Data acquisition method, data acquisition apparatus, and program

Abstract

To provide a technique for simply designing a desired cushioning material.SOLUTION: A data acquisition method includes: a first step of acquiring input data including at least one of shock response data related to a shock response spectrum and stress-strain data related to a stress-strain curve; and a second step of executing at least one of (i) a first acquisition step of acquiring first object data representing an acceleration waveform according to the acquired shock response data, by using first association data obtained by associating the acceleration waveform representing a shock acceleration of an object protected from a cushioning material with a shock response spectrum of the object, and (ii) a second acquisition step of acquiring second object data representing shape data according to the input stress-strain data, by using second association data obtained by associating shape data of the cushioning material with a stress-strain curve of the cushioning material.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present disclosure relates to a data acquisition method, a data acquisition device, and a program.

Background Art

[0002] Patent Document 1 discloses a technique for enhancing the shock absorption performance of a shock absorber. In the technique of Patent Document 1, the shock absorber is configured such that in the stress-strain curve when the shock absorber is compressed, a so-called plateau region where the stress remains substantially constant even as the strain increases becomes wider.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In recent years, the objects to be protected by shock absorbers and the usage situations where shock absorbers are used have diversified, and the need to design various shock absorbers has increased. Therefore, a technique for more easily designing a desired shock absorber is desired.

Means for Solving the Problems

[0005] According to a first aspect of the present disclosure, a data acquisition method is provided. This data acquisition method includes a first step of acquiring input data including at least one of impact response data related to an impact response spectrum and stress-strain data related to a stress-strain curve, and a second step of performing at least one of: a first acquisition step of acquiring first target data representing the acceleration waveform corresponding to the acquired impact response data using first associated data associating an acceleration waveform representing the impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition step of acquiring second target data representing the shape data corresponding to the input stress-strain data using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material.

[0006] According to a second aspect of the present disclosure, a data acquisition device is provided. This data acquisition device includes an input data acquisition unit that acquires input data including at least one of impact response data related to an impact response spectrum and stress-strain data related to a stress-strain curve, and a target data acquisition unit that performs at least one of: a first acquisition process of acquiring first target data representing the acceleration waveform corresponding to the acquired impact response data using first associated data associating an acceleration waveform representing the impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition process of acquiring second target data representing the shape data corresponding to the input stress-strain data using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material.

[0007] According to a third aspect of the present disclosure, a program is provided. This program has a function of acquiring input data including at least any one of impact response data related to an impact response spectrum and stress-strain data related to a stress-strain curve, and a function of causing a computer to execute at least one of: a first acquisition process of acquiring first target data representing an acceleration waveform corresponding to the acquired impact response data by using first associated data associating an acceleration waveform representing an impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition process of acquiring second target data representing the shape data corresponding to the input stress-strain data by using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material. BRIEF DESCRIPTION OF THE DRAWINGS

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Embodiments for Carrying Out the Invention

[0009] A. First Embodiment: FIG. 1 is a block diagram showing a schematic configuration of a data acquisition device 100 according to the first embodiment. The data acquisition device 100 is used for designing a buffer material for protecting an object. Specifically, the data acquisition device 100 is used for acquiring data used in the design of the buffer material. The data acquisition device 100 in the present embodiment is used in a design system 50 for designing a buffer material. Hereinafter, the object protected by the buffer material is also referred to as a protection target.

[0010] The protection target of the buffer material may be, for example, various products or a living body such as a person. The protection target in the present embodiment is a product. The product may be any product, for example, various devices such as a printing device, a projection device, a three-dimensional modeling device, and an injection molding device, or various parts. Further, the buffer material may protect the protection target from impact by being housed together with the protection target in a container for housing the protection target. Further, the buffer material may be designed as a container, an outer shell, a protective gear, a base, etc. having a function of protecting the protection target from impact.

[0011] The data acquisition device 100 in the present embodiment is constituted by a computer including one or more processors 101, a storage device 102, an input / output interface 103, and an internal bus 104. The processor 101, the storage device 102, and the input / output interface 103 are connected so as to be communicable bidirectionally via the internal bus 104. An output device 105 and an input device 106 are connected to the input / output interface 103. A program 155, first related data 161, and second related data 162 are stored in the storage device 102 in the present embodiment.

[0012] The first related data 161 is data that associates an acceleration waveform representing the impact acceleration of the object to be protected with the shock response spectrum (SRS) of the object to be protected. The acceleration waveform is waveform data that represents the acceleration generated in the object to be protected in a time series when an impact is applied to the object to be protected. The shock response spectrum is spectrum data that represents the acceleration generated in the object when an impact is applied to the object, for each natural frequency of the elements included in the object. More specifically, the shock response spectrum is data obtained by plotting the maximum value of the shock acceleration for each natural frequency of the elements included in the object. The acceleration waveform and the shock response spectrum included in the first related data 161 can be obtained, for example, by analyzing the behavior of the object to be protected when an impact is applied to the object to be protected, based on simulations or experiments. Hereinafter, the simulation regarding the behavior of the object to be protected when an impact is applied to the object to be protected is also referred to as an impact simulation. In the first related data 161 in the present embodiment, the acceleration waveform and the shock response spectrum based on the same impact simulation are associated with each other.

[0013] The second related data 162 is data that associates the shape data of the cushioning material with the stress-strain curve of the cushioning material. The shape data is, for example, two-dimensional or three-dimensional CAD (Computer Aided Design) data representing the shape of the cushioning material. The stress-strain curve is data representing the relationship between the stress and the strain of the cushioning material when the cushioning material is compressed. In the second related data 162 in the present embodiment, the shape data and the stress-strain curve for the same cushioning material are associated with each other.

[0014] The first generation model 171 is a machine learning model obtained by performing machine learning using the first related data 161 as learning data. Hereinafter, the learning data used for learning the first generation model 171 is referred to as the first learning data. The first generation model 171 in the present embodiment has been learned by supervised learning. Further, in the present embodiment, the first learning data is teacher data including a plurality of impulse response spectra as training data and acceleration waveforms associated with the respective impulse response spectra as correct labels. As a result, when impulse response data is input to the first generation model 171, the first generation model 171 generates and outputs an acceleration waveform corresponding to the input impulse response data. Specifically, the first generation model 171 is generated as a regression model that estimates a function representing an acceleration waveform according to the impulse response data input to the first generation model 171. The first generation model 171 is configured by, for example, a convolutional neural network (CNN). In the machine learning of the CNN as the first generation model 171, for example, the error backpropagation method is used to train the first generation model 171 so that the error between the acceleration waveform generated by the first generation model 171 and the acceleration waveform as the correct label is reduced. In other embodiments, the first generation model 171 may be configured by, for example, a neural network other than CNN, or may be configured by a support vector machine (SVM) or a decision tree. Further, the first generation model 171 is not limited to supervised learning, and may be learned by, for example, unsupervised learning or reinforcement learning.

[0015] The second generation model 172 is a machine learning model obtained by performing machine learning using the second related data 162 as learning data. Hereinafter, the learning data used for learning the second generation model 172 is referred to as second learning data. The second generation model 172 in the present embodiment has been learned by supervised learning. Also, in the present embodiment, the second learning data is teacher data including a plurality of stress-strain curves as training data and shape data associated with each stress-strain curve as a correct label. As a result, when stress-strain data is input to the second generation model 172, the second generation model 172 generates and outputs shape data corresponding to the input stress-strain data. Specifically, the second generation model 172 is generated as a regression model that estimates a function representing shape data according to the stress-strain data input to the second generation model 172. The second generation model 172 is configured by a neural network such as a CNN, for example, in the same manner as the first generation model 171. In the machine learning of the CNN as the second generation model 172, for example, the error backpropagation method is used to learn the second generation model 172 so that the error between the shape data generated by the second generation model 172 and the shape data as the correct label is reduced. Also, in other embodiments, the second generation model 172 may be configured by, for example, an SVM or a decision tree, or may be learned by unsupervised learning or reinforcement learning.

[0016] By executing the program 155 stored in the storage device 102, the processor 101 realizes various functions including the functions as the input data acquisition unit 110 and the target data acquisition unit 120 and the function of executing the data acquisition process described later.

[0017] The input data acquisition unit 110 acquires input data input by the user. The input data includes at least one of impact response data and stress-strain data. The impact response data is data related to the impact response spectrum. As the impact response data, for example, data related to the impact response spectrum corresponding to a cushioning material for which design is desired and the object to be protected by the cushioning material is input. The impact response data may be represented, for example, as an impact response spectrum or as a feature quantity of the impact response spectrum. The stress-strain data is data related to the stress-strain curve. As the stress-strain data, for example, data related to the stress-strain curve corresponding to a cushioning material for which design is desired is input. The stress-strain data may be represented, for example, as a stress-strain curve or as a feature quantity of the stress-strain curve. The input data acquisition unit 110 in the present embodiment is configured to be able to acquire both impact response data and stress-strain data as input data.

[0018] In the present embodiment, the input data acquisition unit 110 acquires the input data input by the user Ur via the input device 106. The input device 106 is constituted by, for example, a mouse or a keyboard. The user Ur can input desired impact response data or desired stress-strain data to the data acquisition device 100 via the input device 106. In the present embodiment, both impact response data and stress-strain data can be input to the input device 106.

[0019] The target data acquisition unit 120 executes at least one of a first acquisition process and a second acquisition process. The first acquisition process is a process of acquiring first target data using the first related data 161. The first target data is data representing an acceleration waveform corresponding to the impact response data acquired by the input data acquisition unit 110. The second acquisition process is a process of acquiring second target data using the second related data 162. The second target data is data representing shape data corresponding to the stress-strain data acquired by the input data acquisition unit 110. Hereinafter, when the first target data and the second target data are not distinguished, each is simply referred to as target data.

[0020] The output device 105 outputs the target data acquired by the target data acquisition unit 120. The output device 105 in the present embodiment is configured as a display device that outputs the target data as visual information. The display device is constituted by, for example, a liquid crystal panel or an organic EL panel. Note that the output device 105 as a display device may be configured as a touch panel capable of receiving a touch operation from the user Ur. In this case, the output device 105 may also serve as, for example, an input device 106. Further, in other embodiments, the output device 105 may be configured as a device that outputs the target data by transmitting it to an external computer or a recording medium, for example.

[0021] FIG. 2 is a diagram for explaining an example of a protection target and a cushioning material. FIG. 2 shows a state in which a box-shaped container BX for housing the product PR is being transported. In the container BX, in addition to the product PR, a cushioning material CM for protecting the product PR from impact is arranged. That is, in the example of FIG. 2, the protection target of the cushioning material CM is the product PR. In the example of FIG. 2, the product PR includes a component group PG constituted by various components. For example, when the product PR is an electronic device such as a printing device or a projection device, the component group PG includes various electronic components and structural components. Note that in FIG. 2, the cushioning material CM is hatched.

[0022] Further, FIG. 2 shows a state in which the container BX1 falls during transportation. Due to the fall of the container BX1, for example, an impact from the ground due to the fall is applied to the product PR housed in the container BX1. The cushioning material CM is designed so as to suppress or prevent damage or breakage of the product PR caused by such an impact. Specifically, the cushioning material CM is designed to reduce the impact acceleration generated in the product PR by the impact.

[0023] The cushioning material CM is preferably designed so as to suppress or prevent damage or breakage caused by the impact acceleration of each component included in the component group PG of the product PR. Here, each component included in the component group PG is most likely to be damaged when an impact acceleration having the same frequency as the natural frequency of each component occurs. Therefore, it is preferable that the cushioning material CM is designed such that the impact acceleration having the same frequency as the natural frequency of each component generated by the impact when the impact is applied to the product PR is less than the acceleration threshold value that can damage each component.

[0024] FIG. 3 is a diagram for explaining an example of the impact simulation in the present embodiment. The impact simulation is executed to obtain an acceleration waveform and an impact response spectrum. The impact simulation is executed, for example, using analysis software for dynamically analyzing the behavior of the product PR when an impact is applied to the product PR. In the impact simulation, for example, at least a part of non-linear structural analysis, linear structural analysis, non-linear structural analysis, eigenvalue analysis, and frequency response analysis is executed. Further, as a method of spatial discretization in the impact simulation, for example, the finite element method, the boundary element method, and the discrete element method are used. Further, the time evolution in the impact simulation is calculated, for example, by the dynamic explicit method.

[0025] Figure 3 shows a state where a drop simulation is executed in which a sample SP is dropped onto the ground. In the drop simulation, the impact acceleration generated in the sample SP when the dropped sample SP collides with the ground is observed at a predetermined observation point OP. The observation point OP is, for example, the center of gravity of the product PR. In the example of Figure 3, the sample SP is the product PR supported from below by a cushioning material CM. In the example of Figure 3, the cushioning material CM is designed as a cushioning material for protecting the product PR from the impact of dropping. In the simulation of Figure 3, as simulation conditions, for example, the dimensions and weight of the product PR, the dimensions and weight of the cushioning material CM, the relative position between the product PR and the cushioning material CM, the height h1 of the sample SP from the ground, and the stress-strain curve of the cushioning material CM are set. The height h1 is preferably determined, for example, according to the actual transportation conditions under which the product PR is transported. The transportation conditions are, for example, the type of transportation equipment and transportation device used for transporting the product PR, the dimensions of the container that houses the product PR during transportation, and the height at which the product PR is located during transportation. The stress-strain curve of the cushioning material CM is obtained, for example, based on an experiment or simulation in which the cushioning material CM is compressed.

[0026] FIG. 4 is a diagram for explaining the relationship between a stress-strain curve, an acceleration waveform, and a shock response spectrum. FIG. 4 shows an example of an acceleration waveform obtained by a shock simulation and an example of a shock response spectrum. Note that the "maximum acceleration" in FIG. 4 means the maximum value of the shock acceleration for each natural frequency. Also, FIG. 4 shows an example of a stress-strain curve used as a simulation condition for the shock simulation. Specifically, in FIG. 4, as an example of the acceleration waveform, a first waveform AW1 for a first sample and a second waveform AW2 for a second sample are shown. The first sample and the second sample are different samples SP, respectively. Also, in FIG. 4, as an example of the shock response spectrum, a first spectrum SR1 for a first sample and a second spectrum SR2 for a second sample are shown. Further, in FIG. 4, as an example of the stress-strain curve, a first curve SC1 for a cushioning material included in the first sample and a second curve SC2 for a cushioning material included in the second sample are shown.

[0027] In the shock simulation according to the present embodiment, as a simulation result, acceleration waveforms such as the first waveform AW1 and the second waveform AW2 are obtained. Note that in the shock simulation for obtaining the first waveform AW1 and the shock simulation for obtaining the second waveform AW2, only the stress-strain curve among the simulation conditions is different. The first spectrum SR1 is calculated based on the first waveform AW1. The second spectrum SR2 is calculated based on the second waveform AW2. The first curve SC1 corresponds to the stress-strain curve set as a simulation condition for obtaining the first waveform AW1. The second curve SC2 corresponds to the stress-strain curve set as a simulation condition for obtaining the second waveform AW2.

[0028] As described above, since the stress-strain curve is used as a simulation condition in the impact simulation, it can be said that the acceleration waveform and the shock response spectrum are calculated based on the stress-strain curve. Also, the shock response spectrum is calculated based on the acceleration waveform. Incidentally, contrary to the impact simulation, it is also possible to calculate the stress-strain curve based on the acceleration waveform. Specifically, for example, based on a certain acceleration waveform and simulation conditions other than the stress-strain curve, the stress-strain curve can be inversely calculated using an analysis algorithm similar to that of the impact simulation. That is, the stress-strain curve thus obtained corresponds to the stress-strain curve set as the simulation condition for obtaining the acceleration waveform. On the other hand, conventionally, it has been difficult to inversely calculate the acceleration waveform based on the shock response spectrum. This is because the shock response spectrum is data in which only the maximum acceleration for each natural frequency is plotted, and is more discrete data compared to the acceleration waveform. Also, conventionally, when designing a cushioning material using the stress-strain curve, it is relatively easy to calculate the stress-strain curve after determining the shape and dimensions of the cushioning material, etc., but it is relatively difficult to inversely calculate the shape and dimensions of the cushioning material from a certain stress-strain curve.

[0029] FIG. 5 is a flowchart of a data acquisition process for realizing the data acquisition method in the present embodiment. For example, each time input data is input to the data acquisition device 100 via the input device 106, the processor 101 starts the data acquisition process shown in FIG. 5.

[0030] In step S105, the input data acquisition unit 110 acquires the input data input via the input device 106. That is, in step S105, at least one of the shock response data and the stress-strain data is acquired. The step of acquiring the input data, as in step S105, is also referred to as the first step.

[0031] In step S110, the input data acquisition unit 110 determines whether the input data obtained in step S105 includes impulse response data. If the input data does not include impulse response data in step S110, the input data acquisition unit 110 proceeds to step S135.

[0032] If the input data includes impulse response data in step S110, from steps S115 to S125, the target data acquisition unit 120 executes the first acquisition process. Steps S115 to S125 in the present embodiment correspond to the first acquisition step. The first acquisition step is a step of acquiring first target data using the first related data 161.

[0033] In the first acquisition process in the present embodiment, first, in step S115, the target data acquisition unit 120 determines whether the corresponding spectrum is included in the first related data 161 by referring to the first related data 161 based on the impulse response data obtained in step S105. The corresponding spectrum is the impulse response spectrum corresponding to the impulse response data obtained in the first step.

[0034] Specifically, the corresponding spectrum is an impulse response spectrum in which the first difference from the impulse response data is equal to or less than a first reference level determined in advance. The first difference is represented, for example, by the sum of the squares of the differences in maximum acceleration, the square root of the sum of the squares of the differences in maximum acceleration, or the arithmetic mean value of these. Further, the first difference may be represented as a value taking into account, for example, the position or number of peaks in the impulse response data or impulse response spectrum. In this case, the first difference is defined such that, for example, the smaller the position or number of peaks is approximated, the smaller the value is.

[0035] If the corresponding spectrum is included in the first related data 161 in step S115, in step S120, the target data acquisition unit 120 extracts the acceleration waveform associated with the corresponding spectrum from the first related data 161. The target data acquisition unit 120 acquires the extracted acceleration waveform as the first target data. When a plurality of corresponding spectra are included in the first related data 161, the target data acquisition unit 120 may, for example, extract the acceleration waveform associated with the corresponding spectrum having the smallest first difference, or randomly select one corresponding spectrum from the plurality of corresponding spectra or based on a predetermined selection criterion, and extract the acceleration waveform associated with the selected corresponding spectrum. Further, the target data acquisition unit 120 may, for example, extract each acceleration waveform associated with a plurality of corresponding spectra as the first target data.

[0036] If the corresponding spectrum is not included in the first related data 161 in step S115, in step S125, the target data acquisition unit 120 generates an acceleration waveform using the first generation model 171. Specifically, in step S125, the target data acquisition unit 120 inputs the impulse response data acquired in step S105 to the first generation model 171 to generate an acceleration waveform corresponding to the impulse response data. The target data acquisition unit 120 acquires the generated acceleration waveform as the first target data.

[0037] In step S130, the target data acquisition unit 120 outputs the first target data acquired in step S120 or step S125 using the output device 105. In step S130, for example, the first difference calculated in step S115 may be output together with the first target data.

[0038] In the design of the cushioning material, the user can use the acceleration waveform acquired as the first target data in steps S120 and S125, or refer to the output result of step S130. As described above, it is possible to calculate a stress-strain curve based on the acceleration waveform as the first target data acquired in steps S120 and S125. By using the stress-strain curve calculated in this way, a cushioning material having a desired protective performance can be designed more simply. Specifically, for example, as input data, impact response data in which damage and breakage of each component included in the product to be protected are suppressed is input. In the first acquisition step, an acceleration waveform corresponding to the impact response data is acquired, and a stress-strain curve can be calculated based on the acquired acceleration waveform. In this way, by using the acquired stress-strain curve, a cushioning material having a protective performance in which damage and breakage of each component included in the object to be protected are suppressed can be designed more simply. The impact response data in which damage and breakage of each component included in the object to be protected are suppressed is prepared, for example, by setting the value of the maximum acceleration in the impact response data based on a damage boundary curve (DBC: Damage Boundary Curve). The damage boundary curve is obtained, for example, based on simulations and experiments for analyzing the drop impact of the object to be protected.

[0039] In step S135, the input data acquisition unit 110 determines whether stress-strain data is included in the input data acquired in step S105. If the stress-strain data is not included in the input data in step S135, the input data acquisition unit 110 ends the data acquisition process.

[0040] If the stress-strain data is included in the input data in step S135, in steps S140 to S150, the target data acquisition unit 120 executes a second acquisition process. Steps S140 to S150 in the present embodiment correspond to a second acquisition step. The second acquisition step is a step of acquiring second target data using the second related data 162. The step of executing at least one of the first acquisition step and the second acquisition step is also referred to as the second step.

[0041] In the second acquisition process according to this embodiment, first, in step S140, the target data acquisition unit 120 determines whether a corresponding curve is included in the second related data 162 by referring to the second related data 162 based on the stress-strain data acquired in step S105. The corresponding curve is a stress-strain curve corresponding to the stress-strain data acquired in the first step.

[0042] Specifically, the corresponding curve is a stress-strain curve in which the second difference from the stress-strain data is equal to or less than a second standard degree determined in advance. The second difference is represented by, for example, the sum of squares of stress differences, the square root of the sum of squares of stress differences, or the arithmetic mean value of these. Further, the second difference may be a degree taking into account the position and number of peaks in the stress-strain data and the stress-strain curve. In this case, the second difference is defined so as to take a smaller value as the position and number of peaks are closer.

[0043] If the corresponding curve is included in the second related data 162 in step S140, in step S145, the target data acquisition unit 120 extracts the shape data associated with the corresponding curve from the second related data 162. The target data acquisition unit 120 acquires the extracted shape data as the second target data. When a plurality of corresponding curves are included in the second related data 162, the target data acquisition unit 120 may extract, for example, the shape data associated with the corresponding curve having the smallest second difference, or may randomly select one corresponding curve from the plurality of corresponding curves or based on a predetermined selection criterion, and extract the acceleration waveform associated with the selected corresponding curve. Further, the target data acquisition unit 120 may extract, for example, each shape data associated with a plurality of corresponding curves as the second target data.

[0044] If the corresponding curve is not included in the second related data 162 in step S140, in step S150, the target data acquisition unit 120 generates shape data using the second generation model 172. Specifically, in step S150, the target data acquisition unit 120 inputs the stress-strain response data acquired in step S105 into the second generation model 172 to generate shape data corresponding to the stress-strain data. The target data acquisition unit 120 acquires the generated shape data as the second target data.

[0045] In step S155, the target data acquisition unit 120 outputs the second target data acquired in step S145 or step S150 using the output device 105. Note that in step S155, the second difference calculated in step S140, for example, may be output together with the second target data.

[0046] The user can use the shape data acquired as the second target data in steps S145 and S150 or refer to the output result of step S155 in the design of the buffer material. In this way, in the design of the buffer material, particularly, the shape and dimensions of the buffer material can be determined more easily.

[0047] Note that the stress-strain data included in the input data acquired in step S105 may be data related to the stress-strain curve calculated based on the acceleration waveform as the first target data. In this way, based on the desired shock response data, the shape data of the buffer material having the protection performance for realizing the behavior represented by the shock response data can be acquired. By using the shape data acquired in this way, a buffer material having preferable protection performance can be designed more easily.

[0048] According to the data acquisition method in the present embodiment described above, a first acquisition step of acquiring, as first target data, an acceleration waveform corresponding to impact response data using first associated data 161 associating an acceleration waveform with an impact response spectrum, and a second acquisition step of acquiring, as second target data, a shape data corresponding to stress-strain data using second associated data 162 associating shape data with a stress-strain curve are executed, at least one of which is performed. By doing so, an acceleration waveform corresponding to desired impact response data and a shape data corresponding to desired stress-strain data can be acquired as target data, and a cushioning material can be designed using the acquired acceleration waveform and shape data. Therefore, a desired cushioning material can be designed more simply.

[0049] Further, in the present embodiment, in the first acquisition step, a corresponding spectrum is extracted from the first associated data 161, and the extracted corresponding spectrum is acquired as the first target data. Also, in the second acquisition step, a corresponding curve is extracted from the second associated data 162, and the extracted corresponding curve is acquired as the second target data. Therefore, the first target data and the second target data can be easily acquired by extracting them from the first associated data 161 and the second associated data 162.

[0050] Also, in the present embodiment, in the first acquisition step, the first target data is acquired by generating an acceleration waveform corresponding to the impact response data using the first generation model 171 obtained by machine learning using the first related data 161. Specifically, in the present embodiment, in the first acquisition step, when the corresponding spectrum is not included in the first related data 161, the first target data is acquired by generating an acceleration waveform corresponding to the impact response data using the first generation model 171. Further, in the second acquisition step, the second target data is acquired by generating shape data corresponding to the stress-strain data using the second generation model 172 obtained by machine learning using the second related data 162. Specifically, in the present embodiment, in the second acquisition step, when the corresponding spectrum is not included in the second related data 162, the second target data is acquired by generating shape data corresponding to the stress-strain data using the second generation model 172. By doing so, even when the impact response spectrum corresponding to the desired impact response data is not included in the first related data 161 or the stress-strain curve corresponding to the desired stress-strain data is not included in the second related data 162, the first target data and the second target data can be acquired by generating them. Therefore, the first target data and the second target data can be acquired more reliably.

[0051] B. Second Embodiment: FIG. 6 is a block diagram showing a schematic configuration of the data acquisition device 100b in the second embodiment. Different from the first embodiment, the storage device 102b of the data acquisition device 100b in the present embodiment does not store the first generation model 171 and the second generation model 172, but stores the first calculation model 181 and the second calculation model 182. Among the configurations of the data acquisition device 100b and the design system 50 in the second embodiment, points not particularly described are the same as those in the first embodiment.

[0052] The first calculation model 181 is a machine learning model that is machine-learned using the acceleration waveform of the object to be protected and the shock response spectrum corresponding to the acceleration waveform as learning data. The first calculation model 181 is used for calculating the first similarity. The first similarity is the similarity between the shock response data input to the first calculation model 181 and the shock response spectrum included in the first related data 161. Hereinafter, the learning data for training the first calculation model 181 is also referred to as the third learning data.

[0053] As the first calculation model 181, for example, a machine learning model that extracts feature quantities of the shock response data and the shock response spectrum input to the first calculation model 181 can be used. In this case, the first similarity is calculated by comparing the feature quantity extracted by inputting the shock response data to the first calculation model 181 with the feature quantity extracted by inputting the shock response spectrum to the first calculation model 181. In this case, the first similarity may be calculated, for example, by comparing the Euclidean distance, Manhattan distance, or Hamming distance between the feature quantities, or by comparing the feature quantity vectors representing the respective feature quantities. As the feature quantities of the shock response data and the shock response spectrum, for example, feature quantities representing the slope, minimum value, and maximum value of the maximum acceleration, feature quantities representing the positions of the peaks, and feature quantities representing the number of peaks can be extracted.

[0054] As the first calculation model 181 for extracting feature quantities of impact response data and impact response spectra, for example, a part of the layer structure of a CNN that has been learned by supervised learning using third learning data can be used for the correspondence relationship between the acceleration waveform and the impact response spectrum. Specifically, as the first calculation model 181, for example, the layer structure up to the fully connected layer lower than the output layer in the layer structure of a CNN that has been learned using third learning data can be used. In this case, the third learning data is, for example, training data that includes a plurality of impact response spectra as training data and includes the acceleration waveform corresponding to each impact response spectrum as the correct label. As the third learning data, for example, the first related data 161 may be used. Also, as the first calculation model 181, a part of the structure of a machine learning model similar to the first generation model 171 may be used. In other embodiments, the first calculation model 181 may be configured by, for example, a neural network other than a CNN, or may be configured by an SVM or a decision tree. Also, the first calculation model 181 is not limited to supervised learning, and may be learned by, for example, unsupervised learning or reinforcement learning.

[0055] The second calculation model 182 is a machine learning model that is machine-learned using the shape data of the cushioning material and the stress-strain curve of the cushioning material as learning data. The second calculation model 182 is used for calculating the second similarity. The second similarity is the similarity between the stress-strain data input to the second calculation model 182 and the stress-strain curve included in the second related data 162. Hereinafter, the learning data for learning the second calculation model 182 is also referred to as the fourth learning data.

[0056] As the second calculation model 182, for example, a machine learning model that extracts feature quantities of stress-strain data or stress-strain curves input to the second calculation model 182 can be used. In this case, similar to the first similarity, the second similarity is calculated by comparing feature quantities. As the feature quantities of the stress-strain data or stress-strain curves, for example, feature quantities representing the slope, minimum value, or maximum value of stress, feature quantities representing the position of a peak, or feature quantities representing the number of peaks can be extracted. Further, as the second calculation model 182 of the stress-strain data or stress-strain curves, for example, a part of the layer structure of a CNN that has learned the correspondence between shape data and stress-strain by supervised learning using fourth learning data can be used. Specifically, as the second calculation model 182, for example, the layer structure up to the fully connected layer lower than the output layer in the layer structure of a CNN that has been learned using fourth learning data can be used. In this case, the fourth learning data is, for example, training data that includes the stress-strain curve of a cushioning material as training data and includes the shape data of the cushioning material as a correct label. As the fourth learning data, for example, the second related data 162 may be used. Further, as the second calculation model 182, a part of the structure of a machine learning model similar to the second generation model 172 may be used. Note that in other embodiments, the second calculation model 182 may be configured by, for example, a neural network other than a CNN, or may be configured by an SVM or a decision tree. Further, the second calculation model 182 is not limited to supervised learning, and may be learned by, for example, unsupervised learning or reinforcement learning.

[0057] FIG. 7 is a flowchart of data acquisition processing for realizing the data acquisition method in the second embodiment. In FIG. 7, the same steps as those in FIG. 5 are denoted by the same reference numerals as those in FIG. 5.

[0058] In the present embodiment, when the corresponding spectrum is not included in the first related data 161 in step S115, in step S123, the target data acquisition unit 120b extracts an acceleration waveform from the first related data 161 according to the first similarity using the first calculation model 181. The target data acquisition unit 120b acquires the extracted acceleration waveform as first target data.

[0059] Specifically, in step S123, the target data acquisition unit 120b extracts an acceleration waveform associated with an impact response spectrum having a first similarity equal to or greater than a predetermined similarity from the first associated data 161. In step S123, the target data acquisition unit 120b first calculates the first similarity using the first calculation model 181. Next, the target data acquisition unit 120b identifies a similar spectrum, which is an impact response spectrum having a first similarity equal to or greater than a first threshold value predetermined among the impact response spectra included in the first associated data 161. Then, the target data acquisition unit 120b acquires the first target data by extracting the acceleration waveform associated with the identified similar spectrum as the first target data. In the present embodiment, among the impact response spectra having a first similarity equal to or greater than the first threshold value predetermined, the impact response spectrum having the highest first similarity is identified as the similar spectrum. In other embodiments, for example, a plurality of impact response spectra having a first similarity equal to or greater than the first threshold value may be identified as the similar spectrum.

[0060] In step S130b, the target data acquisition unit 120b outputs the first target data, that is, the acceleration waveform extracted in step S120 or step S123, using the output device 105. In step S130b, for example, the first similarity calculated in step S123 may be output together with the first target data.

[0061] Also, when the corresponding curve is not included in the second related data 162 in step S140, in step S148, the target data acquisition unit 120b extracts shape data from the second related data 162 according to the second similarity using the second calculation model 182. Specifically, in step S148, the target data acquisition unit 120b extracts shape data associated with a stress-strain curve with a second similarity equal to or greater than a predetermined similarity from the second related data 162. In step S123, the target data acquisition unit 120b first calculates the second similarity using the second calculation model 182. Next, the target data acquisition unit 120b identifies a similar curve that is a stress-strain curve with a second similarity equal to or greater than a predetermined second threshold among the stress-strain curves included in the second related data 162. Then, the target data acquisition unit 120b obtains the second target data by extracting the shape data associated with the identified similar curve as the second target data. In the present embodiment, among the stress-strain curves with a second similarity equal to or greater than a predetermined second threshold, the stress-strain curve with the highest second similarity is identified as the similar curve. In other embodiments, for example, each stress-strain curve with a second similarity equal to or greater than the second threshold may be identified as the similar curve.

[0062] In step S155b, the target data acquisition unit 120b outputs the second target data, that is, the shape data extracted in step S145 or step S148, using the output device 105. In step S155b, for example, the second similarity calculated in step S148 may be output together with the second target data.

[0063] According to the data acquisition method in the second embodiment described above, in the first acquisition step, the first similarity is calculated using the first calculation model 181 obtained by machine learning with the acceleration waveform and the shock response spectrum as learning data, and the acceleration waveform associated with the shock response spectrum whose first similarity is equal to or greater than a predetermined similarity is extracted from the first related data 161, thereby obtaining the first target data. Specifically, in the present embodiment, in the first acquisition step, when the corresponding spectrum is not included in the first related data 161, the first target data is obtained by extracting the acceleration waveform using the first calculation model 181. Further, in the second acquisition step, the second similarity is calculated using the second calculation model 182 obtained by machine learning with the shape data and the stress-strain curve as learning data, and the shape data associated with the stress-strain curve whose second similarity is equal to or greater than a predetermined similarity is extracted from the second related data 162, thereby obtaining the second target data. Specifically, in the present embodiment, in the second acquisition step, when the corresponding spectrum is not included in the second related data 162, the second target data is obtained by extracting the shape data using the second calculation model 182. By doing so, even when the corresponding spectrum is not included in the first related data 161 or the corresponding curve is not included in the second related data 162, the first target data and the second target data can be obtained by extracting them using the first calculation model 181 and the second calculation model 182. Therefore, the first target data and the second target data can be obtained more reliably.

[0064] C. Third Embodiment: FIG. 8 is a block diagram showing a schematic configuration of the data acquisition device 100c in the third embodiment. Different from the first embodiment, in the storage device 102c of the data acquisition device 100c in the present embodiment, in addition to the first generation model 171 and the second generation model 172, the first calculation model 181 and the second calculation model 182 are stored. Among the configurations of the data acquisition device 100c and the design system 50 in the third embodiment, points not particularly described are the same as those in the first embodiment.

[0065] FIG. 9 is a flowchart of data acquisition processing for realizing the data acquisition method in the third embodiment. In FIG. 9, the same steps as those in FIGS. 5 and 7 are denoted by the same reference numerals as those in FIGS. 5 and 7.

[0066] After step S123 is executed, in step S124, the target data acquisition unit 120c determines whether an acceleration waveform has been extracted in step S123. If it is determined in step S124 that the acceleration waveform has been extracted, the target data acquisition unit 120c proceeds with the process to step S130c. If it is determined in step S124 that the acceleration waveform has not been extracted, in step S125, the target data acquisition unit 120c generates an acceleration waveform using the first generation model 171 and acquires the generated acceleration waveform as the first target data. That is, in the present embodiment, the first correspondence process for acquiring the acceleration waveform associated with the corresponding spectrum, the first similarity process for acquiring the acceleration waveform associated with the similar spectrum, and the first generation process for acquiring the acceleration waveform generated by the first generation model 171 are executed in this order of priority. Note that the case where the acceleration waveform is not extracted in step S123 specifically corresponds to the case where the similar spectrum is not included in the first related data 161. In step S130c, the target data acquisition unit 120c outputs the first target data, that is, the acceleration waveform extracted in step S120 or step S123, or the acceleration waveform generated in step S125, using the output device 105.

[0067] After step S148 is executed, in step S149, the target data acquisition unit 120c determines whether shape data was extracted in step S148. If it is determined in step S149 that shape data was extracted, the target data acquisition unit 120c proceeds with the process to step S155c. If it is determined in step S149 that shape data was not extracted, in step S150, the target data acquisition unit 120c generates shape data using the second generation model 172 and acquires the generated shape data as second target data. That is, in the present embodiment, the second correspondence process for acquiring shape data associated with the corresponding curve, the second similarity process for acquiring shape data associated with the similar curve, and the second generation process for acquiring shape data generated by the second generation model 172 are executed in this order of priority. Note that the case where shape data is not extracted in step S148 specifically corresponds to the case where the second related data 162 does not include a similar curve. In step S155c, the target data acquisition unit 120c outputs the second target data, that is, the shape data extracted in step S145 or step S148, or the shape data generated in step S150, using the output device 105.

[0068] According to the data acquisition method in the third embodiment described above, in the first acquisition step, when the corresponding spectrum and the similar spectrum are not included in the first related data 161, the first target data is acquired by generating an acceleration waveform using the first generation model 171. Also, in the second acquisition step, when the corresponding curve and the similar curve are not included in the second related data 162, the second target data is acquired by generating shape data using the second generation model 172. Therefore, the first target data and the second target data can be acquired more reliably.

[0069] D. Fourth Embodiment: FIG. 10 is a flowchart of data acquisition processing for realizing the data acquisition method in the fourth embodiment. In FIG. 10, the same steps as those in FIG. 5 are denoted by the same reference numerals as those in FIG. 5. In this embodiment, different from the first embodiment, step S107 is executed. Also, in step S115d, it is determined whether or not the first corresponding spectrum is included in the first related data 161 according to a first range described later. Among the configurations of the data acquisition device 100 and the design system 50 in the fourth embodiment, points not particularly described are the same as those in the first embodiment.

[0070] In step S107, the input data acquisition unit 110 acquires information on the protection target. In this embodiment, the information on the protection target is type information indicating the type of the protection target. For example, when the protection target is a printing device, in step S107, type information indicating that the protection target is a printing device is acquired. Note that in this embodiment, the information on the protection target is acquired based on a designation by the user via the input device 106. In other embodiments, the information on the protection target may include, for example, in addition to the type information, or instead of the type information, information indicating the dimensions of the protection target or information indicating the mass of the protection target.

[0071] The corresponding spectrum in this embodiment is defined as an impact response spectrum in which a first difference is about a first reference in a first range that is a predetermined frequency range. Therefore, in step S115d, the target data acquisition unit 120 determines whether or not an impact response spectrum in which the first difference in the first range is about the first reference or less is included in the first related data 161. The first range is determined according to the protection target. Specifically, the first range is determined based on the information on the protection target acquired in step S107. That is, in this embodiment, the first range is determined based on a designation by the user. Note that when the protection target is an office device, the buffer material is generally designed such that the acceleration at about 200 Hz is low. Therefore, when the protection target is an office device, it is preferable that the first range is determined to be a range including the natural frequency of 200 Hz.

[0072] In step S115d, the first range is obtained, for example, by referring to a range database in which the information of the protection target and the first range are associated based on the information of the protection target obtained in step S107. The range database may be stored, for example, in the storage device 102, or may be stored in an external computer or a recording medium of the data acquisition device 100.

[0073] According to the data acquisition method in the fourth embodiment described above, the corresponding spectrum is an impact response spectrum in which the first difference is equal to or less than the first reference level in the first range, which is a frequency range determined according to the protection target. Therefore, by determining an appropriate first range for each protection target, appropriate first target data can be acquired according to the protection target.

[0074] Also, in this embodiment, the first range is determined based on the designation by the user. Therefore, appropriate first target data can be acquired according to the desired protection target in a simple method.

[0075] E. Fifth Embodiment: FIG. 11 is a flowchart of a data acquisition process for realizing the data acquisition method in the fifth embodiment. In FIG. 11, the same steps as those in FIGS. 5, 7, and 10 are denoted by the same reference numerals as those in FIGS. 5, 7, and 10. In this embodiment, similar to the fourth embodiment, steps S107 and S115b are executed. Also, in this embodiment, different from the first embodiment and the fourth embodiment, in step S123d, it is determined whether or not the first similar spectrum is included in the first related data 161 according to a second range described later. Among the configurations of the data acquisition device 100 and the design system 50 in the fifth embodiment, points not particularly described are the same as those in the first embodiment.

[0076] In step S123d, the target data acquisition unit 120 calculates the first similarity in a second range which is a predetermined frequency range. In the present embodiment, in step S123d, the target data acquisition unit 120 determines the first similarity by comparing the maximum acceleration in the second range of the impact response data with the maximum acceleration in the second range of the impact response spectrum included in the first related data 161. The second range is determined according to the protection target. In the present embodiment, the second range is determined by the information of the protection target in the same way as the first range. That is, in the present embodiment, the second range is determined based on the designation by the user. Also, the second range in the present embodiment is the same frequency range as the first range.

[0077] Specifically, in step S123d, the second range is obtained, for example, by referring to a range database in which the information of the protection target is associated with the second range in substantially the same way as the first range. Also, in step S123d, the target data acquisition unit 120 inputs the data in the second range of the impact response data acquired in step S105 and the data in the second range of each impact response spectrum included in the first related data 161 into the first calculation model 181. As a result, the feature amount of the impact response data in the second range and the feature amount of the impact response spectrum are extracted respectively. Then, by comparing the extracted feature amounts with each other, the first similarity is calculated in the second determination range.

[0078] According to the data acquisition method in the fifth embodiment described above, in the first acquisition step, the first calculation model 181 calculates the first similarity in a second range predetermined according to the protection target. Therefore, it is possible to increase the possibility of acquiring more appropriate first target data according to the first similarity.

[0079] Also, in the present embodiment, the second range is determined based on the designation by the user. Therefore, it is possible to increase the possibility of acquiring appropriate first target data according to the desired protection target in a simple method.

[0080] F. Other Embodiments: (F-1) In each of the above embodiments, the input data includes impulse response data and stress-strain data, but it may include at least either the impulse response data or the stress-strain data.

[0081] (F-2) In each of the above embodiments, the second step includes a first acquisition step and a second acquisition step, but it may include at least either the first acquisition step or the second acquisition step.

[0082] (F-3) In each of the above embodiments, in the first acquisition step, only any one of the first correspondence process, the first similarity process, and the first generation process may be executed, or only any two of the processes may be executed, or all three processes may be executed. When two or three processes are executed, the priority order in which each process is executed may be arbitrary. For example, in the first embodiment above, the first correspondence process is executed preferentially over the first generation process, but the first generation process may be executed preferentially over the first similarity process. Also, in the second embodiment above, the first correspondence process is executed preferentially over the first similarity process, but the first similarity process may be executed preferentially over the first correspondence process. Also, the first similarity process may be executed preferentially over the first generation process, or the first generation process may be executed preferentially over the first similarity process. When the priority of the first generation process is higher than that of the first correspondence process or the first similarity process, for example, an evaluation step for evaluating the acceleration waveform generated by the first generation process may be provided, and it may be determined whether to execute the first correspondence process or the first similarity process according to the evaluation result of the evaluation step. In this case, the evaluation in the evaluation step may be executed, for example, by receiving an input of the evaluation result from the user via the input device 106, or may be executed using an evaluation model for evaluating the generated acceleration waveform.

[0083] In each of the above-described embodiments, in the second acquisition step, any one of the second correspondence process, the second similarity process, and the second generation process may be executed, or any two of them may be executed, or all three processes may be executed. When two or three processes are executed, the priority order in which each process is executed may be arbitrary. For example, in the first embodiment described above, the second correspondence process is preferentially executed over the second generation process, but the second generation process may be preferentially executed over the second similarity process. Also, in the second embodiment described above, the second correspondence process is preferentially executed over the second similarity process, but the second similarity process may be preferentially executed over the second correspondence process. Further, the second similarity process may be preferentially executed over the second generation process, or the second generation process may be preferentially executed over the second similarity process. When the priority order of the second generation process is higher than that of the second correspondence process or the second similarity process, for example, an evaluation step for evaluating the shape data generated by the second generation process may be provided, and it may be determined whether to execute the second correspondence process or the second similarity process according to the evaluation result of the evaluation step. In this case, the evaluation in the evaluation step may be executed, for example, by receiving an input of the evaluation result from the user via the input device 106, or may be executed using an evaluation model for evaluating the generated acceleration waveform.

[0084] (F-5) In each of the above embodiments, the first similarity is calculated using a first calculation model 181, which is a machine learning model obtained by machine learning a plurality of acceleration waveforms and a plurality of shock response spectra as learning data. In contrast, the first similarity may be calculated using, for example, a machine learning model obtained by machine learning a plurality of shock response spectra as learning data. Even in this case, the machine learning model for calculating the first similarity can be configured, for example, as a machine learning model that extracts feature amounts of shock response data and shock response spectra. Further, the first similarity may be calculated without using a machine learning model. In this case, for example, the feature amounts of the shock response data and the feature amounts of the shock response spectra are extracted by a feature amount extraction algorithm, and the similarity may be calculated by comparing the extracted feature amounts. As the feature amount extraction algorithm, for example, various algorithms such as AKAZE (Accelerated KAZE), KAZE, SIFT (Scale-Invariant Feature Transform), and ORB (Oriented FAST and Rotated BRIEF) can be used. Similarly, the second similarity may be calculated without using a machine learning model.

[0085] (F-6) In the fourth and fifth embodiments, the first range, which is the frequency range in which the first correspondence process is executed, and the second range, which is the frequency range in which the first similarity process is executed, are determined according to the protection target. In contrast, the frequency range in which the first correspondence process is executed and the frequency range in which the first similarity process is executed do not necessarily have to be determined according to the protection target, and may be determined, for example, simply based on a designation by the user. In this case, the user may input, for example, numerical values representing the first range and the second range to the data acquisition device 100 via the input device 106.

[0086] (F-7) In the above-described fourth and fifth embodiments, the first range, which is the frequency range in which the first corresponding process is executed, and the second range, which is the frequency range in which the first similar process is executed, are determined based on the designation by the user. In contrast, the frequency range in which the first corresponding process is executed and the frequency range in which the first similar process is executed may not be determined based on the designation by the user, and may be obtained, for example, by analyzing the impulse response data included in the input data.

[0087] (F-8) In each of the above embodiments, the cushioning material CM is designed as a cushioning material for protecting the object to be protected from the impact caused by dropping. In contrast, for example, it may be designed as a cushioning material for protecting the object to be protected from an impact in any mode other than dropping, for example, an impact caused by a collision with any object or an impact caused by vibration.

[0088] G. Other forms: The present disclosure is not limited to the above-described embodiments, and can be realized in various forms without departing from the gist thereof. For example, the present disclosure can also be realized by the following forms. The technical features in the above embodiments corresponding to the technical features in each of the following forms can be appropriately replaced or combined in order to solve some or all of the problems of the present disclosure or to achieve some or all of the effects of the present disclosure. Further, if the technical feature is not described as essential in this specification, it can be appropriately deleted.

[0089] (1) According to a first aspect of the present disclosure, a data acquisition method is provided. This data acquisition method includes a first step of acquiring input data including at least one of impact response data related to an impact response spectrum and stress-strain data related to a stress-strain curve, and a second step of performing at least one of: a first acquisition step of acquiring first target data representing an acceleration waveform corresponding to the acquired impact response data by using first associated data associating (i) an acceleration waveform representing an impact acceleration of an object protected by a buffer material and (ii) the impact response spectrum of the object; and a second acquisition step of acquiring second target data representing shape data corresponding to the input stress-strain data by using second associated data associating (i) shape data of the buffer material and (ii) the stress-strain curve of the buffer material. According to this aspect, acceleration waveforms corresponding to desired impact response data and shape data corresponding to desired stress-strain data can be acquired as target data, and a buffer material can be designed by using the acquired acceleration waveforms and shape data. Therefore, a desired buffer material can be designed more simply.

[0090] (2) In the above aspect, in the first acquisition step, when a corresponding spectrum, which is the impact response spectrum corresponding to the impact response data, is included in the first associated data, the first target data is acquired by extracting the acceleration waveform associated with the corresponding spectrum from the first associated data. In the second acquisition step, when a corresponding curve, which is the stress-strain curve corresponding to the stress-strain data, is included in the second associated data, the second target data is acquired by extracting the shape data associated with the corresponding curve from the second associated data. According to this aspect, the first target data and the second target data can be easily acquired by extracting the corresponding spectrum and the corresponding curve from the first associated data and the second associated data.

[0091] (3) In the above-described embodiment, the corresponding spectrum may be an impact response spectrum in which the difference from the impact response data is equal to or less than a predetermined level within a vibration frequency range predetermined according to the object. According to this embodiment, it is possible to increase the likelihood of obtaining appropriate first target data according to the protection target.

[0092] (4) In the above-described embodiment, in the first acquisition step, using a first calculation model obtained by machine learning the acceleration waveform and the impact response spectrum as learning data, a first similarity that is the similarity between the impact response data input to the first calculation model and the impact response spectrum included in the first related data is calculated, and the acceleration waveform associated with the impact response spectrum having the first similarity equal to or greater than a predetermined similarity is extracted from the first related data to obtain the first target data. In the second acquisition step, using a second calculation model obtained by machine learning the stress-strain data and the shape data as learning data, a second similarity that is the similarity between the stress-strain data input to the second calculation model and the stress-strain curve included in the second related data is calculated, and the shape data associated with the stress-strain curve having the second similarity equal to or greater than a predetermined similarity is extracted from the second related data to obtain the second target data. According to this embodiment, for example, even when the corresponding spectrum is not included in the first related data or the corresponding curve is not included in the second related data, the first target data and the second target data can be obtained by extracting the first target data and the second target data using the first calculation model and the second calculation model.

[0093] (5) In the above-described embodiment, in the first acquisition step, the first similarity may be determined within a vibration frequency range predetermined according to the object. According to this embodiment, it is possible to increase the likelihood of obtaining appropriate first target data according to the protection target.

[0094] (6) In the above-described embodiment, the frequency range may be determined based on a designation by the user. According to this embodiment, it is possible to increase the possibility of obtaining appropriate first target data according to a desired protection target in a simple manner.

[0095] (7) In the above-described embodiment, in the first acquisition step, the first target data is obtained by generating the acceleration waveform corresponding to the impact response data using a first generation model that has been machine-learned using the first related data as learning data, and in the second acquisition step, the second target data is obtained by generating the shape data corresponding to the stress-strain data using a second generation model that has been machine-learned using the second related data as learning data. According to this embodiment, for example, even when the corresponding spectrum is not included in the first related data or when the corresponding curve is not included in the second related data, the first target data and the second target data can be obtained by generating them.

[0096] (8) In the above-described embodiment, in the first acquisition step, when the impulse response spectrum corresponding to the impulse response data is not included in the first related data, using a first calculation model obtained by machine learning with the acceleration waveform and the impulse response spectrum as learning data, a first similarity, which is the similarity between the impulse response data input to the first calculation model and the impulse response spectrum included in the first related data, is calculated, and the acceleration waveform associated with the impulse response spectrum whose first similarity is equal to or greater than a predetermined similarity is extracted from the first related data to obtain the first target data. In the second acquisition step, when the stress-strain curve corresponding to the stress-strain data is not included in the second related data, using a second calculation model obtained by machine learning with the stress-strain data and the shape data as learning data, a second similarity, which is the similarity between the stress-strain data input to the second calculation model and the stress-strain curve included in the second related data, is calculated, and the shape data associated with the stress-strain curve whose second similarity is equal to or greater than a predetermined similarity is extracted from the second related data to obtain the second target data. According to this embodiment, the first target data and the second target data can be obtained more reliably.

[0097] (9) In the above-described embodiment, in the first acquisition step, when the impulse response spectrum is not extracted by the first calculation model, using a first generation model obtained by machine learning with the first related data as learning data, the acceleration waveform corresponding to the impulse response data is generated to obtain the first target data. In the second acquisition step, when the stress-strain curve is not extracted by the second calculation model, using a second generation model obtained by machine learning with the second related data as learning data, the shape data corresponding to the stress-strain data is generated to obtain the second target data. According to this embodiment, the first target data and the second target data can be obtained even more reliably.

[0098] (10) In the above-described embodiment, in the first acquisition step, when the impulse response spectrum corresponding to the impulse response data is not included in the first related data, the first target data is acquired by generating the acceleration waveform corresponding to the impulse response data using a first generation model obtained by performing machine learning using the first related data as learning data. In the second acquisition step, when the stress-strain curve corresponding to the stress-strain data is not included in the second related data, the second target data may be acquired by generating the shape data corresponding to the stress-strain data using a second generation model obtained by performing machine learning using the second related data as learning data. According to this embodiment, the first target data and the second target data can be acquired more reliably.

[0099] (11) According to a second embodiment of the present disclosure, a data acquisition device is provided. The data acquisition device includes an input data acquisition unit that acquires input data including at least one of impulse response data related to an impulse response spectrum and stress-strain data related to a stress-strain curve, and a target data acquisition unit that executes at least one of: (i) a first acquisition process of acquiring first target data representing the acceleration waveform corresponding to the acquired impulse response data using first related data associating an acceleration waveform representing an impact acceleration of an object protected by a cushioning material with the impulse response spectrum of the object; and (ii) a second acquisition process of acquiring second target data representing the shape data corresponding to the input stress-strain data using second related data associating the shape data of the cushioning material with the stress-strain curve of the cushioning material.

[0100] (12) According to a third aspect of the present disclosure, a program is provided. This program has a function of acquiring input data including at least either impact response data regarding an impact response spectrum or stress-strain data regarding a stress-strain curve, and a function of causing a computer to execute at least one of: a first acquisition process of acquiring first target data representing an acceleration waveform corresponding to the acquired impact response data by using first associated data associating an acceleration waveform representing an impact acceleration of an object protected by a shock absorber with the impact response spectrum of the object; and a second acquisition process of acquiring second target data representing shape data corresponding to the input stress-strain data by using second associated data associating shape data of the shock absorber with the stress-strain curve of the shock absorber.

[0101] In addition to the above aspect, the present disclosure can be implemented in forms such as a shock absorber design system and a shock absorber design method.

Description of Reference Numerals

[0102] 50… Design system, 100, 100b, 100c… Data acquisition device, 101… Processor, 102, 102b, 102c… Storage device, 103… Input / output interface, 104… Internal bus, 105… Output device, 106… Input device, 110… Input data acquisition unit, 120, 120b, 120c… Target data acquisition unit, 155… Program, 161… First associated data, 162… Second associated data, 171… First generation model, 172… Second generation model, 181… First calculation model, 182… Second calculation model

Claims

1. A first step of obtaining input data including at least one of impact response data related to an impact response spectrum and stress-strain data related to a stress-strain curve; A second step of performing at least one of: a first acquisition step of obtaining first target data representing the acceleration waveform corresponding to the obtained impact response data by using first associated data associating an acceleration waveform representing an impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition step of obtaining second target data representing the shape data corresponding to the input stress-strain data by using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material. A data acquisition method.

2. The data acquisition method according to claim 1, In the first acquisition step, when a corresponding spectrum, which is the impact response spectrum corresponding to the impact response data, is included in the first associated data, the first target data is obtained by extracting the acceleration waveform associated with the corresponding spectrum from the first associated data. In the second acquisition step, when a corresponding curve, which is the stress-strain curve corresponding to the stress-strain data, is included in the second associated data, the second target data is obtained by extracting the shape data associated with the corresponding curve from the second associated data. A data acquisition method.

3. The data acquisition method according to claim 2, The corresponding spectrum is an impact response spectrum in a vibration frequency range predetermined according to the object, and the difference between the impact response spectrum and the impact response data is equal to or less than a predetermined level. A data acquisition method.

4. The data acquisition method according to claim 1, In the first acquisition step, A first calculation model obtained by machine learning using the acceleration waveform and the impact response spectrum as learning data is used to calculate a first similarity, which is the similarity between the impact response data input to the first calculation model and the impact response spectrum included in the first associated data. The first target data is obtained by extracting the acceleration waveform associated with the impact response spectrum having the first similarity equal to or greater than a predetermined similarity from the first associated data. In the second acquisition step, Using a second calculation model obtained by machine learning with the stress-strain data and the shape data as learning data, calculate a second similarity, which is the similarity between the stress-strain data input to the second calculation model and the stress-strain curve included in the second related data. A data acquisition method for obtaining the second target data by extracting the shape data associated with the stress-strain curve whose second similarity is equal to or greater than a predetermined similarity from the second related data. **Claim 5** The data acquisition method according to claim 4, In the first acquisition step, the first similarity is determined within a predetermined frequency range according to the object. **Claim 6** The data acquisition method according to claim 3 or 5, The frequency range is determined based on a designation by a user. **Claim 7** The data acquisition method according to claim 1, In the first acquisition step, the first target data is obtained by generating the acceleration waveform corresponding to the impact response data using a first generation model obtained by machine learning with the first related data as learning data. In the second acquisition step, the second target data is obtained by generating the shape data corresponding to the stress-strain data using a second generation model obtained by machine learning with the second related data as learning data. **Claim 8** The data acquisition method according to claim 2, In the first acquisition step, when the impact response spectrum corresponding to the impact response data is not included in the first related data, Using a first calculation model obtained by machine learning with the acceleration waveform and the impact response spectrum as learning data, calculate a first similarity, which is the similarity between the impact response data input to the first calculation model and the impact response spectrum included in the first related data. The first target data is obtained by extracting the acceleration waveform associated with the impact response spectrum whose first similarity is equal to or greater than a predetermined similarity from the first related data. In the second acquisition step, when the stress-strain curve corresponding to the stress-strain data is not included in the second related data, Using a second calculation model obtained by machine learning with the stress-strain data and the shape data as learning data, calculate a second similarity, which is the similarity between the stress-strain data input to the second calculation model and the stress-strain curve included in the second related data. By extracting the shape data associated with the stress-strain curve whose second similarity is equal to or greater than a predetermined similarity from the second related data, the second target data is obtained. Data acquisition method.

9. The data acquisition method according to claim 8, In the first acquisition step, when the shock response spectrum is not extracted by the first calculation model, using a first generation model obtained by machine learning with the first related data as learning data, generate the acceleration waveform corresponding to the shock response data, thereby obtaining the first target data. In the second acquisition step, when the stress-strain curve is not extracted by the second calculation model, using a second generation model obtained by machine learning with the second related data as learning data, generate the shape data corresponding to the stress-strain data, thereby obtaining the second target data. Data acquisition method.

10. The data acquisition method according to claim 2, In the first acquisition step, when the shock response spectrum corresponding to the shock response data is not included in the first related data, using a first generation model obtained by machine learning with the first related data as learning data, generate the acceleration waveform corresponding to the shock response data, thereby obtaining the first target data. In the second acquisition step, when the stress-strain curve corresponding to the stress-strain data is not included in the second related data, using a second generation model obtained by machine learning with the second related data as learning data, generate the shape data corresponding to the stress-strain data, thereby obtaining the second target data. Data acquisition method.

11. An input data acquisition unit that acquires input data including at least one of shock response data related to a shock response spectrum and stress-strain data related to a stress-strain curve. A data acquisition device comprising: an objective data acquisition unit that executes at least one of: a first acquisition process of acquiring first objective data representing an acceleration waveform corresponding to the acquired impact response data by using first associated data associating an acceleration waveform representing an impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition process of acquiring second objective data representing the shape data corresponding to the input stress-strain data by using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material. **Claim 12** A function of acquiring input data including at least one of impact response data regarding an impact response spectrum and stress-strain data regarding a stress-strain curve A program causing a computer to realize: a function of executing at least one of: a first acquisition process of acquiring first objective data representing an acceleration waveform corresponding to the acquired impact response data by using first associated data associating an acceleration waveform representing an impact acceleration of an object protected by a buffer material with the impact response spectrum of the object; and a second acquisition process of acquiring second objective data representing the shape data corresponding to the input stress-strain data by using second associated data associating the shape data of the buffer material with the stress-strain curve of the buffer material.

Images