Computer system and method for assisting data analysis

Abstract

This computer system manages manufacturing history information for storing a manufacturing history including data items related to the manufacturing process and quality of a product, executes a plurality of statistical analyses using the manufacturing history so as to identify a manufacturing history including a data item having a statistical abnormality, generates a data item list for storing data in which the identified manufacturing history, the data item having the statistical abnormality, and the type of the statistical abnormality are associated with each other, and outputs the data item list as information used for analysis of the manufacturing history.

Description

Computer system and data analysis support method

[0001] The present invention relates to a system for supporting the analysis of big data.

[0002] Advances in IoT have made it possible to acquire and store a variety of data. Techniques that utilize this data include those described in Patent Documents 1 and 2.

[0003] JP 2018-156346 A JP 2020-87110 A

[0004] Data contains a wide variety of data items, and the number of items is large. Conventionally, data items to be analyzed have been selected manually. This has resulted in the problem of high costs involved in selecting data items. Furthermore, if the selected data items are inappropriate, accurate results cannot be obtained. Another issue is that analysis based on the relationships between unknown data items is not possible.

[0005] An object of the present invention is to provide a system and method for automatically extracting data items of interest in data analysis.

[0006] A representative example of the invention disclosed in the present application is as follows: That is, a computer system includes a computer having a processor, a storage device connected to the processor, and a connection interface connected to the processor, manages manufacturing history information that stores manufacturing history including data items related to the manufacturing process and quality of a product, performs multiple statistical analyses using the manufacturing history to identify the manufacturing history including the data item with a statistical anomaly, generates a data item list that stores data associating the identified manufacturing history, the data item with the statistical anomaly, and the type of statistical anomaly, and outputs the data item list as information to be used in analyzing the manufacturing history.

[0007] According to the present invention, it is possible to automatically extract data items that are useful for data analysis. Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments.

[0008] 1 is a diagram illustrating an example of a system configuration of Example 1. FIG. 1 is a diagram illustrating an example of a hardware configuration of a computer configuring the management system of Example 1. FIG. 2 is a diagram illustrating an example of a data structure of manufacturing history information of Example 1. FIG. 3 is a diagram illustrating an example of a data structure of quality inspection information of Example 1. FIG. 4 is a diagram illustrating an example of a data structure of reference information of Example 1. FIG. 5 is a diagram illustrating an example of a data structure of manufacturing process information of Example 1. FIG. 6 is a flowchart illustrating a flow of processing executed by the management system of Example 1. FIG. 7 is a flowchart illustrating an example of a data item filtering processing executed by the management system of Example 1. FIG. 8 is a diagram illustrating an example of a data structure of filtering information of Example 1. FIG. 9 is a diagram illustrating an example of a data structure of a data item list of Example 1. FIG. 10 is a diagram illustrating an example of a data item list of Example 1. FIG. 11 is a flowchart illustrating an example of a data item extraction processing (first aspect) executed by the management system of Example 1. FIG. 12 is a flowchart illustrating an example of a data item extraction processing (second aspect) executed by the management system of Example 1. FIG. 13 is a flowchart illustrating an example of a data item extraction processing (third aspect) executed by the management system of Example 1. FIG. 14 is a diagram illustrating an example of a screen presented by the management system of Example 1. FIG. 15 is a flowchart illustrating an example of an analysis data generation processing executed by the management system of Example 1. FIG. 16 is a diagram illustrating an example of a data structure of an analysis data list generated by the management system of Example 1. 1 is a diagram illustrating an example of a causal relationship model generated by a causal inference unit of Example 1. FIG. 2 is a flowchart illustrating an example of an analysis data generation process executed by the management system of Example 1. FIG. 3 is a diagram illustrating an example of data item extraction by the management system of Example 1.

[0009] Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the present invention should not be construed as being limited to the description of the embodiments shown below. Those skilled in the art will readily understand that the specific configuration can be changed without departing from the spirit or intent of the present invention.

[0010] In the configuration of the invention described below, the same or similar configurations or functions are denoted by the same reference numerals, and redundant explanations will be omitted.

[0011] In this specification, the terms "first," "second," "third," etc. are used to identify components and do not necessarily limit the number or order.

[0012] To facilitate understanding of the invention, the position, size, shape, range, etc. of each component shown in the drawings etc. may not represent the actual position, size, shape, range, etc. Therefore, the present invention is not limited to the position, size, shape, range, etc. disclosed in the drawings etc.

[0013] Fig. 1 is a diagram illustrating an example of a system configuration according to Example 1. Fig. 2 is a diagram illustrating an example of a hardware configuration of a computer that configures the management system according to Example 1.

[0014] The system is composed of a management system 100 and a terminal 101. The number of terminals 101 may be two or more. The terminal 101 is connected to the management system 100 via a network 102 such as a WAN or LAN. The network 102 may be connected via either a wired or wireless method.

[0015] The terminal 101 is a terminal operated by a user who uses the management system 100. The terminal 101 is, for example, a personal computer or a smartphone.

[0016] The management system 100 manages information acquired during product manufacturing operations and provides information for improving the manufacturing operations using the information. The management system 100 is configured with a computer 200 shown in FIG. 2. The computer 200 has a processor 201, a network interface 202, a main memory device 203, and a secondary memory device 204. The hardware elements are connected via a bus 205. The computer 200 may also have input devices such as a mouse, a keyboard, and a touch panel, as well as an output device such as a display.

[0017] The processor 201 operates as a functional unit (module) that realizes a specific function by executing processing in accordance with a program stored in the main memory device 203. In the following description, when a processing is described using a functional unit as the subject, it indicates that the processor 201 is executing a program that realizes the functional unit.

[0018] The main memory 203 is a storage device, such as a volatile or nonvolatile memory, that stores programs executed by the processor 201 and information executed by the programs. The main memory 203 is also used as a work area. The secondary memory 204 is a large-capacity storage device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

[0019] The programs and information stored in the main memory device 203 may be stored in the secondary memory device 204. In this case, the processor 201 reads the programs and information from the secondary memory device 204 and loads them into the main memory device 203.

[0020] The management system 100 has, as functional units, a data item extraction unit 110, a causal inference unit 111, and a simulation unit 112. The management system 100 also stores a user data DB 113 and an analysis DB 114.

[0021] The user data DB 113 is a database for managing various information related to manufacturing operations. The user data DB 113 stores, for example, manufacturing history information 120, quality inspection information 121, reference information 122, and manufacturing process information 123.

[0022] The manufacturing history information 120 is information for managing information related to the manufacturing of a product (manufacturing history). The quality inspection information 121 is information for managing the inspection results of the quality of the product. The standard information 122 is information for managing the quality and design standards of the product. The manufacturing process information 123 is information for managing the manufacturing process of the product. Note that a product is manufactured through multiple manufacturing processes.

[0023] The management system 100 does not need to hold the user data DB 113. In this case, an external system may manage the user data DB 113, and the management system 100 may access this system.

[0024] The analysis DB 114 is a database for managing various information obtained by analyzing information related to manufacturing operations. The analysis DB 114 stores filtering information 130, a data item list 131, causal relationship information 132, and model information 133.

[0025] The filtering information 130 is information for identifying data items to be analyzed. The data item list 131 is a list for managing candidates for data items to be used in data analysis. The causal relationship information 132 is information for managing causal relationships between data items (KPIs) that are important in managing manufacturing operations. The model information 133 is information for managing models to be used in simulations.

[0026] FIG. 3 is a diagram illustrating an example of the data structure of the manufacturing history information 120 according to the first embodiment.

[0027] The manufacturing history information 120 stores the manufacturing history (record) of a product. One manufacturing history exists for one product. The manufacturing history includes the date and time of manufacturing the product, the product name, the product type, the serial number, and multiple measurement values. Note that the columns included in the manufacturing history are merely examples and are not intended to be limiting. In the following description, the columns included in the manufacturing history will be referred to as data items.

[0028] The measured values ​​include values ​​that represent the characteristics of the product itself and values ​​that represent the characteristics of the manufacturing process of the product. Values ​​that represent the characteristics of the product itself include, for example, the size and performance values ​​of the product. Values ​​that represent the characteristics of the manufacturing process of the product include, for example, the control values ​​of manufacturing equipment. The measured values ​​may be obtained using a measuring device such as a sensor, or may be obtained through quality inspections, etc.

[0029] It should be noted that the manufacturing history may contain data items for which no values ​​exist.

[0030] FIG. 4 is a diagram illustrating an example of the data structure of the quality inspection information 121 according to the first embodiment.

[0031] The quality inspection information 121 stores the results (records) of the quality inspection. The results of the quality inspection include the product name, serial number, and defect code. Note that the columns included in the results of the quality inspection are merely examples and are not limiting.

[0032] FIG. 5 is a diagram illustrating an example of the data structure of the reference information 122 according to the first embodiment.

[0033] The criteria information 122 stores records for managing criteria such as product quality and specifications. The records include product names, target data items, reference values, upper and lower limits. Note that the columns included in the records are merely examples and are not intended to be limiting.

[0034] FIG. 6 is a diagram showing an example of the data structure of the manufacturing process information 123 according to the first embodiment.

[0035] The manufacturing process information 123 stores records for managing the manufacturing process of a product. The records include the product name, the process name, the type of work in the process, the elements (4M) for appropriately performing quality control work in the process, and the type of measurement value related to the process. Note that the columns included in the records are merely examples and are not limited to these. The 4M are Man, Machine, Material, and Method.

[0036] FIG. 7 is a flowchart illustrating the flow of processing executed by the management system 100 according to the first embodiment.

[0037] The management system 100 executes the process described below when an execution instruction is received from a user, or periodically. The execution instruction includes, for example, the product name or product type and the target period. When executing the process periodically, for example, setting information that associates the execution period, the target period, and the product type is set in advance. Here, the description will be given assuming that an execution instruction is received.

[0038] The data item extraction unit 110 acquires the manufacturing history of the processing target from the manufacturing history information 120 (step S101). For example, one month's worth of manufacturing history of the specified product type is acquired.

[0039] The data item extraction unit 110 executes a data item filtering process (step S102). In the data item filtering process, data items to be processed are identified. The details of the data item filtering process will be described later.

[0040] The data item extraction unit 110 executes a data item extraction process (step S103). In the data item extraction process, multiple statistical methods are used to extract data items that are candidates for data analysis. The details of the data item extraction process will be described later.

[0041] The causal inference unit 111 executes a process of generating analytical data to be used in a causal inference process for inferring a causal relationship between the extracted data items and KPIs (step S104). The causal inference unit 111 also executes a causal inference process using the analytical data (step S105). The analytical data generation process and the causal inference process will be described in detail later.

[0042] The simulation unit 112 uses the causal relationships to perform a process of generating analysis data for generating a model to be used in the simulation (step S106). The simulation unit 112 also generates a model using the analysis data and performs a simulation using the generated model (step S107). The analysis data generation process and the simulation will be described in detail later.

[0043] Each time the process of FIG. 7 is executed, filtering information 130, a data item list 131, causal relationship information 132, and model information 133 are generated.

[0044] The above process flow allows for continuous and automatic execution of a cycle of extracting data items of interest for data analysis, acquiring causal relationships (knowledge) between the extracted data items and KPIs, and performing simulations using the causal relationships. Conventionally, a person had to manually select the data items to be analyzed and generate the analysis data. This made it impossible to achieve the above-described continuous and automatic cycle. The present invention enables the above-described continuous and automatic cycle to be achieved through the various processes described below.

[0045] First, the data item filtering process will be described. Fig. 8 is a flowchart illustrating an example of the data item filtering process executed by the management system 100 according to the first embodiment. Fig. 9 is a diagram illustrating an example of the data structure of the filtering information 130 according to the first embodiment.

[0046] The data item extraction unit 110 generates empty filtering information 130 (step S201).

[0047] The data item extraction unit 110 selects one data item from among the data items included in the manufacturing history to be processed (step S202).

[0048] The data item extraction unit 110 calculates statistics for the selected data items using the manufacturing history of the processing target (step S203). For example, the calculated statistics include the missing rate, variance, mean, median, maximum value, minimum value, standard deviation, and quartile. If the data type of the data item is not numeric, the process skips to step S203 and proceeds to step S204. In this embodiment, data items whose data type is not numeric are excluded.

[0049] At this time, the data item extraction unit 110 adds a record to the filtering information 130. The record includes a data item 901, a data type 902, a count 903, a missingness rate 904, an average 905, a median 906, and an analysis flag 907. Note that the fields included in the record are merely examples and are not intended to be limiting.

[0050] The data item 901 is a field that stores identification information of a data item included in the manufacturing history. The data type 902 is a field that stores the data type of the data item. The count 903 is a field that stores the number or percentage of manufacturing history entries in which valid values ​​are stored. The missing rate 904, the mean 905, and the median 906 are fields that store statistical quantities of the data item. The analysis flag 907 is a flag that indicates whether or not the data item is a target for processing. If the data item is a target for processing, "TRUE" is stored in the analysis flag 907, and if the data item is not a target for processing, "FALSE" is stored in the analysis flag 907. Note that the target data item for processing refers to a data item that is the target of processing from step S103 onwards.

[0051] The data item extraction unit 110 sets values ​​for the data item 901, data type 902, count 903, missing rate 904, average 905, and median 906 of the added record. Note that the analysis flag 907 of the record added in step S203 is left blank.

[0052] The data item extraction unit 110 determines whether or not processing has been completed for all data items included in the manufacturing history to be processed (step S204).

[0053] If processing has not been completed for all data items included in the manufacturing history to be processed, the data item extraction unit 110 returns to step S202.

[0054] When processing has been completed for all data items included in the manufacturing history to be processed, the data item extraction unit 110 selects a data item to be processed based on the statistics of each data item (step S205). Thereafter, the data item extraction unit 110 ends the data item filtering process.

[0055] In this embodiment, it is assumed that conditions for selecting data items to be processed are set in advance. For example, the following conditions are possible. If all of the following conditions are met, the data item extraction unit 110 determines that the data item is a data item to be processed. [Conditions] (1) Data type = Numeric (2) Count ≠ 0 (3) Value type ≠ 1 (4) Variance ≠ 0 (5) Missing rate < 50%

[0056] The data item extraction unit 110 selects data items that match the conditions as targets for processing. At this time, the data item extraction unit 110 sets "TRUE" to the analysis flag 907 of the record of the data item selected as the target for processing, and sets "FALSE" to the analysis flag 907 of the record of the data item that is not the target for processing.

[0057] The management system 100 does not have to perform the data item filtering process.

[0058] Next, the data item extraction process will be described. In the data item extraction process, statistical analysis is performed based on three perspectives: (first perspective) variations in product quality in the manufacturing process; (second perspective) differences in the manufacturing history of the same type of product with different characteristics; and (third perspective) deviations from the product specifications or design, and data items to be analyzed are extracted. The extraction results are stored in the data item list 131.

[0059] Here, a description will be given of the data item list 131. Figures 10A, 10B, and 10C are diagrams showing an example of the data structure of the data item list 131 according to the first embodiment.

[0060] The data item list 131 stores records including a serial number 1001, a data item 1002, an abnormality type 1003, and a remark 1004. The serial number 1001 is the same column as the serial number included in the manufacturing history. The data item 1002 is a column that stores the name of the data item (column) included in the manufacturing history. The abnormality type 1003 is a column that stores the type of statistical abnormality. The remark 1004 is a column that stores information about the statistical abnormality.

[0061] First, a method for extracting data items based on the first aspect will be described. Fig. 11 is a flowchart illustrating an example of a data item extraction process (first aspect) executed by the management system 100 according to the first embodiment.

[0062] The data item extraction unit 110 selects one data item from among the data items included in the manufacturing history to be processed (step S301).

[0063] The data item extraction unit 110 refers to the filtering information 130 and determines whether the selected data item is a data item to be processed (step S302). Specifically, the data item extraction unit 110 determines whether the analysis flag 907 of the record corresponding to the selected data item is "TRUE."

[0064] If the selected data item is not the data item to be processed, the data item extraction unit 110 proceeds to step S308.

[0065] If the selected data item is the data item to be processed, the data item extraction unit 110 uses the manufacturing history of the data item to be processed to generate a control chart for quality control in the manufacturing process for the selected data item (step S303). For example, a Shewhart control chart is generated.

[0066] The data item extraction unit 110 selects an anomaly determination rule for the control chart (step S304) and determines whether the anomaly determination rule applies (step S305). It is assumed that the anomaly determination rule is set in advance. For example, it is possible to use the eight anomaly determination rules specified in JIS Z 9020-2.

[0067] If the selected abnormality determination rule is not met, the data item extraction unit 110 proceeds to step S307.

[0068] If the selected anomaly determination rule is met, the data item extraction unit 110 adds the record to the data item list 131 (step S306), and then proceeds to step S307.

[0069] Specifically, the data item extraction unit 110 identifies manufacturing histories included in a group that exhibits characteristics corresponding to the anomaly determination rule. The data item extraction unit 110 adds the same number of records as the identified manufacturing histories to the data item list 131. The data item extraction unit 110 sets the serial number included in the manufacturing histories in the serial number 1001 of each record, sets the selected data item in the data item 1002 of each record, and sets the name of the selected anomaly determination rule, etc. in the anomaly type 1003 of each record. The remarks 1004 are left blank.

[0070] In step S307, the data item extraction unit 110 determines whether or not processing has been completed for all of the anomaly determination rules (step S307).

[0071] If the processing has not been completed for all the abnormality determination rules, the data item extraction unit 110 returns to step S304. If the processing has been completed for all the abnormality determination rules, the data item extraction unit 110 proceeds to step S308.

[0072] In step S308, the data item extraction unit 110 determines whether or not the processing has been completed for all data items (step S308).

[0073] If the processing has not been completed for all data items, the data item extraction unit 110 returns to step S301. If the processing has been completed for all data items, the data item extraction unit 110 ends the processing.

[0074] By the above processing, a record such as that shown in FIG. 10A is added to the data item list 131.

[0075] Next, a method for extracting data items based on the second aspect will be described. Fig. 12 is a flowchart illustrating an example of a data item extraction process (second aspect) executed by the management system 100 according to the first embodiment.

[0076] The data item extraction unit 110 standardizes the value of each data item of the manufacturing history to be processed (step S401). For example, the data item extraction unit 110 converts the value of the data item into a standard deviation value (T score).

[0077] The data item extraction unit 110 classifies the manufacturing history data to be analyzed based on predetermined classification criteria (step S402). In this embodiment, the data item extraction unit 110 classifies the manufacturing history data into manufacturing histories of non-defective products and manufacturing histories of defective products by referring to the quality inspection information 121. The classification criteria can be set arbitrarily.

[0078] The data item extraction unit 110 calculates the average value of each data item for the manufacturing history of non-defective products (step S403).

[0079] The data item extraction unit 110 selects one data item from the data items included in the manufacturing history to be processed (step S404).

[0080] The data item extraction unit 110 refers to the filtering information 130 and determines whether the selected data item is a data item to be processed (step S405). The process of step S405 is the same as the process of step S302.

[0081] If the selected data item is not the data item to be processed, the data item extraction unit 110 proceeds to step S408.

[0082] If the selected data item is the data item to be processed, the data item extraction unit 110 calculates the difference between the average value of the selected data item and the value of the selected data item in each manufacturing history of the defective product (step S406).

[0083] The data item extraction unit 110 adds to the data item list 131 the records of the manufacturing history whose difference is greater than the threshold value (step S407), and then proceeds to step S408.

[0084] Specifically, the data item extraction unit 110 identifies manufacturing history records whose difference is greater than a threshold. The data item extraction unit 110 adds the same number of records as the identified manufacturing history records to the data item list 131. The data item extraction unit 110 sets the serial number included in the manufacturing history records to the serial number 1001 of each record, sets the selected data item to the data item 1002 of each record, and sets the abnormality type 1003 of each record to "value difference abnormality." The data item extraction unit 110 also sets the difference to the remarks 1004 of each record.

[0085] In step S408, the data item extraction unit 110 determines whether or not the processing has been completed for all data items (step S408).

[0086] If the processing has not been completed for all data items, the data item extraction unit 110 returns to step S404. If the processing has been completed for all data items, the data item extraction unit 110 ends the processing.

[0087] By the above processing, a record such as that shown in FIG. 10B is added to the data item list 131.

[0088] Next, a method for extracting data items based on the third aspect will be described. Fig. 13 is a flowchart illustrating an example of data item extraction processing (third aspect) executed by the management system 100 according to the first embodiment.

[0089] The data item extraction unit 110 selects one data item from among the data items included in the manufacturing history to be processed (step S501).

[0090] The data item extraction unit 110 refers to the filtering information 130 and determines whether the selected data item is a data item to be processed (step S502). The process of step S502 is the same as the process of step S302.

[0091] If the selected data item is not the data item to be processed, the data item extraction unit 110 proceeds to step S505.

[0092] If the selected data item is the data item to be processed, the data item extraction unit 110 calculates an index indicating the degree of deviation from the standard for the selected data item of each manufacturing history (step S503). For example, the data item extraction unit 110 calculates a statistical distance from the standard based on the distribution of values ​​of the data item of each manufacturing history and the standard information 122.

[0093] The data item extraction unit 110 adds the records of the manufacturing history that deviate greatly from the standard to the data item list 131 (step S504), and then proceeds to step S505.

[0094] Specifically, the data item extraction unit 110 identifies manufacturing history records whose distance is greater than a threshold. The data item extraction unit 110 adds the same number of records as the identified manufacturing history records to the data item list 131. The data item extraction unit 110 sets the serial number included in the manufacturing history records to the serial number 1001 of each record, sets the selected data item to the data item 1002 of each record, and sets the abnormality type 1003 of each record to "reference difference." The data item extraction unit 110 also sets an index to the remarks 1004 of each record.

[0095] A plurality of thresholds may be set, in which case the value of the abnormality type 1003 is set for each range determined by the thresholds.

[0096] In step S505, the data item extraction unit 110 determines whether or not the processing has been completed for all data items (step S505).

[0097] If the processing has not been completed for all data items, the data item extraction unit 110 returns to step S501. If the processing has been completed for all data items, the data item extraction unit 110 ends the processing.

[0098] By the above processing, a record such as that shown in FIG. 10C is added to the data item list 131.

[0099] As explained above, the data item extraction process extracts data items with statistical anomalies based on three perspectives. This makes it possible to automatically narrow down the data items to be analyzed, reducing the cost of data analysis and improving accuracy.

[0100] After the data item extraction process is completed, the management system 100 may present the user with the data item list 131. The management system 100 may also present information indicating the frequency of statistical abnormalities in the manufacturing process, using the data item list 131 and the manufacturing process information 123.

[0101] 14 is a diagram showing an example of a screen presented by the management system 100 of the first embodiment. Screen 1400 is a screen that presents the occurrence frequency of statistical anomalies in a manufacturing process and includes display fields 1401 and 1402. Display field 1401 displays a heat map showing the number of statistical anomalies in a product's manufacturing process. The heat map is generated using the data item list 131 and manufacturing process information 123. Display field 1402 displays details of the statistical anomaly in the manufacturing process (box 1410) selected in display field 1401.

[0102] It is also possible to compile the frequency of statistical abnormalities in products from the perspective of the 4Ms and display it on a screen.

[0103] Next, we will explain data analysis (causal inference and simulation) using the results of data item extraction.

[0104] First, causal inference will be described. Fig. 15 is a flowchart illustrating an example of an analytical data generation process executed by the management system 100 of the first embodiment. Fig. 16 is a diagram illustrating an example of the data structure of an analytical data list generated by the management system 100 of the first embodiment.

[0105] The causal inference unit 111 reads out the generation conditions (step S601). The generation conditions include the data items to be used as KPIs, the analysis period, and the like. The analysis conditions are assumed to be set in advance. The user may set the generation conditions by referring to the screen 1400 or the like. At this time, the generation conditions may include the anomaly type.

[0106] The causal inference unit 111 selects a manufacturing history to be analyzed based on the generation conditions and the data item list 131 (step S602).

[0107] Specifically, the causal inference unit 111 identifies the manufacturing history records whose records are registered in the data item list 131, and selects the manufacturing history records that match the generation conditions from among the manufacturing history records.

[0108] The causal inference unit 111 selects data items to be used for analysis from among the data items to be analyzed (step S603).

[0109] Specifically, the causal inference unit 111 refers to the data item list 131 and identifies data items based on records related to the selected manufacturing history. If the generation conditions include information specifying data items, the causal inference unit 111 selects data items to use from the identified data items. If the generation conditions do not include information specifying data items, the causal inference unit 111 selects all identified data items as data items to use.

[0110] The causal inference unit 111 generates the selected manufacturing history including the selected data items and data items corresponding to the KPIs as analysis data (step S604).

[0111] The above process generates, for example, an analysis data list 1600 as shown in Fig. 16. Fig. 16 shows analysis data in which measurement value (1) and measurement value (3) are selected as the data items to be used and measurement value (x) is designated as a KPI.

[0112] The causal inference unit 111 generates a causal relationship model by executing a causal inference process using the analysis data. For example, it is conceivable to generate a causal relationship model using the technology described in Patent Document 1. Note that the present invention is not limited to a method for generating a causal relationship model. Note that the causal inference process may use a manufacturing history of a processing target that is not registered in the data item list 131.

[0113] 17 is a diagram illustrating an example of a causal relationship model generated by the causal inference unit 111 of the first embodiment. The causal relationship model is expressed as a graph connecting nodes corresponding to data items. Note that black nodes correspond to KPIs. A table of conditional probabilities with data items as variables is generated as the causal relationship information 132.

[0114] Next, a simulation using a causal relationship model will be described.

[0115] Fig. 18 is a flowchart illustrating an example of analysis data generation processing executed by the management system 100 of Example 1. Fig. 19 is a diagram illustrating an example of data item extraction by the management system 100 of Example 1.

[0116] The simulation unit 112 reads out the generation conditions (step S701). The generation conditions include a probability threshold value that indicates the relationship with the KPI.

[0117] The simulation unit 112 extracts important factors (data items) for the KPI based on the causal relationship information 132 and the generation conditions (step S702).

[0118] For example, the simulation unit 112 searches for factors (data items) whose probabilities are greater than a threshold value, starting from the KPI, based on a table of conditional probabilities. For example, as shown in Fig. 19, the nodes in bold frames are extracted as important factors.

[0119] The simulation unit 112 generates a model in which the important factors are used as explanatory variables and the KPI is used as a response variable (step S703). For example, if measurement value (1), measurement value (3), and measurement value (7) are extracted as important factors, it is possible to generate a linear regression model as shown in equation (1): Y = β １ X １ +β ２ X ３ +β ３ X ７ …(1)

[0120] Y is a variable that represents KPI, and X １ , X ３ , X ７ is a variable representing measurement value (1), measurement value (3), and measurement value (7). β １ , β ２ , β ３ represents a coefficient, which is determined by learning using manufacturing history.

[0121] Note that a neural network or the like that receives important factors as input and outputs KPIs may be generated as a model.

[0122] The simulation unit 112 generates a random data set and executes a simulation using the model to output information for improving the KPI (step S704). For example, if the control value of an equipment is an important factor, the control value that maximizes the KPI is output. If an index representing product quality is an important factor, the simulation unit 112 may output the value of the index as is, or may convert it into a control value for an equipment related to the index and output this.

[0123] By extracting important factors using a causal model and generating a model based on these, it is possible to automatically generate a model with high prediction accuracy, thereby enabling simulations that accurately reflect manufacturing performance.

[0124] The present invention is not limited to the above-described embodiments, but includes various modifications. For example, the above-described embodiments are provided to explain the present invention in detail, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, some of the configurations of each embodiment can be added to, deleted from, or replaced with other configurations.

[0125] Furthermore, some or all of the above-described configurations, functions, processing units, processing means, etc. may be implemented in hardware, for example, by designing them as integrated circuits. The present invention can also be realized by software program code that implements the functions of the embodiments. In this case, a storage medium on which the program code is recorded is provided to a computer, and a processor included in the computer reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium implements the functions of the above-described embodiments, and the program code itself and the storage medium on which it is stored constitute the present invention. Examples of storage media for providing such program code include flexible disks, CD-ROMs, DVD-ROMs, hard disks, solid-state drives (SSDs), optical disks, magneto-optical disks, CD-Rs, magnetic tape, non-volatile memory cards, and ROMs.

[0126] Furthermore, the program code that realizes the functions described in this embodiment can be implemented in a wide range of programming or scripting languages, such as assembler, C / C++, perl, Shell, PHP, Python, and Java.

[0127] Furthermore, the program code of the software that realizes the functions of the embodiments may be distributed via a network and stored in a storage means such as a computer's hard disk or memory, or in a storage medium such as a CD-RW or CD-R, and the processor of the computer may read and execute the program code stored in the storage means or the storage medium.

[0128] In the above-described embodiment, the control lines and information lines are those that are considered necessary for the explanation, and not all control lines and information lines are necessarily shown in the product. All components may be interconnected.

Claims

1. A computer system comprising a computer having a processor, a storage device connected to the processor, and a connection interface connected to the processor, managing manufacturing history information that stores a manufacturing history including data items related to a product's manufacturing process and quality, performing multiple statistical analyses using the manufacturing history to identify the manufacturing history including the data items having statistical anomalies, generating a data item list that stores data associating the identified manufacturing history, the data items having statistical anomalies, and types of statistical anomalies, and outputting the data item list as information to be used in analyzing the manufacturing history.

2. A computer system as described in claim 1, wherein the multiple statistical analyses include a statistical analysis focusing on the variation in product quality in the manufacturing process, a statistical analysis focusing on the differences in the manufacturing history of the same type of products with different characteristics, and a statistical analysis focusing on deviations from the product specifications or design.

3. A computer system as claimed in claim 1, characterized in that: the data items to be analysed are selected based on the data item list; and the manufacturing history registered in the data item list, including the selected data items and data items that are important in managing manufacturing operations, is generated as analysis data.

4. A computer system as described in claim 3, characterized in that by performing causal inference using the analysis data, a causal graph is generated which represents the relationship between the data items which are important in the management of manufacturing operations and selected data items.

5. A computer system as described in claim 4, characterized in that it extracts, based on the causal graph, data items that have a strong relationship with the data items that are important in managing manufacturing operations, generates a model in which the extracted data items are explanatory variables and the data items that are important in managing manufacturing operations are objective variables, executes a simulation using the model, and outputs information for improving the values ​​of the data items that are important in managing manufacturing operations based on the results of the simulation.

6. A computer system as described in claim 1, further comprising: manufacturing process information for managing the manufacturing process; the manufacturing history includes values ​​measured in the manufacturing process; the manufacturing process information stores data correlating the manufacturing process with the type of value measured in the manufacturing process; the computer system uses the manufacturing process information and the data item list to count the number of statistical anomalies detected for each manufacturing process; and based on the results of the counting, generates and outputs information for displaying the frequency of occurrence of statistical anomalies in the manufacturing process of the product.

7. A method of supporting data analysis executed by a computer system, wherein the computer system includes a computer having a processor, a storage device connected to the processor, and a connection interface connected to the processor, and manages manufacturing history information that stores a manufacturing history including data items related to a product's manufacturing process and quality, the data analysis support method comprising: a first step in which the computer system performs multiple statistical analyses using the manufacturing history to identify the manufacturing history including the data item having a statistical anomaly; a second step in which the computer system generates a data item list that stores data associating the identified manufacturing history, the data item having a statistical anomaly, and the type of statistical anomaly; and a third step in which the computer system outputs the data item list as information to be used in analyzing the manufacturing history.

8. A method for supporting data analysis as described in claim 7, characterized in that the multiple statistical analyses include a statistical analysis focusing on the variation in product quality in the manufacturing process, a statistical analysis focusing on differences in the manufacturing history of the same type of products with different characteristics, and a statistical analysis focusing on deviations from the product specifications or design.

9. A method for supporting data analysis as described in claim 7, comprising the steps of: said computer system selecting the data items to be analyzed based on said data item list; and said computer system generating, as analysis data, the manufacturing history registered in said data item list, including the selected data items and data items that are important in managing manufacturing operations.

10. A method for supporting data analysis as described in claim 9, characterized in that the computer system generates a causal relationship graph representing the relationship between the data items that are important in managing manufacturing operations and selected data items by performing causal inference using the analysis data.

11. A method for supporting data analysis as described in claim 10, comprising the steps of: said computer system extracting, based on said causal graph, the data items that have a strong relationship with said data items that are important in managing manufacturing operations; said computer system generating a model in which the extracted data items are explanatory variables and said data items that are important in managing manufacturing operations are objective variables, and running a simulation using said model; and said computer system outputting, based on the results of said simulation, information for improving the values ​​of said data items that are important in managing manufacturing operations.

12. A method for supporting data analysis as described in claim 7, wherein the computer system holds manufacturing process information for managing the manufacturing process, the manufacturing history includes values ​​measured in the manufacturing process, the manufacturing process information stores data correlating the manufacturing process with the types of values ​​measured in the manufacturing process, and the third step includes the steps of: the computer system counting the number of statistical anomalies detected for each manufacturing process using the manufacturing process information and the data item list; and the computer system generating and outputting information for displaying the frequency of occurrence of statistical anomalies in the manufacturing process of the product based on the results of the counting.

Images