JP5311493B2 - Data analysis apparatus, manufacturing apparatus using the same, data analysis method, computer program thereof, and recording medium recording the program - Google Patents

Description

  The present invention relates to a technology for determining the quality of a product being manufactured, and in particular, a data analysis apparatus that performs a quality determination of a product using an MTS (Mahalanobis Taguchi system) method, a manufacturing apparatus using the same, a data analysis method, The present invention relates to the computer program and a recording medium on which the program is recorded.

  In recent years, in various production systems, techniques for determining the quality of products using the MTS method have been actively developed. As technologies related to this, there are inventions disclosed in the following Patent Documents 1 to 3.

  Patent document 1 aims at providing the management method which can determine the abnormal operation | movement of a semiconductor manufacturing apparatus correctly. A plurality of data is sampled for at least one parameter under the normal operation state of the semiconductor manufacturing apparatus, and the Mahalanobis space A is created based on the sampled data group. Then, based on this Mahalanobis space, the Mahalanobis distance D2 is calculated from the measured values of the parameters obtained in the operating state of the semiconductor manufacturing apparatus, and when the Mahalanobis distance exceeds a predetermined value, the semiconductor manufacturing apparatus is abnormal. It is determined that an action has occurred.

  Patent Document 2 makes it possible to accurately determine whether a product is good or bad by dynamically changing items used for determination every manufacturing time using the MTS method in order to cope with a change in phenomenon. For the purpose. In the case of an injection molding machine, n Mahalanobis distances are calculated using the actual values of a plurality of items M for each shot according to a combination of orthogonal tables. Then, the SN ratio for n ways is calculated from the Mahalanobis distance. Then, using the calculated S / N ratio, the level average of each item (the first level and the second level in the orthogonal table) is calculated, and the level difference (first level average−second level average) is large. Find items. Using these optimal m items, m Mahalanobis distances are calculated again, and if any one of the m distances exceeds the reference value, it is determined as defective.

  Patent Document 3 provides a product quality determination method for an injection molding machine that can accurately determine the quality of a product even when the plasticization state of the resin changes due to the passage of time or environmental changes. The purpose is to do. Select two or more molding data that change in relation to the molding data that affects the quality of the product. Combine these molding data and perform calculations using functional expressions such as multiplication and division. Based on the above, the acceptable range of good product molding is determined by the upper limit value and the lower limit value, and is set in the control device of the injection molding machine. In the molding operation, the value of the molding data item is detected for each molding cycle, and the calculation is performed using the same function formula, and the quality of the product for each molding cycle is determined depending on whether the calculation result is within the allowable range. Determine. Since the quality of the product is determined based on the result of the evaluation of multiple molding data combined, the quality of the product is suitable even when the value of the molding data fluctuates over time or due to environmental changes. It can be determined accurately.

JP 2000-114130 A JP 2003-181874 A JP-A-7-108579

  In the product quality determination methods disclosed in Patent Documents 1 to 3 described above, the quality determination is performed at the final stage of the product. Therefore, when the product is determined to be defective, the product is discarded. In addition, there is a problem that the economic efficiency deteriorates because it is necessary to make adjustments.

  In addition, when there are multiple product manufacturing processes and a semi-finished product is passed through the intermediate stage of manufacturing, it is possible to judge the quality of the semi-finished product and the quality of the final product with the manufacturing parameters alone. In addition, there is a problem that correction during the manufacturing process is difficult.

  Further, Patent Documents 2 and 3 relate to quality determination of an injection molded product, and a reference space is created from manufacturing data of good products in injection molding. Then, the quality of the product currently being manufactured is determined using the created reference space. However, if the injection-molded product that is the final product goes through a semi-finished product state, the quality judgment in the semi-finished product state and the quality judgment of the final product cannot be performed by manufacturing parameters alone. It is difficult to correct.

  The present invention has been made in order to solve the above-mentioned problems, and its purpose is data that can stabilize the final characteristics of the product at an early stage and prevent the production of defective products. An analysis apparatus, a manufacturing apparatus using the same, a data analysis method, a computer program thereof, and a recording medium on which the program is recorded are provided.

  According to an aspect of the present invention, the data analysis apparatus includes: a first calculation unit that calculates a first Mahalanobis distance from evaluation characteristic data in an intermediate process having a large correlation with the characteristic of the final product; A first analysis means for specifying evaluation characteristic data, which is an increase factor of the first Mahalanobis distance calculated by the means, by factor analysis; and a second Mahalanobis distance is calculated from the manufacturing parameter data measured in the intermediate process. The manufacturing parameter data, which is an increase factor of the second Mahalanobis distance calculated by the second calculation means, is obtained by factor analysis with respect to the evaluation characteristics specified by the second calculation means and the first analysis means. The second analyzing means is arranged such that the distance between the second analyzing means to be identified and the second Mahalanobis calculated by the second calculating means is reduced. Therefore including an adjustment means for adjusting the identified production parameter item.

  Preferably, the adjustment unit adjusts the manufacturing parameter item specified by the second analysis unit so that the first Mahalanobis distance calculated by the first calculation unit is equal to or less than a threshold value.

  Preferably, the first analysis unit assigns the item of the evaluation characteristic to the orthogonal table, and increases the first Mahalanobis distance according to the first Mahalanobis distance calculated for each combination of the items. The evaluation characteristic data is specified.

  Preferably, the second analysis unit assigns the item of the manufacturing parameter to the orthogonal table, and increases the second Mahalanobis distance according to the second Mahalanobis distance calculated for each combination of the items. The manufacturing parameter item is specified.

  According to another aspect of the present invention, a manufacturing apparatus for determining the quality of a product in an extrusion molding machine, which cools a molding product manufactured by an extrusion molding machine that applies a resin coating to a metal material and an extrusion molding machine. A cooling machine, a first measuring machine for measuring the resin thickness of a molded product produced by the extrusion molding machine, a second measuring machine for measuring the cross-sectional shape of the molded product produced by the extrusion molding machine, and extrusion molding. And a data analysis device for analyzing data collected from the first measurement device and the second measurement device, the data analysis device being measured by at least the first measurement device and the second measurement device. A first calculation means for calculating the first Mahalanobis distance from the evaluation characteristic data obtained, and an evaluation characteristic which is an increase factor of the first Mahalanobis distance calculated by the first calculation means. First analysis means for identifying data by factor analysis; second calculation means for calculating a second Mahalanobis distance from manufacturing parameter data measured at least in an extruder and a cooler; and first analysis means Second analysis means for specifying, by factor analysis, manufacturing parameter data that is an increase factor of the second Mahalanobis distance calculated by the second calculation means for the evaluation characteristics specified by Adjusting means for adjusting the manufacturing parameter item specified by the second analysis means so that the distance of the second Mahalanobis calculated by the means is reduced.

  According to still another aspect of the present invention, there is provided a data analysis method for causing a computer to judge whether a product is good or bad, wherein the computer uses the evaluation characteristic data in an intermediate process having a large correlation with the characteristic of the final product to A step of calculating the Mahalanobis distance, a step of identifying the evaluation characteristic data that causes an increase in the calculated first Mahalanobis distance by factor analysis, and a second Mahalanobis from the manufacturing parameter data measured in the intermediate process A step of calculating a distance; and a step of specifying, by factor analysis, manufacturing parameter data that causes an increase in the distance of the second Mahalanobis calculated by the second calculation means with respect to the specified evaluation characteristic. The specified manufacturing parameter term so that the distance of the second Mahalanobis is reduced. The and a step of adjusting.

  According to still another aspect of the present invention, there is provided a computer program for causing a computer to judge whether a product is good or bad, and the computer is configured to perform evaluation based on evaluation characteristic data in an intermediate process having a large correlation with the characteristic of the final product. A step of calculating a distance of one Mahalanobis, a step of specifying evaluation characteristic data that is an increase factor of the calculated distance of the first Mahalanobis by factor analysis, and a second step based on manufacturing parameter data measured in an intermediate process. A step of calculating a Mahalanobis distance; a step of specifying, by factor analysis, manufacturing parameter data that causes an increase in the second Mahalanobis distance calculated by the second calculation means for the specified evaluation characteristic; Identified so that the calculated second Mahalanobis distance is reduced And a step of adjusting the granulation parameter item.

  According to still another aspect of the present invention, there is provided a computer-readable recording medium having recorded thereon a computer program for causing a computer to judge whether the product is good or bad, and the computer correlates with the characteristics of the final product. A step of calculating the distance of the first Mahalanobis from the evaluation characteristic data in the intermediate process having a large value, a step of specifying the evaluation characteristic data that causes an increase in the calculated distance of the first Mahalanobis by factor analysis, and an intermediate process A step of calculating the second Mahalanobis distance from the measured manufacturing parameter data, and a manufacturing parameter that causes an increase in the second Mahalanobis distance calculated by the second calculating means for the specified evaluation characteristic A step of identifying the data by factor analysis and a calculated second Maha So that the distance Nobis decreases, and a step of adjusting the identified production parameter item.

  According to an aspect of the present invention, the adjustment unit adjusts the manufacturing parameter item specified by the second analysis unit so that the distance of the second Mahalanobis calculated by the second calculation unit is decreased. Thus, the final characteristics of the product can be stabilized at an early stage, and the production of defective products can be prevented in advance.

  Also, the adjustment means adjusts the manufacturing parameter item specified by the second analysis means so that the first Mahalanobis distance calculated by the first calculation means is equal to or less than the threshold value. It is possible to adjust the corresponding manufacturing parameter item.

  In addition, the first analysis unit assigns the evaluation characteristic item to the orthogonal table, and causes the first Mahalanobis distance to increase according to the first Mahalanobis distance calculated for each combination of items. Since the evaluation characteristic data is specified, it is possible to accurately specify the evaluation characteristic data.

  Further, the second analysis unit assigns the item of the manufacturing parameter to the orthogonal table, and causes the second Mahalanobis distance to increase according to the distance of the second Mahalanobis calculated for each combination of items. Since the manufacturing parameter item is specified, it is possible to accurately specify the manufacturing parameter item.

It is a figure for demonstrating the concept of the quality determination method of the data analyzer in embodiment of this invention. It is a block diagram which shows an example of the hardware constitutions of the data analysis apparatus in embodiment of this invention. It is the block diagram which showed the data analyzer in embodiment of this invention by the functional structure. It is a flowchart for demonstrating the process sequence of the data analyzer in embodiment of this invention. It is a figure which shows an example of the orthogonal table | surface used when performing a factor analysis. It is a figure for demonstrating the loss function in case there exists target value. It is a figure which shows the specific example of the manufacturing apparatus using the data analysis apparatus in embodiment of this invention. It is a figure which shows the cross-sectional shape of the molded product manufactured by the extrusion molding machine. It is a figure which shows the time-dependent change of MD1 in a manufacture process. It is a graph which shows the result of having implemented factor analysis at P1 time shown in FIG. Change over time of Mahalanobis distance MD2 obtained by using 12 items of manufacturing parameter data, such as resin temperature, mold temperature, cooling water temperature, etc., for the resin thickness 3, when the region where the coating thickness 3 is stable is used as a reference space. FIG. It is a graph which shows the result of having performed the factor analysis at the time of implementation of the parameter adjustment shown in FIG.

  FIG. 1 is a diagram for explaining the concept of a quality determination method for a data analysis apparatus according to an embodiment of the present invention. For example, it is assumed that there are evaluation characteristics A to M of an intermediate product (semi-finished product) having a large correlation with the characteristic Y required for the final product. In addition, it is assumed that items a to m of manufacturing parameters exist in the manufacturing process of the intermediate product.

  First, the Mahalanobis distance MD1 is calculated from evaluation data A to M in an intermediate process (manufacturing process of semi-finished product) having a large correlation with the characteristic Y of the final product as shown in FIG. At this time, data in a region where the characteristic Y of the final product is stable is used for the reference space.

  FIG. 1B is a graph showing the relationship between the Mahalanobis distance MD1 and the characteristic Y of the final product, and shows that the correlation between MD1 and the characteristic Y is large. FIG. 1C is a graph showing the calculation result of MD1.

  Further, the Mahalanobis distance MD2 is calculated from the data a to m of the manufacturing parameter items of the intermediate product as shown in FIG. At this time, a region where the evaluation characteristics A to M are stable is set as a reference space, and MD2 is calculated for each. FIG. 1E is a graph showing the calculation result of MD2.

  Next, the value of MD1 in the intermediate process is monitored, and an item of evaluation data that greatly contributes to an increase in MD1 is specified by factor analysis. Here, as shown in FIG. 1 (f), for example, it is assumed that item A of the evaluation data is specified as an item of evaluation data that greatly contributes to an increase in MD1.

  The Mahalanobis distance MD2 is calculated using the manufacturing parameter item data a to m using the stable region of the item A as a reference space. Then, a factor analysis is performed on MD2, and a manufacturing parameter item that greatly contributes to an increase in MD2 of item A is specified. FIG.1 (g) is a factor analysis figure for every manufacturing parameter item am. The manufacturing parameter item c is identified as a manufacturing parameter item that greatly contributes to the increase in MD2 of item A.

  Finally, the manufacturing parameter item c is adjusted in the direction in which MD2 becomes smaller. As MD2 decreases, MD1 also decreases. As a result, the characteristic Y of the final product can be stabilized.

  FIG. 2 is a block diagram illustrating an example of a hardware configuration of the data analysis apparatus according to the embodiment of the present invention. This data analysis device is realized by a general computer, and includes a computer main body 1, a display device 2, an FD drive 3 in which an FD (Flexible Disk) 4 is mounted, a keyboard 5, a mouse 6, and a CD-ROM. A CD-ROM device 7 on which (Compact Disc-Read Only Memory) 8 is mounted and a network communication device 9 are included.

  The data analysis program is supplied by a recording medium such as FD4 or CD-ROM8. The data analysis program is executed by the computer main body 1 to perform data analysis. The data analysis program may be supplied to the computer main body 1 from another computer via a communication line.

  The recording medium is not limited to the FD 4 and the CD-ROM 8, but may be a magnetic tape, an MO (Magnetic Optical disk), an MD (Mini Disc), a DVD (Digital Versatile Disc), an IC (Integrated Circuit) card, or the like. May be.

  The computer main body 1 includes a CPU 10, a ROM (Read Only Memory) 11, a RAM (Random Access Memory) 12, and a hard disk 13. The CPU 10 performs processing while inputting / outputting data to / from the display device 2, FD drive 3, keyboard 5, mouse 6, CD-ROM device 7, network communication device 9, ROM 11, RAM 12 or hard disk 13. The data analysis program recorded on the FD 4 or the CD-ROM 8 is stored in the hard disk 13 by the CPU 10 via the FD drive 3 or the CD-ROM device 7. The CPU 10 performs data analysis by appropriately loading a data analysis program from the hard disk 13 into the RAM 12 and executing it.

  FIG. 3 is a block diagram showing the data analysis apparatus according to the embodiment of the present invention by a functional configuration. This data analysis apparatus includes a first Mahalanobis distance calculation unit 21, a first factor analysis unit 22, a second Mahalanobis distance calculation unit 23, a second factor analysis unit 24, and a manufacturing parameter item adjustment unit 25. These configurations are realized by the CPU 10 executing the data analysis program. The data analysis device further includes a data storage unit 27.

  The data storage unit 27 stores evaluation characteristics measured by sensors arranged in the manufacturing process, data of manufacturing parameter items, and the like. The data storage unit 27 corresponds to the hard disk 13, RAM 12, and the like shown in FIG.

  The first Mahalanobis distance calculation unit 21 acquires evaluation data stored in the data storage unit 27 in an intermediate process (manufacturing process of a semi-finished product) having a large correlation with the characteristics of the final product, and the Mahalanobis distance MD1 from the evaluation data. Calculate

  The first factor analysis unit 22 monitors the value of MD1 calculated by the first Mahalanobis distance calculation unit 21. Then, by performing factor analysis on MD1 while referring to data stored in the data storage unit 27, an item of evaluation data that greatly contributes to an increase in MD1 is specified.

  The second Mahalanobis distance calculation unit 23 acquires the data of the manufacturing parameter item of the intermediate product stored in the data storage unit 27, and calculates the Mahalanobis distance MD2 from the data of the manufacturing parameter item.

  The second factor analysis unit 24 identifies a manufacturing parameter item that greatly contributes to an increase in MD2 by performing factor analysis on MD2 while referring to data stored in the data storage unit 27.

  The manufacturing parameter item adjustment unit 25 adjusts the manufacturing parameter item in the direction in which MD2 becomes smaller with respect to the manufacturing parameter item determined to have a large contribution to the increase in MD2 by the second factor analysis unit 24.

  FIG. 4 is a flowchart for explaining the processing procedure of the data analysis apparatus according to the embodiment of the present invention. Here, the final product Z is manufactured through a plurality of processes and has a required product characteristic Y. The semi-finished products in the intermediate process of the final product Z are V, W, and X. Also, evaluation characteristics having a high correlation with the product characteristics Y required for the final product Z are denoted by A to M.

  First, the first Mahalanobis distance calculator 21 calculates the Mahalanobis distance MD1 for each of the semi-finished products V, W, and X using the data of the evaluation characteristics A to M when the product characteristic Y is stable (S11). .

Here, a method for calculating the Mahalanobis distance will be briefly described.
The data obtained by the sensors arranged in the manufacturing process is Y _ip , the number of data items is i = 1 to k, the number of data sets is p = 1 to n, and the average value m _i of the data Y _ip for each data item And the standard deviation σ _i is calculated.

Next, the standardized value y _ip is calculated by the following equation (1) using the average value _mi and the standard deviation σ _i .

Next, using the normalized value y _ip , a correlation matrix R of the normalized values is created by the following equations (2) and (3).

  Next, an inverse matrix A of the correlation matrix R is created by the following equation (4).

Finally, to calculate the Mahalanobis distance D ² by the following equation (5).

  Returning to the description of the processing procedure shown in FIG. Next, the first factor analysis unit 22 monitors each MD1 of the semi-finished products V, W, and X calculated by the first Mahalanobis distance calculation unit 21, and evaluates characteristics that greatly contribute to an increase in MD1 by factor analysis. Specify (S12).

Here, a method of factor analysis will be briefly described.
FIG. 5 is a diagram illustrating an example of an orthogonal table used when factor analysis is performed. Assuming that there are k items used for the calculation of the Mahalanobis distance MD, the items I _{1 to} I _k are assigned to the 1 to k columns of the orthogonal table. The orthogonal table used at this time is the smallest one among the orthogonal tables in which the number of columns is larger than the number of items. As shown in FIG. 5, when the number of items m = 9, an L12 orthogonal table is used.

  Here, for the two levels of the orthogonal table, the first level = use the item, and the second level = do not use the item, the items used for the MD calculation for each row of the orthogonal table are described. To do.

  Next, for each combination of items used for the MD calculation selected from the orthogonal table, the average value, the standard deviation, and the inverse matrix of the correlation matrix are calculated using the reference space data.

Next, for the designated n sampling data groups y _i = (y _i1 , y _i2 , y _i3 ,..., Y _ik ) (i = 1 to n), D ² _ji which is an MD value (J = 1 to L, i = 1 to n) is calculated for each combination of items set in the orthogonal table.

Then, the SN ratio η = (η ₁ , η ₂ ,..., Η _L ) for each combination of items is obtained by the following equation (6).

Then, for each item I ₁ ~I _k, η _j the mean value of the row is level 1, computes the eta _j average value of the row is level 2, respectively, to the values and respective SN ratios. At this time, the S / N ratio is also calculated for the orthogonal table column to which no item is assigned. For example, if a table entry I ₂ shown in FIG. 5, as follows.

Level 1 SN ratio = (η ₁ + η ₂ + η ₃ + η ₇ + η ₈ + η ₉ ) / 6 (7)
Level 2 SN ratio = (η ₄ + η ₅ + η ₆ + η ₁₀ + η ₁₁ + η ₁₂ ) / 6 (8)
Here, it can be estimated that a large value of (level 1 SN ratio) − (level 2 SN ratio) is a cause item for an increase in MD.

  Returning to the description of the processing procedure shown in FIG. Next, the second Mahalanobis distance calculation unit 23 acquires the data a to m of the manufacturing parameter items measured in the manufacturing process of the semi-finished products V, W, and X stored in the data storage unit 27, and manufacturing parameter data To calculate the Mahalanobis distance MD2 (S13). At this time, MD2 is calculated using the above formulas (1) to (5), with each area of the evaluation characteristics A to M being a stable area as a reference space.

  Next, the second factor analysis unit 24 specifies a manufacturing parameter item that causes an increase in MD2 by factor analysis for an evaluation characteristic highly contributing to an increase in MD1 (S14). At this time, the same method as the cause analysis described with reference to FIG. 5 is used.

  Next, the manufacturing parameter item adjustment unit 25 adjusts the value of the manufacturing parameter with respect to the manufacturing parameter item specified by the second factor analysis unit 24 so that MD2 decreases (S15).

  Then, the value of MD1 is calculated again, and it is determined whether MD1 is equal to or less than the threshold value (S16). If MD1 is larger than the threshold value (S16, No), the process returns to step S15 to adjust the manufacturing parameter item. If MD1 is equal to or less than the threshold value (S16, Yes), the process is terminated.

As the threshold value of MD1, the loss L is calculated using the loss function of the following equation (9), and the value at which L reaches A ₀ (consumer loss) is used.

Where Δ is the tolerance of the target characteristic of the product, A is the loss (yen) when the target characteristic is outside the tolerance, B is the inspection cost (yen) per characteristic of the process, and C is the process The characteristic correction cost (yen) at n, n ₀ indicates the current measurement interval of the characteristic in the process, and u ₀ indicates the current average correction interval of the characteristic in the process.

In addition, n is the optimum measurement interval of the characteristic in the process, u is the optimum average correction interval of the characteristic in the process, D ₀ is the characteristic correction limit in the process, D is the optimum correction limit of the characteristic in the process, and l is the time lag of the characteristic in the process. , Σ _m represents a standard deviation of measurement errors of characteristics in the process.

  The optimum measurement interval n, optimum average correction interval u, and optimum correction limit D of the characteristics in the process are expressed by the following equations (10) to (12).

FIG. 6 is a diagram for explaining a loss function when there is a target value. In FIG. 6, the target value m is 0 (yen), and the loss L (yen) increases as the tolerance increases. In the present embodiment, the value of MD1 when loss L reaches consumer loss A ₀ (allowable limit upper limit m + Δ ₀ ) is set as a threshold value.

(Concrete example)
FIG. 7 is a diagram showing a specific example of a manufacturing apparatus using the data analysis apparatus in the embodiment of the present invention. This manufacturing apparatus includes a data analysis device 1, an extrusion molding machine 31, a data collection device 32, a parameter control unit 33, a preheating machine 34, a cooling machine 35, a resin thickness measuring machine 36, and an outer shape measurement. Machine 37.

  This manufacturing apparatus applies a resin coating to a metal material a plurality of times by an extrusion molding machine 31, creates a composite material of the metal material and resin, prevents contact wear and fatigue deterioration with other parts, and lengthens the molded product. It will make it last longer.

  The extrusion forming machine 31 coats and coats a polyethylene resin (PE resin) 38 on the surface of the metal material to be coated. At this time, a crosslinking reaction material 39 is used for curing. In order to continuously produce products, the total length of one product reaches several hundred meters to several thousand meters.

  The preheating machine 34 preheats a metal material or the like and conveys it to the extrusion molding machine 31. The cooler 35 cools the intermediate product and the final product molded by the extrusion molding machine 31. The resin thickness measuring device 36 measures the coating thickness of the intermediate product and the final product. Further, the outer shape measuring machine 37 measures the outer shape of the intermediate product or the final product.

  The data collection device 32 includes a preheating temperature by the preheating machine 34, a molding temperature in the extrusion molding machine 31, an extrusion speed, a molding pressure, a motor current, a resin temperature, a crosslinking material amount, a cooling temperature by the cooling machine 35, and a film thickness measuring machine. The product specifying data and the like, which are the measurement results obtained by 36 and the outer shape measuring machine 37, are collected and output to the data analysis apparatus 1. Note that sensors are provided in the preheating machine 34, the extrusion molding machine 31, the cooling machine 35, and the measurement results by the sensors are output to the data collection device 32.

  The parameter control unit 33 controls manufacturing parameter items of the extrusion molding machine 31 by a feedback signal output from the data analysis apparatus 1.

  FIG. 8 is a view showing a cross-sectional shape of a molded product manufactured by the extrusion molding machine 31. FIG. 8A shows an intermediate product after the PE resin 38 is coated once. As shown in FIG. 8B, the final product is manufactured by coating the surface with a metal material and repeating the coating with the PE resin 38 three times.

  In the manufacturing apparatus as shown in FIG. 7, the tensile strength after manufacturing the final product and the tensile strength after the bending fatigue test are the evaluation characteristics of the final product. However, the correlation between the tensile strength, which is the final product characteristic, and the resin parameter, mold temperature, cooling temperature, etc., which are manufacturing parameter data of the intermediate product (semi-finished product), is low.

  In this embodiment, as shown in FIG. 8, when coating with PE resin 38, the thickness of the coating is measured at four locations, and the cross-sectional shape of the product after coating is measured at two locations, and these are used as evaluation characteristic data. Data collection device 32 collects. This evaluation characteristic data has a high correlation with a correlation coefficient of 0.7 or more with the tensile strength and the tensile strength after the bending fatigue test, which are actual product characteristics.

  The first Mahalanobis distance calculation unit 21 calculates the Mahalanobis distance MD1 using the coating thickness data and the cross-sectional shape data after coating. Here, as the data used for the MD1 reference space, the evaluation characteristic data when the tensile strength and the tensile strength after the bending fatigue test are high are used.

  FIG. 9 is a diagram showing a change with time of MD1 in the manufacturing process. As shown in FIG. 9, MD1 tends to increase as the elapsed time increases.

  FIG. 10 is a graph showing the results of the factor analysis performed at time P1 shown in FIG. As shown in FIG. 10, when the resin thickness 3 is used as an item (level 1) and when it is not used (level 2), the difference in MD value is large, and it is estimated that it is a cause item.

  FIG. 11 shows the Mahalanobis distance obtained when the region where the coating thickness 3 is stable is defined as the reference space using the manufacturing parameter data of 12 items such as the resin temperature, the mold temperature and the cooling water temperature. It is a figure which shows the time-dependent change of MD2. As shown in FIG. 11, a tendency to increase with elapsed time is also observed for MD2.

  FIG. 12 is a graph showing the results of factor analysis performed at the time of parameter adjustment shown in FIG. As shown in FIG. 12, as a result of analyzing the cause items of the increase in MD2, it is estimated that the resin pressure, the screw rotation speed of the extrusion molding machine 31, and the cooling temperature are the cause items. From the result, the manufacturing parameter item adjustment unit 25 adjusts the value of the manufacturing parameter item so that the resin pressure, the screw rotation speed of the extrusion molding machine 31, and the cooling temperature are close to the initial values. Then, the parameter control unit 33 is controlled so that the manufacturing parameter item is adjusted by the feedback signal.

  As shown in FIG. 11, the value of MD2 decreased after the parameter adjustment. Moreover, as shown in FIG. 9, the value of MD1 also decreased. As described above, even in the final product characteristics, a good product can be stably produced without producing a defective product.

  As described above, according to the data analysis apparatus of the present embodiment, the manufacturing parameter item that causes the increase in MD2 is identified by factor analysis for the evaluation characteristics that contribute to the increase in MD1, and the manufacturing parameter is determined. Since the items are adjusted, the final characteristics of the product can be stabilized at an early stage, and the production of defective products can be prevented in advance.

  In addition, even when the final product is manufactured through multiple intermediate processes (semi-finished products) and it takes a long time to finalize the product, it is possible to stabilize the characteristics of the final product at an intermediate stage of manufacturing. It became.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  DESCRIPTION OF SYMBOLS 1 Computer main body, 2 Display apparatus, 3 FD drive, 4 FD, 5 Keyboard, 6 Mouse, 7 CD-ROM apparatus, 8 CD-ROM, 9 Network communication apparatus, 10 CPU, 11 ROM, 12 RAM, 13 Hard disk, 21 1st Mahalanobis distance calculation part, 22 1st factor analysis part, 23 2nd Mahalanobis distance calculation part, 24 2nd factor analysis part, 25 Manufacturing parameter item adjustment part, 27 Data storage part, 31 Extrusion machine, 32 Data collection device , 33 Parameter control unit, 34 Preheating machine, 35 Cooling machine, 36 Resin thickness measuring machine, 37 Outer shape measuring machine.

Claims (8)

A first calculating means for calculating a first Mahalanobis distance from evaluation characteristic data in an intermediate process having a large correlation with the characteristics of the final product;
First analysis means for specifying, by factor analysis, evaluation characteristic data that is an increase factor of the first Mahalanobis distance calculated by the first calculation means;
Second calculating means for calculating a second Mahalanobis distance from the manufacturing parameter data measured in the intermediate step;
A second analysis that specifies, by factor analysis, manufacturing parameter data that is an increase factor of the second Mahalanobis distance calculated by the second calculation unit with respect to the evaluation characteristic specified by the first analysis unit. Means,
A data analysis apparatus comprising: an adjustment unit that adjusts a manufacturing parameter item specified by the second analysis unit such that a distance of the second Mahalanobis calculated by the second calculation unit decreases.   The said adjustment means adjusts the manufacturing parameter item specified by the said 2nd analysis means so that the distance of the 1st Mahalanobis calculated by the said 1st calculation means may be below a threshold value. The data analysis apparatus according to 1.   The first analysis means assigns the item of the evaluation characteristic to the orthogonal table, and causes an increase in the distance of the first Mahalanobis according to the distance of the first Mahalanobis calculated for each combination of items. The data analysis device according to claim 1 or 2, wherein evaluation characteristic data is specified.   The second analysis means assigns the item of the manufacturing parameter to the orthogonal table, and causes an increase in the distance of the second Mahalanobis according to the distance of the second Mahalanobis calculated for each combination of items. The data analysis device according to any one of claims 1 to 3, wherein a manufacturing parameter item is specified. A manufacturing apparatus for determining the quality of a product in an extrusion molding machine,
An extrusion molding machine for applying a resin coating to a metal material;
A cooler for cooling the molded product produced by the extrusion molding machine;
A first measuring machine for measuring a resin thickness of a molded product produced by the extrusion molding machine;
A second measuring machine for measuring a cross-sectional shape of a molded product produced by the extrusion molding machine;
A data analysis device for analyzing data collected from the extrusion molding machine, the cooling machine, the first measuring machine and the second measuring machine,
The data analysis device includes: a first calculation unit that calculates a distance of the first Mahalanobis from evaluation characteristic data measured by at least the first measuring instrument and the second measuring instrument;
First analysis means for specifying, by factor analysis, evaluation characteristic data that is an increase factor of the first Mahalanobis distance calculated by the first calculation means;
A second calculating means for calculating a second Mahalanobis distance from manufacturing parameter data measured at least in the extruder and the cooler;
A second analysis that specifies, by factor analysis, manufacturing parameter data that is an increase factor of the second Mahalanobis distance calculated by the second calculation unit with respect to the evaluation characteristic specified by the first analysis unit. Means,
And an adjusting unit that adjusts the manufacturing parameter item specified by the second analyzing unit so that the second Mahalanobis distance calculated by the second calculating unit decreases. A data analysis method for causing a computer to judge the quality of a product,
Causing the computer to calculate a first Mahalanobis distance from evaluation characteristic data in an intermediate process having a large correlation with the characteristics of the final product;
Causing the computer to specify, by factor analysis, evaluation characteristic data that is an increase factor of the calculated first Mahalanobis distance;
Causing the computer to calculate a second Mahalanobis distance from manufacturing parameter data measured in the intermediate process;
Causing the computer to specify, by factor analysis, manufacturing parameter data that causes an increase in the second Mahalanobis distance calculated by the second calculating means for the specified evaluation characteristics;
Causing the computer to adjust the identified manufacturing parameter item such that the calculated second Mahalanobis distance is reduced. A computer program for causing a computer to judge the quality of a product,
Causing the computer to calculate a first Mahalanobis distance from evaluation characteristic data in an intermediate process having a large correlation with the characteristics of the final product;
Causing the computer to specify, by factor analysis, evaluation characteristic data that is an increase factor of the calculated first Mahalanobis distance;
Causing the computer to calculate a second Mahalanobis distance from manufacturing parameter data measured in the intermediate process;
Causing the computer to specify, by factor analysis, manufacturing parameter data that causes an increase in the second Mahalanobis distance calculated by the second calculating means for the specified evaluation characteristics;
Causing the computer to adjust the identified manufacturing parameter item such that the calculated second Mahalanobis distance is reduced. A computer-readable recording medium having recorded thereon a computer program for causing a computer to judge the quality of a product,
Causing the computer to calculate a first Mahalanobis distance from evaluation characteristic data in an intermediate process having a large correlation with the characteristics of the final product;
Causing the computer to specify, by factor analysis, evaluation characteristic data that is an increase factor of the calculated first Mahalanobis distance;
Causing the computer to calculate a second Mahalanobis distance from manufacturing parameter data measured in the intermediate process;
Causing the computer to specify, by factor analysis, manufacturing parameter data that causes an increase in the second Mahalanobis distance calculated by the second calculating means for the specified evaluation characteristics;
And causing the computer to adjust the identified manufacturing parameter item so that the calculated second Mahalanobis distance is reduced.

Images