Defect quality diagnosis device and poor quality diagnosis method

Abstract

A quality defect diagnosis device for diagnosing quality defects occurring in a product is provided with: a data division unit for two-dimensionally dividing product data to be diagnosed, said product data being composed of job data and quality data, for each predetermined block; a data collection unit that extracts job data for each predetermined classification of quality defects on the basis of the tendency of occurrence of quality defects in the divided blocks, and collects the extracted job data for each block and for each region including two or more blocks; a quality defect determination unit that determines the presence or absence of occurrence of quality defects using a plurality of prediction models generated for each classification of quality defects and having learned the relationship between the aggregated job data and quality data; and a main cause estimation unit that estimates the main cause of the quality defect on the basis of the determination result of the quality defect.

Description

Technical Field

[0001] This invention relates to a device and method for diagnosing defects in quality. Background Technology

[0002] In Patent Document 1, as a method for predicting and diagnosing the quality of a product, the following method is disclosed: considering the quality determination location on the product, the working condition data during product manufacturing is associated with the quality determination data in the final process, and a prediction model learned from the data is used to predict the quality.

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent No. 6953990 Summary of the Invention

[0006] The problem that the invention aims to solve

[0007] However, when predicting product quality, if quality data is used uniformly without relying on its characteristics, information related to the various quality defects arising from different mechanisms is missing. Therefore, even if a predictive model is generated based on such quality data, it may still fail to achieve sufficient predictive performance.

[0008] The present invention was made in view of the above circumstances, and its object is to provide a defect diagnosis device and a defect diagnosis method that can accurately predict product defects and presume the main causes of such defects.

[0009] Technical solutions for solving the problem

[0010] To address the aforementioned issues and achieve the objectives, the present invention relates to a defect diagnosis device for diagnosing defects generated in a product. The device comprises: a data segmentation unit that segments product data (consisting of work data and quality data) into two-dimensional blocks; a data aggregation unit that extracts work data according to a predetermined defect classification based on the tendency of defects to occur in the segmented blocks, and aggregates the extracted work data according to each block and each region containing two or more blocks; a defect determination unit that uses multiple prediction models generated according to each defect classification and learned the relationship between the aggregated work data and the quality data to determine whether a defect has occurred; and a primary cause estimation unit that estimates the primary cause of the defect based on the defect determination result.

[0011] Furthermore, in the defect diagnosis device of the present invention, the main cause estimation unit estimates the work data required to suppress the occurrence of the defect when it determines that a specific product that is the subject of diagnosis has a defect.

[0012] Furthermore, in the defect diagnosis device of the present invention, the main cause estimation unit estimates the operational data as the main cause of the defect when a specific product has a defect.

[0013] Furthermore, in the defect diagnosis device of the present invention, the main cause estimation unit estimates the work data as the main cause of the defect when defective products occur in multiple products.

[0014] Furthermore, in the defect diagnosis device of the present invention, the prediction model is generated as follows: product data of past performance consisting of work data and quality data is divided into two-dimensional blocks according to each specified block; based on the tendency of the defects to occur in the divided blocks, the defects contained in the quality data are classified; the work data extracted according to each defect classification is aggregated according to each block and each region containing two or more blocks; and the relationship between the aggregated work data and the quality data is learned by machine learning according to each defect classification.

[0015] Furthermore, in the defect diagnosis device of the present invention, the defect classification unit classifies the defect into any two or more of the following: front end, tail end, central part, width edge part, single random, and group random.

[0016] Furthermore, in the defective quality diagnostic device of the present invention, the data collection unit calculates one or more of the average value, maximum value, minimum value, standard deviation, variance, and the difference between the maximum and minimum values ​​of the work data in block units or product units, thereby collecting the work data.

[0017] Furthermore, the defect diagnostic device of the present invention, in the above invention, obtains the correlation between the quality data and a template pre-generated according to the tendency of each defect to occur, and classifies the defects based on the magnitude of the correlation.

[0018] To address the aforementioned issues and achieve the objectives, the present invention relates to a defect diagnosis method for diagnosing quality defects occurring in a product. The defect diagnosis method includes: a data segmentation step, wherein a computer-equipped data segmentation unit segments product data (consisting of work data and quality data) into two-dimensional blocks; a data aggregation step, wherein the computer-equipped data aggregation unit extracts work data according to each predetermined defect classification based on the tendency of defect occurrence in the segmented blocks, and aggregates the extracted work data according to each block and each region containing two or more blocks; a defect determination step, wherein the computer-equipped defect determination unit uses multiple prediction models generated according to each defect classification and having learned the relationship between the aggregated work data and the quality data to determine whether a quality defect has occurred; and a primary cause estimation step, wherein the computer-equipped primary cause estimation unit estimates the primary cause of the quality defect based on the defect determination result.

[0019] Invention Effects

[0020] According to the defect diagnosis device and method of the present invention, the relationship between work data and defect is generated and used as a prediction model according to each classification corresponding to the tendency of defect generation, thereby enabling high-precision prediction of product defect and estimation of the main cause of the defect. Attached Figure Description

[0021] Figure 1 This is a diagram showing the schematic structure of the information processing device for implementing the defect diagnosis device according to the embodiments of the present invention.

[0022] Figure 2 This is a diagram illustrating in detail the data segmentation steps performed by the data segmentation unit of the defect diagnostic device according to an embodiment of the present invention.

[0023] Figure 3 This is a diagram illustrating in detail the quality defect classification steps performed by the quality defect classification unit of the quality defect diagnostic device according to an embodiment of the present invention.

[0024] Figure 4 This is a diagram illustrating the collection of work data from the front end during the data collection step performed by the data collection unit of the defect diagnostic device according to an embodiment of the present invention.

[0025] Figure 5This is a diagram illustrating the collection of work data at the tail end during the data collection step performed by the data collection unit of the defect diagnosis device according to an embodiment of the present invention.

[0026] Figure 6 This is a diagram illustrating the collection of operational data in the central section during the data collection step performed by the data collection unit of the defect diagnostic device according to an embodiment of the present invention.

[0027] Figure 7 This is a diagram illustrating the collection of work data at the width edge during the data collection step performed by the data collection unit of the defect diagnosis device according to an embodiment of the present invention.

[0028] Figure 8 This is a diagram illustrating the collection of single-shot random and group-shot random work data in the data collection step performed by the data collection unit of the defect diagnosis device according to an embodiment of the present invention.

[0029] Figure 9 This is a flowchart illustrating the process of generating a prediction model used in the defect diagnosis method according to an embodiment of the present invention.

[0030] Figure 10 This is a flowchart illustrating the process of a method for diagnosing quality defects according to an embodiment of the present invention. Detailed Implementation

[0031] The defect diagnosis apparatus and method according to embodiments of the present invention will be described with reference to the accompanying drawings. The defect diagnosis apparatus and method are used to diagnose defects in a product. In this embodiment, a "product" can be exemplified by a steel product (e.g., a coil) manufactured through multiple processes. Furthermore, "defects" in this embodiment include, for example, surface defects, internal defects, and material characteristic values ​​(e.g., the determination result of mechanical properties).

[0032] [Information processing device]

[0033] Figure 1 This describes the structure of the information processing device 1, which implements the defect diagnosis device according to the embodiment. For example... Figure 1 As shown, the information processing device 1 includes an input unit 10, a storage unit 20, an arithmetic unit 30, and an output unit 40.

[0034] The defect diagnosis device according to the embodiment can be implemented using structural elements of the information processing device 1 other than the defect classification unit 32 and the model generation unit 34 of the calculation unit 30. Furthermore, the device for generating the prediction model used in the defect diagnosis device according to the embodiment can be implemented using structural elements of the information processing device 1 other than the defect determination unit 35 and the main cause estimation unit 36 ​​of the calculation unit 30. The structural elements of the information processing device 1 will be described below.

[0035] The input unit 10 is an input unit for the arithmetic unit 30, implemented, for example, through an input device such as a keyboard, mouse pointer, or number keys. The input unit 10 inputs information required for various processing operations in the arithmetic unit 30.

[0036] Storage unit 20 comprises recording media such as EPROM (Erasable Programmable ROM), hard disk drive (HDD), and removable media. Examples of removable media include USB (Universal Serial Bus) storage, CD (Compact Disc), DVD (Digital Versatile Disc), and BD (Blu-ray Disc). Storage unit 20 can store operating systems (OS), various programs, various tables, various databases, etc. Product database 21 is stored in storage unit 20.

[0037] Product DB21 stores past manufacturing data. This product data includes operational data and quality data. "Operational data" measures the equipment operating conditions, product condition, or data inferred through some means during product manufacturing. "Quality data" is data indicating the quality of a product based on measurements or inferences made through some means. Additionally, the aforementioned product data includes information used to generate predictive models (see [reference]). Figure 9 The product data used is based on "past performance," and when diagnosing quality defects (refer to...). Figure 10 The product data used for the "diagnostic object".

[0038] In addition to product DB21, the storage unit 20 also stores, for example, product data corresponding to each block through the data segmentation unit 31, information related to the classification of quality defects by the quality defect classification unit 32, and work data aggregated by block unit or product unit through the data aggregation unit 33. Furthermore, the storage unit 20 may also store the quality defect determination results by the quality defect judgment unit 35, and the main cause estimation results of quality defects by the main cause estimation unit 36.

[0039] The arithmetic unit 30 is implemented, for example, by a processor consisting of a CPU (Central Processing Unit) and a memory consisting of RAM (Random Access Memory) and ROM (Read Only Memory) (main storage unit).

[0040] The arithmetic unit 30 loads the program into the working area of ​​the main storage unit and executes it. Through program execution, it controls various structural units, thereby achieving the intended function. The arithmetic unit 30 functions as the data segmentation unit 31, the defect classification unit 32, the data collection unit 33, the model generation unit 34, the defect determination unit 35, and the primary cause estimation unit 36 ​​through the execution of the aforementioned program. Furthermore, in... Figure 1 The example shown is an example of implementing the functions of each part through a single computer (computation unit), but the means of implementing the functions of each part are not particularly limited. For example, the functions of each part can also be implemented separately through multiple computers.

[0041] The data segmentation unit 31 divides the product data into two-dimensional blocks according to each defined area. For example, the data segmentation unit 31... Figure 2 As shown, the product data is divided into p blocks along its length and q blocks along its width. In this diagram, the leftmost block along the product length is designated as the first block, and the rightmost block as the qth block. Similarly, the topmost block along the product width is designated as the first block, and the bottommost block as the pth block. The data segmentation unit 31 segments the product data, composed of work data and quality data, along with... Figure 2 Each block obtained by dividing the product in that way is correspondingly stored in the storage unit 20.

[0042] When generating the prediction model, the data segmentation unit 31 (refer to...) Figure 9 The data segmentation unit 31 performs segmentation of product data based on past performance. Additionally, when performing quality defect diagnosis (refer to...), the data segmentation unit 31... Figure 10The product data for the diagnostic target is segmented. Furthermore, when generating the predictive model and performing defective quality diagnosis, it is preferable to use the same data segmentation method (number of segments, area of ​​each block, etc.).

[0043] The number of segments (values ​​of q and p) for product data is not particularly limited, but at least three segments are preferred (values ​​of q and p > 3). Furthermore, the area of ​​each segmented block can be consistent for each product data item, or it can be different. For example, in the classification of quality defects described later (refer to...). Figure 3 In the context of data, blocks classified as width edge or central area can also be areas that vary depending on the data for each product.

[0044] The quality defect classification unit 32 classifies the quality defects contained in the quality data into two or more categories based on the tendency of quality defects to occur in the segmented blocks. Here, "tendency of quality defects to occur" refers to, for example, the segmented blocks (see reference 32). Figure 1 The location and distribution of quality defects in the product.

[0045] Specifically, the quality defect classification unit 32 classifies quality defects in the quality data into defects occurring at specific locations along the product's length (e.g., the front end, rear end, center, and width edge) and defects occurring singly or in clusters along the entire length and width of the product. For example, the quality defect classification unit 32... Figure 2 As shown, the quality defects in the quality data are classified into any two or more of the following categories: front end, tail end, central part, width edge part, single random, and group random.

[0046] That is, when the quality defect classification unit 32 divides the product into blocks along its length, if a quality defect occurs in the first block starting from the beginning of the product's length, the defect is classified as "front end". Furthermore, when the quality defect classification unit 32 divides the product into blocks along its length, if a quality defect occurs in the p-th block starting from the beginning of the product's length, the defect is classified as "tail end".

[0047] Furthermore, regarding the block division along the product length direction, if a quality defect occurs in any block from the second to the p-1th block from the beginning of the product length direction, and in any block from the second to the q-1th block from the beginning of the product width direction, the quality defect is classified as "central area". Additionally, regarding the block division along the product width direction, if a quality defect occurs in the first or qth block, the quality defect is classified as "width edge area".

[0048] Furthermore, when the quality defect classification unit 32 divides the product into blocks along the length and width directions, it classifies a single defect as "single random" if a block contains only one defect. Conversely, when the quality defect classification unit 32 divides the product into blocks along the length and width directions, it classifies a block containing two or more defects as "group random".

[0049] Furthermore, where p is an integer greater than or equal to 3, and q is an integer greater than or equal to 3, the values ​​of p and q are arbitrarily determined beforehand. In this case, the quality defects for each block classification can be repeated in more than two ways, or the classification criteria can be assigned a priority order (e.g., front end, tail end > width edge, etc.) to limit it to a single type.

[0050] For example, in Figure 3 In the defective quality classification shown, the front end, tail end, and central portion are not repeated, nor are the central portion and the width edge portion. However, the front end and the width edge portion, and the tail end and the width edge portion, may overlap. Furthermore, single-shot random and multi-shot random do not overlap with each other, but single-shot random or multi-shot random may overlap with other classifications (front end, tail end, central portion, width edge). The defective quality classification section 32 will be as follows... Figure 3 The information related to the classification of defective products shown is stored in the storage unit 20.

[0051] When generating the prediction model, the quality defect classification department 32 (refer to) Figure 9 This involves categorizing past performance quality data. Furthermore, when diagnosing quality defects (refer to...),... Figure 10 The operational data of the diagnostic objects are collected by using the classification of quality defects generated by the prediction model.

[0052] Here, the defect classification unit 32 can also obtain the correlation between quality data and a template pre-generated according to the tendency of each defect to occur, and classify the defects based on the magnitude of the correlation. In this case, a template for each defect classification (front end, tail end, central part, width edge part, single random, group random) is prepared in advance, and the defect classification associated with the template with the highest correlation is adopted.

[0053] As a template, one could cite, for example, the distribution of defect locations based on the marking and statistical analysis of past product defect data. In this case, within the product being diagnosed, the type of defect that corresponds to the template type with the highest occurrence rate in the block containing the defective area can be identified. By using this method, defect classification can be performed quickly on large amounts of product data.

[0054] The data collection unit 33 collects operational data by block unit or product unit. The data collection unit 33 first categorizes the data by defective quality (see reference). Figure 3 The process involves extracting job data. The defect classification used here is the one assigned by the defect classification unit 32 during the generation of the prediction model. Next, the data aggregation unit 33 aggregates the extracted job data according to each block and each region containing two or more of those blocks. The data aggregation unit 33 aggregates the job data as follows, for example.

[0055] Data collection section 33, for example, Figure 4 As shown, for quality defects classified as "front end", only the operation data of the first block from the beginning of p blocks along the product's length direction is extracted, and the extracted operation data is aggregated in block units. Thus, for the front end, since it is a part particularly prone to quality defects, the operation data is aggregated in block units.

[0056] In addition, the data collection unit 33, for example, Figure 5 As shown, for quality defects classified as "tail end", only the operation data of the p-th block from the beginning is extracted from the p blocks along the product length direction, and the extracted operation data is aggregated in block units. Thus, the tail end, like the front end, is a part that is particularly prone to quality defects, so the operation data is aggregated in block units.

[0057] In addition, the data collection unit 33, for example, Figure 6 As shown, for quality defects classified as "central section," only the work data for blocks from the second to the p-1th block along the product's length and from the second to the q-1th block along the product's width are extracted. Then, the data collection unit 33 collects the extracted work data by product unit. That is, in this diagram, the block enclosed by the thick line is considered as one data point.

[0058] In addition, the data collection unit 33, for example, Figure 7 As shown, for quality defects classified as "width edge," the operation data for all blocks in the first length direction along the product width and the operation data for all blocks in the qth length direction along the product width are extracted. Then, the data aggregation unit 33 aggregates the extracted operation data by product unit. That is, in this figure, the block enclosed by the thick line is considered as one data point.

[0059] In addition, the data collection unit 33, for example, Figure 8 As shown, for quality defects classified as "single-shot random", the job data of blocks that conform to this single-shot randomness and the job data of blocks that did not produce quality defects are extracted. Then, the extracted job data are aggregated in blocks.

[0060] In addition, the data collection unit 33, for example, Figure 8 As shown, for quality defects classified as "mass random," job data for blocks that conform to this mass randomness and job data for blocks that did not produce quality defects are extracted. Then, the extracted job data are aggregated in blocks.

[0061] As a method for data aggregation in the data aggregation unit 33, basic statistics representing the values ​​of each position within a block or product can be used. Specifically, these are values ​​such as the mean, maximum, minimum, standard deviation, variance, and the range represented by the difference between the maximum and minimum values. Thus, the data aggregation unit 33 calculates one or more of the mean, maximum, minimum, standard deviation, variance, and the difference between the maximum and minimum values ​​for the work data in block units or product units, thereby aggregating the work data. Furthermore, while at least one basic statistic is used when a work data is aggregated, multiple basic statistics can also be used. The data aggregation unit 33 stores the work data aggregated in block units or product units in the storage unit 20.

[0062] When generating the prediction model, the data collection unit 33 (refer to...) Figure 9 The data collection unit 33 collects past performance data. Additionally, when performing quality defect diagnosis (see...), the data collection unit 33... Figure 10 ), to collect operational data of the diagnostic objects.

[0063] The model generation unit 34 generates multiple prediction models corresponding to the categories of defective products. For each category of defective product, the model generation unit 34 uses machine learning to learn the relationship between the aggregated job data and quality data, thereby generating multiple prediction models corresponding to the categories of defective products.

[0064] That is, the model generation unit 34 constructs a predictive model representing the relationship between work data aggregated in block units or product units and the quality defects generated in the corresponding blocks or products, according to the classification of quality defects. For example, the model generation unit 34 generates six predictive models for quality defects at the beginning, end, center, width edge, single random occurrence, and cluster random occurrence. In generating these predictive models, various machine learning algorithms, such as deep neural networks, decision trees, and support vector machines, as well as other statistical algorithms, can be used.

[0065] The defect determination unit 35 uses multiple prediction models generated by the model generation unit 34 to determine whether a defect has occurred. The defect determination unit 35 determines whether a defect has occurred in each category by inputting the work data of the diagnostic object into the prediction model for each category. Here, the work data input into the prediction model is data that has been segmented in the data segmentation unit 31 and aggregated in the data aggregation unit 33, and aggregated in block units or product units (see [reference]). Figure 10 ).

[0066] The quality defect determination performed by the quality defect determination unit 35 is carried out, for example, before or during product manufacturing. When performed before manufacturing, the set values ​​of the pre-processing manufacturing conditions are input into the prediction model as work data for the product to be predicted, thereby predicting quality defects in the manufactured product. Conversely, when performed during manufacturing, the actual values ​​of the completed manufacturing conditions and the set values ​​of the pre-processing manufacturing conditions are input into the prediction model as work data for the product to be predicted, thereby predicting quality defects in the manufactured product.

[0067] When the output unit 40 outputs the quality defect determination results performed by the quality defect determination unit 35, for example, on a screen where information related to the product being diagnosed can be viewed, a product top view categorized by product unit or block unit is displayed according to the quality defect classification. Furthermore, a circle (○) can be marked on the corresponding part of the product top view if there is no predicted quality defect, and an × mark can be marked if there is a predicted quality defect. Alternatively, different colors can be used to display the results depending on whether there is a predicted quality defect. Alternatively, only when there is a predicted quality defect, the output unit 40 can output a statement such as "The product being diagnosed (or a specific block of the product being diagnosed) is predicted to have a quality defect."

[0068] In the defect determination unit 35, the work data of the entire input diagnostic object is used to determine the prediction models for the front end, rear end, central part, width edge, single random, and group random defects. However, it is also possible to input only a portion of the work data of the diagnostic object into the prediction models. For example, if the work data of the diagnostic object is for the "corner," that is, the data of the first block from the beginning of the product length direction and the first block from the beginning of the product width direction, the resulting defect is limited to any one of the front end, width edge, single random, and group random defects. Therefore, in this case, it is sufficient to input only the work data of the diagnostic object into the prediction models for the front end, width edge, single random, and group random defects.

[0069] The main cause estimation unit 36 ​​estimates the main cause of the quality defect based on the quality defect determination results performed by the quality defect determination unit 35. The "main cause of the quality defect" refers to, for example, operational data that is an important factor in the quality defect of the product.

[0070] The primary cause estimation unit 36, for example, can estimate the operational data needed to suppress the occurrence of a quality defect in a specific product identified as a diagnostic target, when such a defect is detected. That is, if a quality defect is predicted to occur in a product before or during manufacturing, the operational data that needs to be changed / adjusted is estimated to prevent the occurrence of such a defect. Therefore, it is possible to study countermeasures to suppress the occurrence of quality defects.

[0071] Furthermore, the primary cause estimation unit 36 ​​can estimate the work data that is the primary cause of a quality defect in a specific product. That is, it estimates the work data that is the primary cause of a quality defect when it exists in a manufactured product. Therefore, the primary cause of the quality defect can be determined.

[0072] Furthermore, the primary cause estimation unit 36 ​​can estimate the work data that is the main cause of quality defects when defects occur in multiple products. That is, when multiple manufactured products have quality defects, work data that is the main cause of those defects can be estimated. Therefore, the main cause of the quality defects can be determined.

[0073] When the main cause estimation unit 36 ​​determines that a specific product targeted for diagnosis has a quality defect, and when estimating the operational data required to suppress the occurrence of the quality defect, it calculates an index for the product whose contribution is equivalent to that of each factor in the above determination. Then, factors with large values ​​for this index are estimated to be important factors. At this time, a general method for calculating the predictive importance factor, which is independent of the predictive model's generative algorithm and is represented by, for example, a method called "SHAP (SHapley Additive exPlanations)," is used.

[0074] Furthermore, when a quality defect occurs in a specific product, the main cause estimation unit 36 ​​calculates an index for that product that is equivalent to the contribution of each factor in the aforementioned determination, based on the work data estimated to be the main cause of the defect. Then, factors with large values ​​for this index are estimated as important factors. At this time, a general method for calculating important factors, such as SHAP described above, which does not rely on a prediction model, is used.

[0075] Furthermore, when multiple products experience quality defects, and when operational data is presumed to be the primary cause of these defects, the main cause estimation unit 36 ​​calculates an index that represents the influence of each factor in the overall prediction model. Then, factors with large values ​​for this index are presumed to be important factors.

[0076] Alternatively, the main reason estimation section 36 may use, for example, a predictive importance factor calculation method specific to the generative predictive model algorithm, represented by the importance of Gini impurity based on the decision tree algorithm. Or, it may use a general predictive importance factor calculation method that is independent of the generative algorithm, represented by "Permutation Importance".

[0077] The output unit 40 is implemented by a display device such as an LCD display or a CRT display. Based on the display signal input from the arithmetic unit 30, the output unit 40 displays, for example, the quality defect judgment result performed by the quality defect judgment unit 35, the main cause estimation result of the quality defect performed by the main cause estimation unit 36, etc., in the form of text or graphics.

[0078] [Methods for generating prediction models]

[0079] Reference Figure 9 The method for generating the prediction model used in the defect diagnosis method according to the implementation method will be described. The method for generating the prediction model includes a block segmentation step (step S1), a defect classification step (step S2), a data collection step (step S3), and a model generation step (step S4).

[0080] In the block segmentation step, the data segmentation unit 31 divides the past performance product data into two-dimensional blocks according to each specified block (step S1). Next, in the defect classification step, the defect classification unit 32 classifies the defects contained in the quality data according to the tendency of defects to occur in the segmented blocks (step S2).

[0081] Next, in the data aggregation step, the data aggregation unit 33 aggregates the job data extracted according to each defect category by block and each region containing two or more blocks (step S3). Next, in the model generation step, the model generation unit 34 learns the relationship between the aggregated job data and the quality data by machine learning according to each defect category, thereby generating a predictive model (step S4).

[0082] [Methods for diagnosing poor quality]

[0083] Reference Figure 10 The method for diagnosing quality defects involved in the implementation method will be described. In the method for generating the prediction model, the steps include block segmentation (step S11), data collection (step S12), quality defect determination (step S13), and main cause estimation (step S14).

[0084] Additionally, the following description envisions the use of... Figure 9Steps S1-S4 involve generating predictive models in advance at different time points, but can also be used to diagnose quality issues. In other words, it can be implemented at different time points. Figure 9 Steps S1~S4 and Figure 10 Steps S11~S14, or you can continue... Figure 9 Steps S1 to S4 are performed Figure 9 Steps S11 to S14.

[0085] In the block segmentation step, the data segmentation unit 31 segments the product data of the diagnostic target in a two-dimensional manner according to each prescribed block (step S11). Next, in the data aggregation step, the data aggregation unit 33 aggregates the work data extracted according to each predetermined defect category, according to each block and each region containing two or more blocks (step S12). Furthermore, in step S12, for example, using… Figure 9 The information related to the classification of quality defects generated in step S2 is used to extract the work data.

[0086] Next, in the defect determination step, the defect determination unit 35 uses multiple prediction models to determine whether a defect has occurred (step S13). Next, in the main cause estimation step, the main cause estimation unit 36 ​​estimates the main cause of the defect based on the defect determination result (step S14).

[0087] In the defect diagnosis apparatus and method described above, the location of the defect in the product is classified, and a prediction model incorporating the classification results is used to diagnose the defect. Specifically, in the defect diagnosis apparatus and method described in these embodiments, the relationship between work data and defect is generated and used as a prediction model according to each classification corresponding to the tendency of defect occurrence. Therefore, product defect can be predicted with high accuracy, and the main cause of the defect can be estimated.

[0088] The foregoing has provided a detailed description of the defective quality diagnosis device and method of the present invention through embodiments and methods for carrying out the invention. However, the scope of the present invention is not limited to these descriptions and should be interpreted broadly based on the scope of the claims. Furthermore, various modifications and alterations based on these descriptions are also included within the scope of the present invention, which is self-evident.

[0089] Marker description

[0090] 1. Information processing device;

[0091] 10 Input Section;

[0092] 20 storage units;

[0093] 21 Product DB;

[0094] 30. Arithmetic Unit;

[0095] 31 Data Segmentation Section;

[0096] 32. Defective Product Classification Department;

[0097] 33 Data Collection Department;

[0098] 34. Model Generation Department;

[0099] Quality Defect Judgment Department;

[0100] 36. Presumed Main Causes;

[0101] 40 Output section.

Claims

1. A defective product diagnostic device for diagnosing quality defects in a product, wherein, The defective quality diagnostic device includes: The data segmentation department divides the product data of the diagnostic object, which consists of operational data and quality data, into two-dimensional blocks according to each specified area. The data collection unit extracts the job data according to each predetermined category of quality defects based on the tendency of quality defects to occur in the segmented blocks, and collects the extracted job data according to each block and each region containing two or more blocks. The quality defect determination unit uses multiple prediction models generated according to the classification of each quality defect and learned the relationship between the aggregated work data and the quality data to determine whether a quality defect has occurred. as well as The Main Cause Deduction Department deduces the main cause of quality defects based on the results of the quality defect determination.

2. The defective quality diagnostic device according to claim 1, wherein, When a specific product identified as a diagnostic target experiences a quality defect, the primary cause estimation unit estimates the operational data required to suppress the occurrence of the quality defect.

3. The defective quality diagnostic device according to claim 1, wherein, The main cause estimation section estimates operational data as the main cause of a product's quality defect when such a defect occurs.

4. The defective quality diagnostic device according to claim 1, wherein, The main cause estimation department estimates operational data as the main cause of quality defects when multiple products experience quality defects.

5. The defective quality diagnostic device according to claim 1, wherein, The prediction model was generated as follows: Product data representing past performance, consisting of operational and quality data, is divided into two-dimensional blocks. Based on the tendency of quality defects to occur in the segmented blocks, the quality defects contained in the quality data are classified. The job data extracted according to each of the aforementioned defective quality categories will be aggregated according to each of the aforementioned blocks and each region containing two or more of the aforementioned blocks. Based on each of the aforementioned defect categories, machine learning is used to learn the relationship between the aggregated job data and the quality data.

6. The defective quality diagnostic device according to claim 1, wherein, The defective quality categories are any two or more of the following: front end, tail end, central part, width edge part, single random, and group random.

7. The defective quality diagnostic device according to claim 1, wherein, The data collection unit calculates one or more of the following in terms of block unit or product unit: average value, maximum value, minimum value, standard deviation, variance, and the difference between the maximum and minimum values ​​of the operation data, thereby collecting the operation data.

8. The defective quality diagnostic device according to claim 3, wherein, Obtain the correlation between the quality data and a pre-generated template according to the tendency of each quality defect to occur, and classify the quality defects based on the magnitude of the correlation.

9. A method for diagnosing quality defects, wherein, The method for diagnosing quality defects includes: In the data segmentation step, the computer's data segmentation unit divides the product data of the diagnostic object, which consists of operational data and quality data, into two-dimensional blocks according to each specified area. In the data collection step, the data collection unit of the computer extracts the job data according to each predetermined classification of quality defects based on the tendency of quality defects to occur in the segmented blocks, and collects the extracted job data according to each block and each region containing two or more blocks. The defect determination step involves a defect determination unit in the computer using multiple prediction models generated according to the classification of each defect and having learned the relationship between the aggregated job data and the quality data to determine whether a defect has occurred; and The main cause estimation step involves the main cause estimation unit of the computer estimating the main cause of the quality defect based on the quality defect determination result.

Images