JP7270454B2 - Design support system, design support method and design support program - Google Patents

Description

The present invention relates to a design support system, a design support method, and a design support program.

Conventionally, techniques for supporting design work have been proposed.

For example, Patent Literature 1 discloses “design support technology for generating an analysis model for CAE (Computer Aided Engineering) from CAD design data using a learning model”.

Further, Patent Literature 2 discloses “design support technology for classifying and narrowing down similar design data based on feature data extracted from design data and displaying related design information”.

JP 2019-003324 A JP 2008-176464 A

In Patent Literature 1, a data group is prepared as learning data by assigning a user's pass/fail evaluation as a teacher value to past design data and an analysis model. Machine learning of the learning model is performed using this learning data. With such a learning model, if machine learning is performed using a sufficient amount of learning data, it is easy to obtain good output results. However, in the initial stage when learning data is insufficient, machine learning is not sufficient and good output results cannot be obtained. In Patent Document 1, there is no specific disclosure regarding such a problem and its solution.

In Patent Document 2, design data is classified into groups of past similar design data by similarity search of feature data. However, a group classified as having high similarity may not be the appropriate group. In particular, in a similarity search based on feature data, there may be groups to be referred to in relationships other than the feature data, and there is a problem that the designer overlooks these groups. In Patent Document 2, there is no specific disclosure regarding such a problem and its solution.

Accordingly, an object of the present invention is to provide a new technique for supporting design.

In order to solve the above problems, one typical design support system of the present invention includes a requirement specification input unit for entering design requirement specifications, and A classification unit that obtains a group to which the requirement specification is classified (hereinafter referred to as "belonging group"), and a calculation result display unit that acquires information about the group to which it belongs and makes it possible to present it as design support information, and this calculation result display unit obtains information about a group (hereinafter referred to as "related group") into which design results other than the requirement specifications are classified for design candidates designed based on the requirement specifications, and makes it possible to present the information as design support information.

According to the present invention, a technique for supporting design is provided.
Problems, configurations, and effects other than those described above will be clarified by the following description of the embodiments.

FIG. 1 is an overall configuration diagram of the first embodiment. FIG. 2 is a diagram showing the processing procedure (phase 1) of the first embodiment. FIG. 3 is a diagram showing the processing procedure (phase 2) of the first embodiment. FIG. 4 is a diagram showing the processing procedure (phase 3) of the first embodiment. FIG. 5 is a diagram showing an example of an input screen for calculation conditions. FIG. 6 is a diagram showing an example of a display screen for analysis results. FIG. 7 is a diagram showing an example of an input screen for required specifications. FIG. 8 is a diagram showing an example of a design support display screen. FIG. 9 is a diagram showing an example of a design result input screen. FIG. 10 is a diagram showing an example of a design result display screen. FIG. 11 is a diagram showing an example of a network structure display screen without networking. FIG. 12 is a diagram showing an example of a network structure display screen when networking. FIG. 13 is a diagram showing an example of a design support display screen after networking.

Examples will be described below with reference to the drawings.

[1] Configuration of First Embodiment FIG. 1 is a diagram showing the configuration of a design support system according to a first embodiment.
In the figure, the design support system comprises a requirement specification input section 101, a classification section 102, a calculation result display section 103, a design result input section 104, and a network structure calculation section 105. FIG. More specifically, a calculation condition input unit 111 , a database 112 , a data analysis unit 113 , a learning model generation unit 114 and a machine learning unit 115 are provided inside the classification unit 102 .

The required specification input unit 101 provides the operator with input means for inputting the required design specifications. Here, "required specifications" are specifications such as items, data, and specifications that an object created by design should satisfy, and means specifications that should be specified in order to receive design support. As a specific example, the required specifications include dimensions, shape, components, characteristics, etc., but are not limited to these.

The classification unit 102 obtains a group to which the required specifications should be classified from among the groups classified by the past design histories, and sets it as a "belonging group".

The calculation result display unit 103 acquires the information of the group to which it belongs and provides it to the operator as design support information.
Based on this design support information, the operator once determines design candidates manually or automatically.

For this design candidate, the operator carries out measurement experiments on prototypes, cost calculations, specification calculations by simulation, etc., and obtains effective design results for judging the appropriateness of adopting the design candidate. At least part of this design result is data of a different type from the previous requirement specification. As a specific example, the design result is the manufacturing cost of the design candidate, the test result (hardness), the test result (tensile), etc., but is not limited to this.
The design result input unit 104 provides the operator with means for inputting these design results.

The network structure calculation unit 105 obtains groups into which the input design result should be classified, and sets them as related groups. Preferably, the network structure calculation unit 105 acquires the design results of the design history belonging to the group to which it belongs, and determines the divergence from the design results of the design candidates. The network structure calculation unit 105 obtains a group into which the diverging design results are classified, and sets it as a related group.

The calculation result display unit 103 can provide the operator with information based on related groups as design support information in addition to information based on belonging groups. Preferably, the calculation result display unit 103 creates a combined group (hereinafter referred to as a "networked group") by combining belonging groups and related groups, and presents information based on the networked group as design support information.

The calculation condition input unit 111 provides the operator with input means for accepting calculation conditions for "data items used as required specifications" and "number of groups to be classified".

In the database 112, a plurality of data sets including required specifications and design results used or collected in the past are accumulated as design history.

The data analysis unit 113 classifies a set of these data sets into a plurality of groups using a clustering technique such as a self-organizing map based on the calculation conditions (required specifications, number of groups) of the calculation condition input unit 111 .

Based on the calculation conditions (required specifications, number of groups) of the calculation condition input unit 111, the learning model generation unit 114 gives the required specifications to the input layer, and generates a learning model in the initial state with the classification result of the number of groups as the output layer. Generate.

Note that when the network structures of the belonging group and the related groups are known, the learning model generation unit 114 adds the network structures of the related groups to the output layer of the learning model.

Based on the design history in the database 112, the machine learning unit 115 creates learning data by assigning the group classification results of the data analysis unit 113 as teacher values to past required specifications.

Note that when the network structure of the affiliated group and the related group is known, the machine learning unit 115 adds the network structure of the related group to the learning data as a teacher value.
The machine learning unit 115 performs machine learning of the learning model using the created learning data.

The classification unit 102 gives the required specifications input by the required specification input unit 101 to the input of the learning model subjected to machine learning. Based on the processing result of this learning model, the classification unit 102 estimates the group to which the required specifications should be classified.

Note that when machine learning of the learning model has been completed for the network structure of the related group, the classification unit 102 estimates the network structure of the related group in addition to estimating the belonging group based on the processing result of this learning model. You can do it together.

Such a design support system is realized by a computer system 120 having a CPU (Central Processing Unit), memory, etc. as hardware. By executing the design support program by this hardware, the computer system 120 functions as various configurations of the design support system described above.

Note that part or all of the computer system 120 includes a dedicated device, general-purpose machine learning machine, DSP (Digital Signal Processor), FPGA (Field-Programmable Gate Array), GPU (Graphics Processing Unit), PLD (Programmable Logic device), etc.

Also, by centralizing or distributing part or all of the hardware to a server on the network and arranging it in the cloud, one or more client terminals may use the design support system exclusively or jointly via the network. .

[2] Operation of Embodiment 1 The operation of the design support system of Embodiment 1 configured as described above will be described.
2 to 4 are diagrams for explaining the processing procedure of the design support system divided into three phases.
The first phase shown in FIG. 2 is the input of calculation conditions necessary for machine learning and the phase of machine learning by analyzing past design information.
The second phase shown in FIG. 3 is a phase in which required specifications are input and design support is provided.
The third phase shown in FIG. 4 is a phase in which the results of the design designed in phase 2 are input, the group is networked, and additional learning of the learning model is performed accordingly.

A method of calculating the manufacturing cost of the metal material, the drawing number used, the performance, etc. will be described below, taking the manufacturing design of the metal material constituting the mechanical structure as an example.

[2-1] Processing procedure of phase 1 The operation of phase 1 will be described according to the step numbers shown in FIG.
First, in steps S101 to S104, determination of calculation conditions for performing machine learning is performed.

Step S<b>101 : The calculation condition input unit 111 acquires past design information (design history) from the database 112 .

Step S102: The calculation condition input unit 111 creates an input screen for calculation conditions required for machine learning based on the data items of the design history, and provides it to the operator.
FIG. 5 is a diagram showing an example of an input screen for calculation conditions.

For example, the operator inputs (or selects) "metallic material" as the learning model name on this input screen. Furthermore, the input screen has two configurations, the first being "selection of required specification items (variables)" and the second being "the number of groups into which the design history is divided".

In "selection of required specification item (variable)", the item name of the past design history obtained in step S101 is displayed. For example, "longitudinal dimension", "lateral dimension", "radius", "component", "melting drawing NO", "hot drawing NO", "cold drawing NO", "test result (hardness)", " "Test result (tensile)" and "Manufacturing cost" are displayed.

The operator selects requirements specification items (variables) for data analysis of the past design history.

Here, "vertical dimension", "horizontal dimension", "radius" and "component" are selected.
In "Number of Groups", enter the number of groups for grouping the past design history. For example, "10" is entered.

Step S103: The calculation condition input unit 111 acquires the information on the calculation condition input in step S102 via the input screen.

Step S<b>104 : The calculation condition input unit 111 registers the information acquired in step S<b>103 in the database 112 .

Subsequently, in steps S201 to S207, data analysis as preparation for machine learning and machine learning based on the analysis are performed.

Step S<b>201 : The data analysis unit 113 acquires the past design history from the database 112 . This design history is a set of data sets including data values of required specifications, manufacturing costs of designed products, data values of design results, and the like.

Step S202: The data analysis unit 113 acquires from the database 112 the calculation conditions necessary for machine learning input in step S103.
Here, the data analysis unit 113 acquires the item names of “vertical dimension”, “horizontal dimension”, “radius”, and “component” and the number of divisions of the group “10” as the calculation conditions.

Step S203: The data analysis unit 113 performs data analysis based on the acquired information, and groups past design histories.
A plurality of grouping methods are known as clustering techniques and are not particularly limited, but a method called a self-organizing map is used here.
Self-organizing maps are a class of machine learning techniques, such as neural networks, that model the visual cortex of the cerebral cortex.

In the self-organizing map, as many weight vectors as the number of group divisions are randomly initially arranged on the map, and one input vector is prepared based on the requirement specifications included in the design history data set. Calculate the similarity with the input vector for all weight vectors on the map. Euclidean distance is used for similarity.
Each time we find the one with the smallest distance between the vectors, we change the weight vector in the neighborhood by the following formula.

Here, Wu is the weight vector, θ is the neighborhood radius, α is the learning coefficient, U is the input vector, n is the number of iterations, and t is the processing number of the input vector.
In this way, the self-organizing map is used to divide the past design history data set into groups with high similarity of requirement specifications.

Step S204: The calculation result display unit 103 provides the operator with the analysis result of the grouping in step S203.

FIG. 6 shows an example of a display screen for grouping analysis results.
In the same figure, the past design histories of "vertical dimension", "horizontal dimension", "radius", and "component" are displayed on the display screen. Displayed in groups of 10 collected.
If the result is acceptable, the operator presses the "determine" button, and if the operator wants to start over, presses the "redo" button to review the calculation conditions from step S101.

Step S205: The learning model generation unit 114 generates a learning model before machine learning.
Although the learning model is not particularly limited, a neural network is used here.

A neural network is a mathematical model that aims to express the characteristics of the brain, which consists of a large number of nerve cells, by computer simulation. A neural network is given by the following recurrence formula, where X _i is the i-th layer composed of one or more artificial neurons.

Here, A _i and B _i are a weight parameter and a bias parameter, respectively. f is the activation function. In the case of three layers, _X1 is the input layer, _X2 is the intermediate layer, and _X3 is the output layer. A network with multiple intermediate layers is called a deep neural network.

For example, "vertical dimension", "horizontal dimension", "radius", and "component" are used as an input layer as required specifications, and the same grouping results obtained by the data analysis unit 113 are used as an output layer. A learning model with a plurality of is created.

Step S206: The machine learning unit 115 creates learning data for machine learning a learning model.

For example, by assigning the result of grouping in step S203 as a teacher value to the data values ("vertical dimension", "horizontal dimension", "radius", "component") of the required specifications extracted from the past design history, , to generate a data group for training. Learning data is generated by collecting these learning data groups.

The machine learning unit 115 gives learning data to the learning model created in step S205, performs machine learning such as the error backpropagation method, and determines a weight parameter and a bias parameter for each artificial neuron.

Step S<b>207 : The machine learning unit 115 registers the information generated in Phase 1 in the database 112 .

The learning model used by the classifier 102 is ready by the operation of Phase 1 described above.

[2-2] Processing Procedure of Phase 2 Next, the operation of Phase 2 will be described according to the step numbers shown in FIG.
First, in steps S301 to S306, design support is carried out based on the required specifications input by the operator.

Step S301: The required specification input unit 101 provides the operator with a required specification input screen. The operator determines and inputs a rough target for the desired required specifications.
FIG. 7 shows an example of an input screen for required specifications.
In the figure, the operator inputs required specifications for manufacturing metal materials. Here, the vertical dimension of the metal billet is "400.0", the horizontal dimension is "600.0", the radius is "0.0", and the components of the metal material are carbon "0.12%" and chromium "4.1%". ” and molybdenum “4.25%” are entered.

Step S302: The required specification input unit 101 acquires the required specification input in step S301.

Step S303: The classification unit 102 provides the required specifications input by the operator to the input layer of the learning model created in Phase 1. Based on the processing result of the learning model, the classification unit 102 estimates a group to which the required specifications are similar among the groups A to J. FIG.

Step S304: If the learning model has already learned the network structure (for example, related groups, groups of related groups, and their cooperative relationships, etc.), the classification unit 102 estimates the network structure based on the processing result of the learning model. . Details of this operation will be described later.
In the first stage, the learning model does not estimate the network structure because it has not learned about the network structure.

Step S305: The calculation result display unit 103 acquires from the database 112 the design support information related to the group to which it belongs estimated in step S303 and, if possible, the design support information related to the related group having a network relationship, and displays them.

FIG. 8 shows an example of a display screen of design support information.
In the figure, the belonging group having a high degree of similarity to the required specification input by the operator is assumed to be "B" here, and "belonging group: B" is displayed. Furthermore, a scatter diagram is displayed for the group to which the user belongs, with the past design results as the coordinate axes. In this scatter diagram, the data group belonging to the group "B" is plotted with, for example, the X-axis "manufacturing cost" and the Y-axis "test result (hardness)" as coordinate axes. The operator selects a plot that is close to the desired design result on this scatter diagram by a click operation or the like. The calculation result display unit 103 acquires past design history information corresponding to the selected plot from the database 112 and displays it on the display screen. In FIG. 8, melting drawing NO "A144", hot drawing NO "B247", cold drawing NO "C369", manufacturing cost "142", test result (hardness) "244", test result (tensile) " 724” is displayed. Note that the required specifications collected in S301 are also displayed. The operator once determines the required specifications of the desired design candidate by comparing or referring to the design support information and the required specifications.

Step S<b>306 : The classification unit 102 registers the information obtained in phase 2 in the database 112 .

By the operation of phase 2 described above, the design support of the first stage by the design support system and the required specifications of the design candidate by the operator are decided once.

[2-3] Processing Procedure of Phase 3 Next, the operation of Phase 3 will be described according to the step numbers shown in FIG.
In steps S401 to S409, further design support for design candidates and additional learning of network structure for learning models are performed.

Step S401: With regard to the design candidate once decided in phase 2, the operator conducts measurement experiments on prototypes, calculates costs, calculates specifications by simulation, etc., and conducts an effective process for judging the appropriateness of adoption of the design candidate. Obtain data (hereinafter referred to as “design results”). The design result input unit 104 provides the operator with means for inputting these design results.

FIG. 9 shows an example of a design result input screen.
In the same figure, the input screen has fields for inputting required specifications such as the vertical dimension, horizontal dimension, radius, and component of the design candidate, as well as melting drawing number, hot drawing number, cold drawing number, manufacturing cost, and test results (hardness). There are fields for inputting design results such as tensile strength) and test results (tensile).
Step S402: The design result input unit 104 acquires information (required specifications, design results) input by the operator on the input screen.

Step S403: The classification unit 102 inputs the input requirement specifications to the learning model, and estimates the group to which the requirement specifications of the design candidate should be classified based on the processing result of the learning model.

Step S404: If the learning model has learned the network structure, the classification unit 102 estimates the network structure based on the processing result of the learning model. Details of this operation will be described later.
At the stage of this description, the learning model does not estimate the network structure because it has not learned about the network structure.

Step S405: The calculation result display unit 103 acquires from the database 112 the design support information related to the group to which it belongs estimated in step S403, and if possible, the design support information related to the related group having a network relationship. Display with design results.

FIG. 10 shows an example of a design result display screen.
Here, redundant description of the same display elements as in FIG. 8 will be omitted.
In the scatter diagram shown in FIG. 10, in addition to the group of plots "○" corresponding to the design results of the belonging group B, plots "●" corresponding to the design results of newly input design candidates are additionally displayed. be done.

FIG. 10 shows a result (hereinafter referred to as “unexpected”) in which the manufacturing cost data values in the design results of the design candidates are far from the manufacturing cost data group of group B to which it belongs.

The network structure calculator 105 automatically detects unexpected design results from the group to which it belongs, such as manufacturing costs, by calculating distances on a scatter diagram. In this case, the calculation result display unit 103 highlights the manufacturing cost column shown in FIG. 10 with a "check mark" to notify the operator of the unexpected design result.

Also, the operator can judge unexpected design results by visually checking the scatter diagram and enter a "check mark" as in the manufacturing cost column shown in FIG. In this case, the calculation result display unit 103 acquires information on the unexpected design result determined by the operator via the display screen.

The operator determines whether or not further design support (design support by network learning) is necessary for unexpected design results, and presses the buttons on the display screen ("learn with network", "learn without network", "Cancel") selectively.

Step S406: When "learning without networking" is selected in the previous step S405, the calculation result display unit 103 displays the network structure of the group to which it belongs estimated in step S403.

FIG. 11 shows an example of a network structure display screen without networking.
Since there is no networking, only the belonging group "B" is highlighted as shown in FIG.

On the other hand, when "networking and learning" is selected in step S405, the network structure calculation unit 105 calculates the unexpected design result (manufacturing cost in FIG. 10) selected in step S405 and the design result in the group ( For example, a group having a small divergence from the average value, mode, median, deviation, etc. of manufacturing costs is specified and set as a related group. When there are a plurality of unexpected design results, the network structure calculation unit 105 may determine the divergence of the design results based on the Euclidean distance or weighted evaluation value. Note that the network structure calculation unit 105 may set one group having the smallest divergence from the unexpected design result as the related group. Further, the network structure calculation unit 105 may specify a plurality of groups whose divergence from the unexpected design result is smaller than a predetermined value, and set them as a group of related groups.

The calculation result display unit 103 displays the network structure of the affiliated group and related groups obtained in this way.

FIG. 12 shows an example of a network structure display screen for networking.
Here, the group "I" closest to the manufacturing cost, which is an unexpected design result, is calculated and the relevant group is identified as "I". The calculation result display unit 103 displays the cooperation relationship of this network structure by an arrow mark from belonging group "B" to related group "I".

At this stage, the design support screen shown in FIG. 13 may be displayed based on the processing result of the network structure calculation unit 105 .
In FIG. 13, the calculation result display unit 103 displays the information of the networked group "B, I" obtained by synthesizing the affiliated group B and the related group I as the design support information at once.

Step S407: The learning model generation unit 114 reflects the operation of design support up to the present in the learning model. (If networking is not used, steps S205 and S206 are the same as described above except that the latest learning data is used, so redundant explanations will be omitted here.)

Here, the network structure of the belonging group and related groups is known by the process of step S406. Therefore, the learning model generation unit 114 adds an artificial neuron for estimating the network structure to the existing learning model.

As a result, the learning model generation unit 114 uses the required specifications (“vertical dimension”, “horizontal dimension”, “radius”, “component”) as an input layer of the learning model. Furthermore, the learning model generation unit 114 adds the network structure of related groups (for example, any of related groups, groups of related groups, collaboration relationships, etc.) to the output layer, in addition to the output layer showing the classification result of the group to which it belongs. be the new output layer of the learning model. A network of artificial neurons is added to the intermediate layer according to the newly added output layer.

On the other hand, the machine learning unit 115 further adds the network structure of the related group as a teacher value to the learning model similar to step S206 described above.

Step S408: The machine learning unit 115 gives the latest learning data to the latest learning model, performs machine learning such as error backpropagation, and determines a weight parameter and a bias parameter for each artificial neuron.

Step S<b>409 : The machine learning unit 115 registers the information generated in phase 3 in the database 112 .

[2-4] Operation of the learning model that has learned the network structure

In Phase 3, described above, the learning model is enabled to machine-learn the network structure of the affiliate group by collecting a sufficient amount of training data to machine-learn the network structure of the affiliate group.

The operation of the learning model that has learned the network structure in this way will be described with respect to steps S304 and S305 (S404 and S405) described above.

Step S304 (S404): The classification unit 102 inputs the required specifications input to the required specification input unit 101 to the input layer of the learning model that has learned the network structure. The classification unit 102 estimates the classification of the belonging group and the network structure of related groups based on the processing results of the learning model.

Step S305 (S405): The calculation result display unit 103 acquires from the database 112 the design support information related to the group to which it belongs estimated in the previous step and the design support information based on the network structure of the related groups, and displays them.

FIG. 13 is a diagram showing an example of a display screen for design support by a networking group.
In the same figure, redundant description of display elements similar to those in FIG. 8 will be omitted.
In FIG. 13, the group to which the requirement specification input by the operator has a high degree of similarity is estimated as "B" by the learning model. Furthermore, the learning model estimates that the relevant group, which is estimated to have similar design results based on the required specifications, is "I".
The calculation result display unit 103 displays “belonging group: B” and “related group: I” as the estimation result of the learning model.

Furthermore, a scatter diagram with the past design results as the coordinate axes is displayed for the networked group "B, I" that combines the affiliated group B and the related group I. FIG. In this scatter diagram, the data group of the networked group "B, I" is plotted with, for example, the X-axis "manufacturing cost" and the Y-axis "test result (hardness)" as coordinate axes.

The operator selects a plot that is close to the desired design result on this scatter diagram by a click operation or the like. The calculation result display unit 103 acquires past design history information corresponding to the selected plot from the database 112 and displays it on the display screen.

In FIG. 13, melting drawing NO "A154", hot drawing NO "B264", cold drawing NO "C144", manufacturing cost "240", test result (hardness) "228", test result (tensile) " 710” is displayed. Note that the previously collected requirement specifications are also displayed. The operator refers to the design support information from these networking groups "B, I" and the individual requirement specifications to determine the requirement specifications of the desired design candidate from a wider range of options.

[3] Effect of Embodiment 1 (1) In Embodiment 1, the classification unit 102 obtains the group to which the required specifications are classified as the group to which it belongs, from among the groups into which the past design histories are classified. The calculation result display unit 103 makes it possible to present information based on belonging groups as design support information. Furthermore, the calculation result display unit 103 obtains information about a group (hereinafter referred to as “related group”) into which design results other than the requirement specifications are classified, for design candidates designed based on the requirement specifications, and performs the design. Make it possible to present it as supporting information.
Therefore, it is possible to utilize not only the information of the group to which the user belongs but also the information of the related group as the design support information. Therefore, in the first embodiment, it is possible to reduce oversight of the design support information by the information of the related group compared to the conventional technology, and to provide the operator with a wider range of design support information.

(2) In the first embodiment, the calculation result display unit 103 synthesizes a group to which it belongs and a related group to create a networked group, and provides information based on this networked group to the operator as design support information ( See Figure 13).
Therefore, the operator can list the design support information of the entire networked group (that is, obtain all at once) without being conscious of the difference between the group to which he belongs and the related group.

(3) In the first embodiment, the network structure calculation unit 105 obtains related groups based on the input design results. At this stage, no learning model is needed yet.
Therefore, even in the initial stage of system operation when the collection of learning data is insufficient and the learning model cannot be sufficiently machine-learned, the provision of the network structure calculation unit 105 enables design support using related groups.

(4) In the first embodiment, the network structure calculation unit 105 acquires the design results of the design history belonging to the group to which it belongs, and determines the divergence from the design results of the design candidates. The network structure calculation unit 105 obtains, as a related group, a group indicating a design result close to (that is, not diverging from) the design result determined to diverge.
Therefore, the related group is searched only when there is a design result that diverges from the group to which it belongs, and the search for the related group does not expand unnecessarily. Therefore, it becomes possible to provide the operator with more necessary design support information.

(5) In the first embodiment, the data analysis unit 113 clusters a set of data sets including past requirement specifications and design results into a plurality of groups based on the requirement specifications. The learning model generation unit 114 generates a learning model for performing this group classification. The machine learning unit 115 uses the classification result of the data analysis unit 113 as a teacher value, creates learning data with past required specifications as input values, and performs machine learning of a learning model.
Therefore, the classification unit 102 can estimate the group to which the requirement specification should be classified based on the processing result of the learning model by giving the requirement specification acquired by the requirement specification input unit 101 to the input of the learning model. become.

(6) In the first embodiment, the learning model generation unit 114 receives the required specifications as input and generates a learning model having as output the network structure of the group to which it belongs and related groups. The machine learning unit 115 further assigns the network structure of the related group by the network structure calculation unit to the learning data as a teacher value, and performs machine learning of the learning model.
Therefore, the classification unit 102 supplies the requirement specifications acquired by the requirement specification input unit 101 to the input of the learning model, thereby estimating the belonging group to which the requirement specification should be classified and the network structure of related groups. Become.
In particular, in Example 1, it is possible to estimate the network structure of related groups even when the design results are unknown. As a result, it is possible to save the operator's labor for obtaining the design result.

[Supplementary matter of the embodiment]
In the embodiment, a self-organizing map is used for grouping, but the present invention is not limited to this. For example, other clustering techniques such as K-means may be used.

Furthermore, in the embodiments, grouping, machine learning, and the like are described as being performed by a single computer, but the present invention is not limited to this. By using a network environment, a plurality of pieces of hardware or a plurality of pieces of software may share the design support work.

Also, one or more artificial neurons for estimating the design result input in step S402 may be arranged in the middle layer of the learning model. Such an artificial neuron may perform machine learning using the required specifications as an input and the design result input in step S402 as a teacher value. By arranging the artificial neurons that estimate the design result in the middle layer in this way, it becomes possible to add the estimation result of the design result to the estimation of the network structure of the related group, and the estimation accuracy of the network structure of the related group is improved. can be increased.

Furthermore, in the embodiment, the case where one related group "I" is obtained as the network structure of related groups has been described for the sake of simplicity of explanation. However, the invention is not so limited. For a group to which the user belongs, a plurality of related groups spread horizontally may be obtained as a network structure. Further, a further related group may be obtained for the related group. Furthermore, a group of related groups in a tree structure may be obtained for the group to which it belongs. It is preferable that these are appropriately selected according to the circumstances of the object to be designed.

Moreover, in the embodiment, a neural network is used as a learning model. However, the invention is not so limited. For example, such learning models include decision tree learning, association rule learning, genetic programming, inductive logic programming, support vector machines, clustering, Bayesian networks, reinforcement learning, representation learning, principal component analysis, and extreme learning machines. , and/or other machine learning techniques.

101... Required specification input unit 102... Classification unit 103... Calculation result display unit 104... Design result input unit 105... Network structure calculation unit 111... Calculation condition input unit 112... Database 113... Data analysis unit 114...Learning model generation unit, 115...Machine learning unit

Claims (12)

a requirement specification input unit for inputting the requirement specification of the design;
a classification unit that obtains the group to which the requirement specification is classified (hereinafter referred to as "belonging group") from a group of groups into which past design histories are classified;
a calculation result display unit that acquires information about the group to which it belongs and makes it possible to present it as design support information,
The calculation result display unit
Acquiring design results for design candidates designed based on the required specifications, and making it possible to present them as design support information;
For design candidates designed based on the requirement specifications, information about the group (hereinafter referred to as "related group") into which the design results other than the requirement specifications are classified can be obtained and presented as design support information. A design support system characterized by: In the design support system according to claim 1,
The calculation result display unit
A design characterized by creating a composite group (hereinafter referred to as a "networking group") by combining the belonging group and the related group, and presenting information based on the networking group as the design support information. support system. In the design support system according to any one of claims 1 and 2,
a design result input unit for inputting the design results for the design candidates designed based on the required specifications;
and a network structure calculation unit that obtains the related group into which the input design result is classified. In the design support system according to claim 3,
The network structure calculation unit
obtaining the design result of the design history belonging to the belonging group, determining the deviation from the design result of the design candidate, and classifying the group into which the design result having a deviation smaller than a predetermined value is classified as the related group; A design support system characterized by obtaining as In the design support system according to any one of claims 1 to 4,
The classification unit
a database holding, as the design history, a set of data sets including the required specifications and the design results in the past;
a data analysis unit that classifies the set of data sets into a plurality of the groups to which they belong based on the required specifications by a clustering technique such as a self-organizing map;
a learning model generation unit that generates a learning model that receives the required specifications as an input and outputs the classification of the group to which it belongs;
a machine learning unit that performs machine learning of the learning model by creating learning data in which the classification of the group to which the data analysis unit belongs is given as a teacher value for the required specifications of the data set,
providing the requirement specifications acquired by the requirement specification input unit to the input of the learning model, and estimating the group to which the requirement specification belongs based on the processing result of the learning model. system. In the design support system according to claim 5,
The learning model generation unit receives the required specifications as an input and generates the learning model having as an output a network structure of the belonging group and the related group,
The machine learning unit further assigns the network structure of the related group as a teacher value to the learning data to perform machine learning of the learning model,
The classification unit supplies the requirement specifications acquired by the requirement specification input unit to the input of the learning model, and based on the processing result of the learning model, the belonging group and the related group to which the requirement specification is classified. A design support system characterized by estimating a network structure. A design support method using a design support system comprising a requirement specification input unit, a classification unit, and a calculation result display unit,
a requirement specification input step for the requirement specification input unit to input the requirement specification of the design;
a classification step in which the classification unit obtains the group to which the requirement specification is classified (hereinafter referred to as "belonging group") from among groups classified into past design histories;
a calculation result display step in which the calculation result display unit acquires information about the group to which it belongs and enables presentation as design support information;
The calculation result display step includes:
Acquiring design results for design candidates designed based on the required specifications, and making it possible to present them as design support information;
For design candidates designed based on the requirement specifications, information about the group (hereinafter referred to as "related group") into which the design results other than the requirement specifications are classified can be obtained and presented as design support information. A design support method characterized by: In the design support method according to claim 7,
The calculation result display step includes:
A design characterized by creating a composite group (hereinafter referred to as a "networking group") by combining the belonging group and the related group, and presenting information based on the networking group as the design support information. how to help. In the design support method according to any one of claims 7 and 8,
The design support system includes a design result input unit and a network structure calculation unit,
a design result input step for the design result input unit to input the design result for a design candidate designed based on the required specifications;
and a network structure calculation step in which the network structure calculation unit obtains the related group into which the input design result is classified. In the design support method according to any one of claims 7 to 9,
The classification unit comprises a database, a data analysis unit, a learning model generation unit, and a machine learning unit,
The classification step includes:
a data holding step in which the database holds, as the design history, a set of data sets including the past requirement specifications and the design results;
a data analysis step in which the data analysis unit classifies the set of data sets into a plurality of the groups to which they belong based on the required specifications by a clustering technique such as a self-organizing map;
a learning model generation step in which the learning model generating unit generates a learning model having the required specifications as an input and the classification of the group to which the learning model belongs as an output;
The machine learning unit creates learning data in which the classification of the group to which the data analysis step belongs is given as a teacher value for the required specifications of the data set, and performs machine learning of the learning model. a step and
providing the requirement specifications acquired in the requirement specification input step as an input to the learning model, and estimating the group to which the requirement specification belongs based on the processing result of the learning model. Method. In the design support method according to claim 10,
The learning model generating step generates the learning model with the required specifications as an input and a network structure of the belonging group and the related group as an output,
The machine learning step further assigns the network structure of the related group as a teacher value to the learning data to perform machine learning of the learning model,
The classification step supplies the requirement specifications acquired by the requirement specification input step to the input of the learning model, and determines the belonging group and the related group to which the requirement specification is classified based on the processing result of the learning model. A design support method characterized by estimating a network structure. A design support program that causes a computer system to function as the design support system according to any one of claims 1 to 6.

Images