System, information processing device, information processing method, program, and recording medium

Abstract

It provides a technology that is advantageous for conducting searches more efficiently. [Solution] The system executes multiple search cycles. The information processing device receives input of search conditions and, in each of the multiple search cycles, executes a decision process to determine the sample preparation conditions. In each of the multiple search cycles, the prototype device executes a preparation process to prepare a sample using the preparation conditions determined in the decision process, and an evaluation process to evaluate the sample prepared in the preparation process and output the evaluation results. In the decision process of the first search cycle of the multiple search cycles, the information processing device determines the preparation conditions for the preparation process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle of the multiple search cycles. The information processing device changes the search conditions to be used in the third search cycle and subsequent search cycles following the first search cycle of the multiple search cycles.

Description

Technical Field

[0001] The present disclosure relates to a system, an information processing apparatus, an information processing method, a program, and a recording medium.

Background Art

[0002] In the development of materials and the like, a sample is prepared by selecting raw materials, determining the blending amount of the raw materials, and performing processes such as stirring or heating the materials. Also, in development, the prepared sample is evaluated to obtain an evaluation value of the sample. The preparation conditions of the sample are explored so that this evaluation value satisfies the target. In such development, data-driven development that focuses on improving the performance of materials and development efficiency is known. Data-driven development derives the preparation conditions under which a desired evaluation value can be obtained from a data set including the preparation conditions of the sample and the evaluation value. Specifically, the preparation conditions of the sample are determined, the sample is prepared under those preparation conditions, the prepared sample is evaluated to obtain an evaluation value. Then, by analyzing the data set including the preparation conditions and evaluation results of the samples obtained so far, the preparation conditions of the next sample are determined. This series of operations is called a search cycle, and the search cycle is repeatedly executed to derive the preparation conditions under which a desired evaluation value can be obtained.

[0003] In determining the preparation conditions of the sample, data analysis techniques such as Bayesian optimization are often used, and the preparation and evaluation of the sample are often performed manually.

[0004] Patent Document 1 discloses automating a series of search cycles by automatically performing the preparation and evaluation of a sample using a robot or the like.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The user initially sets the search conditions for operating a series of search cycles, such as the number of iterations, the search range for sample preparation conditions, or the target evaluation value. However, if the search conditions are not set appropriately, the improvement in the sample's evaluation value per search cycle may be small, resulting in low search efficiency. [Means for solving the problem]

[0007] This disclosure provides a technology that is advantageous for conducting searches more efficiently.

[0008] A first aspect of this disclosure is a system comprising a prototype device and an information processing device, configured to perform multiple search cycles, wherein the information processing device is configured to receive input of search conditions and to perform a determination process to determine the conditions for preparing a sample in each of the multiple search cycles, the prototype device is configured to perform a preparation process to prepare the sample using the preparation conditions determined in the determination process, and an evaluation process to evaluate the sample prepared in the preparation process and output an evaluation result, the information processing device is configured to determine the preparation conditions in the preparation process of the first search cycle in the determination process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier in the multiple search cycles prior to the first search cycle, and the information processing device is configured to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle in the multiple search cycles.

[0009] A second aspect of this disclosure is an information processing device configured to perform multiple search cycles together with a prototype device, wherein the device is configured to accept input of search conditions, to perform a decision process to determine production conditions for producing a sample in each of the multiple search cycles, to cause the prototype device to perform a production process to produce the sample using the production conditions, and an evaluation process to evaluate the produced sample and output an evaluation result in each of the multiple search cycles, the device is configured to determine the production conditions in the production process of the first search cycle in the decision process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier in the multiple search cycles prior to the first search cycle, and to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle in the multiple search cycles.

[0010] A third aspect of this disclosure is an information processing method for an information processing device configured to perform multiple search cycles together with a prototype device, characterized in that: it receives input of search conditions; in each of the multiple search cycles, it performs a decision process to determine manufacturing conditions for manufacturing a sample; in each of the multiple search cycles, it causes the prototype device to perform a manufacturing process to manufacture the sample using the manufacturing conditions, and an evaluation process to evaluate the manufactured sample and output an evaluation result; in the decision process of the first search cycle of the multiple search cycles, it determines the manufacturing conditions in the manufacturing process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle of the multiple search cycles; and it changes the search conditions to be used in the third search cycle and subsequent search cycles following the first search cycle of the multiple search cycles. [Brief explanation of the drawing]

[0011] [Figure 1] This is a block diagram of the manufacturing evaluation system according to the first embodiment. [Figure 2] (a) is an explanatory diagram of some of the search conditions according to the first embodiment. (b) is an explanatory diagram of some of the search conditions according to the first embodiment. (c) is an explanatory diagram of some of the search conditions according to the first embodiment. [Figure 3] This is a flowchart of the processing of the manufacturing evaluation system according to the first embodiment. [Figure 4] This is an explanatory diagram illustrating an example of the search situation according to the first embodiment. [Figure 5] This is an explanatory diagram illustrating an example of the search situation according to the first embodiment. [Figure 6] This is an explanatory diagram illustrating an example of the search situation according to the first embodiment. [Figure 7] This is an explanatory diagram illustrating an example of the search situation according to the first embodiment. [Figure 8] This is a flowchart of the processing of the manufacturing and evaluation system according to the second embodiment. [Figure 9] This is an explanatory diagram of the search situation related to Example 1. [Figure 10] (a) is a diagram showing the search conditions for Example 2. (b) is a diagram showing the evaluation value parameters for Example 2. [Figure 11] (a) and (b) are figures showing the estimation accuracy of the model according to Example 2. [Figure 12] This figure shows the relationship between processing time and the number of chips in Example 2. [Modes for carrying out the invention]

[0012] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the drawings. The embodiments shown below are illustrative, and for example, the details of the configuration can be modified as appropriate by those skilled in the art without departing from the spirit of the present technology. In the drawings referenced in the following description of embodiments, elements indicated by the same reference numerals shall have the same function unless otherwise specified.

[0013] [First Embodiment] FIG. 1 is a block diagram of a production evaluation system 100 according to the first embodiment. The production evaluation system 100 is an example of a system. The production evaluation system 100 includes an information processing device 109, an automatic prototyping device 103, an input device 101, and a display device 104. The automatic prototyping device 103 is an example of a prototyping device.

[0014] In the development of materials and the like, the production evaluation system 100 automatically produces a sample 124 based on production conditions 128, automatically evaluates the sample 124, and repeats a search cycle of determining the next production conditions 128 from a data set 127 including production conditions 126 and an evaluation value 125 which is the evaluation result. The production evaluation system 100 searches for production conditions 128 under which a desired sample 124 can be obtained by executing a plurality of search cycles. However, in a conventional production evaluation system, the amount of improvement in the evaluation value of a sample per search cycle may be small even when the search cycle is repeated, the evaluation value of the sample may not reach a value desired by the user, or the efficiency of the search may be low.

[0015] The production conditions 128(126) are conditions necessary for producing the sample 124, and include raw material conditions such as the type, characteristics, and amount of raw materials, and processing conditions such as stirring, heating, cooling, and curing of the raw materials. That is, the production conditions 128(126) may include information on the raw materials used to produce the sample 124, information on the blending amount of the raw materials used to produce the sample 124, and information on the processing temperature (for example, curing temperature) in the production process of the sample 124. If there are other conditions necessary for producing the sample 124, they may also be included in the production conditions 128(126). Note that the production conditions 128 are command values (target values) commanded to the automatic prototyping device 103. Also, the production conditions 126 are measured values obtained from the automatic prototyping device 103.

[0016] The evaluation value 125 is a value indicating the performance of production conditions 126. For example, it may include the value obtained by evaluating the physical properties of sample 124, the resources required for the production of sample 124, the cost required for the production of sample 124, the time required for the production of sample 124, and the environmental load required for the production of sample 124. The physical properties of sample 124 are the performance of the produced sample 124, such as adhesive strength, breaking energy, elastic modulus, viscosity, optical transmittance, or electrical resistance, and these physical properties may also be included in the evaluation value 125. Also, the obtained evaluation value 125 may be only one type of evaluation value or multiple types of evaluation values. Thus, the evaluation value 125 may use not only the above-described indicators but also other indicators, and at least one of the above-described indicators and other indicators may be used. The data corresponding these production conditions 126 and the evaluation value 125 is accumulated in the dataset 127 and analyzed to determine the next production conditions 128. For the determination of the next production conditions 128, it is preferable to use Bayesian optimization, but other optimization methods such as the response surface method, regression analysis, or genetic algorithm may be used.

[0017] The automatic prototyping device 103 is a prototyping device that executes the production and evaluation of samples in a plurality of search cycles, and includes a production unit 105 and an evaluation unit 106. The information processing device 109 is a device that receives the input of the search conditions 121, executes a determination process for determining the production conditions 128, and causes the automatic prototyping device 103 to execute the production and evaluation of samples. The production unit 105 of the automatic prototyping device 103 executes a production process for producing the sample 124 under the given production conditions 128. The evaluation unit 106 of the automatic prototyping device 103 executes an evaluation process for evaluating the sample 124 produced by the production unit 105. Each of the plurality of search cycles includes a determination process, a production process, and an evaluation process.

[0018] The information processing device 109 is configured with one or more computers. The following description will use the case where the information processing device 109 has one computer, i.e., one processor, as an example. The information processing device 109 includes a CPU, which is an example of a processor; RAM, which is a temporary storage device; ROM and SSD, which are non-temporary storage devices (recording media); and I / O, which is an interface. The non-temporary storage device is an example of a recording media readable by the computer, and stores programs that cause the CPU to execute information processing methods such as control processing and arithmetic processing that control various parts of the entire device.

[0019] In addition to the above-described configuration, the information processing device 109 having a processor may also be configured with, for example, a PLD (Programmable Logic Device) such as an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), a general-purpose or dedicated computer with a program installed, or a combination of all or part of these.

[0020] The information processing device 109 has the functions of a control unit 102, a recording unit 107, and a manufacturing condition determination unit 108. The control unit 102 creates a control command 122 based on the manufacturing conditions 128 and controls the automatic prototyping device 103 according to the control command 122.

[0021] The manufacturing unit 105 of the automated prototyping apparatus 103 is a device that manufactures a sample 124 based on manufacturing conditions 128 corresponding to the control command 122. In the manufacture of the sample 124, processes such as weighing and dispensing raw materials, stirring the materials, degassing, heating and cooling, pressurizing and unloading, and hardening are performed. Note that the manufacture of the sample 124 is not limited to the processes listed here and may include any other manufacturing processes (manufacturing treatments) necessary for the manufacture of the sample 124. These manufacturing processes (manufacturing treatments) are specifically carried out using robots or automated machines.

[0022] The fabrication unit 105 fabricates the sample 124 based on the commanded fabrication conditions 128. However, depending on the control accuracy of the fabrication unit 105, there may be an error between the commanded fabrication conditions 128 and the measured fabrication conditions 126. For this reason, the fabrication unit 105 may output the measured fabrication conditions 126 as result 129 to the information processing device 109. If the error between fabrication conditions 126 and fabrication conditions 128 is smaller than a predetermined value, fabrication conditions 128 may be treated as result 129.

[0023] The evaluation unit 106 of the automated prototyping device 103 is a device that evaluates (measures) the sample 124. The evaluation unit 106 evaluates the sample 124 and outputs an evaluation value 125, which is an example of an evaluation result (measurement result). The evaluation value 125 is a value that indicates the performance of the manufacturing conditions 126, and may include, for example, the resources required to manufacture the sample 124, the cost required to manufacture the sample 124, the time required to manufacture the sample 124, and the environmental burden required to manufacture the sample 124. The evaluation value 125 may also include physical properties of the sample 124, such as adhesive strength, fracture energy, elastic modulus, viscosity, optical transmittance, or electrical resistance. The evaluation value 125 may also be a value obtained by quantifying an image such as a microscope image, as long as it is a target for development. Thus, the evaluation value 125 may use not only the indicators described above, but also other indicators, and at least one of the indicators described above and other indicators may be used. The evaluation value 125 is output to the information processing device 109 as a result 129.

[0024] The evaluation value 125 is preferably a quantitatively evaluated value, but it may also be a value obtained through sensory evaluation. Furthermore, pre-processing necessary for evaluation may also be included in the evaluation unit 106. These evaluation processes can be performed, for example, using a robot or automated machine.

[0025] The recording unit 107 acquires the result 129 and records it in the dataset 127, associating it with the manufacturing conditions 126 and the evaluation value 125. If the error between manufacturing conditions 126 and manufacturing conditions 128 is smaller than a predetermined value, the dataset 127 may include manufacturing conditions 128 instead of manufacturing conditions 126. That is, the manufacturing conditions included in the dataset 127 include measured values ​​obtained by the manufacturing process of the manufacturing unit 105, or command values ​​obtained by the determination process of the manufacturing condition determination unit 108. The dataset 127 is stored in a storage device such as an SSD. The recording unit 107 may be included in a server, for example, or implemented on a server on the cloud that processes data remotely.

[0026] Dataset 127 stores not only the values ​​from a single exploration cycle, but also the manufacturing conditions 126 (or 128) and evaluation values ​​125 obtained from previous exploration cycles. Furthermore, if it is the first exploration cycle and no data has been stored, Dataset 127 may use manufacturing conditions 126 (or 128) and evaluation values ​​125 obtained from previous explorations (developments) unrelated to the current series of exploration cycles as initial data. In this case, the initial data may also be retained in Dataset 127 when performing a second exploration cycle.

[0027] When operating the manufacturing condition determination unit 108, it is necessary to define its operation, and the conditions used to define the operation are called search conditions 121. The search conditions 121 include search settings related to the manufacturing conditions 128, such as the amount of raw materials and the range for searching processing conditions, search settings related to the tolerance values ​​of evaluation values, and search settings related to the overall operation, such as the termination conditions of the search cycle and the search algorithm.

[0028] The search conditions 121 are initially determined by the user. The information processing device 109 then repeatedly performs the search cycle, but modifies the search conditions 121 as appropriate during the execution of multiple search cycles.

[0029] The input device 101 is a device that the user operates to input the search conditions 121 into the information processing device 109. The input device 101 may include a keyboard and a mouse, but it can be any device that can perform input. For example, the input device 101 may include a touch panel or a touchpad.

[0030] The manufacturing condition determination unit 108 performs a decision process to determine the manufacturing conditions 128 using the search conditions 121 and the dataset 127. The search algorithm used to determine the manufacturing conditions 128 is preferably Bayesian optimization, but other optimization methods such as response surface analysis, regression analysis, or genetic algorithms may also be used. The manufacturing condition determination unit 108 may be included in, for example, a server, or it may be implemented on a server on the cloud that processes data remotely.

[0031] The display device 104 displays images such as the search status 123 output from the automated prototyping device 103, and is specifically a display. The search status 123 includes data analysis results based on the dataset 127, such as the update status of the evaluation value 125.

[0032] The search conditions 121 will now be explained in detail. Figures 2(a), 2(b), and 2(c) are explanatory diagrams of some of the search conditions 121 according to the first embodiment. The search conditions 121 only need to be defined in a way that is necessary to operate the automated prototyping device 103, and are not limited to those described below. The search conditions 121 include search settings 1211 related to the manufacturing conditions 128, search settings 1212 related to the evaluation of the sample 124, and search settings 1213 related to the overall operation. Search setting 1211 is an example of a first search setting, search setting 1212 is an example of a second search setting, and search setting 1213 is an example of a third search setting.

[0033] The search settings 1211 related to the manufacturing conditions define the search range and other elements corresponding to the search parameters. That is, the search settings 1211 include the following candidate manufacturing conditions 128.

[0034] The search parameters are the parameters (items) of the manufacturing conditions to be explored. For example, these include the main raw materials used to prepare sample 124, the amount of main raw materials used (e.g., the ratio of main raw materials), and the processing temperature (e.g., the curing temperature). The ratio of main raw materials is, for example, the ratio of the amount of curing agent to the amount of main raw materials.

[0035] The search range is the range (candidate manufacturing conditions) in which the manufacturing conditions corresponding to the search parameters can be changed. For example, if the search parameters are things that can be expressed numerically, such as the amount of raw materials used (e.g., ratio) or the processing temperature (e.g., curing temperature), then the search range is set to one or more numerical values ​​or a continuous range of numerical values. Furthermore, a search range may be set not only for each of multiple search parameters, but also for combinations of multiple search parameters. For example, the search range may be set to "the sum of the amounts of raw material A001 and raw material A002 is 100". Also, the search range may be discrete values ​​rather than continuous values, and in the case of discrete values, in addition to the upper and lower limits of the range, intervals or divisions of the discrete values ​​may be set. Furthermore, the search range may also be information that cannot be expressed numerically, such as raw materials A001, A002, A003. In this case, the manufacturing condition 128 is determined from among raw materials A001, A002, and A003.

[0036] In this way, the desired evaluation value is searched for by changing the manufacturing conditions within the search range. Manufacturing conditions that are not changed are defined as fixed values ​​in the search setting 1211, and these fixed values ​​are used as the manufacturing conditions 128 for sample 124.

[0037] The search settings 1212 related to the evaluation of sample 124 define the evaluation value parameters to be searched, the target direction of the evaluation value, and the tolerance value of the evaluation value.

[0038] The evaluation parameter defines the desired evaluation parameter (item) in the development of materials, etc., and may be singular or plural. In other words, it is sufficient to explore at least one type of evaluation value. For example, the evaluation parameter may include the resources required to prepare sample 124, the cost required to prepare sample 124, the time required to prepare sample 124, and the environmental impact required to prepare sample 124. It may also include physical properties of sample 124, such as adhesive strength, fracture energy, elastic modulus, viscosity, optical transmittance, or electrical resistance.

[0039] The target direction for the evaluation value defines the direction in which to adjust the evaluation value, such as increasing (maximizing), decreasing (minimizing), or approaching a certain value.

[0040] When there are multiple evaluation parameter values, that is, when there are multiple types of evaluation values ​​to target, the evaluation values ​​may trade off with each other, and the optimal manufacturing conditions may not be uniquely determined. To address this trade-off, tolerance values ​​are defined for each evaluation value. The tolerance value of an evaluation value is, for example, the lower limit or upper limit of the tolerance range for the evaluation value. If the tolerance value of an evaluation value is the lower limit of the tolerance range for the evaluation value, it is a value that is lower than the target value but is still acceptable. In the exploration cycle, conditions that are estimated to result in an evaluation value lower than the tolerance value will not be determined as the next manufacturing conditions, and prototypes of sample 124 that result in an evaluation value lower than the tolerance value will not be produced.

[0041] The overall search settings 1213 define at least one of the following: the search cycle termination condition, the search algorithm, and the number of samples 124 produced at one time in one search cycle. The search cycle termination condition may be that the process ends after a specified number of search cycles have been repeated, or it may be that the process ends when the evaluation value 125 reaches a specified target value.

[0042] A search algorithm is an algorithm that determines the next generation conditions, and defines the optimization method and the internal algorithms used in that optimization method. Optimization methods include Bayesian optimization, response surface analysis, regression analysis, or genetic algorithms. When Bayesian optimization is used, internal algorithms such as kernel functions, acquisition functions, and internal constraints are defined.

[0043] The number of samples prepared at one time is the number of samples 124 prepared in one exploration cycle. Preparing and evaluating multiple samples at once may be more efficient than doing so one at a time.

[0044] Figure 3 is a flowchart of the processing of the manufacturing evaluation system 100 according to the first embodiment. In step S1, the recording unit 107 determines whether or not it has received input of initial data, such as data acquired in past developments that can be used in the current development.

[0045] If step S1 is YES, that is, if initial data has been received, in step S2, the recording unit 107 records the initial data in the dataset 127. If step S1 is NO, that is, if initial data has not been received, the recording unit 107 leaves the dataset 127 empty and skips the process in step S2.

[0046] Next, in step S3, the control unit 102 accepts the input of the search condition 121. Note that the order of steps S1-S2 and step S3 may be reversed.

[0047] The series of processes from step S4 to step S11 constitutes a search cycle, and the manufacturing evaluation system 100 executes multiple search cycles.

[0048] In step S4, the manufacturing condition determination unit 108 determines the manufacturing condition 128 based on the search condition 121 and the data set 127. The manufacturing condition 128 is the command value, i.e., the target value.

[0049] Specifically, in step S4, the manufacturing condition determination unit 108 determines the type of main raw material, the amount of main raw material (e.g., ratio), and the processing temperature (e.g., curing temperature) from among the candidates included in the search setting 1211 to determine the following manufacturing conditions 128. As mentioned above, the candidates may include information on multiple types of raw materials, information on the range of raw material amounts, and information on the range of processing temperatures. At that time, the manufacturing condition determination unit 108 determines the manufacturing conditions 128 based on the search algorithm in order to optimize the evaluation value 125.

[0050] If the dataset 127 is empty, the preparation condition determination unit 108 may randomly determine the preparation conditions 128, determine the preparation conditions 128 based on an orthogonal array of experimental design, or accept the user's specification of the preparation conditions 128.

[0051] Next, in step S5, the control unit 102 creates a control command 122 based on the search conditions 121 and the manufacturing conditions 128, and outputs the control command 122 to the automatic prototyping device 103. The manufacturing unit 105 of the automatic prototyping device 103, having received the control command 122, manufactures the sample 124 according to the control command 122. That is, the manufacturing unit 105 executes a manufacturing process to manufacture the sample 124 using the manufacturing conditions 128. At this time, the manufacturing unit 105 outputs the measured manufacturing conditions 126.

[0052] Next, in step S6, the evaluation unit 106 performs an evaluation process to evaluate the sample 124 prepared by the fabrication unit 105. At this time, the evaluation unit 106 outputs an evaluation value 125, which is the evaluation result.

[0053] Next, in step S7, the recording unit 107 records the result (data) 129, including the evaluation value 125 and the manufacturing conditions 126, into the dataset 127. Next, in step S8, the control unit 102 displays the image corresponding to the search status 123 on the display device 104. Note that the order of the processes in step S7 and step S8 may be reversed.

[0054] Incidentally, in the early stages of the search cycle, the user may not know the appropriate search conditions 121. As the search cycle progresses, the user may come to know new appropriate search conditions 121 through the search situation 123. For example, when the user sees an image corresponding to the search situation 123, the user may decide that it is necessary to change the search conditions 121. In this embodiment, the control unit 102 is in a state where it can accept changes to the search conditions 121 by the user during the execution of multiple search cycles. That is, the user inputs the changed search conditions 121 to the information processing device 109 using the input device 101 at any timing in steps S4 to S8.

[0055] In this embodiment, the exploration status 123 is the result of data analysis using the dataset 127, and is at least one of the following: the trend of the evaluation value 125, the estimation accuracy of the model, the relationship between the fabrication conditions 126 and the evaluation value 125, and the distribution of the evaluation value 125.

[0056] Figure 4 is an explanatory diagram of an example of the search status 123 according to the first embodiment. The search status 123 in Figure 4 shows the progression of the evaluation value 125. The control unit 102 causes the display device 104 to display an image showing the progression of the evaluation value 125 as shown in Figure 4. In the image displayed on the display device 104, the evaluation value 125 against the number of search cycles is plotted as points. Each plotted point is the evaluation value 125 evaluated in each search cycle.

[0057] The solid line in Figure 4 represents the best value of the evaluation score 125 obtained at the time of the executed search cycle. This allows the user to understand the areas where the evaluation score 125 has been updated and areas where it has stagnated.

[0058] If the search remains stagnant, it is possible that at least one of the search conditions 121—the search algorithm, search parameters, search range, and the acceptable value for the evaluation value 125—is inappropriate. By changing at least one of the search conditions—the search algorithm, search parameters, search range, and the acceptable value for the evaluation value 125—the user can expect an improvement (update) in the evaluation value 125 in subsequent search cycles.

[0059] Similarly, if the search remains stagnant, the user can immediately terminate the search by changing the search cycle termination condition in search condition 121 to the current number of search cycles. This allows the user to immediately move on to another consideration.

[0060] If there are multiple types of evaluation values, the control unit 102 may create a separate graph for each of the multiple types of evaluation values, as shown in Figure 4, and present it to the user. Alternatively, it may create a single index that combines the quality of multiple types of physical properties, such as hypervolume, and present it to the user.

[0061] Figure 5 is an explanatory diagram of an example of the search status 123 according to the first embodiment. The search status 123 in Figure 5 shows the estimation accuracy of the model. The control unit 102 causes the display device 104 to display an image showing the estimation accuracy of the model as shown in Figure 5. In the image displayed on the display device 104, the predicted values ​​of the evaluation values ​​are plotted as points against the measured values ​​of the evaluation values.

[0062] The measured value of evaluation value 125 is a value included in dataset 127. When determining the next production condition 128, the production condition determination unit 108 uses dataset 127 to create a model that estimates evaluation value 125 from past production conditions 126.

[0063] Next, the manufacturing condition determination unit 108 derives manufacturing conditions 128 that yield a good evaluation value based on this model, and outputs the next manufacturing conditions 128. Therefore, if the accuracy of the model is low, the accuracy of the next manufacturing conditions 128 will also be low, and a good evaluation value 125 cannot be obtained. If the termination condition is that the evaluation value is above a threshold, the number of search cycles will increase.

[0064] If the model accuracy is low, it is possible that at least the search parameters among the search parameters and search range are inappropriate. By changing at least the search parameters among the search conditions 121, the user can expect to improve the model accuracy, reduce the difference between the predicted and actual values ​​of the evaluation, and expect improvements (updates) in the evaluation values ​​in subsequent search cycles.

[0065] If there are multiple types of evaluation values, the control unit 102 may create a separate graph for each of the multiple types of evaluation values, as shown in Figure 5, and present it to the user. Furthermore, the estimation accuracy may be represented as a change over the number of search cycles. In this case, the estimation accuracy for an arbitrary number of search cycles may be displayed, the estimation accuracy for different search cycle numbers at different time points may be displayed, or the estimation accuracy for different search cycle numbers may be displayed side by side.

[0066] Figure 6 is an explanatory diagram of an example of a search situation 123 according to the first embodiment. The search situation 123 in Figure 6 shows the relationship between two manufacturing conditions and evaluation values. The control unit 102 causes the display device 104 to display an image showing the relationship shown in Figure 6. In the image displayed on the display device 104, the evaluation values ​​125 for the two manufacturing conditions are represented by contours. Each point represents data included in the dataset 127. The manufacturing condition determination unit 108 estimates evaluation values ​​from various manufacturing conditions using a model, and the control unit 102 causes the display device 104 to display the results as a contour map.

[0067] If the evaluation value of 125 stagnates and the unexplored area shown in Figure 6 is large, the user can change the search algorithm for search condition 121 to an algorithm that prioritizes the unexplored area. Furthermore, once the user has identified promising areas with good evaluation values ​​in Figure 6, they can change the search algorithm for search condition 121 to an algorithm that prioritizes the promising areas, thereby limiting the search range to those areas. These changes can be expected to improve (update) the evaluation value.

[0068] Furthermore, if the user wishes to change the search range, they may use an input device 101, such as a mouse, to draw a box around the display screen shown in Figure 6, thereby changing the search range to the boxed-up area. Also, if there are three or more production conditions, the recording unit 107 may display graph images as shown in Figure 6 on the display device 104 for the two production conditions selected by the user, or it may display graph images representing multiple dimensions, such as a parallel coordinate plot, on the display device 104. If there are multiple types of evaluation values, the control unit 102 may create a separate graph as shown in Figure 6 for each of the multiple types of evaluation values ​​and present it to the user.

[0069] Figure 7 is an explanatory diagram of an example of the search status 123 according to the first embodiment. The search status 123 in Figure 7 shows the distribution of two types of evaluation values ​​125. The control unit 102 causes the display device 104 to display an image showing a scatter plot of the two types of evaluation values ​​included in the dataset 127 shown in Figure 7.

[0070] When there are two types of evaluation values, the user can understand the relationship between the two types of evaluation values ​​by viewing a scatter plot. If the user sees a proportional relationship between the two types of evaluation values ​​by looking at the image shown in Figure 7, the user can remove the evaluation value parameter with the lower priority from the evaluation value parameters of search condition 121. This allows for the expectation of improvement (updating) of the evaluation values ​​in subsequent search cycles.

[0071] Furthermore, if the evaluation value is stagnating and is sufficiently close to the target value, it is preferable for the user to change the acceptable value of the evaluation value to bring it closer to the target value. This will allow for an improvement (update) in the evaluation value.

[0072] Furthermore, if the user changes the tolerance value of the evaluation value, the tolerance value of the evaluation value may be changed to the position entered using an input device 101 such as a mouse on the display screen shown in Figure 7. Also, if there are three or more types of evaluation values, the control unit 102 may display graph images on the display device 104, such as those shown in Figure 7, for the two types of evaluation values ​​selected by the user. In addition, graph images representing multiple dimensions, such as a three-dimensional scatter plot or a parallel coordinate plot, may be displayed on the display device 104.

[0073] Furthermore, the images corresponding to the search status 123 based on dataset 127 are not limited to these; any image that clearly shows the search status 123 to the user is acceptable.

[0074] Furthermore, it is preferable that the control unit 102 displays an image showing the search conditions 121 on the display device 104. This allows the user to understand the already entered search conditions 121 and to confirm which items of the search conditions 121 need to be changed.

[0075] Furthermore, when the control unit 102 displays an image corresponding to the search status 123 on the display device 104, it may also display an image on the display device 104 indicating a candidate for changing the search condition 121. The control unit 102 may then change the search condition 121 to the candidate selected by the user.

[0076] In step S9, the control unit 102 determines whether or not it has accepted the change to the search condition 121. If step S9 is YES, that is, if the change to the search condition 121 has been accepted, in step S10, the control unit 102 changes the search condition 121 by overwriting the set search condition 121. If step S9 is NO, that is, if the change to the search condition 121 has not been accepted, the control unit 102 skips the process in step S10.

[0077] Next, in step S11, the control unit 102 determines whether the termination condition is met. If step S11 is NO, i.e., the termination condition is not met, the control unit 102 returns to the process in step S4 and executes the next search cycle. If step S11 is YES, i.e., the termination condition is met, the control unit 102 terminates the process.

[0078] As described above, the manufacturing evaluation system 100 executes multiple search cycles. During the execution of these multiple search cycles, the currently executing search cycle is designated as the first search cycle, the search cycle executed before the first search cycle (immediately before in this embodiment) is designated as the second search cycle, and the search cycle executed after the first search cycle is designated as the third search cycle.

[0079] In step S7 of the second search cycle, the recording unit 107 records the evaluation value 125 obtained in the second search cycle and the data (result 129) including the manufacturing conditions 126 obtained in the second search cycle into the dataset 127. As a result, the dataset 127 will include data (result 129) acquired from the automated prototyping device 103 prior to the second search cycle. Note that if manufacturing condition 128 is used instead of manufacturing condition 126 in the data recorded in the dataset 127, the data acquired from the automated prototyping device 103 will only be the evaluation value 125.

[0080] In the first embodiment, in step S4 of the first search cycle following the second search cycle, the manufacturing condition determination unit 108 of the information processing device 109 determines the manufacturing condition 128 for the manufacturing process of the first search cycle using the search condition 121 and a dataset 127 that includes data (result 129) from before the second search cycle. The dataset 127 includes the manufacturing condition 126 (or 128) from the manufacturing process before the second search cycle and the evaluation value 125 which is the evaluation result from the evaluation process.

[0081] Next, in step S5 of the first search cycle, the control unit 102 creates a control command 122 based on the search conditions 121 and the manufacturing conditions 128, and outputs the control command 122 to the automatic prototyping device 103, so that the manufacturing unit 105 of the automatic prototyping device 103 executes a manufacturing process to produce the sample 124.

[0082] Next, in step S6 of the first search cycle, the evaluation unit 106 performs an evaluation process to evaluate the sample 124 prepared by the fabrication unit 105, and outputs an evaluation value 125, which is the evaluation result.

[0083] Then, if step S9 of the first search cycle is YES, in step S10, the control unit 102 changes the search conditions 121 to be used in the next search cycles, starting from the third search cycle.

[0084] Thus, the control unit 102 of the information processing device 109 is configured to change the search conditions 121 during the execution of multiple search cycles. In the first embodiment, the control unit 102 of the information processing device 109 is configured to receive instructions to change the search conditions 121. When the control unit 102 of the information processing device 109 receives such instructions, it changes the search conditions 121 in accordance with those instructions.

[0085] As described above, the search condition 121 can be changed by the user during the execution of multiple search cycles, allowing the user to obtain production conditions close to their desired evaluation value in fewer search cycles.

[0086] Furthermore, even if the appropriate search conditions 121 are not initially set, the search conditions 121 can be changed by the user according to the search status 123, allowing for the discovery of better production conditions in fewer search cycles and enabling more efficient exploration (development).

[0087] [Second Embodiment] A second embodiment will now be described. Hereinafter, elements denoted by the same reference numerals as those in the first embodiment will have substantially the same configuration and function as those described in the first embodiment unless otherwise specified. The differences from the first embodiment will be primarily described. The configuration of the manufacturing evaluation system in the second embodiment is the same as that of the manufacturing evaluation system 100 in the first embodiment, and therefore its description will be omitted.

[0088] In the first embodiment described above, the user changed the search conditions 121, but in the second embodiment, the information processing device 109 automatically changes the search conditions 121.

[0089] Figure 8 is a flowchart of the processing of the manufacturing evaluation system according to the second embodiment. The difference between the flowchart in Figure 8 and the flowchart in Figure 3 of the first embodiment is that steps S8 and S9 are omitted, and step S12 is added between steps S7 and S10. Note that the process of displaying the image of the search status 123 on the display device 104 in step S8 does not have to be omitted.

[0090] In step S12, the control unit 102 automatically determines the changes to the search conditions 121 based on the search status 123 derived from the dataset 127. Note that the changes to the search conditions 121 include cases where there are no changes. Then, in step S10, the control unit 102 changes the search conditions 121 by overwriting the set search conditions 121. In this way, the control unit 102 automatically changes the search conditions 121 based on the search status 123. For example, the control unit 102 may change the search conditions 121 to predetermined content when the value based on the search status 123 exceeds a predetermined threshold.

[0091] To explain with a specific example, if the evaluation value changes without changing over a predetermined search cycle, and the estimation accuracy expressed by the coefficient of determination is less than a predetermined value, the control unit 102 may calculate the importance of the search parameters to the evaluation value using methods such as Permutation Importance, which derives the importance of the search parameters by analyzing the model, and remove the search parameters with low importance from the search conditions 121. This improves the estimation accuracy and can be expected to improve (update) the evaluation value.

[0092] Furthermore, if the evaluation value does not change over a predetermined search cycle and data exists between the acceptable value and the target value of the evaluation value, the control unit 102 may change the acceptable value of the evaluation value to the same value as that data. If multiple data exist, the acceptable value of the evaluation value may be set to the same value as any of the data, but it is preferable to set it to the same value as the median data. This will propose manufacturing conditions that are expected to be close to the target value, and an improvement (update) in the evaluation value can be expected. Note that the method of changing the search conditions 121 is not limited to those described above, and any method that can automatically change the search settings based on the search status is acceptable.

[0093] [Example 1] Using the manufacturing evaluation system 100 of the first embodiment, we explored manufacturing conditions for a sample 124 that balances both adhesive strength and fracture energy of the adhesive.

[0094] Sample 124 is a cured adhesive product obtained by heating an adhesive mixture made from the main raw materials and a curing material. The manufacturing unit 105 is a combination of a robot hand and a rotary mixer. In the manufacturing process, the manufacturing unit 105 uses the robot hand to weigh the main raw materials and the curing material, and then uses the rotary mixer to stir them.

[0095] The fabrication unit 105, in order to prepare a sample for measuring adhesive strength, uses a robotic hand to operate a syringe and drops the stirred adhesive solution onto an aluminum plate. After inserting shims to ensure a uniform thickness, the adhesive is sandwiched between an aluminum plate and a stainless steel plate.

[0096] Furthermore, in order to prepare a sample for measuring fracture energy, the fabrication unit 105 drops stirred liquid adhesive onto a glass plate coated with a release film, inserts shims to ensure a certain thickness, and then sandwiches the adhesive between two glass plates. The fabrication unit 105 places this bonded body into an oven using a robotic hand and heats it for 30 minutes. The fabrication unit 105 then uses the robotic hand to detach the sample for measuring fracture energy from the glass plate and cuts the detached sample into a dumbbell shape.

[0097] The evaluation unit 106 combines a robotic hand with a tensile indentation testing machine. The evaluation unit 106 uses the robotic hand to set the sample prepared in the manufacturing unit 105 into the tensile indentation testing machine. The adhesive strength is measured by the indentation test, and the fracture energy is measured by the tensile test. The measured values ​​are obtained from the tensile indentation testing machine.

[0098] In Example 1, since there were no prior studies, the initial dataset 127 was left empty. The initial search conditions 121 were as follows: the search cycle termination condition was set to 100 iterations. The search algorithm was multi-objective Bayesian optimization using Expected Hypervolume Improvement. The number of samples prepared at one time was one each for measuring adhesive strength and fracture energy.

[0099] Furthermore, if dataset 127 is empty, the creation condition 128 is randomly proposed within the search range.

[0100] The experimental settings 1211 related to the manufacturing conditions were as shown in the example in Figure 2(a). The experimental parameters were the main raw material, the ratio of the amount of curing agent to the amount of the main raw material, and the curing temperature.

[0101] The main raw material was searched for within three candidate materials (A001, A002, and A003). The ratio of the amount of hardener to the amount of the main raw material was searched within a numerical range of 0.8 to 2.0. The curing temperature was searched within a numerical range of 40 to 120°C.

[0102] The search settings 1212 related to the evaluation value were as shown in the example in Figure 2(b). The evaluation value parameters were adhesive force and fracture energy. The target direction for the evaluation value was to maximize both adhesive force and fracture energy.

[0103] The allowable lower limit for the evaluation value of adhesive strength is 20 MPa, and the allowable lower limit for the evaluation value of breaking energy is 20 kJ / m 2 The target adhesive strength was set at 30 MPa or higher, and the target fracture energy was set at 120 kJ / m². 2 That concludes my explanation.

[0104] The exploration cycle was repeated multiple times using the fabrication and evaluation system 100. As the exploration status 123, an image like the one shown in Figure 9 was displayed on the display device 104. Figure 9 is an explanatory diagram of the exploration status 123 related to Example 1. The exploration status 123 in Figure 9 shows the hypervolume of the evaluation values ​​of two physical properties (adhesion strength and fracture energy) for each exploration cycle.

[0105] In the exploration status 123 shown in Figure 9, a stagnation was observed in updating the hypervolume. Therefore, in the 13th exploration cycle, the exploration condition 121 was changed. The change involved excluding raw material A002, which had been considered unpromising in the previous exploration cycles, from the exploration range. The exploration cycles from the 14th onward were then carried out, and an update in the hypervolume was observed in the 16th exploration cycle. The hypervolume continued to update, and the target value was achieved in the 20th exploration cycle. Therefore, the termination condition in the exploration condition 121 was changed to the number of prototypes (exploration cycles) at 20, and the exploration cycle was terminated.

[0106] As shown in Figure 2(c), although the initial number of prototypes (exploration cycles) was set to 100, by changing the exploration conditions 121 according to the exploration status 123 midway through, it was found that the evaluation values ​​of both adhesive strength and fracture energy properties reached the target values ​​in 20 exploration cycles, demonstrating that efficient development can be carried out.

[0107] [Example 2] While Example 1 described material development, Example 2 is also effective for various design and development processes other than material development. This Example 2 describes an example in which a glass dicing method was developed using the manufacturing evaluation system 100 of the first embodiment. In glass dicing, it is preferable to perform the processing in a short time, but if dicing is done too quickly, the amount of chipping, which is a defect, may increase and the yield may decrease. In Example 2, processing conditions that result in a short processing time and minimal chipping were explored.

[0108] In the fabrication unit 105, sample 124 was prepared as follows: Using a robotic hand, the flat glass plate before processing was moved to the dicing position, and then the dicing blade was rotated and applied to the flat glass plate, moving the glass plate to cut it. During this process, cutting fluid was used for dicing. After that, the glass was moved to a high-pressure washer using the robotic hand for cleaning and drying. The resulting cut glass sample 124 was then transported to the evaluation unit 106 using the robotic hand.

[0109] The evaluation unit 106 is a device that automatically counts chipping. The evaluation unit 106 acquires images of the cross-section using a microscope and counts the chipping. By performing this process while scanning, the chipping of all cross-sections was counted. At this time, chipping was determined by image processing, and defects of a certain size or larger were determined to be chips.

[0110] In Example 2, the evaluation value 125 represents the number of chips and the processing time taken for cutting.

[0111] In Example 2, since there were no prior studies, the initial dataset 127 was left empty. The initial search conditions 121 were as follows: the search cycle termination condition was set to 50 iterations. The search algorithm was multi-objective Bayesian optimization using Expected Hypervolume Improvement. The number of samples prepared at one time was limited to one.

[0112] Furthermore, if dataset 127 is empty, the creation condition 128 is randomly proposed within the search range.

[0113] The search settings 1211 related to the manufacturing conditions were as shown in the example in Figure 10(a). The search parameters were the blade rotation speed, blade feed rate, cutting fluid flow rate, and blade abrasive particle size.

[0114] The search range for blade rotation speed was set to 20,000 to 30,000 rpm. The search range for blade feed rate was set to 1 to 5 mm / s. The search range for cutting fluid flow rate was set to 0.5 to 1.5 liters / minute. The search range for blade abrasive grain size was set to three candidates (3 μm, 5 μm, and 7 μm).

[0115] The search settings 1212 related to the evaluation value were as shown in the example in Figure 10(b). The evaluation value parameters were machining time and the number of chips. The target direction for the evaluation value was to minimize both the machining time and the number of chips.

[0116] In Example 2, the target values ​​were set to three or fewer chips and a processing time of 80 seconds or less.

[0117] In Example 2, the tolerance values ​​for evaluation were changed midway through the cycle. Initially (1st to 20th exploration cycles), the tolerance values ​​were set loosely in order to understand the relationship between manufacturing conditions and evaluation values ​​over a wide range. Specifically, the upper limit of the tolerance value for evaluation values ​​relative to processing time was 140 seconds, and the upper limit of the tolerance value for evaluation values ​​relative to the number of chips was 100. After the change (21st to 30th exploration cycles), the tolerance values ​​were set more strictly in order to approach the desired target value. Specifically, the upper limit of the tolerance value for evaluation values ​​relative to processing time was 80 seconds, and the upper limit of the tolerance value for evaluation values ​​relative to the number of chips was 10.

[0118] The manufacturing evaluation system 100 was used to repeat the search cycle multiple times. As the search status 123, images as shown in Figures 11(a), 11(b), and 12 were displayed on the display device 104. Figures 11(a), 11(b), and 12 are explanatory diagrams of the search status 123 related to Example 2. Figures 11(a) and 11(b) show the estimation accuracy of the model. The search status 123 in Figure 11(a) shows the estimation accuracy of the model obtained from the data at the end of the 6th search cycle, and the search status 123 in Figure 11(b) shows the estimation accuracy of the model obtained from the data at the end of the 20th search cycle. The search status 123 in Figure 12 shows the processing time and the number of chips obtained so far.

[0119] When the number of exploration cycles is small, the model accuracy is low because there is also little data. As an example, Figure 11(a) shows the estimated accuracy of the model obtained from the data at the end of the 6th exploration cycle. As the model accuracy was low, the exploration cycle was continued without changing the exploration condition 121. Figure 11(b) shows the estimated accuracy of the model obtained from the data at the end of the 20th exploration cycle. As shown in Figure 11(b), the model accuracy improved as the amount of data increased, and it became possible to accurately understand the relationship between the manufacturing conditions and the evaluation value. On the other hand, looking at Figure 12, which shows the processing time and the number of chips at the end of the 20th exploration cycle, although a value better than the acceptable value was obtained, it did not reach the target value.

[0120] When the tolerance value is far from the solution's Pareto front, the entire broad Pareto front is searched, resulting in low efficiency. Since the model's accuracy was already sufficient, the tolerance value was adjusted to be closer to the target value range for efficiency, and the search cycle was continued. From the 21st search cycle onward, as shown in Figure 12, values ​​better than the modified tolerance value were obtained at a high rate, and the target value was reached in the 30th search cycle. In multi-objective Bayesian optimization using the Expected Hypervolume Improvement search algorithm, the fabrication conditions are determined with the expectation that the result will be better than the tolerance value. However, in actual fabrication and evaluation, the evaluation value may be lower, so the obtained data is not always better than the tolerance value. Since the target value was reached, the termination condition among the search conditions 121 was changed to 30 prototypes (search cycles), and the search cycle was terminated.

[0121] Initially, the number of prototypes (exploration cycles) was set to 50, but the exploration conditions 121 were changed midway through the process according to the exploration status 123. As a result, it was found that both the processing time and the number of chips reached the target values ​​in 30 exploration cycles, demonstrating that efficient development is possible.

[0122] [Other examples] This disclosure is also valid for designs and developments other than those described in Examples 1 and 2, as long as the design is difficult to achieve through simulation. For example, this includes shape design, layout design, or electronic circuit design to improve durability or manufacturing variability.

[0123] In shape design and layout design, numerical parameters such as dimensions are adjusted as manufacturing conditions 126. In the manufacturing unit 105, samples are automatically manufactured using a CNC (Computer Numerical Control) machine or a 3D printer. The evaluation unit 106 uses a device that automatically obtains evaluation values ​​that meet the development requirements.

[0124] In the design of electronic circuits, numerical parameters such as the constants of circuit elements and the dimensions of wiring are adjusted as fabrication conditions 126. In the fabrication unit 105, samples are automatically fabricated using the placement and soldering of circuit elements with a robotic hand. The evaluation unit 106 uses a device that automatically obtains evaluation values ​​that meet the development requirements.

[0125] In summary, this disclosure provides a technology that is advantageous for conducting searches more efficiently.

[0126] [Other variations] This disclosure is not limited to the embodiments described above, and many variations are possible within the technical concept of this disclosure. For example, at least two of the embodiments and examples described above may be combined. Furthermore, the effects described in the embodiments are merely a list of the most preferred effects arising from the embodiments of this disclosure, and the effects of the embodiments of this disclosure are not limited to those described in the embodiments.

[0127] (Other examples) This disclosure can also be implemented by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be implemented by a circuit (e.g., an ASIC) that implements one or more functions.

[0128] The above disclosure of embodiments includes the following sections.

[0129] (Section 1) A system comprising a prototype device and an information processing device, configured to perform multiple search cycles, The information processing device is configured to receive input of search conditions and to perform a decision process to determine the sample preparation conditions in each of the multiple search cycles. The prototype device is configured to perform, in each of the multiple search cycles, a manufacturing process in which the sample is manufactured using the manufacturing conditions determined in the determination process, and an evaluation process in which the sample manufactured in the manufacturing process is evaluated and an evaluation result is output. The information processing device is configured to determine the production conditions in the production process of the first search cycle in the determination process of the first search cycle, using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle, among the multiple search cycles. The information processing device is configured to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle among the multiple search cycles. A system characterized by the following features.

[0130] (Section 2) When the information processing device receives an instruction to change the search conditions, it changes the search conditions in accordance with the instruction. The system according to item 1, characterized in that

[0131] (Section 3) The information processing device automatically changes the search conditions based on the search status based on the dataset. The system according to item 1, characterized in that

[0132] (Section 4) The dataset includes the measured manufacturing conditions obtained by the manufacturing process, or the manufacturing conditions obtained by the determination process. The system according to any one of items 1 to 3, characterized in that

[0133] (Section 5) The search conditions include a first search setting that includes candidate manufacturing conditions, The aforementioned information processing device is In the aforementioned determination process, the manufacturing conditions for the manufacturing process are determined from among the candidates included in the first search setting. The system according to any one of items 1 to 4, characterized in that

[0134] (Section 6) The aforementioned preparation conditions include information on the raw materials used to prepare the sample. The first search setting includes information on multiple types of raw materials as candidates. The system according to item 5, characterized in that

[0135] (Section 7) The aforementioned preparation conditions include information on the proportions of the raw materials used to prepare the sample. The first search setting includes, as a candidate, information on the range of the blending amounts, The system according to item 5 or 6, characterized in that it is a system according to item 5 or 6.

[0136] (Section 8) The manufacturing conditions include information on the processing temperature in the manufacturing process. The first search setting includes, as a candidate, information on the range of the processing temperature, The system according to any one of claims 5 to 7, characterized in that

[0137] (Section 9) The search conditions include a second search setting relating to at least one physical property of the sample, The information processing device determines the manufacturing conditions based on a search algorithm to optimize the physical properties in the decision process. The system according to any one of claims 1 to 8, characterized in that

[0138] (Section 10) The search conditions include a third search setting that includes at least one of the following: the termination condition for the search cycle, the search algorithm used in the decision process, and the number of samples produced in one search cycle in the fabrication process of the prototype device. A system according to any one of items 1 to 9, characterized in that it is a system according to any one of items 1 to 9.

[0139] (Section 11) The aforementioned sample is a cured product of an adhesive. The system according to any one of claims 1 to 10, characterized in that

[0140] (Section 12) Equipped with an additional display device, The information processing device causes the display device to display an image corresponding to the search status based on the dataset. The system according to any one of items 1 to 11, characterized in that

[0141] (Section 13) Equipped with an additional display device, The information processing device causes the display device to display an image indicating the search conditions. The system according to any one of claims 1 to 12, characterized in that

[0142] (Section 14) Equipped with an additional display device, The information processing device causes the display device to display an image showing a candidate for changing the search condition. The system according to item 2, characterized in that

[0143] (Section 15) The prepared sample is evaluated using at least one of the following: the physical properties of the sample and the resources necessary for preparing the sample. The system according to any one of items 1 to 14, characterized in that

[0144] (Section 16) The aforementioned preparation conditions include information on processing conditions used when preparing the sample. The first search setting includes information on the range of the processing conditions as a candidate, The system according to item 5, characterized in that

[0145] (Section 17) The aforementioned search status is the progression of the evaluation results. The system according to item 12, characterized in that

[0146] (Section 18) The aforementioned search status is the accuracy of the decision process. The system according to item 12, characterized in that

[0147] (Section 19) The aforementioned search status is the relationship between the manufacturing conditions and the evaluation results. The system according to item 12, characterized in that

[0148] (Section 20) The aforementioned search status is a relationship between multiple evaluation results. The system according to item 12, characterized in that

[0149] (Section 21) An information processing device configured to perform multiple search cycles together with a prototype device, The system is configured to accept input of search conditions and, in each of the multiple search cycles, execute a decision process to determine the conditions for preparing the sample. In each of the aforementioned multiple exploration cycles, the prototype device is configured to perform a fabrication process to produce the sample under the aforementioned fabrication conditions, and an evaluation process to evaluate the produced sample and output the evaluation results. In the determination process of the first search cycle among the multiple search cycles, the system is configured to determine the production conditions in the production process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle among the multiple search cycles. The system is configured to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle among the multiple search cycles mentioned above. An information processing device characterized by the following:

[0150] (Section 22) An information processing method for an information processing device configured to perform multiple search cycles together with a prototype device, The system accepts input of search conditions and, in each of the multiple search cycles, executes a decision process to determine the preparation conditions for preparing the sample. In each of the aforementioned multiple exploration cycles, the prototype device is instructed to perform a fabrication process to produce the sample under the fabrication conditions, and an evaluation process to evaluate the produced sample and output the evaluation results. In the determination process of the first search cycle among the multiple search cycles, the production conditions in the production process of the first search cycle are determined using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle among the multiple search cycles. The search conditions used in the third search cycle and subsequent search cycles following the first search cycle are changed among the multiple search cycles mentioned above. An information processing method characterized by the following:

[0151] (Section 23) A program that causes a computer to execute the information processing method described in item 22.

[0152] (Section 24) A computer-readable recording medium on which the program described in item 23 is recorded. [Explanation of symbols]

[0153] 100... Manufacturing and evaluation system (system), 103... Automatic prototyping device (prototyping device), 104... Display device, 109... Information processing device, 124... Sample

Claims

1. A system comprising a prototype device and an information processing device, configured to perform multiple search cycles, The information processing device is configured to receive input of search conditions and to perform a decision process to determine the sample preparation conditions in each of the multiple search cycles. The prototype device is configured to perform, in each of the multiple search cycles, a manufacturing process in which the sample is manufactured using the manufacturing conditions determined in the determination process, and an evaluation process in which the sample manufactured in the manufacturing process is evaluated and an evaluation result is output. The information processing device is configured to determine the production conditions in the production process of the first search cycle in the determination process of the first search cycle among the multiple search cycles, using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle among the multiple search cycles. The information processing device is configured to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle among the multiple search cycles. A system characterized by the following features.

2. When the information processing device receives an instruction to change the search conditions, it changes the search conditions in accordance with the instruction. The system according to feature 1.

3. The information processing device automatically changes the search conditions based on the search status based on the dataset. The system according to feature 1.

4. The dataset includes the measured manufacturing conditions obtained by the manufacturing process, or the manufacturing conditions obtained by the determination process. The system according to feature 1.

5. The search conditions include a first search setting that includes candidate manufacturing conditions. The aforementioned information processing device is In the aforementioned determination process, the manufacturing conditions for the manufacturing process are determined from among the candidates included in the first search setting. The system according to feature 1.

6. The aforementioned preparation conditions include information on the raw materials used to prepare the sample. The first search setting includes information on multiple types of raw materials as candidates. The system according to claim 5, characterized in that it is the same as described in claim 5.

7. The aforementioned preparation conditions include information on the proportions of the raw materials used to prepare the sample. The first search setting includes, as a candidate, information on the range of the blending amounts, The system according to claim 5, characterized in that it is the same as described in claim 5.

8. The manufacturing conditions include information on the processing temperature in the manufacturing process. The first search setting includes, as a candidate, information on the range of the processing temperature, The system according to claim 5, characterized in that it is the same as described in claim 5.

9. The search conditions include a second search setting relating to at least one physical property of the sample, The information processing device determines the manufacturing conditions based on a search algorithm to optimize the physical properties in the decision process. The system according to feature 1.

10. The search conditions include a third search setting that includes at least one of the following: the termination condition for the search cycle, the search algorithm used in the decision process, and the number of samples produced in one search cycle in the fabrication process of the prototype device. The system according to any one of claims 1 to 9.

11. The aforementioned sample is a cured product of an adhesive. The system according to any one of claims 1 to 9.

12. Equipped with an additional display device, The information processing device causes the display device to display an image corresponding to the search status based on the dataset. The system according to any one of claims 1 to 9.

13. It also has a display device, The information processing device causes the display device to display an image indicating the search conditions. The system according to any one of claims 1 to 9.

14. Equipped with an additional display device, The information processing device causes the display device to display an image showing a candidate for changing the search condition. The system according to feature 2.

15. The prepared sample is evaluated using at least one of the following: the physical properties of the sample and the resources necessary for preparing the sample. The system according to feature 1.

16. The aforementioned preparation conditions include information on processing conditions used when preparing the sample. The first search setting includes information on the range of the processing conditions as a candidate, The system according to claim 5, characterized in that it is the same as described in claim 5.

17. The aforementioned search status is the progression of the evaluation results. The system according to feature 12.

18. The aforementioned search status is the accuracy of the decision process. The system according to feature 12.

19. The aforementioned search status is the relationship between the manufacturing conditions and the evaluation results. The system according to feature 12.

20. The aforementioned search status is a relationship between multiple evaluation results. The system according to feature 12.

21. An information processing device configured to perform multiple search cycles together with a prototype device, The system is configured to accept input of search conditions and, in each of the multiple search cycles, execute a decision process to determine the preparation conditions for preparing the sample. In each of the aforementioned multiple exploration cycles, the prototype device is configured to perform a fabrication process to produce the sample under the aforementioned fabrication conditions, and an evaluation process to evaluate the produced sample and output the evaluation results. In the determination process of the first search cycle among the multiple search cycles, the system is configured to determine the production conditions in the production process of the first search cycle using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle among the multiple search cycles. The system is configured to change the search conditions used in the third search cycle and subsequent search cycles following the first search cycle among the multiple search cycles mentioned above. An information processing device characterized by the following:

22. An information processing method for an information processing device configured to perform multiple search cycles together with a prototype device, The system accepts input of search conditions and, in each of the multiple search cycles, executes a decision process to determine the preparation conditions for preparing the sample. In each of the aforementioned multiple exploration cycles, the prototype device is instructed to perform a fabrication process to produce the sample under the fabrication conditions, and an evaluation process to evaluate the produced sample and output the evaluation results. In the determination process of the first search cycle among the multiple search cycles, the production conditions in the production process of the first search cycle are determined using the search conditions and a dataset including data acquired in the second search cycle or earlier, prior to the first search cycle among the multiple search cycles. The search conditions used in the third search cycle and subsequent search cycles following the first search cycle are changed among the multiple search cycles. An information processing method characterized by the following:

23. A program for causing a computer to execute the information processing method described in claim 22.

24. A computer-readable recording medium having the program described in claim 23 recorded on it.

Images