Information processing methods, recording media containing programs, and information processing devices

Abstract

An information processing method, a recording medium containing a program, and an information processing device are proposed. The information processing method includes: creating a first table (S101); using the first table containing the target variable obtained when the experimental conditions of the first control factor are fixed at a first level value, calculating a first response surface (S103); if the first response surface does not contain the target value related to the target variable ("No" in S104), adding experimental conditions of the first control factor with a first level value to the first table (S105); creating a second table by adding combinations of experimental conditions of multiple control factors with a first plurality of level values ​​and the first control factor with a second plurality of level values ​​to the first table (S109); calculating a second response surface containing the aforementioned target value using the second table (S115); and outputting the second response surface (S116).

Description

Technical Field

[0001] This disclosure relates to a method for processing design process parameters, a recording medium containing a program for executing the information processing method on a computer, and an apparatus for processing design process parameters. Background Technology

[0002] In recent years, due to the increasing complexity of control processes, it has become necessary to set multiple process parameters (also known as control parameters) for various controlled objects. For example, in automobile engine control, semiconductor device manufacturing, or pharmaceutical manufacturing, it is necessary to optimize and implement multiple control parameters. To set such control parameters, the following operation is performed: experimental planning is used to search for the optimal conditions of control parameters for each controlled object.

[0003] In this case, for example, in the experimental planning and setting method for engine control parameters, there are situations where normal data cannot be obtained due to engine misfire or other reasons. A known method exists that, even in the case of missed measurement points caused by engine misfire or other reasons, can efficiently ensure the accuracy of the model for each characteristic by effectively utilizing normal data already obtained through experiments and performing fewer additional experiments (see, for example, Patent Document 1).

[0004] Here, if the number of control parameters increases, the number of experimental candidate points increases dramatically. For example, with 5 control parameters, the number of experimental points in the central composite programming is 29, and the total number of candidate points is 5 if each parameter has 5 levels. 5 -29 = 3096. Furthermore, if 3 out of the 29 experimental points in the central composite programming were missed, and 6 additional experimental points were selected to replace those missed points, then a combination of 6 points must be chosen from the 3096 candidate points, i.e., from 3096C6 = 10. 18 Choose one from the options. If the number of control parameters becomes 6, then the combination can be further increased.

[0005] To address this problem, an experimental planning and setting method (see, for example, Patent Document 2) has been considered that can efficiently and accurately set control parameters for additional experimental points. The experimental planning and setting method described in Patent Document 2 includes: a step of determining whether there are any missed test points among a first given number of experimental points; and a step of setting a second given number of additional experimental points. Regarding how to change multiple control parameters based on the missed test points, the step of pre-assigning multiple search directions to multiple priority orders and setting additional experimental points includes: applying the control parameters of the missed test points sequentially from the search direction with the highest priority among the multiple search directions, setting candidate points for additional experimental points by changing them until a necessary number is reached; and a step of selecting additional experimental points from the necessary number of candidate points. Thus, additional experimental points can be set efficiently and with high accuracy.

[0006] Prior art literature

[0007] Patent documents

[0008] Patent Document 1: Japanese Patent Application Publication No. 2006-17698

[0009] Patent Document 2: Japanese Patent Application Publication No. 2008-241337 Summary of the Invention

[0010] One aspect of this disclosure relates to an information processing method comprising: creating a first table, based on first plurality of level values ​​set for each of a plurality of control factors, using an experimental planning method, representing combinations of experimental conditions for each of the plurality of control factors used to experimentally determine a target variable; recording the target variable obtained based on the created first table, i.e., the target variable obtained when the experimental conditions of the first control factor not included in the plurality of control factors are fixed to one of a first level value and a second level value larger than the first level value; using the first table recording the target variable, calculating a first response surface relating to the target variable for the plurality of control factors; if the calculated first response surface does not contain the target value relating to the target variable, adding experimental conditions of the first control factor set to one of the first level value and the second level value to the combinations of experimental conditions in the first table recording the target variable; setting the first plurality of level values ​​for each of the plurality of control factors; and setting the first plurality of level values ​​for each of the plurality of control factors. A second plurality of level values ​​are set for the first control factor, including a level value that is different from the first level value and a level value that is different from the first level value and the second level value; based on the first plurality of level values ​​set for each of the plurality of control factors and the second plurality of level values ​​set for the first control factor, the combination of the plurality of control factors and the experimental conditions of each of the first control factor is added to the first table for the experimental conditions of the first control factor, thereby creating a second table representing the combination of the experimental conditions of each of the plurality of control factors and the first control factor by experimental planning method; the target variable obtained based on the created second table is recorded in the second table; using the second table recording the target variable, the second response surface involving the target variable, i.e., the second response surface containing the target value, is calculated for the plurality of control factors and the first control factor; and the calculated second response surface is output.

[0011] One aspect of this disclosure relates to an information processing apparatus comprising: a processor; and a memory, wherein the processor performs the following operations by executing a program stored in the memory: based on first plurality of level values ​​set for each of a plurality of control factors, a first table is created using an experimental planning method to represent combinations of experimental conditions for each of the plurality of control factors used to experimentally determine a target variable; the first table records a target variable obtained based on the created first table, i.e., a target variable obtained when the experimental conditions of a first control factor not included in the plurality of control factors are fixed to a level value of either a first level value or a second level value larger than the first level value; using the first table containing the target variable, a first response surface relating to the target variable is calculated for the plurality of control factors; if the calculated first response surface does not contain a target value relating to the target variable, a first response surface is added to the combinations of experimental conditions in the first table containing the target variable, with a level value set to either the first level value or the second level value. Experimental conditions for control factors; setting a second plurality of level values ​​for the first control factor, including a level value of the first level value and a level value of the second level value that differs from the first level value and the second level value, and a third level value that differs from the first level value and the second level value; based on the first plurality of level values ​​set for each of the plurality of control factors and the second plurality of level values ​​set for the first control factor, adding combinations of the experimental conditions for each of the plurality of control factors and the first control factor to the first table with added experimental conditions, thereby creating a second table representing combinations of the experimental conditions for each of the plurality of control factors and the first control factor using experimental planning method; recording the target variable obtained based on the created second table in the second table; using the second table with the recorded target variable, calculating the second response surface involved in the target variable, i.e., the second response surface containing the target value, for the plurality of control factors and the first control factor; and outputting the calculated second response surface. Attached Figure Description

[0012] Figure 1 This is a structural diagram illustrating an example of an information processing apparatus according to an embodiment.

[0013] Figure 2 This is a flowchart illustrating an example of an information processing method involved in an implementation.

[0014] Figure 3 This is an example of the first table created using the experimental planning method.

[0015] Figure 4 This is an example table showing the first table containing the target variable.

[0016] Figure 5A This is a diagram used to illustrate an example where the target value involved in the target variable is not included in the first response surface.

[0017] Figure 5B This is a diagram used to illustrate another example of the case where the target value involved in the target variable is not included in the first response surface.

[0018] Figure 6 This is an example of Table 1, which shows the experimental conditions with an additional first control factor.

[0019] Figure 7 This is a table that shows an example of Table 3.

[0020] Figure 8 It is a third level value used to illustrate the existence of three modes.

[0021] Figure 9 This is a diagram used to illustrate the method of using experimental planning numbers.

[0022] Figure 10A This is the first example of the second table, which shows the target variable.

[0023] Figure 10B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0024] Figure 10C This is a table showing the number of experimental plans up to the point of high precision in the first example.

[0025] Figure 11A This is the second example of the second table, which shows the target variable.

[0026] Figure 11B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0027] Figure 11C This is a table showing the number of experimental plans up to the point of becoming a high-precision experimental plan in Example 2.

[0028] Figure 12A This is the third example of the second table, which shows the target variable.

[0029] Figure 12B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0030] Figure 12C This is a table showing the number of experimental plans up to the point of high precision in example 3.

[0031] Figure 13 This is a flowchart illustrating an example of an information processing method involved in a variation.

[0032] Explanation of reference numerals in the attached figures

[0033] 1: Information processing device; 120: Computer; 122: Monitor; 124: Printer; 126: Keyboard; 128: Mouse; 140: CPU; 142: Bus; 144: ROM; 146: RAM; 148: Hard disk drive; 150: DVD drive; 152: Semiconductor memory port; 154: Network interface; 160: Semiconductor memory; 162: DVDROM; 164: Network. Detailed Implementation

[0034] In the case of the conventional methods described in Patent Document 1 or Patent Document 2, the understanding of the process has progressed to some extent, and the premise is that there are fewer experimental points that are difficult to measure.

[0035] On the other hand, when dealing with complex processes, it is difficult to measure a large number of experimental points if the experimental scope is to be expanded sufficiently, raising concerns that adding additional experimental points would increase the overall size of the experiment. Furthermore, if the experimental scope is set too narrowly, there are concerns that optimal conditions cannot be obtained within the generated response surface, and that the experimental plan must be redesigned to conduct research outside the experimental scope.

[0036] Therefore, the purpose of this disclosure is to provide an information processing method, procedure, and apparatus that can efficiently set up experimental plans, such as those for complex processes.

[0037] One aspect of this disclosure relates to an information processing method comprising: creating a first table, based on first plurality of level values ​​set for each of a plurality of control factors, using an experimental planning method to represent combinations of experimental conditions for each of the plurality of control factors used to experimentally determine a target variable; recording the target variable obtained based on the created first table, i.e., the target variable obtained when the experimental conditions of the first control factor not included in the plurality of control factors are fixed to one of a first level value and a second level value larger than the first level value; using the first table recording the target variable, calculating a first response surface related to the target variable for the plurality of control factors; if the calculated first response surface does not contain the target value related to the target variable, adding experimental conditions of the first control factor set to one of the first level value and the second level value to the combinations of experimental conditions recorded in the first table recording the target variable; setting the first plurality of level values ​​for each of the plurality of control factors; and setting the first plurality of level values ​​for each of the plurality of control factors. A second plurality of level values ​​are set for the first control factor, including a level value that is different from the first level value and a level value that is different from the first level value and the second level value; based on the first plurality of level values ​​set for each of the plurality of control factors and the second plurality of level values ​​set for the first control factor, the combination of the plurality of control factors and the experimental conditions of each of the first control factor is added to the first table for the experimental conditions of the first control factor, thereby creating a second table representing the combination of the experimental conditions of each of the plurality of control factors and the first control factor by experimental planning method; the target variable obtained based on the created second table is recorded in the second table; using the second table recording the target variable, the second response surface involving the target variable, i.e., the second response surface containing the target value, is calculated for the plurality of control factors and the first control factor; and the calculated second response surface is output.

[0038] For example, in an initial experimental plan based on a first table containing multiple control factors with set first-multiple level values, if the first response surface does not contain the target value related to the target variable (i.e., the experimental result as the target cannot be obtained), it is assumed that the level value of the first control factor not included in the multiple control factors needs to be adjusted by adding it. In this case, adding the first control factor to the multiple control factors and re-performing the experiment from scratch increases the number of experimental plans, making it inefficient. In contrast, in an experimental plan based on a second table, for combinations of experimental conditions containing the first control factor with set fixed level values ​​(i.e., combinations of experimental conditions in the first table), the experiment is not re-performed; instead, a replanning is performed for combinations of experimental conditions for the multiple control factors with set first-multiple level values ​​and the first control factor with set second-multiple level values. Thus, the experimental result as the target can be obtained with fewer replannings. Therefore, according to this disclosure, experimental plans, such as those for complex processes, can be efficiently designed.

[0039] For example, it is also possible to create a third table representing the combination of experimental conditions for each of the plurality of control factors and the first control factor, based on the first plurality of level values ​​set for each of the plurality of control factors and the first control factor, and to add the combination of experimental conditions when creating the second table until the value of the average prediction variance calculated for the second table becomes smaller than the value of the average prediction variance calculated for the third table.

[0040] Therefore, the average prediction variance calculated for Table 2 becomes smaller than the average prediction variance calculated for Table 3, thus enabling efficient and accurate experimental planning in experimental planning based on Table 2.

[0041] For example, the setting of the third level value of the first control factor in the second table can be performed through any of the following modes: a first mode, setting the third level value to be smaller than the first level value; a second mode, setting the third level value to be larger than the first level value and smaller than the second level value; and a third mode, setting the third level value to be larger than the second level value. The second table is created for a first selection mode selected from the first, second, and third modes, and the table is created through the first selection mode. If the second table cannot calculate a second response surface containing the target value, the second table is created for any second selection mode other than the first selection mode selected from the first mode, the second mode, and the third mode. If the second table created in the second selection mode cannot calculate a second response surface containing the target value, the second table is created for a third selection mode other than the first selection mode and the second selection mode selected from the first mode, the second mode, and the third mode.

[0042] Therefore, even if the second table created by selecting from the first, second, and third modes as the mode of setting the third level value cannot calculate the second response surface containing the target value of the target variable, it is possible to try to calculate the second response surface containing the target value of the target variable in other modes, thus increasing the possibility of outputting the second response surface containing the target value of the target variable.

[0043] For example, the first table can also be created based on the central composite programming method.

[0044] Therefore, by using central composite programming as an experimental programming method, experimental planning can be set more efficiently.

[0045] One aspect of this disclosure relates to a program for causing a computer to perform the information processing method described above.

[0046] One aspect of this disclosure involves storing the aforementioned procedures on a recording medium.

[0047] Therefore, it is possible to provide a procedure that can efficiently set up experimental plans, such as those for complex processes.

[0048] One aspect of this disclosure relates to an information processing apparatus comprising: a processor; and a memory, wherein the processor performs the following operations by executing a program stored in the memory: based on first plurality of level values ​​set for each of a plurality of control factors, a first table is created using an experimental planning method to represent combinations of experimental conditions for each of the plurality of control factors used to experimentally determine a target variable; the first table records a target variable obtained based on the created first table, i.e., a target variable obtained when the experimental conditions of a first control factor not included in the plurality of control factors are fixed to a level value of either a first level value or a second level value larger than the first level value; using the first table containing the target variable, a first response surface relating to the target variable is calculated for the plurality of control factors; if the calculated first response surface does not contain a target value relating to the target variable, a first level value set to a level value of either the first level value or the second level value is added to the combinations of experimental conditions in the first table containing the target variable. Experimental conditions for control factors; setting a second plurality of level values ​​for the first control factor, including a level value of the first level value and a level value of the second level value that differs from the first level value and the second level value, and a third level value that differs from the first level value and the second level value; based on the first plurality of level values ​​set for each of the plurality of control factors and the second plurality of level values ​​set for the first control factor, adding the combination of the plurality of control factors and the experimental conditions of each of the first control factor to the first table with added experimental conditions for the first control factor, thereby creating a second table representing the combination of the experimental conditions of each of the plurality of control factors and the first control factor using experimental planning method; recording the target variable obtained based on the created second table in the second table; using the second table recording the target variable, calculating the second response surface involved in the target variable, i.e., the second response surface containing the target value, for the plurality of control factors and the first control factor; and outputting the calculated second response surface.

[0049] Therefore, it is possible to provide an information processing device that can efficiently set up experimental plans, such as those for complex processes.

[0050] Furthermore, these general or specific methods can be implemented through systems, methods, integrated circuits, computer programs, or recording media such as computer-readable CD-ROMs, or through any combination of systems, methods, integrated circuits, computer programs, and recording media.

[0051] The embodiments will now be described in detail with reference to the accompanying drawings.

[0052] Furthermore, the embodiments described below are either general or specific examples. The numerical values, shapes, materials, constituent elements, the arrangement and location of constituent elements, connection methods, steps, and the order of steps shown in the following embodiments are examples and are not intended to limit this disclosure.

[0053] (Implementation Method)

[0054] <device>

[0055] Figure 1 This is a structural diagram illustrating an example of the information processing apparatus 1 according to an embodiment. Additionally, in Figure 1 In addition to the information processing device 1, a semiconductor memory 160, a DVD-ROM (Digital Versatile Disk Read-Only Memory) 162, and a network 164 are also shown.

[0056] The information processing apparatus 1 described in the embodiment can be implemented using computer system hardware and a program executed on the computer system. Furthermore, the information processing apparatus 1 shown herein is merely an example, and it can also be implemented using other structures.

[0057] Reference Figure 1 The information processing device 1 includes a computer 120, a monitor 122 connected to the computer 120, a keyboard 126, a mouse 128, and a printer 124. Alternatively, the information processing device 1 may not include the monitor 122, printer 124, keyboard 126, and mouse 128.

[0058] Computer 120 includes DVD drive 150 and semiconductor memory port 152.

[0059] like Figure 1 As shown, the computer 120 also includes a bus 142 connected to a DVD drive 150 and a semiconductor memory port 152, a CPU 140 connected to the bus 142, and a ROM 144 storing the boot program of the computer 120.

[0060] In addition, computer 120 also includes: RAM 146, which provides a working area for use by CPU 140 and serves as a storage area for programs executed by CPU 140; hard disk drive 148, which stores initial experimental planning data, experimental data, simulation data, additional experimental planning data, optimal point setting data, and optimal point calculation data; and network interface 154, which provides a connection to network 164.

[0061] The software of the information processing apparatus 1 according to the implementation method is distributed in the form of object code or script recorded in a medium such as DVD-ROM 162 or semiconductor memory 160, and provided to computer 120 via a read device such as DVD drive 150 or semiconductor memory port 152, and stored in hard disk drive 148. When CPU 140 executes a program, the program is read from hard disk drive 148 and loaded into RAM 146. Instructions are fetched from an address specified by a program counter (not shown) and executed. CPU 140 reads the data to be processed from hard disk drive 148 and stores the processing result in hard disk drive 148. The optimized combination of experimental conditions is output by printer 124.

[0062] The general operation of computer 120 is well known, so detailed explanations are omitted.

[0063] Regarding the distribution method of the software, the software does not necessarily have to be fixed on a storage medium. For example, the software can also be distributed from other computers connected to network 164. Alternatively, a portion of the software can be stored on hard disk 148, and the remaining portion can be retrieved from network 164 and merged during execution.

[0064] Furthermore, the distribution method of software is not limited to object code. As mentioned earlier, it can also be in the form of scripts, or it can be supplied as source code and converted into object code by a suitable compiler installed on computer 120.

[0065] Typically, modern computers utilize general functions provided by the computer's operating system (OS) to achieve their functions in a controlled manner according to a desired purpose. Therefore, even programs that do not contain general functions that can be provided from the OS or third parties but only specify the execution order of general functions are clearly included within the scope of this disclosure, as long as the program as a whole has a control structure to achieve the desired purpose.

[0066] <Process>

[0067] Next, based on Figure 2 The operation of the information processing device 1 according to the embodiment will be explained.

[0068] Figure 2 This is a flowchart illustrating an example of an information processing method according to an embodiment. Furthermore, since the information processing method is executed by the information processing device 1 (computer 120, specifically, processor (CPU 140)), Figure 2 This is also a flowchart illustrating an example of the operation of the information processing device 1.

[0069] <s101>

[0070] First, in step S101, the information processing device 1 creates a first table representing the combination of experimental conditions for each of the multiple control factors to be experimentally determined by experimental planning, based on the first multiple level values ​​set for each of the multiple control factors.

[0071] Design of Experiments (DOE) includes classic orthogonal programming, central composite programming, space-filling programming, and various other programming methods depending on the objective. Orthogonal programming has weak alternation effects, while space-filling programming tends to require more experiments. Considering that alternation effects are also strong in complex processes, and that experiments or simulations require time, in this embodiment, the creation of Table 1 is based on central composite programming.

[0072] A control factor is a controllable process parameter or planning parameter, such as temperature, humidity, pressure, or speed. A level value is a value set for the control factor; for example, when the control factor is temperature, the level value could be 0°C, 100°C, or 200°C. For instance, when the temperature of the chamber in a semiconductor film deposition process is used as the control factor, the target variable becomes the semiconductor film thickness. A specific example of the first table created in step S101 is shown below. Figure 3 .

[0073] Figure 3 This is an example of the first table created using the experimental planning method. X1, X2, X3, and X4 are control factors, specifically, the temperature of the chamber. For example, the first multiple level values ​​are -1, 0, and 1, where -1 represents 80°C, 0 represents 90°C, and 1 represents 100°C. Figure 3 Table 1 shows the combination of 21 experimental conditions for X1 to X4.

[0074] <s102>

[0075] Next, in step S102, the information processing device 1 records the target variable obtained based on the created first table in the first table. This target variable is obtained when the experimental conditions of the first control factor (not included in the multiple control factors) are fixed at either a first level value or a second level value larger than the first level value. For example, the information processing device 1 fixes the experimental conditions of the first control factor at the first level value and records the target variable obtained through experimentation or simulation in the first table of a database (hereinafter referred to as "DB") constructed within the computer 120. A specific example of the first table containing the target variable is shown below. Figure 4 .

[0076] Figure 4 This is an example table showing the first table containing the target variable. X5 is the first control factor not included among the multiple control factors (X1 to X4), specifically, the temperature of a chamber different from X1 to X4. The column for X5 is indicated by a dashed line because X5 was not recorded in the first table when the target variable was recorded. However, X5 is managed by fixing it to one of the first and second level values. Here, the first level value is set to 0, the second level value is set to 1, and X5 is fixed as the first level value, whichever is higher. Alternatively, X5 can also be fixed as the second level value.

[0077] <s103>

[0078] Next, in step S103, the information processing device 1 uses the first table recording the target variable to calculate the first response surface related to the target variable for multiple control factors.

[0079] <s104>

[0080] Next, in step S104, the information processing device 1 determines whether the calculated first response surface contains the target value involved in the target variable (hereinafter, the target value involved in the target variable is referred to as the target value). To this end, it uses... Figure 5A as well as Figure 5B Please provide an explanation.

[0081] Figure 5A This is a diagram used to illustrate an example where the target value is not included in the calculated first response surface.

[0082] Figure 5B This is a diagram used to illustrate another example where the target value is not included in the first response surface.

[0083] exist Figure 5A as well as Figure 5B In the diagram, the first response surface is shown using solid and dashed lines. The solid lines represent the first response surface calculated using the first table containing the target variables, while the dashed lines represent the first response surface inferred from the first table containing the target variables. In other words, the first response surface is inferred from the calculated first response surface.

[0084] For example, the optimal point setting is recorded in the memory of the information processing device 1. The information processing device 1 calculates candidate optimal points based on the first response surface calculated in step S103, compares the recorded optimal point setting with the calculated candidate optimal points, and determines whether the optimal point can be achieved through multiple control factors that do not include the first control factor (in other words, whether the calculated first response surface contains a target value). For example, the target value can be a peak of the first response surface or a given value. Figure 5A The diagram shows the optimal level value corresponding to the target value when the target value is the peak of the first response surface. Figure 5B The diagram shows the optimal level value corresponding to the target value, given that the target value is a given value. For Figure 5A as well as Figure 5B If any of these conditions are met, it can be concluded that the target value is not included in the calculated first response surface (solid line).

[0085] If the calculated first response surface does not contain the target value ("No" in step S104), the first control factor is used as the control factor for expansion, and the process proceeds to step S105. If the calculated first response surface contains the target value ("Yes" in step S104), the process ends because the desired experimental result can be obtained in the initial experimental plan. Furthermore, the target value is not limited to the peak of the first response surface; it can also be the bottom. Additionally, the given value can be 0.

[0086] <s105>

[0087] Next, in step S105, if the calculated first response surface does not contain the target value, the information processing device 1 adds experimental conditions with a first control factor set as either a first level value or a second level value to the combination of experimental conditions in the first table recording the target variable. A specific example of the first table of experimental conditions with the added first control factor is shown below. Figure 6 .

[0088] Figure 6 This is an example of a table showing the experimental conditions with an added first control factor. As described above, here the first level value (0) is set for X5 as the first control factor, which is the level value of either the first level value or the second level value. The experimental conditions with X5 set to 0 are added to the first table.

[0089] <s106>

[0090] Next, in step S106, the information processing device 1 evaluates the accuracy of the experimental planning for the target. Various metrics can be used to evaluate the accuracy of the experimental planning, but using the average prediction variance yields relatively good results. For example, the information processing device 1 creates a third table representing combinations of experimental conditions for multiple control factors and each of the first control factors, based on first multiple level values ​​set for each of the multiple control factors and the first control factor. A specific example of the third table is shown below. Figure 7 .

[0091] Figure 7 This is an example table showing Table 3. For example... Figure 7 Therefore, in Table 3, for X5, which is the first control factor, the same first multiple level values ​​are set as for X1 to X4, which are multiple control factors, namely -1, 0, and 1. Considering the number of control factors in Table 1 with such an added X5, and assuming a general response surface methodology is applied, the average prediction variance of Table 3 is calculated. Hereafter, the average prediction variance of Table 3 will be referred to as Vall_0.

[0092] <s107>

[0093] Next, in step S106, the information processing device 1 sets a first plurality of level values ​​(e.g., -1, 0, and 1) for each of the plurality of control factors, and sets a second plurality of level values ​​for the first control factor, including a level value (e.g., a second level value) that differs from one of the first level value (e.g., 0) and a second level value (e.g., 1), and a third level value that differs from both the first and second level values. Furthermore, if the level value of one of the first and second level values ​​is the second level value, the level value that differs from the first level value becomes the first level value. The second table, described later, contains the plurality of control factors and the first control factor; the setting here refers to the setting of the level value for the first control factor in the second table.

[0094] The third level value in Table 2 is set using any one of Mode 1, Mode 2, and Mode 3. For this, use... Figure 8 Please provide an explanation.

[0095] Figure 8 It is a third level value used to illustrate the existence of three modes.

[0096] like Figure 8 As shown, Mode 1 sets the third level value to be smaller than the first level value. In Mode 1, the third level value is set to, for example, -1. Mode 2 sets the third level value to be larger than the first level value and smaller than the second level value. In Mode 2, the third level value is set to, for example, 0.5. Mode 3 sets the third level value to be larger than the second level value. In Mode 3, the third level value is set to, for example, 2. For example, the user can select any of Mode 1, Mode 2, and Mode 3.

[0097] <s108>

[0098] Next, in step S108, the information processing device 1 sets the additional planning number.

[0099] <s109>

[0100] Next, in step S109, the information processing device 1, based on the first plurality of level values ​​set for each of the plurality of control factors and the second plurality of level values ​​set for the first control factor, adds planning to the first table (i.e., the first table of experimental conditions for which the first control factor has been added) which was previously operated on for the combination of experimental conditions for the plurality of control factors and the first control factor. Furthermore, various methods exist for determining the content of the combination of added experimental conditions, depending on the purpose, such as D-optimal planning or I-optimal planning. Compared to I-optimal planning, which aims to minimize the prediction variance across the entire planning area, D-optimal planning focuses on reducing the prediction variance at each planning point. For example, there are more cases where good results are obtained by determining the planning using D-optimal planning; therefore, D-optimal planning is used here.

[0101] <s110>

[0102] Next, in step S110, the information processing device 1 calculates the average prediction variance of the additional experimental plan with the added experimental conditions, and evaluates the accuracy of the additional experimental plan using the average prediction variance, similar to Table 3. Hereafter, the average prediction variance of the additional experimental plan will be referred to as Vall_ADD.

[0103] <s111>

[0104] Next, in step S111, the information processing device 1 evaluates the accuracy of the additional experimental plan by determining whether Vall_0 exceeds Vall_ADD. Figure 9 Step S111 will be explained.

[0105] Figure 9 This is a diagram used to illustrate the method of using experimental planning numbers. Specifically, it is a diagram used to illustrate the method of using experimental planning numbers when the precision of the experimental planning is increased to a high precision.

[0106] like Figure 9 As shown, if the number of plans is increased for additional experimental plans, Vall_ADD decreases, for example, falling below Vall_0 when the number of plans is 36. When Vall_ADD, representing the precision of the additional experimental plan, is lower than Vall_0, representing the precision of the experimental plan based on Table 3 (i.e., Vall_0 > Vall_ADD, which is "Yes" in step S111), the information processing device 1 stops adding combinations of experimental conditions and proceeds to step S113. Furthermore, if Vall_0 is not greater than Vall_ADD (which is "No" in step S111), the process proceeds to step S112.

[0107] <s112>

[0108] If the current number of additional plans is greater than or equal to the given number of plans ("No" in step S112), the information processing device 1 terminates the process. For example, it is common for the given number of plans to be set as the number of plans in the first table (i.e., the initial number of plans). That is, since it is impossible to efficiently set experimental plans when the number of additional plans exceeds the initial number of plans, the process terminates. If the current number of additional plans is less than the given number of plans ("Yes" in step S112), the information processing device 1 performs step S109 again to add combinations of experimental conditions. In addition, the addition of combinations of experimental conditions can be done not one by one, but multiple combinations can be added.

[0109] Furthermore, the following example will be explained here: An additional combination of experimental conditions is performed for the first control factor, setting a third level value for one of the selected modes (mode 1, mode 2, and mode 3), without adding a combination of experimental conditions setting third level values ​​for other unselected modes. Examples where additional combinations of experimental conditions setting third level values ​​for other modes can also be performed will be explained in the variations described later.

[0110] <s113>

[0111] Next, in step S113, the information processing device 1 determines the content of the additional experiment and creates a second table representing the combination of multiple control factors and the experimental conditions of each of the first control factors using the experimental planning method.

[0112] <s114>

[0113] Next, in step S114, the information processing device 1 records the target variables obtained based on the created second table in the second table. A specific example of the second table containing the target variables will be described later. Figure 10A This will be explained in detail later.

[0114] <s115>

[0115] Next, in step S115, the information processing device 1 uses the second table that records the target variable to calculate the second response surface involving the target variable, i.e., the second response surface containing the target value, for multiple control factors and the first control factor.

[0116] <s116>

[0117] Then, in step S116, the information processing device 1 outputs the second response surface.

[0118] <Specific Examples of the Three Patterns>

[0119] Next, specific examples of three modes for setting the third level value will be given to illustrate this disclosure.

[0120] Figures 10A to 10C This is a diagram (Example 1) used to illustrate an example of a combination of experimental conditions where the third level value (i.e., -1) of the first mode is set for X5 as the first control factor.

[0121] Figure 10A This is the first example of the second table, which shows the target variable.

[0122] Figure 10B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0123] Figure 10C This is a table showing the number of experimental plans up to the point of high precision in the first example.

[0124] exist Figure 10A As shown in Table 2, as No. 22 to No. 36, it can be seen that the combination of experimental conditions for multiple control factors (X1 to X4) with a first multiple level value and a first control factor (X5) with a third level value (i.e., -1) containing the first mode is added to Table 1 (No. 1 to No. 21).

[0125] In addition, such as Figure 10B As shown, in the conventional method, the first multiple level values ​​set for multiple control factors (X1 to X4) are also set to the first control factor (X5). The experimental conditions for this first control factor (X5) are added, and the experimental plan is redesigned from scratch. Therefore, in the case of the conventional method, for example, the total number of plans becomes 48. On the other hand, as... Figure 10C As shown, in the case of the method disclosed herein, with 21 planning iterations (initial planning iterations) as the first table, and an additional 6 iterations of experimental conditions for replanning (expansion), a total of 27 planning iterations are performed to obtain the optimal value (that is, the second response surface containing the target value can be calculated). By performing an additional 15 iterations of experimental conditions (expansion), it can be confirmed that Vall_ADD is lower than Vall_0 (0.391), and sufficient experimental planning accuracy is obtained. That is, in the case of the method disclosed herein, with a total of 36 planning iterations, the same result can be obtained with 12 fewer planning iterations than conventional methods (in other words, experimental planning can be set efficiently and with high accuracy with 12 fewer planning iterations than conventional methods). Figure 10C In the evaluation, if Vall_ADD is lower than Vall_0, the evaluation is displayed as "Satisfied". Conversely, if Vall_ADD exceeds Vall_0, the evaluation is displayed as "Dissatisfied". If the optimal value is obtained, the evaluation is displayed as "Satisfied". Conversely, if the optimal value is not obtained, the evaluation is displayed as "Dissatisfied". If both the evaluation of Vall_ADD and the evaluation of the optimal value are satisfied, the overall evaluation is displayed as "Satisfied". If either the evaluation of Vall_ADD or the evaluation of the optimal value is dissatisfied, the overall evaluation is displayed as "Dissatisfied". Figure 10B , Figure 11B , Figure 11C , Figure 12B as well as Figure 12C The same applies.

[0126] also, Figures 11A to 11C This is a diagram used to illustrate an example (Example 2) of the combination of experimental conditions where a third level value (i.e., 0.5) of the second mode is set for X5 as the first control factor.

[0127] Figure 11A This is the second example of the second table, which shows the target variable.

[0128] Figure 11B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0129] Figure 11C This is a table showing the number of experimental plans up to the point of becoming a high-precision experimental plan in Example 2.

[0130] exist Figure 10A As shown in Table 2, as No. 22 to No. 41, it can be seen that the combination of experimental conditions for multiple control factors (X1 to X4) with a first multiple level value and a first control factor (X5) with a third level value of 0.5 including the second mode is added to Table 1 (No. 1 to No. 21).

[0131] In addition, such as Figure 11B As shown, in the conventional method, the first multiple level values ​​set for multiple control factors (X1 to X4) are also set to the first control factor (X5). The experimental conditions for this first control factor (X5) are added, and the experimental plan is redesigned from scratch. Therefore, in the case of the conventional method, for example, the total number of plans becomes 48. On the other hand, as... Figure 11C As shown, in the case of the method disclosed herein, with 21 planning iterations (initial planning iterations) as the first table, and an additional (extended) combination of 8 experimental conditions for replanning, a total of 29 planning iterations are performed to obtain the optimal value (that is, the second response surface containing the target value can be calculated). By performing an additional (extended) combination of 20 experimental conditions, it can be confirmed that Vall_ADD is lower than Vall_0 (0.391), and sufficient experimental planning accuracy is obtained. That is, in the case of the method disclosed herein, with a total of 41 planning iterations, the same result can be obtained with 7 fewer planning iterations than conventional methods (in other words, experimental planning can be set efficiently and accurately with 7 fewer planning iterations than conventional methods).

[0132] at last, Figures 12A to 12C This is a diagram (the third example) used to illustrate an example of the combination of experimental conditions when the third level value (that is, 2) of the third mode is set for X5 as the first control factor.

[0133] Figure 12A This is the third example of the second table, which shows the target variable.

[0134] Figure 12B This is a table showing the number of experimental plans when the experimental plan becomes high-precision in the comparative example.

[0135] Figure 12C This is a table showing the number of experimental plans up to the point of high precision in example 3.

[0136] exist Figure 12A As shown in Table 2, as No. 22 to No. 41, it can be seen that the combination of experimental conditions for multiple control factors (X1 to X4) with a first multiple level value and a first control factor (X5) with a third level value (i.e., 2) including the third mode is added to Table 1 (No. 1 to No. 21).

[0137] In addition, such as Figure 12B As shown, in the conventional method, the first multiple level values ​​set for multiple control factors (X1 to X4) are also set to the first control factor (X5). The experimental conditions for this first control factor (X5) are added, and the experimental plan is redesigned from scratch. Therefore, in the case of the conventional method, for example, the total number of plans becomes 48. On the other hand, as... Figure 12C As shown, in the case of the method disclosed herein, with 21 planning iterations (initial planning iterations) as the first table, and an additional (extended) combination of 8 experimental conditions for replanning, a total of 29 planning iterations are performed to obtain the optimal value (that is, the second response surface containing the target value can be calculated). By performing an additional (extended) combination of 20 experimental conditions, it can be confirmed that Vall_ADD is lower than Vall_0 (0.391), and sufficient experimental planning accuracy is obtained. That is, in the case of the method disclosed herein, with a total of 41 planning iterations, the same result can be obtained with 7 fewer planning iterations than conventional methods (in other words, experimental planning can be set efficiently and accurately with 7 fewer planning iterations than conventional methods).

[0138] in addition, Figure 2 Not all steps in the process shown are necessary. If prior planning and experimentation have been conducted, execution can begin, for example, from step S104.

[0139] <Variation Example>

[0140] Although the following example is given: adding a combination of experimental conditions for setting a third level value of one of the modes selected from the first mode, the second mode, and the third mode for the first control factor, and not adding a combination of experimental conditions for setting a third level value of other modes that were not selected, an example in which a combination of experimental conditions for setting a third level value of other modes can also be added is given as a variation.

[0141] Figure 13 This is a flowchart illustrating an example of an information processing method involved in a modified example. Steps identical to those in the implementation method are omitted; instead, differences are explained.

[0142] <s117>

[0143] In step S117, the information processing device 1 sets a second plurality of level values ​​for the first control factor, including a level value (e.g., the second level value) that is either a first level value (e.g., 0) or a second level value (e.g., 1), and a third level value that is different from the first and second level values. The creation of the second table in step S113 is performed for the first selection mode selected from the first mode, the second mode, and the third mode. For example, here, it is assumed that the first mode is selected as the first selection mode.

[0144] <s118>

[0145] In steps S108 to S113, a second table is created by adding a combination of experimental conditions to the combination of experimental conditions for setting the third level value of the first selection mode (e.g., the first mode) of the first control factor. The information processing device 1 uses the created second table to determine whether the calculated second response surface contains the target value. That is, the information processing device 1 determines whether the second response surface containing the target value can be calculated using the second table created in the first selection mode. If the second response surface containing the target value can be calculated using the second table created in the first selection mode ("Yes" in step S118), the process moves to step S115; if the second response surface containing the target value cannot be calculated using the second table created in the first selection mode ("No" in step S118), the process moves to step S119.

[0146] <s119>

[0147] If the second response surface containing the target value cannot be calculated using the second table created in the first selection mode, the creation of the second table in step S113 is performed for any second selection mode other than the first selection mode selected from the first, second, and third modes. For example, since the first selection mode has already been selected as the first selection mode, it is set here that the second selection mode is selected as the second selection mode.

[0148] Then, similar to the first selection mode, in steps S108 to S113, a second table is created by adding the combination of experimental conditions that set the third level value of the second selection mode (e.g., the second mode) to the first control factor. The information processing device 1 determines whether the second response surface calculated using the created second table contains the target value. That is, the information processing device 1 determines whether the second response surface containing the target value can be calculated using the second table created in the second selection mode. If the second response surface containing the target value can be calculated using the second table created in the second selection mode ("Yes" in step S118), the process moves to step S115; if the second response surface containing the target value cannot be calculated using the second table created in the second selection mode ("No" in step S118), the process moves to step S119.

[0149] If the second table created in the second selection mode cannot calculate the second response surface containing the target value, the second table is created for a third selection mode other than the first and second selection modes selected from the first, second, and third selection modes. Since the first selection mode has already been selected as the first selection mode and the second selection mode has already been selected as the second selection mode, the third selection mode is selected here as the third selection mode.

[0150] Furthermore, if the second response surface containing the target value cannot be calculated using the second table created in the third selection mode, this process ends.

[0151] (Other implementation methods)

[0152] The above describes the information processing method and information processing apparatus 1 related to the embodiments and modifications, but this disclosure is not limited to the above embodiments and modifications.

[0153] For example, the steps in the information processing method can also be executed by a computer (computer system). Furthermore, this disclosure can be implemented as a program for causing a computer to perform the steps included in these methods. Moreover, this disclosure can be implemented as a non-transitory computer-readable recording medium such as a CD-ROM containing the program.

[0154] For example, in the case where this disclosure is implemented by a program (software), the program is executed using hardware resources such as the computer's CPU, memory, and input / output circuits, thereby executing each step. That is, the CPU retrieves data from the memory or input / output circuits and performs calculations, or outputs the calculation results to the memory or input / output circuits, thereby executing each step.

[0155] Furthermore, the processing units included in the information processing apparatus 1 according to the above embodiments and variations are typically implemented as integrated circuits, i.e., LSIs. These can be implemented as single chips independently, or as single chips including some or all of them.

[0156] Furthermore, integrated circuitry is not limited to LSIs; it can also be achieved through dedicated circuits or general-purpose processors. FPGAs (Field Programmable Gate Arrays) that can be programmed after LSI fabrication, or reconfigurable processors that can reconfigure the connections and settings of the circuitry within the LSI, can also be used.

[0157] Furthermore, in the above embodiments and variations, each component can also be implemented using dedicated hardware or by executing software programs suitable for each component. Each component can also be implemented by a program execution unit such as a CPU or processor reading and executing software programs recorded on a recording medium such as a hard disk or semiconductor memory.

[0158] Furthermore, the order in which the steps in the flowchart are executed is an illustrative order for the purpose of specifically illustrating this disclosure, and may also be in an order other than that described above. Additionally, some of the steps described above may be executed simultaneously (in parallel) with other steps.

[0159] The above description illustrates one or more methods of information processing and information processing apparatus 1 based on embodiments and variations. However, this disclosure is not limited to these embodiments and variations. Any modifications to these embodiments or variations that are conceived by those skilled in the art, or any combination of constituent elements from different embodiments, that are constructed without departing from the spirit of this disclosure, may also be included within the scope of one or more embodiments.

[0160] [Industry availability]

[0161] This disclosure can be widely and effectively applied to manufacturing processes and control processes ranging from general electronic components to automotive batteries starting with capacitors, or from machining system processes to chemical system processes.

Claims

1. An information processing method comprising: creating a first table representing a combination of experimental conditions for each of a plurality of control factors for obtaining a target variable by experiment based on a first plurality of level values respectively set for the plurality of control factors by an experimental design method; recording, in the first table, a target variable obtained based on the created first table, when an experimental condition of a first control factor not included in the plurality of control factors is fixed to a level value of one of a first level value and a second level value larger than the first level value; calculating, using the first table in which the target variable is recorded, a first response surface involving the target variable with respect to the plurality of control factors; in a case where a target value involved in the target variable is not included in the calculated first response surface, adding an experimental condition of the first control factor having a level value of one of the first level value and the second level value to the combination of experimental conditions in the first table in which the target variable is recorded; setting the first plurality of level values for each of the plurality of control factors and a second plurality of level values for the first control factor including a level value of the other of the first level value and the second level value different from the level value of the one and a third level value different from the first level value and the second level value; creating a second table representing a combination of experimental conditions for each of the plurality of control factors and the first control factor based on the first plurality of level values respectively set for the plurality of control factors and the second plurality of level values set for the first control factor by adding, to the first table in which the experimental condition of the first control factor is added, a combination of experimental conditions for each of the plurality of control factors and the first control factor, thereby creating the second table representing the combination of experimental conditions for each of the plurality of control factors and the first control factor by the experimental design method; recording, in the second table, a target variable obtained based on the created second table; calculating, using the second table in which the target variable is recorded, a second response surface involving the target variable with respect to the plurality of control factors and the first control factor, that is, a second response surface including the target value; and outputting the calculated second response surface.

2. The information processing method according to claim 1, wherein a third table representing a combination of experimental conditions for each of the plurality of control factors and the first control factor is created based on the first plurality of level values respectively set for the plurality of control factors and the first control factor, the addition of the combination of experimental conditions at the time of creating the second table is performed until a value of an average prediction variance calculated for the second table becomes smaller than a value of an average prediction variance calculated for the third table.

3. The information processing method according to claim 1 or 2, wherein the setting of the third level value of the first control factor in the second table is performed by any one of the following modes: a first mode in which the third level value is set to be smaller than the first level value; a second mode in which the third level value is set to be larger than the first level value and smaller than the second level value; and ​ the third level value is set larger than the second level value, the second table is created for a first selected mode selected from the first mode, the second mode, and the third mode, in a case where a second response surface including the target value cannot be calculated by the second table created in the first selected mode, the second table is created for a second selected mode selected from the first mode, the second mode, and the third mode, other than the first selected mode, in a case where a second response surface including the target value cannot be calculated by the second table created in the second selected mode, the second table is created for a third selected mode selected from the first mode, the second mode, and the third mode, other than the first selected mode and the second selected mode.

4. The information processing method according to claim 1 or 2, wherein the creation of the first table is based on a central composite design method.

5. A recording medium, having recorded a program for causing a computer to execute the information processing method according to any one of claims 1 to 4.

6. An information processing apparatus comprising: a processor; and a memory, the processor, by executing a program stored in the memory, performs the following operations: creating, by an experimental design method, a first table indicating a combination of experimental conditions for each of a plurality of control factors for finding a target variable by experiment, based on a first plurality of level values respectively set for the plurality of control factors; recording, in the first table, a target variable acquired based on the created first table, that is, a target variable acquired when an experimental condition of a first control factor not included in the plurality of control factors is fixed to one of a first level value and a second level value larger than the first level value; calculating, for the plurality of control factors, a first response surface to which the target variable is related, using the first table in which the target variable is recorded; in a case where a target value to which the target variable is related is not included in the calculated first response surface, appending, to the combination of experimental conditions of the first table in which the target variable is recorded, an experimental condition of the first control factor having a level value of one of the first level value and the second level value; setting, for the first control factor, a second plurality of level values including a level value of the other of the first level value and the second level value different from the level value of the one, and a third level value different from the first level value and the second level value; creating, by an experimental design method, a second table indicating a combination of experimental conditions of each of the plurality of control factors and the first control factor, by appending, to the first table to which the experimental condition of the first control factor is appended, a combination of experimental conditions of each of the plurality of control factors and the first control factor, based on the first plurality of level values respectively set for the plurality of control factors and the second plurality of level values set for the first control factor. record the target variable obtained based on the created second table in the second table; using the second table in which the target variable is recorded, calculate a second response surface, i.e., a second response surface including the target value, of the target variable with respect to the plurality of control factors and the first control factor; and output the calculated second response surface.

Images