Display process defect automatic detection method

Abstract

The invention relates to the technical field of display process defect detection, and discloses a display process defect automatic detection method, which comprises the following steps: setting a plurality of detection product categories, and establishing a defect analysis model according to the detection product categories; setting a first defect detection strategy of the to-be-detected product, and obtaining a display process parameter set of the to-be-detected product according to the first defect detection strategy; analyzing the display process parameter set according to the defect analysis model, and generating an initial identifier according to an analysis result, the initial identifier including an unknown identifier, a normal identifier and a defect identifier; and when the identification is the unknown identification, generating a second defect detection strategy, generating a defect evaluation value according to the second defect detection strategy, and judging whether a defect early warning instruction is generated or not according to the defect evaluation value, so that the defect detection precision of the to-be-detected product is improved.

Description

Technical Field

[0001] The present application relates to the technical field of display process defect detection, and in particular to a method for automatically detecting display process defects. Background Art

[0002] Display process defects refer to production quality problems caused by various reasons during the manufacturing process of display devices. These defects may occur at different process stages.

[0003] In the existing technology, manual inspection is usually used to conduct comprehensive inspections of display devices during the production stages. However, manual inspection is inefficient and cannot guarantee the inspection accuracy of the display process at each stage. Therefore, there is an urgent need for an automatic detection method for display process defects to comprehensively improve the defect detection accuracy and efficiency at each stage and effectively reduce the probability of display process defects. Summary of the Invention

[0004] To solve the above technical problems, the present application provides a method for automatic detection of display process defects. By establishing a defect analysis model, an initial identification is generated. When it is an unknown identification, a second defect detection strategy is generated and a defect evaluation value is generated. The display process parameters of each detection stage are accurately detected and defect evaluation is performed, thereby improving the defect detection accuracy of the product to be inspected and reducing the probability of subsequent display process defects.

[0005] In some embodiments of the present application, a method for automatically detecting display process defects is provided, comprising: Set up several categories of testing products and establish defect analysis models based on the categories of testing products; Setting a first defect detection strategy for the product to be inspected, and obtaining a display process parameter set for the product to be inspected according to the first defect detection strategy; Analyze the display process parameter set according to the defect analysis model, and generate an initial identification according to the analysis result, wherein the initial identification includes an unknown identification, a normal identification, and a defect identification; When it is an unknown identifier, a second defect detection strategy is generated, a defect evaluation value is generated according to the second defect detection strategy, and whether to generate a defect warning instruction is determined according to the defect evaluation value.

[0006] In some embodiments of the present application, a defect analysis model is established according to the category of the inspected product, including: Collect basic information of multiple test products, classify the multiple test products according to the basic information, and obtain multiple test product categories; Extract historical inspection logs of the same inspection product category to obtain the historical display process parameter matrix of the corresponding inspection product category; The history display process parameter matrix displays behaviors of a plurality of history display process parameters in the same history detection log, history display process parameters in the same category are listed, each history detection log is mapped with a corresponding history defect identifier, each history display process parameter is mapped with a corresponding history display process image, the history defect identifier includes a defect category and a history defect degree; A detection evaluation value of a corresponding detection product category is generated according to the history display process parameter matrix; A plurality of division nodes of the corresponding detection product category are set according to the detection evaluation value, and a plurality of detection stages are generated based on the division nodes and a process stage of the corresponding detection product category; A history display process parameter sub-matrix of each detection stage is extracted based on the history display process parameter matrix, impact evaluation analysis is performed on the history display process parameter sub-matrix, and a feature parameter set is generated according to analysis results; The feature parameter set is set as a training data package of the corresponding detection stage, and a defect analysis sub-model of the corresponding detection stage is constructed based on the training data package of each detection stage; A detection time interval of the corresponding detection stage is set according to a comprehensive change feature of the feature parameter set, and a sub-detection strategy of the corresponding detection stage is generated in combination with a corresponding detection equipment; The defect analysis sub-model and the sub-detection strategy of each detection product category are sequentially set; A defect analysis model is generated according to all defect analysis sub-models and sub-detection strategies.

[0007] In some embodiments of the present application, a detection evaluation value of a corresponding detection product category is generated according to a history display process parameter matrix, including: A calculation formula of the detection evaluation value is: ; Wherein, J is the detection evaluation value, a1 is a first detection conversion coefficient, Q1 is a first weight coefficient, n1 is a number of product evaluation indexes, is a reference evaluation value of an i1th product evaluation index generated based on a history display process parameter matrix of a corresponding to-be-detected product category, is a weight coefficient of the i1th product evaluation index, a2 is a second detection conversion coefficient, Q2 is a second weight coefficient, n2 is a number of defect evaluation indexes, is a reference evaluation value of an i2th defect evaluation index generated based on the history display process parameter matrix of the corresponding to-be-detected product category, is a weight coefficient of the i2th defect evaluation index.

[0008] In some embodiments of the present application, impact evaluation analysis is performed on the history display process parameter sub-matrix, and a feature parameter set is generated according to analysis results, including: Performing cluster analysis on the historical defect identification mapped by each historical inspection log in the historical display process parameter submatrix of the same inspection stage to obtain multiple analysis matrices; The analysis matrix includes historical display process parameters from a number of historical inspection logs of different historical defect levels for the same defect category; Preset several standard display process parameters for each detection stage, compare the historical display process parameters in each analysis matrix with the corresponding standard display process parameters, and obtain the deviation historical display process parameters, continuous deviation duration, and corresponding deviation value for each row in each analysis matrix; Generate the deviation coefficient of the corresponding deviation history display process parameter based on the continuous deviation duration and the corresponding deviation value, and generate the historical defect coefficient of the corresponding defect category based on the historical defect degree; Map the deviation coefficient of each column of the same deviation history display process parameter in the same analysis matrix and the historical defect coefficient of the corresponding column to the same blank point diagram to obtain the deviation parameter-defect coefficient correlation diagram of the corresponding analysis matrix; Generate an evaluation coefficient for the impact of each deviation history display process parameter on the historical defect degree of the corresponding defect category based on the deviation parameter-defect coefficient correlation diagram; Pre-set impact assessment coefficient thresholds; Constructing a characteristic parameter set based on historical process parameters whose impact assessment coefficients in each analysis matrix of the same detection stage are greater than an impact assessment coefficient threshold; The characteristic parameter set includes several historical display process parameters, and each historical display process parameter is mapped to a corresponding historical display process image and a historical defect identifier.

[0009] In some embodiments of the present application, setting a first defect detection strategy for a product to be inspected includes: Obtain basic information about the product to be tested; Calculate the similarity between the product to be tested and each test product category; The calculation formula of the similarity is: ; Among them, D is the similarity, d0 is the similarity conversion coefficient, and w is the number of feature evaluation indicators; is the weight coefficient of the sth feature evaluation index; The reference value of the sth feature evaluation index generated for the product set to be tested; is the reference value of the sth characteristic evaluation index in the corresponding test product category; Setting the sub-detection strategy of the detection product category with the greatest similarity as the first defect detection strategy for the current product to be detected; The first defect detection strategy includes several stages to be detected of the product to be detected, detection time intervals corresponding to the stages to be detected, and corresponding detection equipment; A display process parameter set corresponding to the product to be inspected is obtained based on the first defect detection strategy, where the display process parameter set includes a plurality of display process parameter subsets of the stages to be inspected.

[0010] In some embodiments of the present application, setting an initial identifier according to the analysis result includes: Analyze a subset of display process parameters of each stage to be inspected based on the defect analysis model, and divide the stages to be inspected into stages with settable identification and stages without settable identification according to the analysis results; Setting the initial identification of the settable identification stage as the defect identification, wherein the defect identification includes the current defect category and the current defect severity; Generate a first defect evaluation value for the detection phase of the corresponding defect identification according to the current defect category and the current defect degree; Generating a second defect evaluation value corresponding to the stage to be inspected according to the displayed process parameter subset of the stage that cannot be set; Presetting a preset second defect evaluation value threshold; If the second defect evaluation value is less than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as a normal identification; If the second defect evaluation value is greater than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as an unknown identification.

[0011] In some embodiments of the present application, before generating the second defect detection strategy, the method further includes: Construct an association model for the stage to be detected, which is initially identified as a defect identifier, and extract the defect category and the corresponding defect degree corresponding to the stage to be detected; Generate the defect category of the corresponding to-be-detected stage and the stage to-be-detected affected by the corresponding defect degree and the degree of influence based on the association model, and set the stage to-be-detected with an influence degree greater than a preset influence degree threshold as the associated to-be-detected stage of the corresponding to-be-detected stage; Determine whether the initial identifiers of the associated to-be-detected stages whose initial identifier is a defect identifier are defect identifiers. If not, replace them with defect identifiers.

[0012] In some embodiments of the present application, generating a second defect detection strategy includes: Screening out the stages to be inspected that are initially identified as unknown, as well as the displayed process parameters corresponding to the stages to be inspected and the corresponding process parameter images; Analyze the process parameter image of the current stage to be inspected based on the image-parameter mapping table, and determine the expected display process parameter range mapped by the corresponding process parameter image according to the analysis result; Determine whether the display process parameter corresponding to the process parameter image is within the expected display process parameter range. If so, set the corresponding display process parameter as a credible parameter; if not, set the corresponding display process parameter as an untrustworthy parameter. Collect several untrustworthy parameters in the current detection stage and perform cluster analysis, determine the parameter categories of several untrustworthy parameters based on the analysis results, and configure corresponding preset detection equipment and corresponding preset detection time intervals for corresponding parameter categories; generating a second defect detection strategy based on a preset detection device and a corresponding preset detection time interval; Untrusted parameters are collected according to the second defect detection strategy until the untrusted parameters are within the corresponding expected process parameter range, and the corresponding display process parameters are replaced.

[0013] In some embodiments of the present application, generating a defect evaluation value according to the second defect detection strategy includes: Generate a defect risk assessment value corresponding to the stage to be inspected based on the trusted parameters of the stage to be inspected that is initially identified as unknown and the replaced display process parameters; Generate a defect assessment value based on a number of defect risk assessment values ​​of the to-be-detected stages initially identified as unknown identifications and a number of first defect assessment values ​​of the to-be-detected stages initially identified as defect identifications; The calculation formula of the defect evaluation value is: ; Among them, P is the defect evaluation value, u is the defect evaluation conversion coefficient, y1 is the weight coefficient of the defect risk evaluation value, is the selection coefficient of the defect risk assessment value of the c1th stage to be detected, when , =1, when , =0, is the defect risk assessment value of the c1th stage to be detected, is the defect risk assessment value threshold, is the weight coefficient of the c1th stage to be detected, m1 is the number of stages to be detected that are initially marked as unknown, m2 is the number of stages to be detected that are initially marked as defective, is the first defect evaluation value of the c2th stage to be detected, is the weight coefficient of the c2th stage to be detected.

[0014] In some embodiments of the present application, determining whether to generate a defect warning instruction based on the defect evaluation value includes: Pre-set defect evaluation value threshold; If the defect evaluation value is greater than the defect evaluation value threshold, a defect warning instruction is generated; The defect warning instruction includes several defect identification stages of the current product to be inspected, defect display process parameters corresponding to the stages to be inspected, and several unknown identification stages to be inspected, and risk display parameters corresponding to the stages to be inspected; If the defect evaluation value is not greater than the defect evaluation value threshold, no defect warning instruction is generated.

[0015] Compared with the prior art, the method for automatically detecting display process defects in the embodiment of the present application has the following advantages: By establishing a defect analysis model and generating an initial identification, when it is an unknown identification, a second defect detection strategy is generated and a defect evaluation value is generated. The display process parameters of each detection stage are accurately detected and defect evaluation is performed, thereby improving the defect detection accuracy of the product to be inspected and reducing the probability of subsequent display process defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for automatically detecting display process defects in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0018] In the description of this application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and 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, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation 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 the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0021] like Figure 1 As shown, a method for automatically detecting display process defects according to an embodiment of the present application includes: Set up several categories of testing products and establish defect analysis models based on the categories of testing products; Setting a first defect detection strategy for the product to be inspected, and obtaining a display process parameter set for the product to be inspected according to the first defect detection strategy; Analyze the display process parameter set according to the defect analysis model, and generate an initial identification according to the analysis result, wherein the initial identification includes an unknown identification, a normal identification, and a defect identification; When it is an unknown identifier, a second defect detection strategy is generated, a defect evaluation value is generated according to the second defect detection strategy, and whether to generate a defect warning instruction is determined according to the defect evaluation value.

[0022] In this embodiment, display process defects include but are not limited to production quality problems such as gluing, exposure, and development. Due to differences in working principles and manufacturing material systems, different display devices have significantly different types of display process defects. Therefore, they are divided into multiple inspection product categories, laying the foundation for subsequent improvement of display process defect detection accuracy.

[0023] In some embodiments of the present application, a defect analysis model is established according to the category of the inspected product, including: Collect basic information of multiple test products, classify the multiple test products according to the basic information, and obtain multiple test product categories; Extract historical inspection logs of the same inspection product category to obtain the historical display process parameter matrix of the corresponding inspection product category; The rows of the historical display process parameter matrix are several historical display process parameters in the same historical inspection log, which are listed as historical display process parameters of the same category, and each historical inspection log is mapped with a corresponding historical defect identifier, and each historical display process parameter is mapped with a corresponding historical display process image, and the historical defect identifier includes a defect category and a historical defect degree; Generate the test evaluation value of the corresponding test product category based on the historical display process parameter matrix; According to the test evaluation value, a number of division nodes corresponding to the test product category are set, and a number of test stages are generated based on the division nodes and the process stages corresponding to the test product category; Extract the historical display process parameter sub-matrix of each detection stage based on the historical display process parameter matrix, perform impact assessment analysis on the historical display process parameter sub-matrix, and generate a feature parameter set based on the analysis structure; The feature parameter set is set as the training data packet of the corresponding detection stage, and the defect analysis sub-model of the corresponding detection stage is constructed based on the training data packet of each detection stage; Setting the detection time interval of the corresponding detection phase according to the comprehensive change characteristics of the characteristic parameter set, and generating the sub-detection strategy of the corresponding detection phase in combination with the corresponding detection equipment; Set the defect analysis sub-model and sub-detection strategy for each inspection product category in turn; A defect analysis model is generated based on all defect analysis sub-models and sub-detection strategies.

[0024] In this embodiment, the larger the detection evaluation value, the greater the probability of quality problems occurring in the production process of the corresponding detection product category, the more likely the process parameters are to deviate, the larger the production quantity of the detection products, and the higher the production cost. Therefore, the more division nodes are set, the process stage is divided into multiple detection stages, and the degree of deviation of the process parameters in each detection stage is accurately detected, laying the foundation for the subsequent construction of the defect analysis model.

[0025] In this embodiment, the characteristic parameter set refers to the main process parameters that will cause defects in the corresponding detection stage and affect the defect type and defect degree. The training data packet of the corresponding detection stage is determined based on the characteristic parameter set, so as to construct a defect analysis model, lay the foundation for automatic defect detection of the product to be inspected, and improve the defect detection accuracy and detection efficiency.

[0026] In some embodiments of the present application, generating a test evaluation value corresponding to a test product category based on a historical display process parameter matrix includes: The calculation formula of the detection evaluation value is: ; Among them, J is the detection evaluation value, a1 is the first detection conversion coefficient, Q1 is the first weight coefficient, and n1 is the number of product evaluation indicators. is the reference evaluation value of the i1th product evaluation index generated based on the historical display process parameter matrix corresponding to the product category to be tested, is the weight coefficient of the i1th product evaluation index, a2 is the second detection conversion coefficient, Q2 is the second weight coefficient, n2 is the number of defect evaluation indicators, is the reference evaluation value of the i2th defect evaluation index generated based on the historical display process parameter matrix corresponding to the product category to be tested, is the weight coefficient of the i2th defect evaluation index.

[0027] In this embodiment, product evaluation indicators include but are not limited to the required cost of the product category, the mass production of the product category, the application coefficient of the product category, the number of product production stages, the production difficulty, etc., and defect evaluation indicators include but are not limited to the defect frequency of the product category and the degree of impact caused by the defect, the loss caused by the defect, and the service life after the defect, etc.

[0028] In this embodiment, the larger the detection evaluation value, the more likely it is that anomalies will occur in the production process of the corresponding detection product category, resulting in a high probability of defects. Therefore, the detection accuracy of the detection products of the corresponding detection product category should be higher to ensure defect recognition accuracy and recognition efficiency.

[0029] In some embodiments of the present application, an impact assessment analysis is performed on a historical display process parameter sub-matrix, and a characteristic parameter set is generated based on the analysis results, including: Performing cluster analysis on the historical defect identification mapped by each historical inspection log in the historical display process parameter submatrix of the same inspection stage to obtain multiple analysis matrices; The analysis matrix includes historical display process parameters from a number of historical inspection logs of different historical defect levels for the same defect category; Preset several standard display process parameters for each detection stage, compare the historical display process parameters in each analysis matrix with the corresponding standard display process parameters, and obtain the deviation historical display process parameters, continuous deviation duration, and corresponding deviation value for each row in each analysis matrix; Generate the deviation coefficient of the corresponding deviation history display process parameter based on the continuous deviation duration and the corresponding deviation value, and generate the historical defect coefficient of the corresponding defect category based on the historical defect degree; Map the deviation coefficient of each column of the same deviation history display process parameter in the same analysis matrix and the historical defect coefficient of the corresponding column to the same blank point diagram to obtain the deviation parameter-defect coefficient correlation diagram of the corresponding analysis matrix; Generate an evaluation coefficient for the impact of each deviation history display process parameter on the historical defect degree of the corresponding defect category based on the deviation parameter-defect coefficient correlation diagram; Pre-set impact assessment coefficient thresholds; Constructing a characteristic parameter set based on historical process parameters whose impact assessment coefficients in each analysis matrix of the same detection stage are greater than an impact assessment coefficient threshold; The characteristic parameter set includes a plurality of historical display process parameters, and each historical display process parameter is mapped to a corresponding historical display process image and a historical defect identifier.

[0030] In the embodiment, the longer the continuous deviation duration and the greater the corresponding deviation value, the greater the corresponding deviation coefficient, and vice versa.

[0031] In the embodiment, the greater the historical defect degree, the greater the corresponding historical defect coefficient, and vice versa.

[0032] In the embodiment, when the deviation of the historical display process parameter in the deviation parameter-defect coefficient correlation diagram causes the corresponding historical defect coefficient to change, and the greater the deviation coefficient, the greater the historical defect coefficient, the greater the impact evaluation coefficient, and vice versa.

[0033] In the embodiment, by constructing the characteristic parameter set, the historical display process parameters in the characteristic parameter set of each detection stage and the corresponding historical display process image are used as training input data, and the corresponding historical defect identifier is used as training output data. The neural network is trained to obtain a sub-defect analysis model, thereby improving the defect detection accuracy and detection efficiency of each detection stage.

[0034] In some embodiments of the present application, a first defect detection strategy of a product to be detected is set, including: obtaining basic information of the product to be detected; calculating the similarity between the product to be detected and each detection product category; The calculation formula of the similarity is: ; Wherein, D is the similarity, d0 is the similarity conversion coefficient, w is the number of characteristic evaluation indexes; is the weight coefficient of the s-th characteristic evaluation index; is the reference value of the s-th characteristic evaluation index generated for the product to be detected; is the reference value of the s-th characteristic evaluation index in the corresponding detection product category; The sub-detection strategy of the detection product category with the greatest similarity is set as the first defect detection strategy of the current product to be detected; The first defect detection strategy includes a plurality of detection stages of the product to be detected and the detection time interval and the corresponding detection equipment of the corresponding detection stage. Based on the first defect detection strategy, a display process parameter set corresponding to the product to be detected is obtained, and the display process parameter set includes a plurality of display process parameter sub-sets of the detection stages.

[0035] In the embodiment, the basic information includes, but is not limited to, working principle, manufacturing material system, display process parameter, production stage, and the like of the product to be monitored.

[0036] In the embodiment, the greater the similarity is, the more suitable the corresponding sub-detection and sub-defect analysis model of the product category to be detected is for the current product to be detected.

[0037] In some embodiments of the present application, the initial identification is set according to the analysis result, including: Based on the defect analysis model, the display process parameter sub-set of each stage to be detected is analyzed, and according to the analysis result, the several stages to be detected are divided into a settable identification stage and an un-settable identification stage; The initial identification of the settable identification stage is set as a defect identification, and the defect identification includes a current defect category and a current defect degree; A first defect evaluation value of the stage to be detected corresponding to the defect identification is generated according to the current defect category and the current defect degree; A second defect evaluation value of the stage to be detected corresponding to the un-settable identification stage is generated according to the display process parameter sub-set of the un-settable identification stage; A preset second defect evaluation value threshold is set in advance; If the second defect evaluation value is less than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as a normal identification; If the second defect evaluation value is greater than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as an unknown identification.

[0038] In the embodiment, the settable identification stage refers to the stage to be detected in which the defect type and the corresponding defect degree can be directly determined according to the defect analysis model, and the un-settable identification stage refers to the stage to be detected in which whether the abnormality exists cannot be accurately judged according to the defect analysis model.

[0039] In the embodiment, the first defect evaluation value is obtained by comparing the display process parameter sub-set of the un-settable identification stage with the standard display process sub-set of the corresponding stage to be detected set in advance, and the smaller the parameter difference between the display process parameter in the display process parameter sub-set and the corresponding standard display process parameter in the standard display process sub-set is, the smaller the corresponding first defect evaluation value is, and vice versa.

[0040] In the embodiment, the identification of each stage to be detected is judged by the defect analysis model, the several stages to be detected with defect identifications are accurately identified, the defect evaluation value of the remaining stages to be detected is calculated, the several stages to be detected with normal identifications are accurately identified, and the detection accuracy of each stage to be detected of the product to be detected is improved.

[0041] In some embodiments of the present application, before generating the second defect detection strategy, the method further includes: Construct an association model for the stage to be detected, which is initially identified as a defect identifier, and extract the defect category and the corresponding defect degree corresponding to the stage to be detected; Generate the defect category of the corresponding to-be-detected stage and the stage to-be-detected affected by the corresponding defect degree and the degree of influence based on the association model, and set the stage to-be-detected with an influence degree greater than a preset influence degree threshold as the associated to-be-detected stage of the corresponding to-be-detected stage; Determine whether the initial identifiers of the associated to-be-detected stages whose initial identifier is a defect identifier are defect identifiers. If not, replace them with defect identifiers.

[0042] In this embodiment, the associated stage to be detected refers to the stage to be detected that is greatly affected by the defect type and defect degree of the corresponding stage to be detected, that is, the association model is constructed based on other stages to be detected that must also have defects due to the different defect types and defect degrees of the current stage to be detected, and is set according to the actual defect type and the corresponding defect degree.

[0043] In some embodiments of the present application, generating a second defect detection strategy includes: Screening out the stages to be inspected that are initially identified as unknown, as well as the displayed process parameters corresponding to the stages to be inspected and the corresponding process parameter images; Analyze the process parameter image of the current stage to be inspected based on the image-parameter mapping table, and determine the expected display process parameter range mapped by the corresponding process parameter image according to the analysis result; Determine whether the display process parameter corresponding to the process parameter image is within the expected display process parameter range. If so, set the corresponding display process parameter as a credible parameter; if not, set the corresponding display process parameter as an untrustworthy parameter. Collect several untrustworthy parameters in the current detection stage and perform cluster analysis, determine the parameter categories of several untrustworthy parameters based on the analysis results, and configure corresponding preset detection equipment and corresponding preset detection time intervals for corresponding parameter categories; generating a second defect detection strategy based on a preset detection device and a corresponding preset detection time interval; Untrusted parameters are collected according to the second defect detection strategy until the untrusted parameters are within the corresponding expected process parameter range, and the corresponding display process parameters are replaced.

[0044] In this embodiment, the optimal detection device and the optimal detection time interval for each parameter category are set in advance. The optimal detection device refers to the detection device with the highest detection accuracy for the corresponding parameter category. The optimal detection device and the optimal detection time interval are set as the preset detection device and the preset detection time interval.

[0045] In the embodiment, the image-parameter mapping table is constructed according to the historical process parameter images and the historical display process parameters with higher accuracy, the process parameter images are analyzed based on the image-parameter mapping table, the consistency degree of the process parameter images and the historical process parameter images is obtained, and the historical display process parameters corresponding to the historical process parameter images with the consistency degree greater than a preset degree threshold are constructed into the expected display process parameter interval.

[0046] In the embodiment, a simulation model of the to-be-detected stage is established, and the simulation is performed on the continuous process parameter images, the untrusted parameters are collected according to the second defect detection strategy, and the untrusted parameters are replaced, so as to improve the detection accuracy of the to-be-detected stage.

[0047] In some embodiments of the present application, the defect evaluation value is generated according to the second defect detection strategy, including: generating a defect risk evaluation value of the to-be-detected stage according to the trusted parameters of the to-be-detected stage initially identified as unknown identification and the replaced display process parameters; generating a defect risk evaluation value of the to-be-detected stage according to the defect risk evaluation value of the to-be-detected stage initially identified as unknown identification and the first defect evaluation value of the to-be-detected stage initially identified as a defect identification; The calculation formula of the defect evaluation value is: ; Wherein, P is the defect evaluation value, u is the defect evaluation conversion coefficient, y1 is the weight coefficient of the defect risk evaluation value, is the selection coefficient of the defect risk evaluation value of the c1th to-be-detected stage, when , =1, when , =0, is the defect risk evaluation value of the c1th to-be-detected stage, is the defect risk evaluation value threshold, is the weight coefficient of the c1th to-be-detected stage, m1 is the number of the to-be-detected stages initially identified as unknown identification, and m2 is the number of the to-be-detected stages initially identified as a defect identification, is the first defect evaluation value of the c2th to-be-detected stage, is the weight coefficient of the c2th to-be-detected stage.

[0048] In some embodiments of the present application, whether a defect warning instruction is generated is determined according to the defect evaluation value, including: a defect evaluation value threshold is preset; if the defect evaluation value is greater than the defect evaluation value threshold, a defect warning instruction is generated; The defect warning instruction includes several defect identification stages of the current product to be inspected, defect display process parameters corresponding to the stages to be inspected, and several unknown identification stages to be inspected, and risk display parameters corresponding to the stages to be inspected; If the defect evaluation value is not greater than the defect evaluation value threshold, no defect warning instruction is generated.

[0049] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and replacements can be made without departing from the technical principles of the present application. These improvements and replacements should also be regarded as the scope of protection of the present application.

Claims

1. A method for automatically detecting display process defects, characterized in that: include: Set up several categories of testing products and establish defect analysis models based on the categories of testing products; Setting a first defect detection strategy for the product to be inspected, and obtaining a display process parameter set for the product to be inspected according to the first defect detection strategy; Analyze the display process parameter set according to the defect analysis model, and generate an initial identification according to the analysis result, wherein the initial identification includes an unknown identification, a normal identification, and a defect identification; When it is an unknown identifier, a second defect detection strategy is generated, a defect evaluation value is generated according to the second defect detection strategy, and whether to generate a defect warning instruction is determined according to the defect evaluation value.

2. The method for automatically detecting display process defects according to claim 1, wherein: Establish defect analysis models based on the product category being tested, including: Collect basic information of multiple test products, classify the multiple test products according to the basic information, and obtain multiple test product categories; Extract historical inspection logs of the same inspection product category to obtain the historical display process parameter matrix of the corresponding inspection product category; The rows of the historical display process parameter matrix are several historical display process parameters in the same historical inspection log, which are listed as historical display process parameters of the same category, and each historical inspection log is mapped with a corresponding historical defect identifier, and each historical display process parameter is mapped with a corresponding historical display process image, and the historical defect identifier includes a defect category and a historical defect degree; Generate the test evaluation value of the corresponding test product category based on the historical display process parameter matrix; According to the test evaluation value, a number of division nodes corresponding to the test product category are set, and a number of test stages are generated based on the division nodes and the process stages corresponding to the test product category; Extract the historical display process parameter sub-matrix of each detection stage based on the historical display process parameter matrix, perform impact assessment analysis on the historical display process parameter sub-matrix, and generate a feature parameter set based on the analysis structure; The feature parameter set is set as the training data packet of the corresponding detection stage, and the defect analysis sub-model of the corresponding detection stage is constructed based on the training data packet of each detection stage; Setting the detection time interval of the corresponding detection phase according to the comprehensive change characteristics of the characteristic parameter set, and generating the sub-detection strategy of the corresponding detection phase in combination with the corresponding detection equipment; Set the defect analysis sub-model and sub-detection strategy for each inspection product category in turn; A defect analysis model is generated based on all defect analysis sub-models and sub-detection strategies.

3. The method for automatically detecting display process defects according to claim 2, wherein: Generates test evaluation values ​​for corresponding test product categories based on the historical display process parameter matrix, including: The calculation formula of the detection evaluation value is: ; Among them, J is the detection evaluation value, a1 is the first detection conversion coefficient, Q1 is the first weight coefficient, and n1 is the number of product evaluation indicators. is the reference evaluation value of the i1th product evaluation index generated based on the historical display process parameter matrix corresponding to the product category to be tested, is the weight coefficient of the i1th product evaluation index, a2 is the second detection conversion coefficient, Q2 is the second weight coefficient, n2 is the number of defect evaluation indicators, is the reference evaluation value of the i2th defect evaluation index generated based on the historical display process parameter matrix corresponding to the product category to be tested, is the weight coefficient of the i2th defect evaluation index.

4. The method for automatically detecting display process defects according to claim 3, wherein: Perform impact assessment analysis on the historical display process parameter sub-matrix and generate a set of characteristic parameters based on the analysis results, including: Performing cluster analysis on the historical defect identification mapped by each historical inspection log in the historical display process parameter submatrix of the same inspection stage to obtain multiple analysis matrices; The analysis matrix includes historical display process parameters from a number of historical inspection logs of different historical defect levels for the same defect category; Preset several standard display process parameters for each detection stage, compare the historical display process parameters in each analysis matrix with the corresponding standard display process parameters, and obtain the deviation historical display process parameters, continuous deviation duration, and corresponding deviation value for each row in each analysis matrix; Generate the deviation coefficient of the corresponding deviation history display process parameter based on the continuous deviation duration and the corresponding deviation value, and generate the historical defect coefficient of the corresponding defect category based on the historical defect degree; Map the deviation coefficient of each column of the same deviation history display process parameter in the same analysis matrix and the historical defect coefficient of the corresponding column to the same blank point diagram to obtain the deviation parameter-defect coefficient correlation diagram of the corresponding analysis matrix; Generate an evaluation coefficient for the impact of each deviation history display process parameter on the historical defect degree of the corresponding defect category based on the deviation parameter-defect coefficient correlation diagram; Pre-set impact assessment coefficient thresholds; Constructing a characteristic parameter set based on historical process parameters whose impact assessment coefficients in each analysis matrix of the same detection stage are greater than an impact assessment coefficient threshold; The characteristic parameter set includes several historical display process parameters, and each historical display process parameter is mapped to a corresponding historical display process image and a historical defect identifier.

5. The method for automatically detecting display process defects according to claim 4, wherein: Set the first defect detection strategy for the product to be inspected, including: Obtain basic information about the product to be tested; Calculate the similarity between the product to be tested and each test product category; The calculation formula of the similarity is: ; Among them, D is the similarity, d0 is the similarity conversion coefficient, and w is the number of feature evaluation indicators; is the weight coefficient of the sth feature evaluation index; The reference value of the sth feature evaluation index generated for the product set to be tested; is the reference value of the sth characteristic evaluation index in the corresponding test product category; Setting the sub-detection strategy of the detection product category with the greatest similarity as the first defect detection strategy for the current product to be detected; The first defect detection strategy includes several stages to be detected of the product to be detected, detection time intervals corresponding to the stages to be detected, and corresponding detection equipment; A display process parameter set corresponding to the product to be inspected is obtained based on the first defect detection strategy, where the display process parameter set includes a plurality of display process parameter subsets of the stages to be inspected.

6. The method for automatically detecting display process defects according to claim 5, wherein: Set initial identification based on the analysis results, including: Analyze a subset of display process parameters of each stage to be inspected based on the defect analysis model, and divide the stages to be inspected into stages with settable identification and stages without settable identification according to the analysis results; Setting the initial identification of the settable identification stage as the defect identification, wherein the defect identification includes the current defect category and the current defect severity; Generate a first defect evaluation value corresponding to the defect identification stage to be detected according to the current defect category and the current defect degree; Generating a second defect evaluation value corresponding to the stage to be inspected according to the displayed process parameter subset of the stage that cannot be marked as settable; Presetting a preset second defect evaluation value threshold; If the second defect evaluation value is less than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as a normal identification; If the second defect evaluation value is greater than the preset second defect evaluation value threshold, the initial identification of the corresponding stage to be detected is set as an unknown identification.

7. The method for automatically detecting display process defects according to claim 6, wherein: Before generating the second defect detection strategy, the following steps are also included: Construct an association model for the stage to be inspected, which is initially identified as a defect identifier, and extract the defect category and the corresponding defect degree corresponding to the stage to be inspected; Generate the defect category of the corresponding to-be-detected stage and the stage to-be-detected affected by the corresponding defect degree and the degree of influence based on the association model, and set the stage to-be-detected with an influence degree greater than a preset influence degree threshold as the associated to-be-detected stage of the corresponding to-be-detected stage; Determine whether the initial identifiers of the associated to-be-detected stages whose initial identifier is a defect identifier are defect identifiers. If not, replace them with defect identifiers.

8. The method for automatically detecting display process defects according to claim 7, wherein: Generate a second defect detection strategy, including: Screening out the stages to be inspected that are initially identified as unknown, as well as the displayed process parameters corresponding to the stages to be inspected and the corresponding process parameter images; Analyze the process parameter image of the current stage to be inspected based on the image-parameter mapping table, and determine the expected display process parameter range mapped by the corresponding process parameter image according to the analysis result; Determine whether the display process parameter corresponding to the process parameter image is within the expected display process parameter range. If so, set the corresponding display process parameter as a credible parameter; if not, set the corresponding display process parameter as an untrustworthy parameter. Collect several untrustworthy parameters in the current detection stage and perform cluster analysis, determine the parameter categories of several untrustworthy parameters based on the analysis results, and configure corresponding preset detection equipment and corresponding preset detection time intervals for corresponding parameter categories; generating a second defect detection strategy based on a preset detection device and a corresponding preset detection time interval; Untrusted parameters are collected according to the second defect detection strategy until the untrusted parameters are within the corresponding expected process parameter range, and the corresponding display process parameters are replaced.

9. The method for automatically detecting display process defects according to claim 8, wherein: Generating a defect evaluation value according to the second defect detection strategy includes: Generate a defect risk assessment value corresponding to the stage to be inspected based on the trusted parameters of the stage to be inspected that is initially identified as unknown and the replaced display process parameters; Generate a defect assessment value based on a number of defect risk assessment values ​​of the to-be-detected stages initially identified as unknown identifications and a number of first defect assessment values ​​of the to-be-detected stages initially identified as defect identifications; The calculation formula of the defect evaluation value is: ； Among them, P is the defect evaluation value, u is the defect evaluation conversion coefficient, y1 is the weight coefficient of the defect risk evaluation value, is the selection coefficient of the defect risk assessment value of the c1th stage to be detected, when , =1, when , =0, is the defect risk assessment value of the c1th stage to be detected, is the defect risk assessment value threshold, is the weight coefficient of the c1th stage to be detected, m1 is the number of stages to be detected that are initially marked as unknown, m2 is the number of stages to be detected that are initially marked as defective, is the first defect evaluation value of the c2th stage to be detected, is the weight coefficient of the c2th stage to be detected.

10. The method for automatically detecting display process defects according to claim 9, wherein: Determine whether to generate a defect warning instruction based on the defect evaluation value, including: Pre-set defect evaluation value threshold; If the defect evaluation value is greater than the defect evaluation value threshold, a defect warning instruction is generated; The defect warning instruction includes several defect identification stages of the current product to be inspected, defect display process parameters corresponding to the stages to be inspected, and several unknown identification stages to be inspected, and risk display parameters corresponding to the stages to be inspected; If the defect evaluation value is not greater than the defect evaluation value threshold, no defect warning instruction is generated.

Images