameters. Historical environmental parameters whose impact exceeds a preset impact threshold are set as the impact factors of the corresponding historical monitoring parameters. Construct an impact factor-historical impact coefficient mapping table for each historical monitoring parameter. The impact factor-historical impact coefficient mapping table includes several impact factors corresponding to the historical monitoring parameter. The historical monitoring parameter includes several historical monitoring data. Each impact factor includes several historical environmental data. Each historical monitoring data corresponds to a historical environmental data for each influencing factor, and each historical environmental data is mapped to a historical influence coefficient. The historical monitoring data of each historical monitoring parameter and the historical environmental data of each corresponding influencing factor are used as training input data, and the historical influence coefficient mapped to each historical environmental data of each corresponding influencing factor is used as training output data. The neural network is trained to obtain the influence model. Input the real-time monitoring data and the real-time environmental data of several influencing factors corresponding to the real-time monitoring data into the influence model to obtain several real-time influence coefficients; A real-time corrected impact coefficient is generated based on several real-time impact coefficients and the degree of influence of the corresponding impact factors. The real-time monitoring data is corrected according to the real-time correction influence coefficient to obtain the real-time corrected monitoring data.

[0039] In this embodiment, the actual values ​​of the same monitoring data in different historical monitoring logs are the same, the monitoring strategies are the same but the historical environmental data are different and the collected historical monitoring data have a large deviation, so as to calculate the degree of influence. The degree of influence refers to the impact of historical environmental data on the collection accuracy of historical monitoring data, resulting in errors between historical monitoring data and actual monitoring data.

[0040] In this embodiment, the historical impact coefficient is calculated based on the data error between historical monitoring data and actual monitoring data under historical environmental data for each impact factor, i.e., actual monitoring data = historical monitoring data * historical impact coefficient.

[0041] In this embodiment, the weight coefficients of the corresponding influencing factors are set according to the degree of influence of several influencing factors of the same historical monitoring parameter, and the weighted mean is calculated in combination with the real-time influence coefficient to obtain the real-time corrected influence coefficient.

[0042] In this embodiment, by calculating several real-time impact coefficients for each real-time monitoring data, the environmental impact in the above monitoring strategy is compensated, the monitoring accuracy of the monitoring data is improved again, the foundation for subsequent calculation of sequence coefficients is laid, and the accuracy of bad spot monitoring in each monitored area is improved.

[0043] In some embodiments of this application, each real-time monitoring data sequence is analyzed, and sequence coefficients are calculated based on the analysis results, including: The real-time corrected monitoring data of the same area to be monitored are classified according to the data type, and several real-time monitoring data sequences are constructed according to the location relationship of the monitoring points. Each real-time corrected monitoring data in the same real-time monitoring data sequence is compared with the corresponding standard monitoring data range. If it is within the standard monitoring data range, the corresponding real-time corrected monitoring data is set as normal data, and the normal coefficient is calculated. If the data is not within the standard monitoring data range, set the corresponding real-time correction monitoring data as abnormal data and calculate the abnormality coefficient. The sequence coefficients of the corresponding real-time monitoring data sequence are calculated based on the number of normal data, the corresponding normal coefficients, the number of abnormal data, and the corresponding abnormal coefficients.

[0044] In some embodiments of this application, the formula for calculating the sequence coefficients is as follows: ; X is the sequence coefficient, r1 is the first transformation coefficient, r2 is the second transformation coefficient, n1 is the number of normal data points, n2 is the number of outlier data points, and Z1i is the normal coefficient of the i-th normal data point. Z1i is the normal coefficient threshold, a1i is the weight coefficient of the i-th normal data, Z2s is the abnormal coefficient of the s-th abnormal data, and a2s is the weight coefficient of the s-th abnormal data.

[0045] In this embodiment, z0 represents the value corresponding to the real-time corrected monitoring data. To correct the intermediate values ​​in the standard monitoring data range in real time, To correct the value of the interval boundary point that is closest to the corresponding real-time correction monitoring data in the standard monitoring data interval, The normalization coefficient is used for example. If the range is [80-120], the median is 100, and the real-time monitoring data value is 110, then (110-100) / (120-100) = 50%, indicating a deviation of 50% from the median. The smaller the percentage deviation from the median, the larger the corresponding normalization coefficient. It is a pre-set minimum normal coefficient.

[0046] In this embodiment, the normality coefficient is used to assess the normality of normal data. The smaller the deviation from the median, the larger the normality coefficient, and vice versa.

[0047] In this embodiment, , The minimum value within the standard monitoring data range. The maximum value within the standard monitoring data range, i.e., the values ​​at both ends of the range. For example, if the standard monitoring data range is [3.5, 5.5] and the real-time corrected monitoring data is 2.5, then (2.5-3.5) / (5.5-3.5) = -50%, which means that the real-time corrected monitoring data is 50% lower than the lower limit of the standard monitoring data range. The abnormality coefficient of each abnormal data is determined based on the preset abnormality coefficient mapped by the preset lower limit deviation percentage and the preset upper limit percentage.

[0048] In this embodiment, the anomaly coefficient is used to assess the degree of anomaly of abnormal data, predict the type of anomaly, and the degree of impact of the anomaly. The larger the percentage deviation of the lower limit or the upper limit, the greater the predicted anomaly type and the greater the degree of impact of the anomaly, and the larger the corresponding anomaly coefficient and the smaller the sequence coefficient, and vice versa.

[0049] In this embodiment, by calculating sequence coefficients, the data status of each data type is accurately evaluated, laying the foundation for subsequent calculation of the regional status of the corresponding monitored area and improving the accuracy of bad spot monitoring for each monitored area.

[0050] In some embodiments of this application, the regional state of the corresponding monitored area is set according to the sequence coefficients, including: The state coefficient of the corresponding monitoring area is generated based on the sequence coefficients of several real-time monitoring data sequences of the same monitoring area and the weight coefficients of the corresponding real-time monitoring data. A first preset state coefficient threshold and a second preset state coefficient threshold are preset. If the state coefficient is not greater than the first preset state coefficient threshold, the corresponding area state is set as having bad points, and the bad point characteristics of the corresponding area to be monitored are determined. An operation and maintenance strategy is generated based on the bad point characteristics. If the state coefficient is between the first preset state coefficient threshold and the second preset state coefficient threshold, the corresponding area state is set as a risk state, and the monitoring characteristics of the corresponding area to be monitored are adjusted. If the state coefficient is not less than the second preset state coefficient threshold, the corresponding area state is set to the state of no bad pixels.

[0051] In this embodiment, the larger the sequence coefficient, the larger the state coefficient, indicating that the state of the corresponding monitored area is better and the probability of bad points is smaller.

[0052] In this embodiment, the characteristics of bad pixels include bad pixel type, bad pixel severity, bad pixel distribution location, etc., and the operation and maintenance strategy is determined based on historical bad pixel operation and maintenance logs.

[0053] In this embodiment, adjusting the monitoring features includes adjusting the monitoring strategy, the monitoring time interval, and the number of monitoring points. When the monitoring points are closer to the first preset state coefficient threshold, the number of monitoring points increases and the monitoring time interval becomes shorter, thereby timely determining whether bad points have occurred and performing maintenance in a timely manner, reducing maintenance costs.

[0054] In some embodiments of this application, determining the characteristics of bad pixels corresponding to the area to be monitored includes: Obtain the sequence coefficients of several real-time monitoring data sequences of the monitored area with defective pixels; Real-time monitoring data sequences with sequence coefficients less than a preset sequence coefficient threshold are selected, and abnormal data in the selected real-time monitoring data sequences are extracted and marked to the corresponding monitoring points in the corresponding monitoring area. Construct a bad pixel feature reference library, which includes several preset bad pixel types. Each preset bad pixel type is mapped to several preset bad pixel data intervals, and each preset bad pixel data interval is mapped to a corresponding preset bad pixel feature. The similarity is obtained by performing a similarity analysis between all abnormal data of the same marked monitoring point and several preset bad point data intervals of each preset bad point type in the bad point feature reference library; If the similarity is greater than the preset similarity threshold, the corresponding preset bad point type will be used as the current bad point type of the corresponding marked monitoring point. If all abnormal data of the marked monitoring point falls within a number of preset bad point data intervals of the current bad point type, then the preset bad point feature mapped to the corresponding preset bad point data interval will be used as the current bad point feature of the corresponding marked monitoring point.

[0055] In this embodiment, the defect feature reference library is constructed based on the historical defect types and corresponding historical defect features of electroluminescent devices.

[0056] In this embodiment, the preset defect types include normally lit defect, normally dark defect, flickering defect, electrical failure defect, etc. The historical defect characteristics include abnormal leakage current under driving voltage, continuous light emission brightness higher than surrounding pixels, abnormally increased slope of IV curve, etc. The historical monitoring data associated with the historical defect characteristics are summarized to obtain the historical defect data range of each historical defect characteristic.

[0057] In this embodiment, by determining the type and characteristics of defects in the monitored area where defects exist, a foundation is laid for determining the subsequent operation and maintenance strategy. Defects are detected in a timely manner, and operation and maintenance strategies are set reasonably to ensure the accuracy of defect monitoring for electroluminescent devices.

[0058] In some embodiments of the present application, Figure 2 As shown, it also includes a defect detection system for electroluminescent devices: The setting module is used to divide the electroluminescent device into multiple monitoring areas and set monitoring characteristics based on the historical data set of each monitoring area. The correction module is used to acquire real-time monitoring data for each area to be monitored according to the monitoring characteristics, generate a real-time correction influence coefficient for each real-time monitoring data, and correct the real-time monitoring data to obtain real-time corrected monitoring data. The analysis module is used to construct several real-time monitoring data sequences based on multiple real-time corrected monitoring data of the same area to be monitored, analyze each real-time monitoring data sequence, and calculate the sequence coefficients based on the analysis results. The determination module is used to set the regional status of the corresponding area to be monitored based on the sequence coefficients. If the regional status is a state with bad pixels, the bad pixel characteristics are determined and an operation and maintenance strategy is generated.

[0059] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. A method for detecting defective pixels in an electroluminescent device, characterized in that, include: The electroluminescent device is divided into multiple monitoring areas, and monitoring characteristics are set according to the historical data set of each monitoring area. Real-time monitoring data for each area to be monitored is obtained according to the monitoring characteristics, a real-time correction influence coefficient for each real-time monitoring data is generated, and the real-time monitoring data is corrected to obtain real-time corrected monitoring data. Several real-time monitoring data sequences are constructed based on multiple real-time corrected monitoring data of the same area to be monitored. Each real-time monitoring data sequence is analyzed, and the sequence coefficients are calculated based on the analysis results. The region status of the corresponding area to be monitored is set according to the sequence coefficient. If the region status is a state with bad pixels, the bad pixel characteristics are determined and an operation and maintenance strategy is generated.

2. The defect detection method for an electroluminescent device as described in claim 1, characterized in that, Monitoring characteristics are defined based on the historical data set of each area to be monitored, including: The monitoring features include the number of monitoring points and the monitoring time interval; Obtain several historical bad spot monitoring logs for each area to be monitored, and extract the historical bad spot monitoring strategy, historical bad spot monitoring data, and historical status data from each historical bad spot monitoring log; A historical data set for each area to be monitored is constructed based on historical defect monitoring data and historical status data. Several defect evaluation indicators and several device evaluation indicators are pre-defined; The historical bad spot monitoring data in the historical data set is evaluated based on several bad spot evaluation indicators to obtain the first monitoring sub-evaluation value; The historical state data in the historical dataset is evaluated based on several device evaluation indicators to obtain the second monitoring sub-evaluation value; Based on the first and second monitoring sub-evaluation values, a monitoring evaluation value is generated for the corresponding area to be monitored. The number of monitoring points and the monitoring time interval are set according to the monitoring evaluation values ​​for the corresponding monitoring area.

3. The defect detection method for an electroluminescent device as described in claim 2, characterized in that, Monitoring characteristics are set based on the historical data set of each area to be monitored, including: The monitoring features also include monitoring strategies; The types of defects in the corresponding monitoring areas are determined based on historical data sets, as well as the attention coefficient for each defect type; Based on the correspondence between the extracted historical defect monitoring strategies and historical defect monitoring data, as well as the correspondence between historical defect monitoring data and defect types, several historical defect monitoring strategies corresponding to each defect type are determined. Based on several preset optimization rules, the optimization analysis is performed on several historical bad point monitoring strategies corresponding to each bad point type, and the historical optimization coefficient of each historical bad point monitoring strategy is determined according to the optimization analysis results. The historical bad point monitoring strategies corresponding to each bad point type are sorted according to the historical optimization coefficient to obtain the sorting results; The number of sorting and filtering steps is set according to the attention coefficient of each bad pixel type, and the sorting results are filtered to obtain the bad pixel monitoring strategy for each bad pixel type. Based on the bad pixel monitoring strategies for all bad pixel types within the same monitoring interval, a monitoring strategy for the corresponding monitoring area is generated.

4. The defect detection method for an electroluminescent device as described in claim 3, characterized in that, Based on the optimization analysis results, the historical optimization coefficients for each historical defect monitoring strategy are determined, including: Several preset optimization rules are set, including reliable rules, anti-interference rules, low-cost rules and low-latency rules, and each preset optimization rule includes several preset optimization intervals, and each preset optimization interval is mapped to a preset sub-optimization coefficient. Based on several preset optimization rules, the optimization analysis of several historical bad point monitoring strategies corresponding to each bad point type is carried out to obtain several historical optimization intervals in which the historical bad point monitoring data corresponding to each historical bad point monitoring strategy is located. Compare the historical optimization intervals with the preset optimization intervals under the same preset optimization rule. If they are in the same preset optimization interval, set the preset sub-optimization coefficient of the corresponding preset optimization interval to the historical sub-optimization coefficient of the corresponding historical optimization interval. If they are not in the same preset optimization interval, determine the interval proportion of the preset optimization interval involved, set the weight coefficient according to the interval proportion, and calculate the historical sub-optimization coefficient of the corresponding historical optimization interval in combination with the preset sub-optimization coefficient of the preset optimization interval involved. The historical optimization coefficients of the corresponding historical bad pixel monitoring strategy are generated based on several historical sub-optimization coefficients of each historical bad pixel monitoring strategy corresponding to each bad pixel type and the weight coefficients of the corresponding preset optimization rules.

5. The defect detection method for an electroluminescent device as described in claim 4, characterized in that, Generate a real-time corrected impact coefficient for each real-time monitoring data point, and correct the real-time monitoring data to obtain real-time corrected monitoring data, including: Obtain historical monitoring logs for each area to be monitored, and calculate the degree of influence of different historical environmental parameters in the historical monitoring logs on the historical monitoring parameters. Historical environmental parameters whose impact exceeds a preset impact threshold are set as the impact factors of the corresponding historical monitoring parameters. Construct an impact factor-historical impact coefficient mapping table for each historical monitoring parameter. The impact factor-historical impact coefficient mapping table includes several impact factors corresponding to the historical monitoring parameter. The historical monitoring parameter includes several historical monitoring data. Each impact factor includes several historical environmental data. Each historical monitoring data corresponds to a historical environmental data for each influencing factor, and each historical environmental data is mapped to a historical influence coefficient. The historical monitoring data of each historical monitoring parameter and the historical environmental data of each corresponding influencing factor are used as training input data, and the historical influence coefficient mapped to each historical environmental data of each corresponding influencing factor is used as training output data. The neural network is trained to obtain the influence model. Input the real-time monitoring data and the real-time environmental data of several influencing factors corresponding to the real-time monitoring data into the influence model to obtain several real-time influence coefficients; A real-time corrected impact coefficient is generated based on several real-time impact coefficients and the degree of influence of the corresponding impact factors. The real-time monitoring data is corrected according to the real-time correction influence coefficient to obtain the real-time corrected monitoring data.

6. The defect detection method for an electroluminescent device as described in claim 5, characterized in that, Each real-time monitoring data sequence is analyzed, and sequence coefficients are calculated based on the analysis results, including: The real-time corrected monitoring data of the same area to be monitored are classified according to the data type, and several real-time monitoring data sequences are constructed according to the location relationship of the monitoring points. Each real-time corrected monitoring data in the same real-time monitoring data sequence is compared with the corresponding standard monitoring data range. If it is within the standard monitoring data range, the corresponding real-time corrected monitoring data is set as normal data, and the normal coefficient is calculated. If the data is not within the standard monitoring data range, set the corresponding real-time correction monitoring data as abnormal data and calculate the abnormality coefficient. The sequence coefficients of the corresponding real-time monitoring data sequence are calculated based on the number of normal data, the corresponding normal coefficients, the number of abnormal data, and the corresponding abnormal coefficients.

7. The defect detection method for an electroluminescent device as described in claim 6, characterized in that, The formula for calculating the sequence coefficients is as follows: ； X is the sequence coefficient, r1 is the first transformation coefficient, r2 is the second transformation coefficient, n1 is the number of normal data points, n2 is the number of outlier data points, and Z1i is the normal coefficient of the i-th normal data point. Z1i is the normal coefficient threshold, a1i is the weight coefficient of the i-th normal data, Z2s is the abnormal coefficient of the s-th abnormal data, and a2s is the weight coefficient of the s-th abnormal data.

8. The defect detection method for an electroluminescent device as described in claim 7, characterized in that, The regional status of the corresponding monitored area is set according to the sequence coefficients, including: The state coefficient of the corresponding monitoring area is generated based on the sequence coefficients of several real-time monitoring data sequences of the same monitoring area and the weight coefficients of the corresponding real-time monitoring data. A first preset state coefficient threshold and a second preset state coefficient threshold are preset. If the state coefficient is not greater than the first preset state coefficient threshold, the corresponding area state is set as having bad points, and the bad point characteristics of the corresponding area to be monitored are determined. An operation and maintenance strategy is generated based on the bad point characteristics. If the state coefficient is between the first preset state coefficient threshold and the second preset state coefficient threshold, the corresponding area state is set as a risk state, and the monitoring characteristics of the corresponding area to be monitored are adjusted. If the state coefficient is not less than the second preset state coefficient threshold, the corresponding area state is set to the state of no bad pixels.

9. The defect detection method for an electroluminescent device as described in claim 8, characterized in that, Determine the characteristics of bad pixels in the corresponding area to be monitored, including: Obtain the sequence coefficients of several real-time monitoring data sequences of the monitored area with defective pixels; Real-time monitoring data sequences with sequence coefficients less than a preset sequence coefficient threshold are selected, and abnormal data in the selected real-time monitoring data sequences are extracted and marked to the corresponding monitoring points in the corresponding monitoring area. Construct a bad pixel feature reference library, which includes several preset bad pixel types. Each preset bad pixel type is mapped to several preset bad pixel data intervals, and each preset bad pixel data interval is mapped to a corresponding preset bad pixel feature. The similarity is obtained by performing a similarity analysis between all abnormal data of the same marked monitoring point and several preset bad point data intervals of each preset bad point type in the bad point feature reference library; If the similarity is greater than the preset similarity threshold, the corresponding preset bad point type will be used as the current bad point type of the corresponding marked monitoring point. If all abnormal data of the marked monitoring point falls within a number of preset bad point data intervals of the current bad point type, then the preset bad point feature mapped to the corresponding preset bad point data interval will be used as the current bad point feature of the corresponding marked monitoring point.

10. A defect detection system for an electroluminescent device, characterized in that, include: The setting module is used to divide the electroluminescent device into multiple monitoring areas and set monitoring characteristics based on the historical data set of each monitoring area. The correction module is used to acquire real-time monitoring data for each area to be monitored according to the monitoring characteristics, generate a real-time correction influence coefficient for each real-time monitoring data, and correct the real-time monitoring data to obtain real-time corrected monitoring data. The analysis module is used to construct several real-time monitoring data sequences based on multiple real-time corrected monitoring data of the same area to be monitored, analyze each real-time monitoring data sequence, and calculate the sequence coefficients based on the analysis results. The determination module is used to set the regional status of the corresponding area to be monitored based on the sequence coefficients. If the regional status is a state with bad pixels, the bad pixel characteristics are determined and an operation and maintenance strategy is generated.

Images