Generation method, estimation method, generation device, and estimation device

Abstract

To estimate properly, information indicating a processing result.SOLUTION: A generation method is configured to: perform experiment of processing according to plan information including, first kind information and second kind information indicating conditions of processing of a device, and acquire third kind information and fourth kind information indicating a result of the experiment (S121); acquire expanded plan information in which a uniformity coefficient between expanded second kind information and expanded third kind information is equal to or greater than a threshold value (S122); perform experiment of processing according to the expanded plan information, for acquiring expanded third kind information and expanded fourth kind information indicating a result of experiment (S123); derive an expanded first relationship being a relationship of the expanded first kind information, expanded second kind information and expanded third kind information, and derive an expanded second relationship being a relationship of the expanded first kind information, the expanded second kind information and expanded fourth kind information (S124); then use the second kind information and third kind information measured in device processing, as input, and use the expanded first relationship and the second relationship, for generating and outputting a model which estimates fourth kind information indicating the processing result (S125).SELECTED DRAWING: Figure 20

Description

Technical Field

[0001] The present invention relates to a generation method, an estimation method, a generation device, and an estimation device.

Background Art

[0002] Conventionally, a model related to device processing has been used. Regarding such a model, many cases have been reported in which parameters of an objective variable (output variable) are estimated from parameters of explanatory variables (input variables). When a physical model based on an actual physical phenomenon can be constructed, high-precision estimation is possible by estimating the parameters of the objective variable using this physical model, and the man-hours required for modeling can also be reduced.

[0003] On the other hand, when it is difficult to construct a physical model, for example, a method is known in which an input-output relationship is assumed by a polynomial model using a large amount of accumulated measurement data and estimated by fitting. An estimation method combining these two methods has also been proposed (see Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, all of the above methods assume that measurement data of the parameters of the explanatory variables are collected in-line during device processing.

[0006] On the other hand, in order to obtain a model that can accurately estimate the target variable, it is assumed that among the explanatory variables, there may be parameters for which measurement data is not collected inline. In this case, in order to make the model usable inline, it is necessary to make arrangements such as combining another experimental design with the parameters for which measurement data is not collected inline as the output variable.

[0007] However, even if such arrangements are made, it is assumed that the input variables of another experimental design cannot generate experimental points as planned due to the nature of the variables. Also, even if experimental points can be generated as planned, since it is necessary to conduct two experimental designs, there is a problem that the number of experiments required to generate the model doubles.

[0008] The present invention has been made in view of such problems of the prior art, and an object thereof is to provide a generation method for generating a model that appropriately estimates information indicating the result of processing.

Means for Solving the Problems

[0009] A model generation method according to an aspect of the present invention is a generation method executed by a processor using a memory. The generation method executed by the processor using the memory includes performing an experiment on the processing according to planned information including first-type information and second-type information indicating processing conditions of a device, thereby obtaining third-type information and fourth-type information indicating the result of the processing experiment. The extended planned information includes extended first-type information obtained by adding new first-type information to the first-type information and extended second-type information obtained by adding new second-type information to the second-type information. The extended planned information is obtained in which the uniformity between the extended second-type information and extended third-type information obtained as a result of the processing experiment performed according to the extended planned information is equal to or greater than a threshold value. By performing the processing experiment according to the extended planned information, the extended third-type information and extended fourth-type information indicating the result of the processing experiment are obtained. An extended first relationship that is a relationship among the extended first-type information, the extended second-type information, and the extended third-type information is derived, and an extended second relationship that is a relationship among the extended first-type information, the extended second-type information, and the extended fourth-type information is derived. A model is generated and output that estimates the fourth-type information indicating the result of the processing using the extended first relationship and the extended second relationship with the second-type information and the third-type information measured during the processing of the device as inputs.

[0010] Note that these general or specific aspects may be implemented in a system, method, integrated circuit, computer program, or recording medium such as a computer-readable CD-ROM, or may be implemented in any combination of a system, method, integrated circuit, computer program, and recording medium.

Advantages of the Invention

[0011] The generation method of the present invention can generate a model that appropriately estimates information indicating the result of processing.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

[0013] A generation method according to one aspect of the present invention is a generation method executed by a processor using a memory, and performs an experiment of the processing according to plan information including first type information and second type information indicating processing conditions of a device, thereby obtaining third type information and fourth type information indicating the results of the processing experiment, and obtaining extended plan information including extended first type information obtained by adding new first type information to the first type information and extended second type information obtained by adding new second type information to the second type information, wherein the extended plan information is such that the uniformity between the extended second type information and extended third type information obtained as a result of the processing experiment performed according to the extended plan information is equal to or greater than a threshold value, performs the processing experiment according to the extended plan information, thereby obtaining the extended third type information and extended fourth type information indicating the results of the processing experiment, derives an extended first relationship which is a relationship among the extended first type information, the extended second type information, and the extended third type information, and derives an extended second relationship which is a relationship among the extended first type information, the extended second type information, and the extended fourth type information, and generates and outputs a model which, taking the second type information and the third type information measured during the processing of the device as inputs, estimates the fourth type information indicating the result of the processing using the extended first relationship and the extended second relationship.

[0014] According to the above aspect, by using the relationships among the first type of information, the second type of information, the third type of information, and the fourth type of information obtained from experiments, a model can be obtained for estimating the fourth type of information during processing from the second type of information and the third type of information obtained during processing. By using this model, even when the fourth type of information cannot be obtained during processing, as long as the second type of information and the third type of information can be obtained during processing, the fourth type of information during processing can be obtained by estimation. Here, since the extended third type of information extended so that the uniformity is equal to or higher than the threshold is used to generate the model for estimation, the accuracy of the fourth type of information estimated by that model can be further improved. Thus, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be generated. Also, when estimating the fourth type of information, there is no need to conduct a new experiment with the fourth type of information as the output variable. Therefore, there is an effect that an increase in the number of experiments due to obtaining information that cannot be obtained during processing can be avoided in advance.

[0015] For example, further, a first relationship that is the relationship among the first type of information, the second type of information, and the third type of information is derived. In the acquisition of the extended plan information, the extended second type of information is obtained by adding the new second type of information to the second type of information, the extended third type of information is obtained by adding new third type of information to the third type of information, and the extended first type of information is obtained by using the first relationship with the extended second type of information and the extended third type of information as inputs, and the extended plan information including the extended first type of information and the extended second type of information may be obtained.

[0016] According to the above aspect, since the extended first type of information obtained by using the extended second type of information and the extended third type of information is used, the extended plan information can be obtained more easily. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be generated more easily.

[0017] For example, in obtaining the expansion plan information, it may be determined whether the uniformity of the obtained second type of expansion information and the third type of expansion information is equal to or greater than a threshold value, and when it is determined that the uniformity of the obtained second type of expansion information and the third type of expansion information is equal to or greater than the threshold value, the expansion plan information may be obtained.

[0018] According to the above aspect, by determining that the uniformity of the second type of expansion information and the third type of expansion information is equal to or greater than the threshold value, it is possible to more easily obtain the third type of expansion information whose uniformity is equal to or greater than the threshold value. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be more easily generated.

[0019] For example, in obtaining the expansion plan information, the expansion plan information may be obtained by adding new third type information to the third type information and thereby adding new first type information belonging to the range from the minimum value to the maximum value among the first type information.

[0020] According to the above aspect, the new first type information added by the addition of the third type information comes to belong to the range in which the first type information indicating the processing conditions is distributed. Therefore, the new first type information belongs to an appropriate range as the processing conditions, in other words, it is possible to avoid the new first type information deviating from the appropriate range as the processing conditions. As a result, the expansion first relationship becomes even more appropriate, and the accuracy of the estimated fourth type information can be further increased. Therefore, according to the above generation method, it is possible to generate a model that more appropriately estimates the information indicating the result of processing.

[0021] For example, in obtaining the expansion plan information, the uniformity between the second type of expansion information and the third type of expansion information may be calculated using the average prediction variance for the evaluation plan information including the second type information and the third type information as factors, and the expansion plan information may be obtained using the calculated uniformity.

[0022] According to the above aspect, since the uniformity between the extended second type of information and the extended third type of information is evaluated by the average prediction variance, the uniformity between the extended second type of information and the extended third type of information can be evaluated more easily. Therefore, according to the above generation method, a model for estimating information indicating the result of processing can be generated.

[0023] For example, in the acquisition of the extended second type of information and the acquisition of the extended third type of information, the extended second type of information and the extended third type of information may be acquired by the D-optimal design method for the evaluation planning information including the second type of information and the third type of information as factors.

[0024] According to the above aspect, since the extended second type of information and the extended third type of information are acquired by the D-optimal design method for the evaluation planning information including the extended second type of information and the extended third type of information as factors, more appropriate extended third type of information can be acquired more easily. Therefore, according to the above generation method, a model for estimating information indicating the result of processing can be generated more easily.

[0025] For example, the extended first relationship is expressed by an extended first formula that outputs the extended third type of information with the extended first type of information and the second type of information as inputs, and the model may include an extended third formula derived from the extended first formula, the extended third formula outputting the first type of information with the second type of information and the third type of information measured during the processing of the device as inputs.

[0026] According to the above aspect, the fourth type of information can be obtained using the extended third formula derived by formula transformation from the extended first formula. Therefore, according to the above generation method, a model for appropriately estimating information indicating the result of processing can be generated more easily.

[0027] For example, the extended second relationship is expressed by an extended second formula that takes the extended first type of information and the second type of information as inputs and outputs the fourth type of information. The model further includes the extended second formula, and uses, as inputs, the second type of information and the third type of information measured during the processing of the device, to obtain the first type of information output by the extended third formula, and the fourth type of information output by the extended second formula using, as inputs, the second type of information measured during the processing of the device. The model may include a model for obtaining the fourth type of information.

[0028] According to the above aspect, the extended first type of information can be easily estimated by the extended third formula from the measured values of the second type of information and the third type of information, and the fourth type of information can be estimated by the extended second formula from the estimated extended first type of information and the measured value of the second type of information. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be generated more easily.

[0029] For example, the first type of information and the fourth type of information may be information defined in advance as information not measured during the processing, and the second type of information and the third type of information may be information defined in advance as information measured during the processing.

[0030] According to the above aspect, when there is information that is not measured in the information indicating the processing conditions and there is information that is not measured in the information indicating the result of processing, the information indicating the result of processing can be appropriately estimated. Therefore, according to the above generation method, even when there is information that is not measured during processing, a model for appropriately estimating the information indicating the result of processing can be generated.

[0031] For example, the processing is laser welding, the first type of information includes the gap width between the plate materials welded in the laser welding, the second type of information includes the laser scan speed in the laser welding, the third type of information includes the surface weld width of the laser weld in the laser welding, and the fourth type of information may include the interface weld width of the laser weld in the laser welding.

[0032] According to the above aspect, a model for appropriately estimating information indicating the result of processing in laser welding can be generated more easily.

[0033] An estimation method according to an aspect of the present invention inputs the second type of information and the third type of information measured during the processing of the device into the model output by the above generation method, and outputs the fourth type of information output by inputting the second type of information and the third type of information into the model as estimation information for estimating the result of the processing.

[0034] According to the above aspect, by inputting the second type of information and the third type of information obtained during processing into the model, information indicating the result of processing can be appropriately estimated.

[0035] A generation device according to an aspect of the present invention includes a processor and a memory connected to the processor. The processor uses the memory to perform the processing experiment according to the plan information including the first type of information and the second type of information indicating the processing conditions of the device, thereby obtaining the third type of information and the fourth type of information indicating the result of the processing experiment, and an extended plan information including extended first type of information obtained by adding new first type of information to the first type of information and extended second type of information obtained by adding new second type of information to the second type of information, wherein a uniformity between the extended second type of information and extended third type of information obtained as a result of the processing experiment performed according to the extended plan information is equal to or greater than a threshold value, obtains the extended third type of information and extended fourth type of information indicating the result of the processing experiment by performing the processing experiment according to the extended plan information, derives an extended first relationship that is a relationship among the extended first type of information, the extended second type of information, and the extended third type of information, and derives an extended second relationship that is a relationship among the extended first type of information, the extended second type of information, and the extended fourth type of information, and is a generation device that generates and outputs a model that estimates the fourth type of information indicating the result of the processing using the extended first relationship and the extended second relationship with the second type of information and the third type of information measured during the processing of the device as inputs.

[0036] According to this, the same effects as those of the above generation method are achieved.

[0037] An estimation device according to an aspect of the present invention includes a processor and a memory connected to the processor. The processor uses the memory to input the fourth type of information output by inputting the second type of information and the third type of information measured during the processing into the model output by the generation device according to claim 12, and outputs the input fourth type of information as estimation information estimating the result of the processing.

[0038] According to this, the same effects as those of the above estimation method are achieved.

[0039] Note that these general or specific aspects may be implemented by a system, an apparatus, an integrated circuit, a computer program, or a recording medium such as a computer-readable CD-ROM, or may be implemented by any combination of a system, an apparatus, an integrated circuit, a computer program, or a recording medium.

[0040] Hereinafter, embodiments will be specifically described with reference to the drawings.

[0041] Note that the embodiments described below are all examples showing general or specific examples. The numerical values, shapes, materials, components, arrangement positions and connection forms of the components, steps, the order of the steps, etc. shown in the following embodiments are merely examples and are not intended to limit the present invention. In addition, among the components in the following embodiments, components not described in the independent claims indicating the most general concept are described as optional components.

[0042] (Embodiment) In the present embodiment, a generation method for generating a model for appropriately estimating information indicating the result of processing will be described.

[0043] First, the process of device processing will be described. Here, as an example of device processing, the laser welding process in a manufacturing line will be described, but the application of this embodiment is not limited thereto.

[0044] Generally, in the process of device processing, the quality of the process is evaluated. The quality evaluation is made by evaluating information indicating the quality of the device processing, more specifically, physical quantities related to the quality of the device processing. However, the above physical quantities are not always measurable, and there are cases where they cannot be measured.

[0045] For example, in the laser welding process in a manufacturing line, joint strength is cited as one of the indicators for evaluating process quality. If the joint strength is measured inline, there is an advantage that it may lead to control for preventing defects in products produced through the process.

[0046] However, it is substantially difficult or impossible to measure the joint strength inline. Therefore, the connection strength has to be evaluated by a joint strength evaluation test performed offline.

[0047] Also, the joint strength is correlated with the interfacial melting area between the plate materials to be welded, and the interfacial melting area can be calculated from the interfacial weld width and the welding distance. Therefore, if the interfacial weld width between the plate materials can be estimated, it can lead to the evaluation of the joint strength. However, at present, the interfacial weld width is not measured inline either.

[0048] The system (also referred to as an estimation system) of this embodiment estimates information indicating the quality of device processing from measurable information among information indicating processing conditions and measurable information among information indicating processing results, thereby enabling the evaluation of the quality of device processing. According to this method, when information indicating the quality of device processing is not directly measured, that information can be obtained by estimation.

[0049] Hereinafter, a model generation method for generating a model that estimates information indicating the quality of device processing from information indicating processing conditions and information indicating processing results, and an estimation method for the above information using the above model will be described.

[0050] FIG. 1 is an explanatory diagram showing the configuration of system 1 according to the present embodiment.

[0051] As shown in FIG. 1, system 1 is an estimation system including a generation device 10 and an estimation device 20. The estimation device 20 is connected to a processing device 29.

[0052] The generation device 10 is a device that generates a model for estimating information indicating the result of device processing. The generation device 10 generates a model (also referred to as an estimation model) for estimating information indicating the result of device processing based on information obtained by conducting an experiment on device processing by the processing device 29. The generation device 10 provides the generated estimation model to the estimation device 20. The generation device 10 executes the above processing offline.

[0053] The estimation device 20 is a device that estimates information indicating the result of device processing. The estimation device 20 acquires information indicating the conditions of device processing and information indicating the result of device processing from the processing device 29, and inputs the acquired information into the estimation model to estimate information indicating the result of device processing. The estimation device 20 executes the above processing inline.

[0054] The processing device 29 is a device that performs device processing. Specifically, device processing includes laser welding of the device or sputtering, etc.

[0055] FIG. 2 is an explanatory diagram showing the processing of system 1 according to the present embodiment.

[0056] As shown in FIG. 2, in step S1, the generation device 10 generates an estimation model for estimating information indicating the result of device processing offline. At this time, the generation device 10 generates the above estimation model using the information obtained by conducting a processing experiment.

[0057] In step S2, the generation device 10 stores the estimation model generated in step S1 in the estimation device 20.

[0058] In step S3, the estimation device 20 estimates and outputs information (parameters) indicating the result of processing in-line using the estimation model stored in step S2. At this time, the estimation device 20 estimates the above information using the information obtained as the result of actually performing device processing.

[0059] Hereinafter, the configurations and processes of the generation device 10 and the estimation device 20 will be described.

[0060] (Generation Device 10) FIG. 3 is a block diagram showing the functional configuration of the generation device 10 according to the present embodiment.

[0061] As shown in FIG. 3, the generation device 10 includes, as functional units, an acquisition unit 11, a derivation unit 12, and a generation unit 13. The generation device 10 can be realized by a general computer device. Each functional unit included in the generation device 10 can be realized by a processor (for example, a CPU (Central Processing Unit)) (not shown) included in the generation device 10 executing a program using a memory (not shown).

[0062] The acquisition unit 11 is a functional unit that conducts experiments on device processing and acquires the first parameter and the second parameter indicating the processing conditions, and the third parameter and the fourth parameter indicating the processing results. The first parameter, the second parameter, the third parameter, and the fourth parameter are also referred to as the first type of information, the second type of information, the third type of information, and the fourth type of information, respectively. The experiment on device processing is an experiment conducted assuming the device processing before the actual device processing, and is carried out offline. The experiment on device processing includes, for example, a physical experiment in which the same process as the processing process in the production line is actually carried out in a separate environment, or a simulation experiment in which a process simulating the processing process in the production line is carried out by computer simulation.

[0063] When acquiring the above parameters by a physical experiment, the acquisition unit 11 may acquire the results of the physical experiment conducted in an experimental device different from the generation device 10 from the device. In that case, the acquisition unit 11 may control the above experimental device.

[0064] When acquiring the above parameters by a simulation experiment, the acquisition unit 11 may execute the simulation experiment using the computer resources (such as a processor and a memory) provided in the generation device 10.

[0065] The acquisition unit 11 also acquires an experimental design model 105 including a set of parameters used in the experiment. The experimental design model 105 is used in the experiment for acquiring the third parameter and the fourth parameter.

[0066] The derivation unit 12 is a functional unit that derives the relationships among the first parameter, the second parameter, the third parameter, and the fourth parameter. Specifically, the derivation unit 12 derives the relationship among the first parameter, the second parameter, and the third parameter (also referred to as the first relationship). The derivation unit 12 also derives the relationship among the first parameter, the second parameter, and the fourth parameter (also referred to as the second relationship).

[0067] Further, the derivation unit 12 derives the relationships of the extended first parameter, the extended second parameter, the extended third parameter, and the extended fourth parameter, which are obtained by respectively extending the first parameter, the second parameter, the third parameter, and the fourth parameter. Specifically, the derivation unit 12 derives the relationship of the extended first parameter, the extended second parameter, and the extended third parameter (also referred to as the extended first relationship). Further, the derivation unit 12 derives the relationship of the extended first parameter, the extended second parameter, and the extended fourth parameter (also referred to as the extended second relationship).

[0068] Here, the extended first parameter is obtained by adding a new first parameter to the first parameter. The extended second parameter is obtained by adding a new second parameter to the second parameter. The extended third parameter is obtained by adding a new third parameter to the third parameter. The extended fourth parameter is obtained by adding a new fourth parameter to the fourth parameter.

[0069] When deriving the extended first relationship and the extended second relationship, the derivation unit 12 obtains an extended experimental model (also referred to as extended plan information) that includes the extended first parameter and the extended second parameter. The extended experimental model is an extended experimental model in which the uniformity between the extended second parameter and the extended third parameter obtained as a result of the experiment performed according to the extended experimental model is equal to or greater than a threshold value. Then, the derivation unit 12 performs an experiment according to the extended experimental model to obtain the extended third parameter and the extended fourth parameter indicating the result of the experiment. Using the extended third parameter and the extended fourth parameter thus obtained, the derivation unit 12 derives the extended first relationship and the extended second relationship.

[0070] Note that the first relationship is expressed, for example, by a first statistical model formula (also referred to as the first formula) that outputs the third parameter with the first parameter and the second parameter as inputs. The second relationship is expressed by a second statistical model formula (also referred to as the second formula) that outputs the fourth parameter with the first parameter and the second parameter as inputs.

[0071] Similarly, the extended first relationship is expressed, for example, by an extended first statistical model formula (also referred to as the extended first formula) that outputs an extended third parameter using the extended first parameter and the extended second parameter as inputs. Further, the extended second relationship is expressed by an extended second statistical model formula (also referred to as the extended second formula) that outputs an extended fourth parameter using the extended first parameter and the extended second parameter as inputs.

[0072] By extending the first parameter and the like performed by the derivation unit 12, the uniformity of the second parameter and the third parameter can be improved. The first parameter and the second parameter included in the experimental design model 105 can be determined in advance to be sufficiently uniform. Here, the uniformity is the degree of uniformity. Further, being sufficiently uniform means that the uniformity of the distribution of the parameter within the range from the minimum value to the maximum value of the parameter is greater than the threshold value. In other words, it corresponds to the density of the parameter being substantially uniform within the above range, or the parameter being evenly present. The uniformity is evaluated, for example, using the average prediction variance. The average prediction variance is an index indicating that the smaller the value, the greater the uniformity of the distribution (that is, the more uniform).

[0073] On the other hand, it is unclear whether the third parameter obtained as a result of the processing is sufficiently uniform, and it may not be sufficiently uniform. In such a case, new first parameters and second parameters are added to the first parameter and the second parameter included in the experimental design model 105 so that the third parameter is sufficiently uniform, thereby extending the first parameter and the second parameter. By doing so, a sufficiently uniform third parameter is obtained by an experiment executed using the extended first parameter and the second parameter. In this way, expanded the uniformity of the first parameter, expanded the second parameter, expanded and the third parameter can be improved. to the tag Therefore, expanded the second parameter, uniformity of and expanded the third parameter can be improved.

[0074] When obtaining the extended experiment model, the derivation unit 12 obtains the extended second parameter by adding a new second parameter to the second parameter. Also, the extended third parameter is obtained by adding a new third parameter to the third parameter. The derivation unit 12 obtains the extended first parameter using the extended third parameter and the first relationship. Then, the derivation unit 12 may obtain an extended experiment model including the obtained extended first parameter and extended second parameter.

[0075] When obtaining the extended experiment model, the derivation unit 12 may obtain an extended experiment model in which the uniformity of the second parameter and the third parameter is equal to or greater than a threshold value by adding a new third parameter and thereby adding a new first parameter belonging to the range from the minimum value to the maximum value among the first parameters.

[0076] The generation unit 13 is a functional unit that generates and outputs an estimation model, which is a model for estimating a fourth parameter indicating the result of processing, using the second parameter and the third parameter measured in-line when the processing device 29 actually processes a device as inputs. The generation unit 13 estimates the fourth information using the extended first relationship and the extended second relationship based on the estimation model.

[0077] When the extended first relationship is expressed by the extended first statistical model formula and the extended second relationship is expressed by the extended second statistical model formula, the estimation model includes an extended third statistical model formula (also referred to as the extended third formula). The extended third statistical model formula is a formula that outputs the extended first parameter using the extended second parameter and the extended third parameter derived from the extended first statistical model formula.

[0078] The estimation model includes the extended second statistical model formula together with the extended third statistical model formula. The estimation model includes a model that obtains the first parameter output by the extended third statistical model formula with the second parameter and the third parameter measured during processing as inputs, and the fourth parameter output by the extended second statistical model formula with the second parameter measured during processing as an input.

[0079] Note that the first parameter and the second parameter may be information defined in advance as information not measured during processing. Also, the second parameter and the third parameter may be information defined in advance as information measured during processing. Information not measured during processing may include, for example, information that can technically be measured during processing but is not actually measured due to constraints such as the cost or time required for measurement. Also, information not measured during processing may include information that is technically difficult or impossible to measure during processing.

[0080] Hereinafter, a method for generating the estimation model by the derivation unit 12 will be described.

[0081] FIG. 4 is an explanatory diagram showing parameters related to the laser welding process. FIG. 5 is an explanatory diagram showing the relationship between the first parameter to the fourth parameter in the experiment according to the present embodiment. The first parameter to the fourth parameter will be described with reference to FIGS. 4 and 5.

[0082] In FIG. 4(a), a state of a laser welding process in which the processing apparatus 29 welds the plate material 9A and the plate material 9B by laser welding is schematically shown. As shown in FIG. 4(a), the plate material 9A and the plate material 9B are arranged so as to partially overlap. The processing apparatus 29 irradiates the region where the plate material 9A and the plate material 9B overlap while scanning the laser beam 91.

[0083] In Fig. 4(b), the cross-sectional state of the plate materials 9A and 9B welded by laser welding using the processing apparatus 29 is schematically shown. As shown in Fig. 4(b), among the plate materials 9A and 9B, the portion irradiated with the laser beam 91 is welded. Among the welded portions of the plate materials 9A and 9B, the width on the upper surface of the plate material 9A (that is, the surface viewed from the +z-axis direction) is referred to as the surface weld width 93, and the width at the interface between the plate materials 9A and 9B is also referred to as the interface weld width 95. Further, there is a minute gap having a gap width 94 between the plate materials 9A and 9B.

[0084] Next, the first parameter 101 to the fourth parameter 104 and the experimental design model 105 used for generating the estimation model will be described with reference to Fig. 5.

[0085] The first parameter 101 is a parameter indicating the processing conditions, and is a parameter for which in-line measurement is not performed due to cost or time constraints, etc. The first parameter 101 is a parameter necessary for accurately estimating the information indicating the result of processing.

[0086] The first parameter 101 includes, for example, the gap width 94 between the plate materials 9A and 9B to be welded. The gap width 94 can be controlled by using a jig in an off-line experiment, and can also be controlled by setting simulation conditions in the case of an experiment by simulation.

[0087] The second parameter 102 is a parameter indicating the processing conditions, and is a parameter for which in-line measurement is performed. The second parameter 102 includes, for example, the laser scan speed 92.

[0088] The experimental design model 105 is information including the parameters (the first parameter 101 and the second parameter 102) to be used in the experiment. The experimental design model 105 is generated based on the preset upper and lower limit values of the values that the first parameter 101 and the second parameter 102 can each take in the experiment. The experimental design model 105 includes setting information of the values (also referred to as experimental point conditions) that the first parameter 101 and the second parameter 102 each take in the experiment. As a result of conducting the experiment under the first parameter 101 and the second parameter 102 set according to the experimental point conditions shown in the experimental design model 105, the third parameter 103 and the fourth parameter 104 are output.

[0089] The third parameter 103 is information indicating the result of processing and is a parameter for which in-line measurement is performed. The third parameter 103 includes, for example, the surface weld width 93 of the laser welded part.

[0090] The fourth parameter 104 is information indicating the result of processing and is a parameter for which in-line measurement is not performed due to cost or time constraints, etc. The fourth parameter 104 is a characteristic parameter related to the quality of processing. The fourth parameter 104 includes, for example, the interface weld width 95 at the interface between the plate materials 9A and 9B of the laser welded part. To directly measure the interface weld width 95, there is a method of, for example, cutting the processed product offline and measuring it on the cut surface, but such measurement is difficult or impossible in-line.

[0091] Next, the first statistical model formula to the third statistical model formula, the extended first statistical model formula to the extended third statistical model formula, and the estimation model will be described.

[0092] FIG. 6 is an explanatory diagram showing the relationship between the first to fourth parameters, the first statistical model formula, and the second statistical model formula, which are derived by the generation device 10 according to the present embodiment. FIG. 7 is an explanatory diagram showing the relationship between the first to third parameters and the third statistical model formula, which are derived by the generation device 10 according to the present embodiment. FIG. 8 is an explanatory diagram showing an estimated model generated by the generation device 10 according to the present embodiment.

[0093] The derivation unit 12 derives the first statistical model formula 111 and the second statistical model formula 112 by performing statistical modeling based on the experimental design model 105 using the set of the first parameter 101 to the fourth parameter 104 obtained by experiments. Here, the first statistical model formula 111 is a model formula that takes the first parameter 101 and the second parameter 102 as input variables (explanatory variables) and the third parameter 103 as an output variable (objective variable). The second statistical model formula 112 is a model formula that takes the first parameter 101 and the second parameter 102 as input variables (explanatory variables) and the fourth parameter 104 as an output variable (objective variable).

[0094] That is, the first statistical model formula 111 and the second statistical model formula 112 can be expressed in the form shown in the following (Equation 1) (see FIG. 6).

[0095] First statistical model formula 111: Third parameter = f1 (First parameter, Second parameter) Second statistical model formula 112: Fourth parameter = f2 (First parameter, Second parameter) (Equation 1)

[0096] By the way, the fourth parameter 104, which is the objective variable, is used for evaluating the process. However, the fourth parameter 104 is a parameter for which in-line measurement is not performed, and thus it will be estimated using the second statistical model formula 112.

[0097] However, since the first parameter 101, which is one of the input variables for the second statistical model formula 112, is also a parameter for which in-line measurement is not performed, the first parameter 101 also needs to be estimated.

[0098] Therefore, as a method for estimating the first parameter 101, the first statistical model equation 111 is used. In the first statistical model equation 111, the first parameter 101 and the second parameter 102 are input variables, and the third parameter 103 is an output variable. By solving the algebraic equation with the first parameter 101 as the unknown, the first statistical model equation 111 can be converted into an equation in which the second parameter 102 and the third parameter 103 are input variables and the first parameter 101 is the output variable. The derivation unit 12 obtains an equation thus converted (corresponding to the third statistical model equation 113, see (Equation 2)) (see FIG. 7). Note that when deriving the third statistical model equation, if the third statistical model equation cannot be determined without appropriate restriction information, appropriate restriction information is introduced.

[0099] Third statistical model equation 113: First parameter = f1 -1 (Second parameter, Third parameter) (Equation 2)

[0100] In this way, the first parameter 101 is calculated by the third statistical model equation 113 including the second parameter 102 and the third parameter 103.

[0101] Then, by substituting (Equation 2) into the first parameter 101 in the second statistical model equation 112 of (Equation 1) (that is, using both the first parameter 101 and the second parameter 102 as input variables), the fourth parameter 104 can be estimated by the second statistical model equation 112.

[0102] That is, the fourth parameter 104 can be expressed in the form of the following (Equation 3) (see FIG. 8).

[0103] Fourth parameter = f2(f1 -1 (Second parameter, Third parameter), Second parameter) (Equation 3)

[0104] A model that can output the fourth parameter 104 when the second parameter 102 and the third parameter 103 are input is called an estimation model 106.

[0105] When the uniformity of the second parameter and the third parameter is equal to or greater than the threshold, the fourth parameter can be estimated with relatively high accuracy by the estimation model 106 using the above (Equation 3). On the other hand, when the uniformity of the second parameter and the third parameter is less than the threshold, the accuracy of the fourth parameter estimated by the estimation model 106 using the above (Equation 3) may be relatively low.

[0106] In this case, using the estimation model 106 derived using the extended first parameter or the like, the fourth parameter having relatively high accuracy can be estimated.

[0107] The extended first statistical model formula, the extended second statistical model formula, and the extended third statistical model formula can be expressed in the form shown in the following (Equation 4) in the same manner as the above (Equation 1) and (Equation 2).

[0108] Extended first statistical model formula: Third parameter = g1(First parameter, Second parameter) Extended second statistical model formula: Fourth parameter = g2(First parameter, Second parameter) Extended third statistical model formula: First parameter = g1 -1 (Second parameter, Third parameter) (Equation 4)

[0109] And the fourth parameter 104 can be expressed in the form shown in the following (Equation 5).

[0110] Fourth parameter = g2(g1 -1 (Second parameter, Third parameter), Second parameter) (Equation 5)

[0111] Note that although g1 is generally different from f1, it is not excluded that g1 is the same as f1. Similarly, although g2 is generally different from f2, it is not excluded that g2 is the same as f2.

[0112] Therefore, if the second parameter 102 and the third parameter 103 obtained by in-line measurement are input to the estimation model 106, the fourth parameter 104 can be estimated as information indicating the result of the processing that was the object of the in-line measurement.

[0113] The processing of the generation device 10 configured as described above will be described.

[0114] FIG. 9 is a flowchart showing the processing executed by the generation device 10 according to the present embodiment. The processing shown in FIG. 9 is the processing included in step S1 of FIG. 2.

[0115] As shown in FIG. 9, in step S101, the acquisition unit 11 acquires an experimental design model 105 (also referred to as a first experimental design model). The factors of the first experimental design model are the first parameter and the second parameter.

[0116] In step S102, the acquisition unit 11 sets the first parameter and the second parameter to be used in the experiment according to the experimental design model 105 (first experimental design model) acquired in step S101.

[0117] In step S103, the acquisition unit 11 executes an experiment using the experimental design model 105 (first experimental design model) set in step S102, that is, the first parameter and the second parameter.

[0118] In step S104, the acquisition unit 11 acquires the third parameter and the fourth parameter output as the result of the experiment executed in step S103.

[0119] In step S105, the derivation unit 12 derives a first statistical model formula using the first parameter and the second parameter set in step S102 and the third parameter acquired in step S104.

[0120] In step S106, the derivation unit 12 derives a second statistical model formula using the first parameter and the second parameter set in step S102 and the fourth parameter acquired in step S104.

[0121] In step S107, the derivation unit 12 derives a third statistical model formula using the first statistical model formula derived in step S105 and the second statistical model formula derived in step S106.

[0122] In step S108, the derivation unit 12 acquires a second experimental design model. The factors of the second experimental design model are the second parameter that is a factor of the first experimental design model and the third parameter output as a result of the experiment executed in step S103.

[0123] In step S109, the derivation unit 12 determines whether or not the uniformity of the factors of the second experimental design model is less than a threshold value (also referred to as the first threshold value). Here, the uniformity of the factors of the second experimental design model is an index indicating the uniformity of the distribution of the factors of the second experimental design model, and the larger the value, the higher the uniformity (that is, the more uniform). Also, the factors of the second experimental design model are, more specifically, the extended second parameter and the extended third parameter included in the second experimental design model. If it is determined that the uniformity of the factors of the second experimental design model is less than the first (Yes in step S109), the process proceeds to step S110, and if not (No in step S109), the series of processes shown in FIG. 9 is terminated.

[0124] The uniformity of the factors of the second experimental design model is evaluated using, for example, the average prediction variance. In this case, when the value obtained by dividing the average prediction variance of the factors of the second experimental design model by the average prediction variance of the factors of the first experimental design model is less than a threshold value (also referred to as the second threshold value), it corresponds to the uniformity of the factors of the second experimental design model being less than the first threshold value.

[0125] Note that the second threshold value can be set to a numerical value within a range from, for example, 1 to 2, and more specifically, the inventors of the present application have obtained the finding that it can be set to 1.05.

[0126] In step S110, the derivation unit 12 expands the second experimental design model. The derivation unit 12 expands the second experimental design model by adding a set of the second parameter and the third parameter to the factors of the second experimental design model. At this time, the first parameter output by inputting the set of the second parameter and the third parameter to be added into the first statistical model formula 111 satisfies the condition of belonging to the range from the minimum value to the maximum value among the plurality of first parameters included in the first experimental design model, and a set of the second parameter and the third parameter is added to the factors of the second experimental design model. That is, the derivation unit 12 obtains a second experimental design model in which the uniformity of the factors is equal to or greater than the first threshold value so that the added first parameter belongs to the range from the minimum value to the maximum value among the plurality of first parameters included in the first experimental design model.

[0127] In other words, the derivation unit 12 obtains an extended second experimental design model by adding a set of the second parameter and the third parameter that satisfies the condition (also referred to as a constraint condition) shown in the following (Equation 6) to the second experimental design model.

[0128] Maximum value (first parameter in the first experimental design model) ≧ f1 -1 (Second parameter, third parameter), and Minimum value (first parameter in the first experimental design model) ≦ f1 -1 (Second parameter, third parameter) (Equation 6)

[0129] Incidentally, there are various methods for determining the combination of the second parameter and the third parameter to be added. Specific examples include methods such as the D-optimal design method or the I-optimal design method. The I-optimal design method is a determination method that attempts to minimize the prediction variance in all planned regions. The D-optimal design method is a determination method that focuses on reducing the prediction variance at each planned point. The inventors of the present application have obtained the finding that adopting the D-optimal design method rather than the I-optimal design method can obtain the second parameter and the third parameter with higher uniformity.

[0130] In step S111, the derivation unit 12 determines whether the uniformity of the distribution of the factors of the second experimental design model extended in step S110 is less than the first threshold. If it is determined that the uniformity of the distribution of the factors of the second experimental design model is less than the first threshold (Yes in step S111), step S110 is executed again. Otherwise (No in step S111), the process proceeds to step S112. Proceeding to step S112 corresponds to the case where the extension of the second experimental design model (step S110) is executed one or more times so that the uniformity of the distribution of the factors of the second experimental design model becomes equal to or greater than the first threshold.

[0131] In step S112, the derivation unit 12 obtains the extended first parameter using the extended second experimental design model and the third statistical model formula.

[0132] In step S113, the derivation unit 12 extends the first experimental design model using the extended first parameter obtained in step S112.

[0133] In step S114, the derivation unit 12 sets the first experimental design model extended in step S113, that is, the extended first parameter and the extended second parameter, as the first parameter and the second parameter to be used in the experiment.

[0134] In step S115, the derivation unit 12 executes an experiment using the first experimental design model extended in step S113, that is, the extended first parameter and the extended second parameter.

[0135] In step S116, the derivation unit 12 acquires the third parameter and the fourth parameter output as the results of the experiment executed in step S115 as the extended third parameter and the extended fourth parameter, respectively.

[0136] In step S117, the derivation unit 12 derives the first statistical model formula (also referred to as the extended first statistical model formula) using the extended first parameter and the extended second parameter set in step S114 and the extended third parameter acquired in step S116 (see the above (Equation 4)).

[0137] In step S118, the derivation unit 12 derives the second statistical model formula (also referred to as the extended second statistical model formula) using the extended first parameter and the extended second parameter set in step S114 and the extended fourth parameter acquired in step S116 (see the above (Equation 4)).

[0138] In step S119, the derivation unit 12 derives the third statistical model formula (extended third statistical model formula) using the extended first statistical model formula derived in step S117 and the extended second statistical model formula derived in step S118 (see the above (Equation 4)).

[0139] Through the series of processes shown in FIG. 9, a model for appropriately estimating the information indicating the result of processing can be generated.

[0140] (Estimation device 20) Next, the estimation device 20 will be described.

[0141] FIG. 10 is a block diagram showing the hardware configuration of the estimation device 20 according to the present embodiment.

[0142] The estimation device 20 is realized by, for example, a computer, and includes a processor 21, a memory 22, an input / output IF 23, a sensor 24, an input device 25, and a display device 26.

[0143] The processor 21 is an arithmetic device that performs parameter estimation processing, and is, for example, a CPU.

[0144] The memory 22 is a storage device that stores programs or data, and is, for example, a RAM (Random Access Memory). The estimated model 106 generated by the generation device 10 is stored in the memory 22.

[0145] The input / output IF 23 is an interface device that exchanges data with each other among the processor 21, the memory 22, the sensor 24, the input device 25, and the display device 26. The input / output IF 23 is connected to each of the above devices. The connection is wired or wireless, or a combination thereof may be used.

[0146] The sensor 24 is installed in the processing device 29 that is the object of in-line measurement. The processing device 29 is, for example, a laser welding device. The sensor 24 is, for example, a laser displacement meter that measures the surface welding width 93 (see (b) of FIG. 4) of the plate material to be welded.

[0147] The input device 25 is a device that receives input of information regarding the first parameter to the fourth parameter, and is, for example, a keyboard or a touch panel.

[0148] The display device 26 is a device that shows information regarding the first parameter to the fourth parameter, and is, for example, an LCD (Liquid Crystal Display) monitor.

[0149] FIG. 11 is a block diagram showing the functional configuration of the estimation device 20 according to the present embodiment.

[0150] As shown in FIG. 11, the estimation device 20 includes, as functional components, an input unit 31, a sensor data acquisition unit 32, a parameter estimation unit 33, an output unit 34, and a storage unit 35.

[0151] The input unit 31 is a functional unit that receives, from a user via the input device 25, the input of determination value information such as standard values regarding the first parameter 101, the second parameter 102, the third parameter 103, and the fourth parameter 104. The timing of the input is, for example, when the model of the processing device 29 is switched, but is not limited thereto. The value input here is registered in the input value storage unit 36 of the storage unit 35.

[0152] The sensor data acquisition unit 32 is a functional unit that acquires measurement data of the second parameter 102 and the third parameter 103 from the sensor 24 connected to the processing device 29. The second parameter 102 is, for example, the scan speed 92 (see FIG. 4(a)), and the third parameter 103 is, for example, the surface weld width 93 of the plate material to be welded (see FIG. 4(b)). The acquisition frequency of the data can be arbitrarily set. In subsequent parameter estimation, the sequentially acquired data may be used each time, or the average value may be calculated from the data acquired multiple times for one workpiece, and the average value may be used as the representative value of the workpiece. The acquired data is recorded in the sensor data storage unit 37. Further, it is determined whether the acquired data conforms to the determination conditions stored in the input value storage unit 36. If it does not conform, quality information indicating a defect (NG) is output. The determination conditions are, for example, conditions indicating standard values or conditions indicating a normal range.

[0153] The parameter estimation unit 33 estimates the fourth parameter 104 by inputting the second parameter 102 and the third parameter 103 recorded in the sensor data storage unit 37 into the estimation model 106 (that is, by using the above formula (3)). The fourth parameter 104 is, for example, the interface welding width 95 between the plate materials (see (b) of FIG. 4). The estimated value of the calculated fourth parameter 104 is recorded in the parameter estimated value storage unit 38. Further, it is determined whether the estimated value of the calculated fourth parameter 104 conforms to the determination conditions stored in the input value storage unit 36, and if it does not conform, quality information indicating a defect (NG) is output. The determination conditions are, for example, conditions indicating standard values or conditions indicating a normal range.

[0154] The output unit 34 is a functional unit that outputs the data recorded in the storage unit 35 or the determination result. The output unit 34 outputs, for example, by displaying the above data or the like on the display device 26. Note that the output unit 34 may output the above data or the like by voice or may output it by transmitting it to another device through communication.

[0155] The storage unit 35 is a functional unit that stores various values and various data. The storage unit 35 includes an input value storage unit 36, a sensor data storage unit 37, and a parameter estimated value storage unit 38. Values or data are stored or read out from the storage unit 35 by the above functional units.

[0156] FIG. 12 is a flowchart showing the processing executed by the estimation device 20 according to the present embodiment. The processing shown in FIG. 12 is the processing included in step S3 of FIG. 2.

[0157] In step S301, the sensor data acquisition unit 32 acquires the measurement data of the second parameter 102 and the third parameter 103 from the sensor 24.

[0158] In step S302, the sensor data acquisition unit 32 stores the measurement data acquired in step S301 in the sensor data storage unit 37.

[0159] In step S303, the sensor data acquisition unit 32 determines whether the measurement data acquired in step S301 conforms to the determination condition. If it conforms to the determination condition (Yes in step S303), step S304 is executed; otherwise (No in step S303), step S311 is executed.

[0160] In step S304, the parameter estimation unit 33 estimates the fourth parameter 104 by inputting the second parameter 102 and the third parameter 103, which are the measurement data recorded in the sensor data storage unit 37, into the estimation model 106 (that is, by using the above formula (3)).

[0161] In step S305, the parameter estimation unit 33 stores the fourth parameter 104 estimated in step S304 in the parameter estimation value storage unit 38.

[0162] In step S306, the parameter estimation unit 33 determines whether the fourth parameter 104 estimated in step S304 conforms to the determination condition. If it conforms to the determination condition (Yes in step S306), step S307 is executed; otherwise (No in step S306), step S312 is executed.

[0163] In step S307, the output unit 34 outputs quality information indicating that it is a good product (OK).

[0164] In step S311, the output unit 34 outputs quality information indicating that it is a defective product (NG) based on the fact that the second parameter 102 or the third parameter 103 does not conform to the determination condition.

[0165] In step S312, the output unit 34 outputs quality information indicating that it is a defective product (NG) based on the fact that the fourth parameter 104 does not conform to the determination condition.

[0166] When the processing of step S307, S311, or S312 is completed, the series of processes shown in FIG. 12 is terminated.

[0167] By the series of processes shown in FIG. 12, in, for example, a laser welding process, it becomes possible to estimate inline the interface welding width 95 between plate materials based on measurement data such as the scan speed 92 of the laser for inline measurement or the surface welding width 93 of the laser welding part. If it were attempted to measure the interface welding width 95, it would be necessary to observe the cross-sectional shape offline, but there is an effect that it becomes possible to estimate the interface welding width 95 inline.

[0168] Hereinafter, examples of the accuracy of estimation by the first experimental design model, the second experimental design model, and the estimation model will be described.

[0169] FIG. 13 is an explanatory diagram showing an example of the first experimental design model acquired by the generation device 10 according to the present embodiment.

[0170] In FIG. 13, as the first experimental design model, for each of 21 cases, parameters A, B, C, D, and E indicating the processing conditions (also referred to as factors) of the device are shown. Here, parameters A, B, C, and D correspond to the second parameters, and parameter E corresponds to the first parameter.

[0171] The first experimental design model shown in FIG. 13 is an example of the first experimental design model acquired by the acquisition unit 11 in step S101 (see FIG. 9).

[0172] FIG. 14 is an explanatory diagram showing an example of the results of an experiment conducted according to the first experimental design model according to the present embodiment.

[0173] In FIG. 14, for each of 21 cases, parameters G and H indicating the results (also referred to as responses) of the processing experiment using the parameters A, B, C, D, and E shown in FIG. 13 are shown. Here, parameter G corresponds to the third parameter, and parameter H corresponds to the fourth parameter.

[0174] The third parameter and the fourth parameter shown in FIG. 14 are examples of the third parameter and the fourth parameter acquired by the acquisition unit 11 in step S104 (see FIG. 9).

[0175] FIG. 15 is an explanatory diagram showing an example of the second experimental design model acquired by the generation device according to the present embodiment.

[0176] In the second experimental design model shown in FIG. 15, the second parameters (that is, parameters A, B, C, and D) and the third parameter (that is, parameter F) shown in FIG. 14 are used as factors, and the first parameter (that is, parameter E) and the fourth parameter (that is, parameter H) are used as responses.

[0177] FIG. 16 is an explanatory diagram showing an example of an extension of the second experimental design model acquired by the generation device according to the present embodiment.

[0178] FIG. 16 shows the first to fourth parameters for 24 cases.

[0179] Specifically, for the 21 cases from case 1 to case 21 in FIG. 16, the second parameters (that is, parameters A, B, C, and D) and the third parameter (that is, parameter F) shown in FIG. 15 are included as factors.

[0180] Also, for the 3 cases from case 22 to case 24 in FIG. 16, the second parameters and the third parameter added so that the uniformity of the factors (that is, the second parameter and the third parameter) of the second experimental design model is equal to or greater than the first threshold value are included (steps S110, S111).

[0181] In other words, the second experimental design model shown in FIG. 16 is an example of the second experimental design model extended one or more times by the derivation unit 12 in steps S110 and S111 (see FIG. 9).

[0182] FIG. 17 is an explanatory diagram showing an example of the expansion of the first experimental design model according to the present embodiment.

[0183] For the 21 cases from case 1 to case 21 in FIG. 17, the first parameter and the second parameter shown in FIG. 13 are shown.

[0184] Also, for the 3 cases from case 22 to case 24 in FIG. 17, the second parameter added to the second experimental design model, and the first parameter calculated by the first statistical model formula from the second parameter and the third parameter added to the second experimental design model are shown.

[0185] The expanded first experimental design model shown in FIG. 17 is an example of the first experimental design model expanded by the derivation unit 12 in step S113 (see FIG. 9).

[0186] FIG. 18 is an explanatory diagram showing an example of the results of an experiment conducted according to the expanded first experimental design model according to the present embodiment.

[0187] For the 21 cases from case 1 to case 21 in FIG. 18, the first parameter to the fourth parameter shown in FIG. 14 are shown.

[0188] Also, for the 3 cases from case 22 to case 24 in FIG. 18, the third parameter and the fourth parameter obtained by the experiment conducted by setting the first parameter and the second parameter added to the second experimental design model are shown.

[0189] The third parameter and the fourth parameter shown in FIG. 18 are examples of the expanded third parameter and the expanded fourth parameter obtained by the derivation unit 12 in step S116 (see FIG. 9).

[0190] After that, using the first parameter to the fourth parameter shown in FIG. 18, the expanded first statistical model formula, the expanded second statistical model formula, and the expanded third statistical model formula are derived by the derivation unit 12 (steps S117 to S119).

[0191] FIG. 19 is an explanatory diagram showing an example of the estimation accuracy of the estimation model according to the present embodiment.

[0192] FIG. 19 shows a graph in which the measured values (vertical axis, denoted as "H measured value") are plotted against the estimated values (horizontal axis, denoted as "H estimated value") for the parameter H, which is the fourth parameter, in the 24 cases shown in FIG. 18 and the like.

[0193] Here, the results based on the estimated values obtained from the estimation model generated using the first parameter and the second parameter included in the first experimental design model before expansion are indicated by black circles. This corresponds to the 21 cases from case 1 to case 21 in FIG. 18.

[0194] Also, the results based on the estimated values obtained from the estimation model generated using the additional first parameter and the additional second parameter included in the expanded first experimental design model are indicated by white circles. This corresponds to the 3 cases from case 22 to case 24 in FIG. 18.

[0195] For the estimated values and the measured values obtained from the estimation model generated using the additional first parameter and the additional second parameter included in the first experimental design model before expansion, the RMSE (root mean square error) is 0.0274.

[0196] Also, for the estimated values and the measured values obtained from the estimation model generated using the additional first parameter and the additional second parameter included in the expanded first experimental design model, the RMSE (root mean square error) is 0.0277.

[0197] Thus, by expanding the first experimental design model, the estimation accuracy of the parameter H by the estimation model is improved.

[0198] (Modification example) In this embodiment, another form of the generation method and the estimation method for generating a model that appropriately estimates information indicating the result of processing will be described.

[0199] FIG. 20 is a flowchart showing the processing (i.e., the generation method) executed by the generation device according to this modification. The processing shown in FIG. 20 is another example of the processing included in step S1 of FIG. 2.

[0200] As shown in FIG. 20, in step S121, the generation device performs an experiment of processing according to the plan information including the first type of information and the second type of information indicating the processing conditions of the device, thereby obtaining the third type of information and the fourth type of information indicating the result of the processing experiment.

[0201] In step S122, the generation device obtains extended plan information including extended first type of information obtained by adding new first type of information to the first type of information and extended second type of information obtained by adding new second type of information to the second type of information, where the uniformity between the extended second type of information and the extended third type of information obtained as a result of the processing experiment performed according to the extended plan information is equal to or greater than a threshold value.

[0202] In step S123, the generation device performs an experiment of processing according to the extended plan information, thereby obtaining the extended third type of information and the extended fourth type of information indicating the result of the processing experiment.

[0203] In step S124, the generation device derives an extended first relationship that is the relationship among the extended first type of information, the extended second type of information, and the extended third type of information, and derives an extended second relationship that is the relationship among the extended first type of information, the extended second type of information, and the extended fourth type of information.

[0204] In step S125, the generation device generates and outputs a model that estimates the fourth type of information indicating the result of processing by using the extended first relationship and the extended second relationship, with the second type of information and the third type of information measured during the processing of the device as inputs.

[0205] As a result, the generation device can generate a model for appropriately estimating information indicating the result of processing.

[0206] FIG. 21 is a flowchart showing the processing (i.e., the estimation method) executed by the estimation device according to this modified example. The processing shown in FIG. 21 is another example of the processing included in step S3 of FIG. 2.

[0207] In step S321, the estimation device inputs the second type of information and the third type of information measured during processing into the model output by the above-described generation device.

[0208] In step S322, the estimation device outputs the fourth type of information output by inputting the second type of information and the third type of information as estimation information obtained by estimating the result of processing.

[0209] As a result, the estimation device can estimate the information indicating the result of processing by using the model for appropriately estimating the information indicating the result of processing.

[0210] As described above, according to the generation method of the present embodiment, by using the relationships among the first type of information, the second type of information, the third type of information, and the fourth type of information obtained from experiments, a model for estimating the fourth type of information during processing can be obtained from the second type of information and the third type of information obtained during processing. By using this model, even when the fourth type of information cannot be obtained during processing, as long as the second type of information and the third type of information can be obtained during processing, the fourth type of information during processing can be obtained by estimation. Here, since the model for estimation is generated by using the extended third type of information extended so that the uniformity is equal to or higher than the threshold value, the accuracy of the fourth type of information estimated by the model can be further increased. Thus, according to the above-described generation method, a model for appropriately estimating the information indicating the result of processing can be generated. Also, when estimating the fourth type of information, it is not necessary to conduct a new experiment using the fourth type of information as an output variable. Therefore, there is an effect that an increase in the number of experiments due to obtaining information that cannot be obtained during processing can be avoided in advance.

[0211] In addition, since the extended first type of information obtained using the extended second type of information and the extended third type of information is used, the extended plan information can be obtained more easily. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be generated more easily.

[0212] Also, by determining that the uniformity of both the extended second type of information and the extended third type of information is equal to or greater than a threshold value, the extended third type of information with a uniformity equal to or greater than the threshold value can be obtained more easily. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the result of processing can be generated more easily.

[0213] In addition, the new first type of information added by the addition of the third type of information comes to belong to the range in which the first type of information indicating the processing conditions is distributed. Therefore, the new first type of information belongs to an appropriate range as a processing condition, in other words, it is possible to avoid the new first type of information deviating from an appropriate range as a processing condition. As a result, the extended first relationship becomes even more appropriate, and the accuracy of the estimated fourth type of information can be further increased. Therefore, according to the above generation method, a model for more appropriately estimating the information indicating the result of processing can be generated.

[0214] In addition, since the uniformity of the extended second type of information and the extended third type of information is evaluated by the average prediction variance, the uniformity of the extended second type of information and the extended third type of information can be evaluated more easily. Therefore, according to the above generation method, a model for estimating the information indicating the result of processing can be generated.

[0215] In addition, since the extended second type of information and the extended third type of information are obtained by the D-optimal design method for the evaluation plan information including the extended second type of information and the extended third type of information as factors, more appropriate extended third type of information can be obtained more easily. Therefore, according to the above generation method, a model for estimating the information indicating the result of processing can be generated more easily.

[0216] Further, the fourth type of information can be obtained by using the extended third formula derived from the extended first formula through formula transformation. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the processing result can be generated more easily.

[0217] Also, the extended first type of information can be easily estimated from the measured values of the second type of information and the third type of information by the extended third formula, and the fourth type of information can be estimated from the estimated extended first type of information and the measured value of the second type of information by the extended second formula. Therefore, according to the above generation method, a model for appropriately estimating the information indicating the processing result can be generated more easily.

[0218] Also, when there is information that is not measured in the information indicating the processing conditions and there is information that is not measured in the information indicating the processing result, the information indicating the processing result can be appropriately estimated. Therefore, according to the above generation method, even when there is information that is not measured during processing, a model for appropriately estimating the information indicating the processing result can be generated.

[0219] Also, a model for appropriately estimating the information indicating the processing result in laser welding can be generated more easily.

[0220] Also, according to the estimation method of the present embodiment, by inputting the second type of information and the third type of information obtained during processing into the model, the information indicating the processing result can be appropriately estimated.

[0221] In the above embodiment, each component may be configured by dedicated hardware or may be realized by executing a software program suitable for each component. Each component may be realized by a program execution unit such as a CPU or a processor reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory. Here, the software for realizing the generation device and the estimation device of the above embodiment is the following program.

[0222] That is, this program causes a computer to execute a generation method that a processor executes using a memory. By conducting an experiment on the processing according to planned information including first-type information and second-type information indicating processing conditions of a device, third-type information and fourth-type information indicating the result of the processing experiment are obtained. The extended planned information includes extended first-type information obtained by adding new first-type information to the first-type information and extended second-type information obtained by adding new second-type information to the second-type information. Extended planned information is obtained in which the uniformity between the extended second-type information and extended third-type information obtained as a result of the processing experiment conducted according to the extended planned information is equal to or greater than a threshold value. By conducting the processing experiment according to the extended planned information, the extended third-type information and extended fourth-type information indicating the result of the processing experiment are obtained. An extended first relationship that is a relationship among the extended first-type information, the extended second-type information, and the extended third-type information is derived, and an extended second relationship that is a relationship among the extended first-type information, the extended second-type information, and the extended fourth-type information is derived. A model is generated and output that, using the extended first relationship and the extended second relationship, estimates the fourth-type information indicating the result of the processing with the second-type information and the third-type information measured during the processing of the device as inputs.

[0223] Also, this program causes a computer to execute an estimation method that inputs the second-type information and the third-type information measured during the processing of the device into the model output by the above generation method, and outputs the fourth-type information output by inputting the second-type information and the third-type information into the model as estimation information estimating the result of the processing.

[0224] As described above, the estimation device and the like according to one or more aspects have been described based on the embodiments. However, the present invention is not limited to these embodiments. As long as the gist of the present invention is not deviated from, various modifications conceived by those skilled in the art applied to these embodiments or forms constructed by combining components in different embodiments may also be included within the scope of one or more aspects.

Industrial Applicability

[0225] The model generation method, parameter estimation method, and system according to the present invention enable parameter estimation of an objective variable with a small number of experiments even when the explanatory variables include parameters for which measurement data is not collected inline, and are useful as a model generation method, parameter estimation method, and system.

Explanation of Signs

[0226] 1 System 9A, 9B Plate material 10 Generation device 11 Acquisition unit 12 Derivation unit 13 Generation unit 20 Estimation device 21 Processor 22 Memory 23 Input / output IF 24 Sensor 25 Input device 26 Display device 29 Processing device 31 Input unit 32 Sensor data acquisition unit 33 Parameter estimation unit 34 Output unit 35 Storage unit 36 Input value storage unit 37 Sensor data storage unit 38 Parameter estimation value storage unit 91 Laser beam 92 Scanning speed 93 Surface welding width 94 Gap width 95 Interface welding width 101 First parameter 102 Second parameter 103 Third parameter 104 Fourth parameter 105 Experiment design model 106 Estimation model 111 First statistical model formula 112 Second statistical model equation 113 Third statistical model equation

Claims

1. A generation method executed by a processor using a memory, comprising: Performing an experiment on the processing according to plan information including first type information and second type information indicating processing conditions of a device, thereby obtaining third type information and fourth type information indicating the result of the processing experiment; Extended plan information including extended first type information obtained by adding new first type information to the first type information and extended second type information obtained by adding new second type information to the second type information, and obtaining extended plan information in which the uniformity of each of the extended third type information obtained as a result of the processing experiment performed according to the extended second type information and the extended plan information is equal to or greater than a threshold value; Performing an experiment on the processing according to the extended plan information, thereby obtaining the extended third type information and extended fourth type information indicating the result of the processing experiment; Deriving an extended first relationship that is a relationship among the extended first type information, the extended second type information, and the extended third type information, and deriving an extended second relationship that is a relationship among the extended first type information, the extended second type information, and the extended fourth type information; Generating and outputting a model that, using the extended first relationship and the extended second relationship, estimates the fourth type information indicating the result of the processing with the second type information and the third type information measured during the processing of the device as inputs; Generation method.

2. Furthermore, deriving a first relationship that is a relationship among the first type information, the second type information, and the third type information; In the acquisition of the extended plan information: Obtaining the extended second type information by adding the new second type information to the second type information; Obtaining the extended third type information by adding new third type information to the third type information; Obtaining the extended first type information using the first relationship with the extended second type information and the extended third type information as inputs; Obtaining the extended plan information including the extended first type information and the extended second type information; The generation method according to claim 1.

3. In the acquisition of the extended plan information: Determining whether the uniformity of the obtained extended second type information and the extended third type information is equal to or greater than a threshold value; When it is determined that the uniformity of the obtained extended second type information and the extended third type information is equal to or greater than the threshold value, obtaining the extended plan information; The generation method according to claim 1 or 2.

4. In the acquisition of the extended plan information: By adding new third-type information to the third-type information, new first-type information belonging to the range from the minimum value to the maximum value among the first-type information is added, thereby obtaining the extended plan information. The generation method according to any one of claims 1 to 3.

5. In the acquisition of the extended plan information, Using the average prediction variance of the evaluation plan information including the second-type information and the third-type information as factors, the uniformity between the extended second-type information and the extended third-type information is calculated. The extended plan information is obtained using the calculated uniformity. The generation method according to any one of claims 1 to 4.

6. In the acquisition of the extended second-type information and the acquisition of the extended third-type information, The extended second-type information and the extended third-type information are obtained by the D-optimal design method for the evaluation plan information including the second-type information and the third-type information as factors. The generation method according to claim 2.

7. The extended first relationship is expressed by an extended first formula that outputs the extended third-type information with the extended first-type information and the second-type information as inputs. The model is An extended third formula derived from the extended first formula, including an extended third formula that outputs the first-type information with the second-type information and the third-type information measured during the processing of the device as inputs. The generation method according to any one of claims 1 to 6.

8. The extended second relationship is expressed by an extended second formula that outputs the fourth-type information with the extended first-type information and the second-type information as inputs. The model is Furthermore, it includes the extended second formula. A model that obtains the fourth-type information output by the extended second formula with the first-type information output by the extended third formula with the second-type information and the third-type information measured during the processing of the device as inputs and the second-type information measured during the processing of the device as inputs. The generation method according to claim 7.

9. The first-type information and the fourth-type information are information defined in advance as information not measured during the processing. The second-type information and the third-type information are information defined in advance as information measured during the processing. The generation method according to any one of claims 1 to 8.

10. The processing is laser welding. The first-type information includes the gap width between the plates to be welded in the laser welding. The second type of information includes the scanning speed of the laser in the laser welding, The third type of information includes the surface welding width of the laser welding part in the laser welding, The fourth type of information includes the interfacial welding width of the laser welding part in the laser welding The generation method according to any one of claims 1 to 9.

11. Input the second type of information and the third type of information measured during the processing of the device into the model output by the generation method according to any one of claims 1 to 10, Output the fourth type of information output by inputting the second type of information and the third type of information into the model as estimated information for estimating the result of the processing Estimation method.

12. Comprising a processor and a memory connected to the processor, The processor uses the memory, By performing the experiment of the processing according to the plan information including the first type of information and the second type of information indicating the processing conditions of the device, the third type of information and the fourth type of information indicating the result of the experiment of the processing are obtained, Extended plan information including extended first type of information obtained by adding new first type of information to the first type of information and extended second type of information obtained by adding new second type of information to the second type of information, and the uniformity of each of the extended third type of information obtained as the result of the experiment of the processing performed according to the extended second type of information and the extended plan information is a threshold value or more, By performing the experiment of the processing according to the extended plan information, the extended third type of information and the extended fourth type of information indicating the result of the experiment of the processing are obtained, Derive an extended first relationship which is the relationship among the extended first type of information, the extended second type of information and the extended third type of information, and derive an extended second relationship which is the relationship among the extended first type of information, the extended second type of information and the extended fourth type of information, Generate and output a model that estimates the fourth type of information indicating the result of the processing by using the extended first relationship and the extended second relationship with the second type of information and the third type of information measured during the processing of the device as inputs Generation device.

13. Comprising a processor and a memory connected to the processor, The processor uses the memory, The fourth type of information output by inputting the second type of information and the third type of information measured during the processing into the model output by the generation device according to claim 12 is output as estimation information estimating the result of the processing. Estimation device.

Images