JPH05135062A - Group analysis method - Google Patents

Abstract

(57) [Summary] [Structure] Regarding the processing performed regarding product failure, information on failure occurrence such as date, phenomenon, and remedy is collected from the field and accumulated in a database, and a part or all of the product is targeted. An analysis condition input method characterized in that, when analyzing a cause of failure, tendency of failure occurrence, etc., an analysis condition of the next stage can be input by using each element on the screen representing the analysis result as an icon, that is, an input device. [Effect] At the workstation, by searching the source data in record units and expanding it in the memory, the analysis for each item of data related to product failure and its repair and all combinations of items are targeted. It is possible to perform the analysis.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of collecting information on product quality of a customer and searching / analyzing the accumulated information.

[0002]

2. Description of the Related Art Regarding the conventional collection and analysis of information on product quality of customers, for example, Japanese Patent Laid-Open No. 63-40962 proposes a method of inputting product failure information by a bar code. -The management method and the search / analysis method are not disclosed. In addition, the failure information is poor in practicability only because it is "defective content".

[0003]

In the above-mentioned prior art, first of all, no consideration has been given to the specific items and means for collecting quality data, and as a system aiming to manage quality, There was a problem in terms of practicality. Secondly, no consideration was given to the method of storing / managing quality data and the method of searching, and there was a problem in terms of system cost / performance and system operability including system scalability. Thirdly, the function of displaying the status of a product failure or quality failure that actually occurs in the field has been left behind, and it has not played a role as a tool for pursuing those factors.
In addition, since the user directly uses the functions on the large-scale computer, no consideration has been given to the detailed condition settings for search and analysis, and there was a problem in terms of the dynamics or flexibility of the editing and analysis functions for the user. ..

An object of the present invention is that a (system) user can set detailed search and analysis conditions based on a product failure or quality defect that actually occurs in the field,
Moreover, it is to provide a means for analyzing quality data, which enables the pursuit of these factors in a short time, as a system with high cost performance.

[0005]

[Means for Solving the Problems] In order to achieve the above object, when a product breaks down under the customer who is the user,
Information about repairs performed mainly by local maintenance personnel such as service personnel performed locally is sent online to the factory, which is the main body of product design, manufacturing, and inspection, through sales departments such as distributors and sales offices. The feedback is performed by a computer network.

Here, the computer network is a system in which general-purpose large-scale computers are connected to each other, and general-purpose large-scale computers are connected to workstations or workstations in a hierarchical manner.

[0007] In addition, each time a customer service representative makes a repair on site, he / she writes in a card with a certain format the product type, failure phenomenon, parts, contents of treatment, etc., and writes the contents of the card. Input from a general-purpose large-scale computer terminal installed in the sales department.

[0008]

According to the present invention, when a product breaks down under the customer who is the user, the information about the repair mainly performed by the maintenance person in charge of the customer such as a service person on site is provided for each repair agent. -It is possible to provide immediate feedback to the factory, which is the main body of product design, manufacturing, and inspection, through sales departments such as sales offices. As a result, it is possible to quickly improve the design of the product under production and the evaluation / inspection method of the used parts, and further improve the reliability in the design of the new product.

Since the search / analysis of the quality data can be performed entirely at the workstation, the sequence of search / analysis request, processing, and output can be executed in real time, so that the turnaround time of analysis work can be reduced.

Further, in the workstation, by searching the source data in record units and expanding it in the memory, analysis for each item of the data relating to the product failure and its repair and all combinations of the items are performed. Targeted analysis is possible.

[0011]

Embodiments of the present invention will be described below with reference to FIGS.

FIG. 1 shows the overall configuration of a system for carrying out the present invention. This figure shows a network from a base to which a customer inputs quality data to a factory. When a product breaks down under the customer who is the user, information about repairs performed mainly by maintenance personnel in charge of the customer, such as a service person, is provided locally through a sales department such as a distributor or sales office. Product design / manufacturing /
It is transmitted online to the factory, which is the main body of inspection. Specifically, the quality data of the customer is the terminal 10 of each distributor.
The result input from 1 and edited by the computer 102 of each business office is sent to the same large-scale computer 3 from all the business offices, and the result edited for each factory there is connected to the large-scale computer 3 It is stored in the storage device. The large-scale storage device 3 has a mailbox function, and can be constantly referred to by the large-scale computers 5 (factory 1), (factory 2), (factory 3), (factory 4), ... Of each factory.

Each time a maintenance person in charge of customer service, such as a service engineer, makes a repair on site, he / she writes in the card having a fixed format the product type, the phenomenon of failure, the parts, the contents of treatment, etc. Input the contents of from the terminal of a general-purpose large-scale computer installed in the sales department. The format of this card is shown in FIG. The content of the card consists of dozens of items. The main items are shown in (1) to (27) below.

(1) Product category: Product type and name (2) Free charge category: Distinction between paid repair and free repair (3) Reason for free charge ... Reason for performing free repair (4) Serial number: First product The number assigned at the time of manufacture to identify each item (5) Part No .... The number of the part to be repaired (6) Card No .... Unique number for identifying the card (7) Joint No .... The serial number assigned to the repaired parts at the same time (8) Reception number: Card number managed by the distributor or retailer (9) Special sales category: Card issuer classification (sales office or distributor o
r Distributor) (10) Phenomenon… Failure phenomenon (11) Occasionally… Timing or condition when failure is reproduced (12) Number of operating months… Number of months from when the customer starts operating the product until it fails (13) Special agent … Distributor dealing with trouble repair (14) Sales office… Sales office to which the distributor in (13) belongs (15) Requester classification… Repair requester classification (customer or dealer or
(16) Work source classification ... Main body of repair work (in-house or outsourced) (17) Year of failure ... Year and month of repair of failure (18) Period classification ... Whether the target product is within the warranty period or not Discrimination as to whether it is outside (19) Purchase date: The date when the customer purchased the product (20) Entry / exit category: Distinction between whether the target product was received and repaired or repaired on a business trip (21) Action content category … Details of actions taken to repair the trouble (22) Adjustments and adjustments… A symbol that identifies whether or not adjustments or adjustments have been made and the location of the adjustment (23) Repair content classification… Repair purposes such as parts replacement and inspection (24 ) Repair amount ... Expense required for repair (divided into labor cost and parts cost) (25) Statistical year ... Year when repair cost was recorded (26) Part name ... Name of parts subject to repair (27 ) Manufacturer: Manufacturer of the parts to be repaired Furthermore, at each factory, for the information consisting of the above items received from the mailbox Hereinafter managing the manufacturing side (28) -
Add the items shown in (40).

(28) Product type ... Product type and model (29) Manufacturing year ... Year when the product was manufactured (year setting varies depending on the product) (30) Manufacture date ... 31) Product year: The year when the product of the same type was first shipped (32) Product number: Identification number for managing the product in lot units at the time of production (33) Circuit NO: To identify the parts on the board No. (34) Bad back number: A code for identifying a fault phenomenon with a particularly high management standard (35) Countermeasures ... (33) Countermeasures for the phenomenon with a bad back number (36) Countermeasures ... Defects caused by responsibility (in that case, free of charge even outside the warranty period) (37) Treatment ... Identification symbol for accident treatment (38) Countermeasure number ... Production number for countermeasures against large defects (39) Operating period classification ... Division of period from manufacturing to failure (40) Repair history ... Past repair history Information on repairs and products of the above failures, product sales record Together the object of analysis. Here, the sales performance is the quantity of products delivered from the sales office to the distributor, the quantity of products delivered from the distributor to the distributor, or the actual sales customer from the distributor by type, month, and sales office. One of the quantity of products delivered to

Next, FIG. 3 shows the relationship between product failures and quality defects that actually occur in the field and their factors. This figure illustrates some of the relationships between the above-mentioned card contents and their factors.

Example 1) Cases in which the distribution of failure occurrences by dealers or sales offices dealing with failure repairs is caused by climate differences due to regional differences such as temperature and humidity and by differences in service systems Example 2) Product manufacturing year Failure occurrence distribution by product number, manufacturing date, product number (lot) is caused by manufacturing history such as mixing of defective parts or changes in manufacturing specifications. Example 3) Failure occurrence distribution by year and month of repair of failures Is caused by climate due to seasonal differences such as temperature and humidity Example 4) Failure distribution by product type, repaired parts (modules), and failure phenomenon is caused by a specific product type and parts Case example 5) The distribution of failure occurrences by the number of months from the start of product operation to failure by the customer or by its class occurs in specific failure patterns such as initial failure, random failure and wear failure. Due to this, FIG. 4 performs a search and analysis of a large computer of each factory that stores and manages the quality data corresponding to each factory by issuing a transfer request to the large scale storage device 103 of FIG. 2 shows a configuration of a workstation. A large-scale storage device 46 is connected to the large-scale computer 45 and stores customer quality data of all past products of all products in each factory. The quality data is searched and analyzed in the workstation 47, and the external storage device 48 is used.
It has a database of quality data and outputs the result of searching and analyzing it to the external storage device 49 or the printer 50. The workstation 47 is connected to the large-scale computer 45 via a high-speed network, and when a request for data that does not exist in the database of the workstation occurs, the part not existing in the database is transferred by this network.

FIG. 5 shows the workstation 4 of FIG.
7 shows a software configuration for searching and analyzing quality data. The respective modules sharing the individual functions shown in FIG. 5 will be described below. (1) The user interface 51 manages starting and ending of each module. (2) The screen display application 52 controls the display of a color graphic monitor using a graphic package.
(3) The screen management module 53 is for starting and ending the screen display application. (4) The application activation / management module 54 activates / ends each module. (5) Failure analysis module 5
Reference numeral 5 has a function of analyzing distribution, correlation, trend and the like. (6) The failure occurrence prediction module 56 has a failure occurrence distribution prediction function based on hazard analysis.
(7) The alarm management module 57 has a function of managing the progress status with respect to the target set value of the budget / parts regarding the defect. (8) Free search module 58
Is an instruction to search the intermediate file or the database according to the specified search item. (9) The intermediate file management module 9 manages intermediate files and activates the database search language. (1
0) The database management module 60 is a database manager for managing / updating and searching the database. (11) The M-2050 communication module 61 is
It controls data communication between a large-scale computer and a workstation using an inter-file communication package.

FIG. 6 shows the processing contents for searching and analyzing the quality data in the workstation. Hereinafter, each procedure shown in FIG. 6 will be described along with a flow in the case of performing search / analysis. A search is performed in procedure 2 based on the search conditions set in procedure 1. Procedure 2 activates the database management function of procedure 3. Procedure 3 searches the database according to the search condition. However, if the data set of the database does not satisfy the search condition, the host access function of procedure 4 is used to increase the size of the part other than the common part between the search condition and the database, that is, the part not existing in the database. Transfer from the computer and register in the database. The result retrieved by the procedure 2 is written in the core memory as the source data set 6. At this time, the data value can be directly referred to for the item selected in the data set item selection of the procedure 5. Analysis is performed on the data set written in the core memory. In procedure 7, the analysis method is specified, then in procedure 8, the parameters to be analyzed are selected, and the analysis procedure 9 is processed based on them. The analysis procedure 9 performs a numerical operation according to the analysis method specified in the procedure 7, and outputs the operation result in the output format 12 according to the analysis method. Here, the numerical operation according to each analysis method is to calculate one of the management items shown in FIG. 7A for each data item shown in FIG. 7B according to the definition.

The database management / retrieval method of the processing flow described above will be described. When a search request is generated by the procedure 2 of FIG. 6 in the workstation, the database management function of procedure 3 of FIG. 6 is activated and the database is searched according to the search conditions. At this time, if the data set of the database of the workstation does not satisfy the search condition, the part other than the common part of the search condition and the database, that is, the part which does not exist in the database is transferred from the large-scale computer by the procedure 4 in FIG. 6 and registered in the database. .. The processing procedure in this case is
It shows in FIG. A conditional expression for the lacking part is created from the search conditional expression, a large-scale computer is searched according to the condition, and the result is transferred. If the workstation's disk has enough space, register the transferred data. If there is not enough space on the disk of the workstation, select the conditional expression that represents the data group that can be deleted from the disk, delete the corresponding data, and then register the data.

FIG. 9 and FIG. 9 covering all cases of the product of the set when calculating the inclusion relation between the data set and the search condition and when creating the conditional expression of the missing part from the search condition expression. Shown in 10. In each case, the product of the condition shown in the first column (leftmost column) and the condition shown in the second column (middle column) is shown in the third column (right column). is there. Below, each case is demonstrated in order.

A: (Upper) The product of the condition of being equal to x and the condition of being equal to x indicates that the condition is equal to x.

(Lower row) The product of the condition of being equal to x and the condition of being equal to y indicates an empty set.

B: The product of the condition that it is equal to x and the condition that it is smaller than y is the condition that it is equal to x if x <y, and that it is an empty set if x> y.

C: The product of the condition that it is equal to x and the condition that it is greater than y indicates that it is equal to x if x> y and becomes an empty set if x <y.

D: The product of the condition of being equal to x and the condition of being larger than y and being smaller than z is an empty set if x <y, an empty set if x> z, and x otherwise. Indicates that the conditions are equal.

E: The product of the condition that it is smaller than x and the condition that it is smaller than y indicates that the condition is smaller than y if x> y and smaller than x if x <y.

F: The product of the condition that it is smaller than x and the condition that it is larger than y is greater than x and y if x> y.
The condition is that it is smaller, and if x <y, it indicates that it is an empty set.

G: The product of the condition of being smaller than x and the condition of being larger than y and smaller than z is an empty set if x <y, and is larger than y and smaller than z if x> z. In the cases other than, the condition is that it is larger than y and smaller than x.

H: The product of the condition that it is larger than x and the condition that it is larger than y indicates that it is a condition that it is larger than y if x <y and that it is larger than x if x> y.

I: The product of the condition of being larger than x and the condition of being larger than y and smaller than z is an empty set if x> z, and is larger than y and smaller than z if x <y. In the cases other than, the condition is that it is larger than x and smaller than z.

J: The product of the condition that it is greater than x and less than y and the condition that it is greater than z and less than w is
If y <z, it becomes an empty set, if w <x, it becomes an empty set, and if x <z and w <y, the condition is larger than z and smaller than w, and if x <z and w> y. The condition is that it is larger than z and smaller than y, and x
> Z and w> y, the condition is larger than x and smaller than y, and if x> z and w <y, the condition is larger than x and smaller than w.

FIG. 11 is a functional block diagram showing the external specifications of the entire system. The conditional expression management program relating to the present invention includes (1) creation of an insufficient conditional expression, (2) deletion of a conditional expression and creation of a deleted conditional expression, (3) addition of a conditional expression, and (4) modification of a resident conditional expression. Consists of registration. Furthermore, (1) creation of the insufficient conditional expression consists of (1-1) reading the conditional expression file, (1-2) calculating the product of the conditional expressions, and (1-3) calculating the product with the complement. ..
(1-2) The calculation of the product of the conditional expression consists of a combination of the calculation of the product of the (1-2-1) elements, and the calculation of the product with the (1-3) complement is (1-3-1 ) It consists of a combination of calculation of the complementary set of elements and calculation of the product of (1-3-2) elements. Further, (2) deletion of conditional expressions and creation of deleted conditional expressions involve (2-1) creation of insufficient conditional expressions.

The configuration of the main functions shown in FIG. 11 is further detailed and expressed in PAD diagrams in FIGS. 12 to 16. 12 shows that when the conditional expression management program is started, depending on the value of the input parameter, (1) creation of a conditional expression, (2) deletion of conditional expression and creation of deletion conditional expression, (3) addition of conditional expression , Or (4) Indicates that one of the functions of resident conditional expression change and registration operates. 13 is the same as FIG.
In (1), the procedure for operating the function of querying conditional expressions and creating insufficient conditional expressions is shown. Follow the steps below
I will show you later.

Step 1: Open the conditional expression file in reference mode
Step2: Create a conditional expression table; step3: Check the input conditional expression; step4: Calculate the product of all conditional expressions in the conditional expression table and each input conditional expression. If the result is empty, the conditional expression table Remove from.

Step 5: The product of each input conditional expression and the complementary set of all conditional expressions in the conditional expression table is calculated, and if the result is empty, the message "included" is output. If the result is not empty, multiply the products incrementally.

FIG. 14 shows a procedure in which the function of (2) deletion of conditional expression and creation of deleted conditional expression in FIG. 12 operates. This procedure will be shown later.

Step1: Open the conditional expression file in update mode
Step2: Delete the input conditional expression from the conditional expression file; step3: (1) Invoke the function of querying the conditional expression and creating the insufficient conditional expression; FIG. 15 shows (3) Addition of conditional expression in FIG. Shows the procedure by which the function of. This procedure will be shown later.

Step 1: Open the conditional expression file in add mode
Step2: Add the input conditional expression to the conditional expression file; step3: Close the conditional expression file; FIG. 16 shows the procedure for the function of (4) Resident conditional expression change and registration in FIG. 12 to operate. There is. This procedure will be shown later.

Step 1: Open the conditional expression file in update mode
Step2: Write the input conditional expression in the conditional expression file; step3: Close the conditional expression file; FIG. 17 stores the search conditional expression that is the target of the operation.
The file specifications of the conditional expression file are shown below. Further, FIG. 18 shows the file specifications of the input / output parameter file for storing the elements of the conditional expression for exchanging data with the upper program.

As a result of the processing described above, each function for performing a numerical operation according to the analysis method specified in the procedure 7 of FIG. 6 and outputting the operation result by the output format 12 of FIG. 6 according to the analysis method. Is shown below.

(1) Transition Diagram FIG. 19 shows an example of the transition diagram. With this function, the number of failures, amount of repairs, and failure rate can be classified into up to five types and plotted in time series. The data values can be seen in the data table.

(2) Bar Graph An example of a bar graph is shown in FIG.
Shown in. With this function, the number of failures, amount of repairs, and failure rate are classified into categories and displayed. In addition, it is possible to display specific category items by classifying them into different categories. The data values can be seen in the data table.

(3) Pareto Chart An example of a Pareto chart is shown in FIG. With this function, the number of failures, repair amount, and failure rate are classified into categories and sorted and displayed in order of size. Furthermore, the cumulative value of the ratio of the value of each item to the whole is plotted. The data values can be seen in the data table.

(4) Composition Ratio Diagram An example of the composition ratio diagram is shown in FIG. With this function, the number of failures, repair amount, and failure rate are classified into two main and sub categories, with the main category on the horizontal axis and the sub category on the vertical axis in the form of composition ratio. The data values can be seen in the data table.

(5) Pie chart An example of a pie chart is shown in FIG. With this function, the number of failures, amount of repairs, and failure rate are classified into categories and displayed. In addition, it is possible to display specific category items by classifying them into different categories. The data values can be seen in the data table.

(6) Scatter diagram FIG. 24 shows an example of the scatter diagram. With this function, a pair of data is displayed as one point. The data values can be seen in the data table.

(7) Schedule A case of the schedule is shown in FIG. For the service card data retrieved by this function, sort and sort items
(Up to 3 levels are possible), specific data is extracted and displayed in a table format.

Numerical calculations are performed according to the specified analysis method as described above, and the individual functions for outputting the calculation results in the output format according to the analysis method are further expanded so that product failures that actually occur in the field and For the purpose of pursuing these factors from quality defects, the process of narrowing down the factors and the output of the calculation results can be executed in parallel.
The contents of this processing are illustrated in FIGS. 26 to 28.

Here, an example of analysis for the product series of air conditioners is shown. First, in the product line of air conditioners, select products that should be subject to design improvements and inspection method improvements.
For this purpose, the number of failures, the repair amount, the failure rate, etc. may be displayed by classifying them by product type. This result is shown in FIG.
Shown in. Here, a product with a format T, which generates a large amount of repair money, is selected as an analysis target. In this case, when the cursor is moved to the bar indicating the repair amount of T in FIG. 26 and the mouse is clicked, only the data of the product in that format can be the analysis target. In the product of type T, select the part that should be the target of design improvement and inspection method improvement. For this purpose, the number of failures, the repair amount, the failure rate, etc. may be displayed by classifying them by parts or phenomena. First, FIG. 27 shows the result of stratifying the repair amount of a product of type T by component. At this time, as shown in the right column of the figure, the conditions for limiting the data to be analyzed are sequentially added to the table as search conditions. Here, the compressor, which is a component that generates a relatively large amount of repair money, is selected as an analysis target. In this case as well, in the same manner as the above-mentioned procedure, if the cursor is moved to the bar showing the repair amount of the compressor in FIG. 27 and the mouse is clicked, only the data of the products of type T in which the compressor has failed are analyzed. It can be narrowed down as a target. Next, in a product of type T in which a failure has occurred in the compressor, a portion to be subjected to design improvement or inspection method improvement is selected. For this reason, the result of stratifying the repair amount by phenomenon is shown in FIG. As a result of the process of narrowing down the factors described above, it is clarified that the product type T and the failure of the compressor having the start failure are one of the most important objects of the design improvement and the inspection method improvement.

The processing method illustrated in FIGS. 26 to 28 will be described below.

FIG. 29 shows a data relation diagram for performing factor analysis. In the factor analysis, the content of a specific factor can be further analyzed and expanded in detail from the analysis data.

The source data management table of FIG. 29 (a) is
It manages the original data for analysis, and each line represents one failure repair. The data relating to one failure repair consists of multiple items, and each item manages independent information. Perform data analysis by performing the specified data processing method on each column of each item. The source data number table is for storing the number of data items stored in the source data management table.

The analysis means management table is for managing information for analyzing the original data, and is composed of analysis items, data processing methods, and analysis methods.
The analysis item is the item number (1 to L) of the source data to be analyzed.
To store. The data processing method stores the type of calculation performed on the specified analysis item. The analysis method stores the display type for displaying the analysis result. The specified data processing method is calculated for the analysis items managed in the analysis means management table. The analysis result is displayed according to the specified analysis method.

The factor analysis management table of FIG. 29 (b) shows the analysis contents management table which is information necessary for performing factor analysis and the content of one item designated by the analysis data displayed according to the analysis method. Is stored as. The factor analysis number table manages the number of data stored in the factor analysis table. When multiple data are stored in the factor analysis management table, one item of data in the source data management table is sequentially compared for the number of times of factor analysis to see if the factor analysis contents of the analysis items match. If the analysis items are the same, it is determined that the data is the target of factor analysis, and the data processing method stored in the analysis means management table is analyzed.

FIG. 30 shows a specific factor analysis processing flow.

In this processing flow, data is set in advance in the source data management table and the source data number table in the data relation diagram shown in FIG.

First, the factor analysis number in the factor analysis number table is cleared. Next, the analysis method, analysis item, and data processing method, which are the information managed in the analysis procedure management table, are designated and set in the table. Analyze the source data according to the information set in the analysis procedure management table. The analysis result is displayed according to the analysis method which is the information of the analysis procedure management table. Then, based on the analysis result, it is determined whether or not to analyze and expand the content in detail for the specific factor. When performing analysis regarding a specific factor, one of the analysis items displayed by the analysis method of the analysis means management table is specified. The designated factor analysis content is saved in the factor analysis management table. Further, the contents of the analysis means management table (analysis method, analysis item, data processing method) showing the current analysis contents are saved in the factor analysis management table. After that, 1 is added to the factor analysis number counter and the saving of the current analysis information is completed. When performing the specific factor analysis, the information managed in the analysis procedure management table is designated. If the specific factor analysis process is not performed, it is determined whether to restore the previous specific factor analysis, which is the content of the previous analysis, and if it is restored, the following process is performed.
The analysis method, analysis item, and data processing method are set in the analysis means management table from the factor analysis management table. 1 is subtracted from the factor analysis count table to restore the analysis data.
Primitive data analysis is performed based on the information set in the analysis procedure management table from the factor analysis management table, and the analysis result is displayed. When the restoration of the specific factor analysis is not performed, the entire processing ends.

FIG. 31 shows a source data analysis processing flow.

This flow carries out a data processing method on the contents of the primitive management table according to the information of the analysis means management table.

First, the original data read counter is cleared. Next, the analysis item contents are read from the source data management table. Whether the factor analysis is performed or not is determined depending on whether the value of the factor analysis frequency table is zero. When performing factor analysis, the factor analysis comparison number counter is cleared. The factor analysis contents are read from the factor analysis management table, and it is determined whether the factor analysis contents and the analysis items stored in the factor analysis management table of the original data are the same. If they are equal, 1 is added to the comparison number counter, and when the value of the comparison number counter is equal to the factor analysis number, the read position data equal to the source data read counter value of the source data management table determines the specific factor analysis data. After that, the analysis items in the analysis means management table are analyzed according to the data processing method. If the source data items and factor analysis contents are not equal, or if factor analysis is not performed, data processing is not performed. Next, 1 is added to the source data read counter, and the analysis item contents are processed according to the source data read counter value from the source data management table until the value in the source data number table becomes equal to the number of times of reading.

Next, a method of estimating the future failure occurrence status from the current failure occurrence and product shipping status will be described. First, the procedure of the method is shown below.

Step1. Total number of operating units N (t) in operating months t
And the number of failures r (t).

Step2. The failure rate is calculated by λ (t) = r (t) / N (t).

Step3. Plot (t, lnΣλ (t)) on the hazard probability paper.

Step4. Fit a straight line or a broken line to the plot points and select the one that has the best fit.

Step5. The parameters (m, η) of the failure distribution (Weibull distribution) are obtained from the slope and intercept of the straight line.

Step6. The cumulative failure rate F (t) of the predicted month t is found using the parameters found in step 5.

Step7. ΣnF (t) (n: number of units sold each month)
The cumulative number of failures in the predicted month is calculated by.

Step8. Σn / F (t) / N (N: N = ΣN
(t) Total sales volume) is used to determine the cumulative failure for the forecast month.

Next, FIG. 32 specifically shows the processing.

As shown in the left column of the figure, when the repair of the failure occurs, the type of the product, the phenomenon of the failure, the parts, the contents of the treatment, etc. are reported by the card, but unlike the life test, The time when each product starts operating is different. Therefore, as shown in the upper left of the right column of the figure, the failure rate is calculated for each number of operating months. The cumulative hazard value is calculated from this failure rate and plotted against the number of operating months. The failure occurrence is estimated based on the slope and intercept obtained by linearly approximating this plot. Weibull distribution is used as a model for estimation. Slope m and intercept m · lnη obtained by linear approximation
The failure occurrence is estimated by the function f (t) shown in the lower left, using as a Weibull parameter. The Weibull distribution is roughly classified into a pattern in which many initial failures occur when m <1 and a pattern in which many wear failures occur when m> 1. When m = 1, the failure occurrence is accidental.

Next, a method will be described in which groups are created for specific items such as product types, parts, and phenomena, and statistical processing and visualization are performed in group units. The contents of the method are illustrated in FIGS. 33 to 43. Hereinafter, this method will be referred to as a group analysis method or a group analysis.

First, a method of performing group classification for performing group analysis will be described. After the database is searched, the screen shown in FIG. 33 can be called by selecting the function of the schedule from the icons shown on the lower left side of the list in which the search results are expanded on the memory.
By selecting the group definition function from the icons shown on the lower left side of this screen, the group classification setting function shown in FIGS. 34 to 39 can be activated.

In the screen shown in FIG. 34, the group number is input for one group, the elements constituting the group are selected from the item list with the mouse, and then the name representing the content of the group is input. The group number is usually set from 1 to serial. Here, group number 1
Corresponding to, a group of abnormal phenomena related to sound is set. Whether or not the group has been set correctly can be displayed on the screen by selecting the confirmation icon, as shown in FIG. 35, the name of the group and the elements constituting the group.

Similarly, FIG. 36 is a screen for setting a group of phenomena related to poor cooling, and FIG. 37 is a screen for confirming the name and constituent elements of the group.
Further, FIG. 38 is a screen for setting a group of phenomena relating to poor heating and cooling, and FIG. 39 is a screen for confirming the name and constituent elements of the group.

Next, a method of analysis based on the set group classification will be described. A function suitable for the purpose is selected from the icon indicating the analysis function on the lower left side of the list in which the database search result shown in FIG. 40 is expanded on the memory. Here, by selecting the bar graph function for analyzing the distribution, the screen for setting the bar graph analysis conditions shown in FIG. 41 can be called up. In order to perform group analysis, an item called group display is set on the horizontal axis. The vertical axis shows the number of failures, repair amount, failure rate, etc.
One is selected from the evaluation criteria of types and set. FIG. 42
Is an example in which the horizontal axis represents the group display and the vertical axis represents the number of failures. FIG. 43 shows an example in which the analysis result is output as a bar graph based on this condition.

The data structure and processing procedure for implementing the group analysis method will be described below. First, FIG. 44 shows the structure of the source data management table.

The source data management table manages source data for analysis, and each row represents one data item. One piece of data is composed of a plurality of items (1 to L), and each item manages independent information. Specified data processing is performed on one vertical column of each item and data analysis is performed. An area for storing a group number is provided in advance at the rear (L + 1) of each line of data and used for group analysis. Figure 44 (b)
The source data number area of is for storing the number of data items stored in the source data management table. The group item number area of FIG. 44 (c) is for storing data item numbers (1 to L) for performing group analysis. The group data management table shown in FIG. 45 (a) is for managing the contents divided into groups, and includes a group name, the number of item codes, a code registration start position, and a code registration end position. The group name stores the name given to one group number divided into groups. The number of item codes stores the number of data contents of items belonging to the same group. The code registration start position stores the storage start position on the group item code table that stores the data of items belonging to the same group. The code registration end position stores the storage end position on the group item code table that stores the data of items belonging to the same group.

The group item code management table shown in FIG. 45 (b) is for storing item content data belonging to the same group when grouping is performed. The item code, the previous code storage position, and the next code are stored. It consists of storage location and group number. The item code stores each item content data of the same group. The previous code storage position stores the storage position of item content data belonging to the same group in the group item code table. For example, -1 is stored as an initial value, and this represents the first data item belonging to the same group. The next code storage position stores the storage position of the item content data belonging to the same group in the group item code management table. For example, if -1 is stored in this area, this indicates the end of the data belonging to the same group. When the table is searched based on the data of the previous code storage position and the next code storage position, all item content data belonging to the same group can be extracted. The group number stores the group number to which the item code belongs. The item code number area of FIG. 45 (c) stores the number of data registered in the group item code management table. The item data management table and group item code management table allow you to easily add and delete item codes,
The table has a bidirectional pointer at the code registration end position and the group number.

Next, FIG. 46 shows a processing procedure for setting a group classification. This processing flow is premised on that data is stored in advance in the source data management table and the source data number area in the data relation diagram shown in FIG.

(1) First, an area (group item number area, group data management table, group item code management table, item code number area) for setting group analysis information is initialized.

(2) Next, among the items stored in the source data management table, the group analysis target items are selected by a method such as a list or keyboard input. The selected item number is stored in the group item number area.

(3) For the items selected in (2),
The contents of the source data management table are totaled and the item code contents list belonging to the target item is created.

(4) A group number to be added to the same group for group analysis is set. This group number also corresponds to the position in the group data management table.

(5) From the list of item code contents created in (3), item codes belonging to the same group are selected by the required number of data. The group item code management table is searched to determine whether a preset item code of the same group and an item code of another group exist.
For the selected codes, after making them unselectable, the item codes are selected from the list of item code contents by the required number of data. The selected item code is sequentially registered in the group item code management table. The number of registered data is stored in the item code number area.

(6) When the processing of (5) is completed and the required number of data item codes have been selected, a group name representing the contents of this group is set. The group name is stored in the group data management table. Also,
The code storage positions of the same group set in (5) are stored at the address of the group item code management table in the number of item codes, the code registration start position, and the code registration end position of the group data management table.

(7) When one grouping is completed, a judgment is made as to whether or not this setting process is completed. When performing the setting again, the processes of (4) to (6) are repeated.

(8) When the setting of the group analysis information is completed, the group number is added to the group number storage position of the source data management table based on the information of the group data management table and the group item code management table. At that time, for item codes not registered in the group item code management table, values representing other item codes are stored. When this process is repeated for the number of data stored in the source data number area, the group number addition process is terminated. The data for which the process of adding the group number to the original data is completed can be processed as the L + 1th item of the original data.

As described above, the setting of group classification, the statistical processing and visualization for each group can be performed.

[0091]

According to the present invention, since the quality data retrieval / analysis can be performed entirely at the workstation,
The search / analysis request, processing, and output sequences are executed in real time, which has the effect of reducing the turnaround time of analysis work.

According to the present invention, when a product breaks down under the customer who is the user, information about repairs mainly performed on-site by a customer service representative such as a service person is provided for each repair. Through sales departments such as distributors and sales offices, it is possible to provide immediate feedback to the factory, which is the main body of product design, manufacturing, and inspection. As a result, it is possible to improve the design and improvement of the product in production and the evaluation / inspection method of the used parts at an early stage, and further improve the reliability in the design of the new product.

Since the search / analysis of the quality data can be performed entirely in the workstation, the sequence of search / analysis request, processing, and output can be executed in real time, so that the turnaround time of analysis work can be reduced.

Further, in the workstation, by searching the source data in record units and expanding it in the memory, analysis for each item of the data relating to the product failure and its repair and all combinations of the items are performed. Targeted analysis is possible.

[Brief description of drawings]

FIG. 1 shows an overall configuration of a system when the present invention is implemented.

FIG. 2 shows a card format for collecting or adding information when the present invention is implemented.

FIG. 3 shows the relationship between product failures and quality defects that actually occur in the field and their factors.

FIG. 4 is a diagram showing a configuration of a large-scale computer of each factory that stores and manages quality data for each factory received from the large-scale computer of FIG. 1 and a workstation that searches and analyzes the quality data.

FIG. 5 shows a software configuration for searching and analyzing quality data in the workstation 7 of FIG.

FIG. 6 shows the contents of processing for searching and analyzing quality data in a workstation when the present invention is implemented.

FIG. 7 shows a management item (a) and a data item (b) that serve as references when performing a numerical operation according to a request for analysis when the present invention is implemented.

FIG. 8 is a diagram showing a processing procedure for searching a large-sized computer and transferring the result when a search request is generated in the workstation according to the procedure 2 of FIG. 6;

9 is a diagram showing cases A to G of products of a set of data generated in the processing of FIG.

10 is a diagram showing cases H to J of products of a set of data generated in the processing of FIG.

FIG. 11 is a diagram showing a functional configuration diagram of the conditional expression management program of FIG. 8;

12 is a diagram showing a processing procedure of the conditional expression management program of FIG.

13 is a diagram showing a processing procedure for inquiring a conditional expression and creating an insufficient conditional expression when the function flag is 1 in FIG.

14 is a diagram showing a processing procedure for deleting a conditional expression and creating a deleted conditional expression when the function flag is 2 in FIG.

15 is a diagram showing a processing procedure for adding a conditional expression when the function flag is 3 in FIG.

16 is a diagram showing a processing procedure for changing and registering a resident conditional expression when the function flag is 4 in FIG.

17 is a diagram showing file specifications of a conditional expression file that stores a search conditional expression in the conditional expression management program of FIG.

18 is a diagram showing a file specification of an input / output parameter file for storing elements of a conditional expression for exchanging data with a higher-level program in the conditional expression management program of FIG.

FIG. 19 is a diagram showing a transition of the repair amount with respect to a failure year, which is an individual output format when performing numerical calculation and outputting the calculation result according to the analysis method in the case of implementing the present invention.

FIG. 20 is a bar graph showing the number of failures for each product type, which is an individual output format when numerical calculation is performed and the calculation result is output according to the analysis method in the case of implementing the present invention.

FIG. 21 is a palette diagram showing the number of failures for each product type, which is an individual output format when numerical calculation is performed and the calculation result is output according to the analysis method in the case of implementing the present invention.

FIG. 22 is a diagram showing the number of failures by product type and by part number, which is an individual output format when performing numerical calculation and outputting the calculation result according to the analysis method in the case of implementing the present invention.

FIG. 23 is a diagram showing a repair amount ratio for each product format, which is an individual output format when a numerical operation is performed and an operation result is output according to an analysis method in the case of implementing the present invention.

FIG. 24 is a diagram showing the number of failures with respect to the number of operating months, which is an individual output format when performing numerical calculation and outputting the calculation result according to the analysis method in the case of implementing the present invention.

FIG. 25 is a diagram showing a detailed list of service cards, which is an individual output format when numerical calculation is performed and a calculation result is output according to an analysis method in the case of implementing the present invention.

FIG. 26 is an output showing the repair amount occurrence status by product type, which is a factor in the process of pursuing these factors from the actual product failure or quality defect that occurs in the field when the present invention is implemented. The figure showing the process which narrows down.

FIG. 27 is an output showing the repair amount occurrence status for each part of the product format T, which is caused by the actual product failure or quality defect in the field when the present invention is carried out.
The figure showing the process which narrows down a factor in the process of pursuing those factors.

FIG. 28 is an output showing the amount of repaired money generated by each phenomenon of the compressor, and a process of pursuing these factors from a product failure or quality defect that actually occurs in the field when the present invention is implemented. The figure showing the process of narrowing down the factors.

FIG. 29 is a data relation diagram for performing the factor analysis process as illustrated in FIGS. 26 to 28.

FIG. 30 shows a flow for performing a factor analysis process as exemplified in FIGS. 26 to 28.

31 shows a flow for processing the contents of the source data management table according to the information of the analysis means management table of FIG.

FIG. 32 shows a specific process of a method of estimating a failure occurrence state centering on hazard analysis.

FIG. 33 shows a list of search results expanded on a memory for activating a group classification setting function when performing group analysis after performing a database search.

FIG. 34 shows an example of setting a group of abnormal phenomena related to sound corresponding to group number 1 for performing group analysis.

FIG. 35 shows a screen for confirming whether the name of the group and the elements forming the group in FIG. 34 have been set correctly.

FIG. 36 shows an example in which a group of phenomena related to poor cooling is set corresponding to group number 2 in order to perform group analysis.

FIG. 37 shows a screen for confirming whether the name of the group and the elements forming the group in FIG. 36 have been set correctly.

FIG. 38 shows an example of setting a group of phenomena related to defective heating and cooling in association with group number 3 for performing group analysis.

FIG. 39 shows a screen for confirming whether the name of the group and the elements forming the group in FIG. 38 have been set correctly.

FIG. 40 shows an icon representing an analysis function for activating a function suitable for the purpose of analysis based on the set group classification.

41 shows a screen for setting conditions for bar graph analysis, which is called up by selecting the bar graph function for analyzing distribution in FIG. 40. FIG.

FIG. 42 shows an example in which a group display is set on the horizontal axis and a failure number is set on the vertical axis for performing group analysis.

43 shows an example in which the analysis results are output as a bar graph based on the conditions set in FIG.

FIG. 44 shows a data structure of a source data management table for implementing the group analysis method.

FIG. 45 shows a data structure for a group data management table and a group item code management table for implementing the group analysis method.

FIG. 46 shows a processing procedure for setting group classification in the group analysis method.

[Explanation of symbols]

1 ... Search condition setting processing, 2 ... Data base function start processing, 3 ... Database search processing according to search conditions, 4 ... Host access processing, 5 ... Data processing item selection processing , 6 ... Processing for writing the search result as a primitive data set into the core memory, 7 ... Analysis method designation processing, 8 ... Analysis item selection processing as parameters, 9 ... Analysis processing.

Claims (1)

[Claims] 1. Regarding processing performed for product failure, information on failure occurrence such as date, phenomenon, and remedy is collected from the field and accumulated in a database, and a part or all of the information is related to the cause of product failure. -A group analysis method characterized in that, when analyzing a tendency of failure occurrence, a group is created for a specific item, and statistical processing and visualization can be performed for each group.