Design support device, design support method, and design support program

Abstract

To easily set target values of characteristic items constituting an objective variable in a manufacturing process of a product, or the like.SOLUTION: A design assistance device 10 is provided with: a data acquisition unit 11 that acquires multiple pieces of performance data each comprising a design parameter group and observation values of a plurality of characteristic items of a product or the like; a model construction unit 12 that constructs, on the basis of the performance data, a prediction model for predicting the observation values of the characteristic items on the basis of the design parameter group; a target value acquisition unit 13 that acquires target values for the observation values of the characteristic items; a score function construction unit 15 for constructing a score function including at least a target attainment probability term indicating an overall achievement probability that is a probability at which the target values of all the characteristic items are achieved and which is calculated using the design parameter group as a variable on the basis of the prediction model; a target attainment probability acquisition unit 16 that acquires the overall attainment probability by optimizing the score function; and an output unit 17 that outputs the overall attainment probability.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] One aspect of this disclosure relates to a design support device, a design support method, and a design support program. [Background technology]

[0002] Product design utilizing machine learning is being researched. In one area of ​​product design, for example, in the design of functional materials, a model is constructed to estimate the properties of a material using machine learning with training data consisting of pairs of raw material mixing ratios and properties for experimentally and already fabricated materials, and the properties for unexperimented raw material mixing ratios are predicted. By planning experiments based on such property predictions, it becomes possible to efficiently optimize parameters such as material properties and raw material mixing ratios, thereby improving development efficiency. Furthermore, Bayesian optimization is known to be an effective method for such optimization, and design devices that output design values ​​using Bayesian optimization are known (see, for example, Patent Document 1). [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-52737 [Overview of the project] [Problems that the invention aims to solve]

[0004] On the other hand, in the development of materials and other products, target values ​​are set for observed values ​​of characteristic items that indicate the characteristics of the product. Setting appropriate target values ​​in product development is important for efficiently advancing product development, deciding whether to continue or discontinue development, and making decisions such as changing development policies. However, when a product has many characteristic items and there are many target values ​​to set for those characteristic items, setting appropriate target values ​​has not been easy.

[0005] Therefore, the present invention has been made in view of the above problems, and aims to easily set target values ​​for characteristic items constituting the objective variable in the manufacturing process of products, work in progress, semi-finished products, parts, or prototypes. [Means for solving the problem]

[0006] A design support device relating to the first aspect of this disclosure is a design support device that assists in setting target values ​​for observed values ​​of multiple characteristic items that indicate the characteristics of a product, work in progress, semi-finished product, part, or prototype manufactured based on a group of design parameters, and comprises: a data acquisition unit that acquires multiple actual data sets consisting of a group of design parameters and observed values ​​of each of the multiple characteristic items for a manufactured product, work in progress, semi-finished product, part, or prototype; a model construction unit that constructs a prediction model based on the actual data that predicts the observed values ​​of the characteristic items as a probability distribution or an approximation or substitute index therefor based on the group of design parameters; a target value acquisition unit that acquires target values ​​for each observed value of the characteristic item; a score function construction unit that constructs a score function that includes at least a target achievement probability term, which is calculated using the group of design parameters as variables based on the prediction model and represents the overall achievement probability, which is the probability that the target values ​​of all characteristic items are achieved; a target achievement probability acquisition unit that acquires the overall achievement probability by optimizing the score function; and an output unit that outputs the overall achievement probability.

[0007] A design support method relating to the first aspect of this disclosure is a design support method in a design support device that supports setting target values ​​for observed values ​​of multiple characteristic items that indicate the characteristics of a product, work in progress, semi-finished product, part, or prototype manufactured based on a set of design parameters, comprising: a data acquisition step of acquiring multiple actual data sets consisting of a set of design parameters and observed values ​​of each of the multiple characteristic items for a manufactured product, work in progress, semi-finished product, part, or prototype; a model construction step of constructing a prediction model based on the actual data that predicts the observed values ​​of the characteristic items as a probability distribution or an approximation or substitute index therefor based on the set of design parameters; a target value acquisition step of acquiring target values ​​for the observed values ​​of each characteristic item; a score function construction step of constructing a score function that includes at least a target achievement probability term, which is the probability that the target values ​​of all characteristic items are achieved, calculated using the set of design parameters as variables based on the prediction model; a target achievement probability acquisition step of acquiring the overall achievement probability by optimizing the score function; and an output step of outputting the overall achievement probability.

[0008] A design support program according to a first aspect of the present disclosure is a design support program for causing a computer to function as a design support apparatus that supports setting of target values for observed values of a plurality of characteristic items indicating characteristics of finished products, work-in-progress, semi-finished products, parts, or prototypes produced based on a design parameter group consisting of a plurality of design parameters, the program causing the computer to implement: a data acquisition function of acquiring a plurality of pieces of performance data each consisting of the design parameter group and observed values of each of the plurality of characteristic items for produced finished products, work-in-progress, semi-finished products, parts, or prototypes; a model construction function of constructing a prediction model that predicts an observed value of a characteristic item as a probability distribution, an approximation thereof, or an alternative index based on the design parameter group, on the basis of the performance data; a target value acquisition function of acquiring target values for observed values of each characteristic item; a score function construction function of constructing a score function including at least a target achievement probability term that represents an overall achievement probability, which is a probability that the target values of all characteristic items are achieved when calculated with the design parameter group as variables based on the prediction model; a target achievement probability acquisition function of acquiring the overall achievement probability through optimization of the score function; and an output function of outputting the overall achievement probability.

[0009] According to this aspect, a prediction model is constructed based on performance data, target values for each characteristic item are acquired, and a score function including an overall achievement probability term is constructed based on the acquired target values and the prediction model. Then, through optimization of the score function, the overall achievement probability is acquired and output. Therefore, by referring to the output overall achievement probability, it can be determined whether the target values are appropriate, which facilitates setting of appropriate target values.

[0010] In the design support apparatus according to a second aspect, in the design support apparatus according to the first aspect, the target value acquisition unit may acquire a target value for at least one characteristic item among the plurality of characteristic items based on an input by a user.

[0011] From this perspective, users can easily set target values ​​for each characteristic item. Therefore, by repeatedly inputting target values ​​and outputting the overall achievement probability while changing the input target values, it becomes possible to easily set appropriate target values.

[0012] In the design support device relating to the third aspect, the target value acquisition unit in the design support device relating to the first or second aspect may acquire a target value for at least one of the multiple characteristic items from a predetermined database that stores the target values ​​of the characteristic items in advance.

[0013] From this perspective, the overall achievement probability can be output using pre-stored target values ​​for each characteristic item. This makes it possible to reduce the effort involved in inputting target values, for example, when there are many characteristic items and the acquisition of target values ​​is based on user input. Furthermore, if a set of target values ​​with a proven track record as appropriate target values ​​is pre-stored in the database, it becomes easier to set appropriate target values ​​by acquiring that set of target values ​​as the initial target values ​​and modifying some of the target values. In addition, it becomes easier to set appropriate targets by pre-storeing proven observed values ​​for each characteristic item in the database and using those proven values ​​as the initial target values.

[0014] In the design support device relating to the fourth aspect, the output unit may output the target value acquired by the target value acquisition unit along with the overall achievement probability in the design support device relating to any one of the first to third aspects.

[0015] From this perspective, the target values ​​of the characteristic items in the optimization of the score function and the overall achievement probability obtained through optimization are output together, for example, in a display manner, making it easy to recognize the correspondence between the target value group and the overall achievement probability.

[0016] In the design support device relating to the fifth aspect, in the design support device relating to the fourth aspect, the target value acquisition unit acquires the range of target values ​​for the observed value of at least one characteristic item among a plurality of characteristic items, the score function construction unit constructs a score function corresponding to each target value within the range of target values, the target achievement probability acquisition unit acquires the overall achievement probability corresponding to each target value within the acquired range of target values ​​by performing optimization of each score function, and the output unit outputs the overall achievement probability corresponding to each target value within the set range of target values ​​in association with each other.

[0017] From this perspective, by referring to the correlation between each target value and the overall achievement probability, it is possible to recognize the change in the overall achievement probability in response to changes in the target value of the characteristic item. Therefore, it becomes easier to set appropriate target values.

[0018] In the design support device relating to the sixth aspect, in the design support device relating to any one of the first to fifth aspects, the target achievement probability acquisition unit may acquire the achievement probability of the target value for each characteristic item, and the output unit may output the achievement probability of the target value for each characteristic item together with the overall achievement probability.

[0019] From this perspective, by outputting the probability of achieving the target value of each characteristic item, for example, through a display, it becomes possible to identify the target value of the characteristic item that is a bottleneck in achieving a high overall success probability. Therefore, it becomes easier to improve the overall success probability by adjusting that target value.

[0020] In the design support device relating to the seventh aspect, the design support device relating to any one of the first to sixth aspects further comprises an optimization range acquisition unit that acquires an optimization range, which is the search range in the optimization of the score function for at least one of the design parameters included in the design parameter group, wherein the optimization range of the design parameter is defined by at least one of the upper limit, lower limit, upper and lower limits, and fixed value of the value of the design parameter, and the target achievement probability acquisition unit may perform the optimization of the score function based on the acquired optimization range.

[0021] From this perspective, it is possible to obtain the overall probability of achieving the design parameters within the optimization range. Therefore, for example, by resetting the optimization range multiple times while checking the overall probability of achieving it, it becomes possible to obtain a suitable range of design parameters for product development.

[0022] In the design support device relating to the eighth aspect, in the design support device relating to any one of the first to seventh aspects, the model construction unit may construct a predictive model for at least one of the multiple characteristic items based on the input of parameters constituting the predictive model.

[0023] From this perspective, part of the calculation of the probability that the target value of a characteristic item will be achieved will not depend on actual data. Therefore, for example, in cases where sufficient actual data is unavailable, it is possible to prevent a decrease in the accuracy of the overall achievement probability caused by a deterioration in the estimation accuracy of the distribution of observed values ​​of the characteristic item in the prediction model.

[0024] In the design support device relating to the ninth aspect, in the design support device relating to any one of the first to eighth aspects, the target achievement probability acquisition unit may acquire a set of design parameters that maximize the overall achievement probability by optimizing the score function, and the output unit may output a set of design parameters that maximize the overall achievement probability.

[0025] From this perspective, by referring to the outputted set of design parameters, it is possible to verify whether the set of design parameters is appropriate for achieving the outputted overall success probability. Furthermore, by referring to the outputted set of design parameters, it is possible to manufacture products and conduct experiments. [Effects of the Invention]

[0026] According to one aspect of this disclosure, it becomes possible to easily set target values ​​for characteristic items that constitute the objective variable in the manufacturing process of products, work in progress, semi-finished products, parts, or prototypes. [Brief explanation of the drawing]

[0027] [Figure 1] This figure shows an overview of the material design process to which the design support device according to this embodiment is applied. [Figure 2] This is a block diagram showing an example of the functional configuration of a design support device according to the embodiment. [Figure 3] This is a hard block diagram of the design support device according to the embodiment. [Figure 4] This figure shows an example of a set of design parameters for a fabricated material. [Figure 5] This figure shows examples of observed values ​​for characteristic parameters related to fabricated materials. [Figure 6] This figure shows examples of target values ​​for the acquired characteristic items. [Figure 7] This diagram schematically shows an example of a display screen (user interface) for inputting target values ​​and outputting the overall achievement probability. [Figure 8] This figure shows an example of the optimization range of the acquired design parameters. [Figure 9] This is a flowchart showing an example of the contents of the design support method in the design support device according to the embodiment. [Figure 10] This diagram schematically illustrates an example of a display screen (user interface) that shows the relationship between the target value and the overall probability of achieving it. [Figure 11] This diagram shows the structure of the design support program. [Modes for carrying out the invention]

[0028] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant descriptions are omitted.

[0029] Figure 1 shows an overview of the material design and development process as an example of the design and development process for products, work-in-progress, semi-finished products, parts, or prototypes to which the design support device according to this embodiment is applied. In the following, "products, work-in-progress, semi-finished products, parts, or prototypes" will be referred to as "products, etc." The design support device 10 of this embodiment can be applied to the design and development process of any product, etc. that has multiple characteristic items indicating the characteristics of the product, etc., and target values ​​for each characteristic item. Specifically, the design support device 10 can be applied to setting suitable target values ​​for characteristic items in a flow consisting of determining design parameters (explanatory variables), manufacturing the product, etc. based on the determined design parameters, and acquiring observed values ​​(dependent variables) of the characteristic items of the product, etc. In addition to material development and design, the design support device 10 can be applied to the design of products such as automobiles and pharmaceuticals, and the optimization of the molecular structure of pharmaceuticals. Furthermore, the design support device 10 of this embodiment can be applied as a communication tool. Specifically, the design support device 10 can be used to reach an agreement between the manufacturer of a product and the client (customer) requesting the manufacture of the product, by referring to the target values ​​of characteristic items output by the design support device 10 and the probability of achieving those target values. In this embodiment, as described above, the design support process by the design support device 10 will be explained using the example of material design and development as an example of product design and development.

[0030] As shown in Figure 1, the design support processing by the design support device 10 is applied, as an example, to the fabrication and experimentation of materials in the plant and laboratory A. That is, materials are fabricated in the plant and laboratory A according to the set design parameter group x, and based on the fabricated materials, observed values ​​y of multiple characteristic items indicating the properties of the materials are obtained. Note that the fabrication and experimentation of materials in the plant and laboratory A may be performed by simulation.

[0031] The design support device 10 assists in setting target values ​​based on actual data consisting of a design parameter group x and observed values ​​y of multiple characteristic items of a material manufactured based on the design parameter group x. Specifically, the design support device 10 constructs a prediction model that predicts the observed values ​​of characteristic items based on the actual data consisting of the design parameter group x and observed values ​​y, constructs a score function that includes an overall achievement probability term that represents the overall achievement probability that the target values ​​of all characteristic items are achieved, calculated based on the prediction model, and obtains and outputs the overall achievement probability by optimizing the score function.

[0032] For example, in the design and development of a material product, when a material is manufactured by mixing multiple polymers and additives, the design support device 10 is used to set target values ​​for multiple characteristic items, using a group of design parameters such as the blending amounts of each polymer and additive as design variables, and observed values ​​of characteristic items such as elastic modulus and thermal expansion coefficient as objective variables.

[0033] Figure 2 is a block diagram showing an example of the functional configuration of a design support device according to this embodiment. The design support device 10 of this embodiment is a device that assists in setting target values ​​for observed values ​​of multiple characteristic items that indicate the properties of a material manufactured based on a group of design parameters consisting of multiple design parameters. As shown in Figure 2, the design support device 10 may include a functional unit configured in the processor 101, a design parameter storage unit 21, an observed value storage unit 22, a target value storage unit 23, and an optimization range storage unit 24. Each functional unit will be described later.

[0034] Figure 3 shows an example of the hardware configuration of the computer 100 that constitutes the design support device 10 according to this embodiment. Note that the computer 100 can constitute the design support device 10.

[0035] As an example, computer 100 includes a processor 101, main memory 102, auxiliary storage 103, and communication control device 104 as hardware components. Computer 100, which constitutes the design support device 10, may further include input devices such as a keyboard, touch panel, and mouse 105, and output devices such as a display 106.

[0036] The processor 101 is a computing unit that executes the operating system and application programs. Examples of processors include CPUs (Central Processing Units) and GPUs (Graphics Processing Units), but the type of processor 101 is not limited to these. For example, the processor 101 may be a combination of a sensor and a dedicated circuit. The dedicated circuit may be a programmable circuit such as an FPGA (Field-Programmable Gate Array), or it may be another type of circuit.

[0037] The main memory 102 is a device that stores programs for realizing the design support device 10, calculation results output from the processor 101, and the like. The main memory 102 is composed of, for example, at least one of ROM (Read Only Memory) and RAM (Random Access Memory).

[0038] The auxiliary storage device 103 is generally a device capable of storing a larger amount of data than the main memory 102. The auxiliary storage device 103 is composed of a non-volatile storage medium such as a hard disk or flash memory. The auxiliary storage device 103 stores the design support program P1 and various data necessary for the computer 100 to function as a design support device 10, etc.

[0039] The communication control device 104 is a device that performs data communication with other computers via a communication network. The communication control device 104 is composed of, for example, a network card or a wireless communication module.

[0040] Each functional element of the design support device 10 is realized by loading the corresponding program P1 onto the processor 101 or main memory 102 and having the processor 101 execute the program. Program P1 contains code for realizing each functional element of the corresponding server. The processor 101 operates the communication control device 104 according to program P1 and performs data reading and writing in the main memory 102 or auxiliary storage device 103. Through such processing, each functional element of the corresponding server is realized.

[0041] Program P1 may be provided permanently recorded on a tangible recording medium such as a CD-ROM, DVD-ROM, or semiconductor memory. Alternatively, at least one of these programs may be provided via a communication network as a data signal superimposed on a carrier wave.

[0042] Referring again to Figure 2, the design support device 10 comprises a data acquisition unit 11, a model construction unit 12, a target value acquisition unit 13, an optimization range acquisition unit 14, a score function construction unit 15, a target achievement probability acquisition unit 16, and an output unit 17. The design parameter storage unit 21, the observed value storage unit 22, the target value storage unit 23, and the optimization range storage unit 24 may be configured in the design support device 10 as shown in Figure 2, or they may be configured as other devices accessible from the design support device 10.

[0043] The data acquisition unit 11 acquires multiple sets of actual data for the fabricated material. The actual data consists of pairs of design parameter sets and observed values ​​for multiple characteristic items. The design parameter storage unit 21 is a storage means that stores the design parameter sets in the actual data, and may be configured as, for example, a main memory device 102 and an auxiliary memory device 103. The observed value storage unit 22 is a storage means that stores the observed values ​​in the actual data.

[0044] FIG. 4 is a diagram illustrating an example of a design parameter group stored in a design parameter storage unit 21. As shown in FIG. 4, the design parameter storage unit 21 stores a design parameter group x for t-th material fabrication (t=1 to T) t The design parameter group x includes, for example, P design parameters v from 1 to P p (p=1 to P). As an example, the design parameter group x may include a blending amount v1 of a raw material A, a blending amount v2 of a raw material B, ..., a design parameter v P and can constitute vector data of a dimension P corresponding to the number of design parameters. In addition to the examples illustrated, the design parameters may be, for example, non-vector data such as molecular structures and images. Further, the design parameters may be data indicating an option among a plurality of molecules.

[0045] FIG. 5 is a diagram illustrating an example of observed values y stored in an observed value storage unit 22. As shown in FIG. 5, the observed value storage unit 22 stores observed values y for a plurality of characteristic items m (m=1 to M) indicating characteristics of a material in t-th material fabrication (t=1 to T) m,t As an example, the characteristic item m includes a glass transition temperature y 1,t , an adhesive strength y 2,t and a characteristic item M (y M,t ). A pair of the design parameter group x t and the observed value y m,t constitutes performance data.

[0046] A model construction unit 12 constructs a prediction model based on the performance data. The prediction model is a model that predicts an observed value y of a characteristic item m as a probability distribution, an approximation thereof, or an alternative index based on the design parameter group x m Any model that can predict the observed value y as a probability distribution, an approximation thereof, or an alternative index may be used to constitute the prediction model, and the type thereof is not limited. The observed value y m Any model that can predict the observed value y as a probability distribution, an approximation thereof, or an alternative index may be used, and the type of model is not limited. The observed value y mPredictive models that use the probability distribution as a surrogate index predict the probability distribution of observed values, for example, by using the distribution of predicted values ​​of predictors constituting an ensemble (random forest), the distribution obtained by Monte Carlo dropout (neural network), or the distribution of predictions of multiple predictors constructed under different conditions (arbitrary machine learning method) as a surrogate index.

[0047] For example, a prediction model takes a set of design parameters x as input and observed values ​​y m The model may be a regression model that outputs a probability distribution. If the prediction model is a regression model, it may be composed of one of the following regression models: Gaussian process regression, random forest, and neural network. The model building unit 12 may build the prediction model using a well-known machine learning method using actual data. The model building unit 12 may also build the prediction model using a machine learning method that applies actual data to the prediction model and updates the parameters of the prediction model.

[0048] Furthermore, the prediction model may be a machine learning model that predicts the probability distribution of observed values ​​or its approximation or alternative indicator using one of the following: the posterior distribution of predicted values ​​based on Bayesian theory, the distribution of predicted values ​​of the predictors constituting the ensemble, the theoretical formulas for the prediction interval and confidence interval of the regression model, Monte Carlo dropout, and the distribution of predictions of multiple predictors constructed under different conditions. The prediction of the probability distribution of observed values ​​or its alternative indicator can be obtained by a model-specific method. The probability distribution of observed values ​​or its approximation or alternative indicator can be obtained based on the posterior distribution of predicted values ​​for Gaussian process regression and Bayesian neural networks, based on the distribution of predictions of the predictors constituting the ensemble for random forests, based on the prediction interval and confidence interval for linear regression, and based on Monte Carlo dropout for neural networks. However, the method for calculating the distribution of observed values ​​or its alternative indicator for each machine learning model is not limited to the above methods.

[0049] Furthermore, any model may be extended to a model that can predict the probability distribution of observed values ​​or an alternative indicator thereof. For example, one such model uses the distribution of predicted values ​​obtained by constructing multiple datasets using methods such as bootstrapping, and then constructing a predictive model for each dataset, as an alternative indicator of the probability distribution of observed values. However, the method of extending a machine learning model to a model that can predict the probability distribution of observed values ​​or an alternative indicator thereof is not limited to the methods described above.

[0050] Furthermore, the prediction model may be constructed using methods such as linear regression, PLS regression, Gaussian process regression, bagging ensemble learning such as random forest, boosting ensemble learning such as gradient boosting, support vector machines, and neural networks.

[0051] In a predictive model constructed as a Gaussian process regression, the probability distribution of observed values ​​is predicted by inputting the design parameter set x from the actual data that constitutes the explanatory variables of the training data, the observed values ​​y that constitute the dependent variable, and the design parameter x of the target to be predicted into the model.

[0052] Furthermore, the model building unit 12 may tune the hyperparameters of the prediction model using well-known hyperparameter tuning methods. That is, the model building unit 12 may update the hyperparameters of the prediction model constructed by Gaussian process regression by maximum likelihood estimation using a vector representing the design parameter group x, which is the explanatory variable in the actual data, and the observed value y, which is the dependent variable.

[0053] Furthermore, the prediction model may be constructed using a classification model. If the prediction model is a classification model, the model construction unit 12 can construct the prediction model using a machine learning method that is capable of evaluating a well-known probability distribution using actual data.

[0054] In this way, the model building unit 12 constructs a predictive model using a predetermined regression model or classification model, making it possible to obtain the probability distribution of observed values ​​of characteristic items based on an arbitrary set of design parameters x.

[0055] Furthermore, the prediction model may be a single-task model that predicts the observed value of one characteristic item as a probability distribution, its approximation, or an alternative indicator, or a multi-task model that predicts the observed values ​​of multiple characteristic items as probability distributions, their approximations, or alternative indicators. In this way, by constructing a prediction model using a multi-task model or single-task model appropriately configured according to the nature of the characteristic items, the accuracy of the prediction of observed values ​​by the prediction model can be improved.

[0056] The model building unit 12 may build a prediction model for at least one of several characteristic items based on the input of parameters that constitute the prediction model. Specifically, the model building unit 12 may build the prediction model based on the input of parameters that constitute the prediction model and contribute to probability calculation.

[0057] For example, if the prediction model is configured to predict observed values ​​of a characteristic item that follows a Gaussian distribution, the model building unit 12 may obtain the standard deviation or variance of the Gaussian distribution, which is a parameter representing the distribution of the observed values, based on user input or the like, and build a prediction model for the characteristic item based on the obtained standard deviation or variance.

[0058] In this way, by inputting parameters that constitute the predictive model and contribute to probability calculation, a portion of the probability calculation becomes independent of actual data. Therefore, for example, in cases where there is insufficient actual data in terms of quantity and accuracy, it is possible to prevent a decrease in the accuracy of the overall achievement probability caused by a deterioration in the estimation accuracy of the observed values ​​of characteristic items in the predictive model.

[0059] The target value acquisition unit 13 acquires target values ​​for the observed values ​​of each characteristic item. Figure 6 shows the target values ​​y of each characteristic item stored in the target value storage unit 23.m(target) This is an example shown in the figure. The target value storage unit 23 is a storage means that stores target values ​​input by the user via a predetermined user interface, or target values ​​that have been set in advance.

[0060] As shown in Figure 6, the target value storage unit 23 stores the target value y of characteristic item m (m=1~M). m(target) It remembers. Target value y m(target) For example, the target value y for the glass transition temperature. 1(target) , target value y of adhesive strength 2(target) and the target value y of characteristic item M M(target) It may include.

[0061] The target value acquisition unit 13 may acquire a target value for at least one of the multiple characteristic items based on user input.

[0062] Figure 7 schematically shows an example of a display screen (user interface) for inputting target values ​​and outputting the overall achievement probability. As shown in Figure 7, the design support device 10 can display the user interface G on a display or the like, which is one aspect of the output device 106. The user interface G displays the target value y m(target) It includes an operation object gt for input. Specifically, the user interface G contains the target value y for each of the characteristic items 1, 2, 3, ..., M. 1(target) ,y 2(target) ,y 3(target) ,···,y M(target) It includes operation objects gt1, gt2, gt3, ...gtm for inputting values. The target value acquisition unit 13 acquires the target value of the characteristic item based on the user's operation input of the operation object gt.

[0063] This method of obtaining target values ​​allows users to easily set target values ​​for each characteristic item. Therefore, by repeatedly inputting target values ​​and outputting the overall achievement probability while changing the input target values, it becomes possible to easily set appropriate target values. The objects ga and gb included in the user interface G will be described later.

[0064] Furthermore, the target value acquisition unit 13 may acquire a target value for at least one of the multiple characteristic items from a predetermined database that has previously stored the target values ​​of the characteristic items.

[0065] For example, a group of target values ​​that have proven to be appropriate may be stored in the target value storage unit 23 in advance, and the target value acquisition unit 13 may acquire the group of proven target values ​​from the target value storage unit 23. By storing an appropriate group of target values ​​in advance in this way, an initial group of target values ​​can be easily acquired when setting and considering target values.

[0066] Alternatively, the target value acquisition unit 13 may acquire the actual values ​​of each characteristic item from the target value acquisition unit 23, after which the actual values ​​of each characteristic item have been pre-stored in the target value acquisition unit 23. By applying the actual values ​​to the initial target values ​​in this way, it becomes easier to set appropriate targets.

[0067] Furthermore, by using pre-stored target values ​​for each characteristic item as initial target values, for example, when there are many characteristic items, the effort involved in obtaining target values ​​based on user input becomes cumbersome, but this process can be streamlined.

[0068] The optimization range acquisition unit 14 acquires the optimization range for at least one design parameter from among the design parameters included in the design parameter group. The optimization range is the search range for each design parameter, which is an explanatory variable in the optimization of the score function.

[0069] The optimization range storage unit 24 stores the design parameter v pThis is a storage means that stores the optimization range. Figure 8 shows the design parameters v stored in the optimization range storage unit 24. p This figure shows an example of the optimization range. As shown in Figure 8, the optimization range storage unit 24 stores each design parameter v p The optimization range is stored. Design parameter v p The optimization range can be set in cases other than when the possible values ​​of the design parameters as explanatory variables are limited to a finite number of candidates.

[0070] Design parameter v p The optimization range of the design parameter is defined by at least one of the following: an upper limit, a lower limit, upper and lower limits, and a fixed value. It may be defined by a combination of these. As shown in Figure 8, the optimization range of design parameter v1 is defined by its upper and lower limits and is set to "0-10". The optimization range of design parameter v2 is defined by a fixed value and is set to "5". The optimization range of design parameter v3 is defined by two upper and lower limits and is set to "0-10, 30-49". P The optimization range is defined by two fixed values, which are set to "2 or 3".

[0071] If there is a desired value for the design parameter v, then that design parameter v p By defining the optimization range of the design parameter v by a fixed value, p Since the optimization is performed with the parameters fixed, the overall probability of achieving the desired design parameters can be obtained as a result of the optimization.

[0072] As a more specific example, consider the design parameter v, which represents the quantity of expensive materials. p By setting the optimization range to "0", it becomes possible to perform optimization under the condition that the material is not used. Furthermore, a certain design parameter v p If, after performing optimization with the optimization range set to a fixed value, a low probability of achieving the target is obtained, then the design parameter v pBy changing the optimization range to, for example, upper and lower limits with a certain range, and performing optimization again to evaluate the probability of achieving the target, it is possible to easily consider a suitable optimization range.

[0073] The optimization range storage unit 24 stores each design parameter v input by the user via the input device 105. p The optimization range may be stored. The optimization range acquisition unit 14 stores each design parameter v input by the user. p The optimization range can be obtained. In this case, the user's design parameter v p The optimization range can be easily set and changed.

[0074] In this way, the score function is optimized within the optimization range of the acquired design parameters, and the overall achievement probability within the specified optimization range is obtained. Therefore, for example, by setting the optimization range to a range that can be easily adopted as design parameters, it becomes possible to obtain the overall achievement probability within the range of easily adoptable design parameters. Furthermore, if the overall achievement probability within the range of easily adoptable design parameters is low, by repeatedly changing the range of design parameters and checking the overall achievement probability, it becomes possible to obtain a range of design parameters that has a high probability of achieving the target and can be adopted relatively easily.

[0075] The score function construction unit 15 constructs a score function that includes at least an overall achievement probability term. The overall achievement probability term represents the overall achievement probability, which is the probability that the target values ​​of all characteristic items are achieved. The overall achievement probability is calculated using the design parameter set as variables based on the prediction model. The structure of the score function is not limited as long as it includes an overall achievement probability term.

[0076] Specifically, the score function construction unit 15 constructs a score function A(x) as shown in equation (1) below. A(x) = g(P(x)) ... (1) In equation (1), g(P(x)) is the target achievement probability term. That is, the score function A(x) includes at least the target achievement probability term g(P(x)). Note that the score function in equation (1) is just one example, and the score function may include terms other than the target achievement probability term.

[0077] The goal achievement probability term can be constructed to include the overall achievement probability P(x). For example, the overall achievement probability P(x) may be defined as shown in equation (2) below, assuming that the goal achievement events for each characteristic item are independent of each other. P(x) = Π 1<=m<=M Pm(x) ···(2) In other words, the overall achievement probability P(x) is the product of the achievement probabilities Pm(x) of each characteristic item m (m=1 to M). Since the prediction model can predict the probability distribution of observed values ​​of the characteristic items based on the design parameter set x, the achievement probability Pm(x) of each characteristic item can be expressed as a function with the design parameter set x, which is based on the prediction model for each characteristic item, as the input variable. Alternatively, the overall achievement probability P(x) may be expressed as a function with the design parameter set x, which is based on the prediction model for all characteristic items, as the input variable, without having to calculate the achievement probability of each characteristic item.

[0078] For example, the target achievement probability term g(P(x)) may consist of the overall achievement probability P(x), as shown in equation (3), or it may consist of the logarithm of the overall achievement probability P(x), as shown in equation (4). ∫(P(x)) = P(x) ... (3) g(P(x)) = log(P(x)) ... (4) Furthermore, the goal achievement probability term may be the overall achievement probability P(x) or a term obtained by multiplying the logarithm of the overall achievement probability P(x) by a coefficient, or it may include a term for the addition of other factors. In an example of a score function constructed in this way, its optimization can be treated as a maximization problem.

[0079] In this example of the embodiment, the prediction model is configured to output the probability distribution of observed values, so that the probability of achieving the target value Pm(x) for each characteristic item m corresponding to the design parameter group x can be obtained. Furthermore, since the overall achievement probability, calculated by the product of the achievement probabilities Pm(x) of each characteristic item m, is included in the target achievement probability term of the score function, the overall achievement probability is appropriately reflected in the score function that is the target of optimization.

[0080] The target achievement probability acquisition unit 16 acquires the overall achievement probability by optimizing the score function. In this embodiment, the score function A(x) is configured as shown in equation (4) based on equations (1) and (3) above. A(x) = P(x) ... (4) The target achievement probability acquisition unit 16 can then obtain the overall achievement probability as the output of the score function A(x) by optimizing (maximizing) the score function A(x).

[0081] In the example of equation (4), the score function outputs the overall achievement probability. However, if the output of the score function is a value obtained by multiplying or adding other factors to the overall achievement probability, the target achievement probability acquisition unit 16 may acquire the overall achievement probability P(x) based on the design parameter group x obtained by optimizing the score function and the target values ​​of each characteristic item.

[0082] The output unit 17 outputs the overall achievement probability obtained by the target achievement probability acquisition unit 16. The output method is not limited, but the output unit 17 may display the overall achievement probability on a display, which is one aspect of the output device 106. As shown in Figure 7, the output unit 17 may display the overall achievement probability obtained by the target achievement probability acquisition unit 16 in the overall achievement probability field ga of the user interface G.

[0083] Furthermore, the output unit 17 may output the target value acquired by the target value acquisition unit 13 along with the overall achievement probability. As mentioned above, the user interface G shown in Figure 7 uses the target value y m(target)Since it includes an operation object gt for input, the output unit 17 can output and display both the set target value and the overall achievement probability by displaying the operation object gt and the overall achievement probability field ga on the display.

[0084] In this way, the target values ​​of the characteristic items in the optimization of the score function and the overall achievement probability obtained through optimization are output together, for example, by displaying them, making it easy to recognize the correspondence between the target value group and the overall achievement probability.

[0085] Furthermore, the output unit 17 may output the probability of achieving the target value (Pm(x)) of each characteristic item m, which can be obtained in the process of obtaining the overall achievement probability by the target achievement probability acquisition unit 16, along with the overall achievement probability. For example, by displaying the probability of achieving the target value of each characteristic item m on a display, the target value of the characteristic item that is a bottleneck in achieving a high overall achievement probability can be recognized, and by adjusting that target value, such as by relaxing it, it becomes easier to improve the overall achievement probability.

[0086] Furthermore, the target achievement probability acquisition unit 16 may further acquire a set of design parameters that maximize the overall achievement probability by optimizing the score function. The output unit 17 may then further output the set of design parameters that maximize the overall achievement probability, which was acquired by the target achievement probability acquisition unit 16. By referring to the output set of design parameters, it is possible to confirm whether the set of design parameters that can achieve the outputted overall achievement probability is appropriate. Furthermore, by referring to the output set of design parameters, it is possible to manufacture products and conduct experiments.

[0087] Furthermore, as shown in Figure 7, the user interface G may include an instruction button gb for receiving instructions to execute an optimization process based on a set target value. When the design support device 10 receives an instruction input from the user via the instruction button gb, the construction and optimization of a score function based on the set target value is performed, and the overall achievement probability is displayed in the overall achievement probability column ga. In this way, the setting of target values, construction and optimization of score functions, and output of overall achievement probabilities can be repeatedly performed by referring to and operating the user interface G, thereby allowing for consideration of appropriate target values.

[0088] Next, the design support processing by the design support device 10 in this embodiment will be described in detail. Figure 9 is a flowchart showing an example of the contents of the design support method in the design support device 10 according to this embodiment. The design support method is executed when the design support program P1 is loaded into the processor 101 and the program is executed, thereby realizing each of the functional units 11 to 17.

[0089] In step S1, the data acquisition unit 11 acquires the design parameter group x t (t is an integer from 1 to T) and the observed value y for each characteristic item m,t Obtain performance data consisting of pairs.

[0090] In step S2, the model building unit 12 constructs a predictive model for each characteristic item m based on the actual data.

[0091] In step S3, the target value acquisition unit 13 acquires target values ​​for the observed values ​​of each characteristic item m.

[0092] In step S4, the score function construction unit 15 constructs a score function based on the prediction model constructed in step S2 and the target values ​​of each characteristic item obtained in step S3.

[0093] In step S5, the target achievement probability acquisition unit 16 performs optimization of the score function. Here, the optimization range acquisition unit 14 calculates each design parameter v p The optimization range is obtained, and the target achievement probability acquisition unit 16 performs optimization of the score function based on the obtained optimization range.

[0094] In step S6, the target achievement probability acquisition unit 16 acquires the overall achievement probability based on the optimization of the score function A.

[0095] In step S7, the output unit 17 outputs (displays) the overall achievement probability obtained by the target achievement probability acquisition unit 16 in step S6.

[0096] Next, with reference to Figure 10, another example of outputting the overall achievement probability will be explained. Figure 10 is a schematic diagram showing an example of a display screen (user interface) that shows the relationship between the target value and the overall achievement probability.

[0097] The target value acquisition unit 13 may acquire a range of target values ​​for the observed value of at least one of the multiple characteristic items. That is, the target value acquisition unit 13 may acquire a numerical range instead of a single value representing the target value. In the example shown in Figure 10, for the sake of explanation, the target value acquisition unit 13 acquires a numerical range (y) of target values ​​for two characteristic items a and b out of at least two characteristic items. a(target) :1.0~4.0,y b(target) An example is shown of how values ​​10-40 are obtained based on user input, etc.

[0098] The score function construction unit 15 constructs a score function corresponding to each target value within the range of target values. Specifically, the score function construction unit 15 constructs a score function for each combination of points obtained by dividing the range of target values ​​for each characteristic item into predetermined intervals.

[0099] The target achievement probability acquisition unit 16 acquires the overall achievement probability corresponding to each combination of target values ​​within the acquired target value range by optimizing each score function. The output unit 17 outputs the overall achievement probability corresponding to each target value within the set target value range, as shown in the example in Figure 10, where the target value y a(target) The horizontal axis is y, and the target value is y. b(target) The vertical axis is set, and the output unit 17 outputs the target value y a(target) and target value y b(target) The program outputs the overall achievement probability corresponding to the combination, and then draws and outputs contour lines based on the outputted overall achievement probability.

[0100] By displaying this screen, users can recognize how the overall achievement probability changes in response to changes in the target values ​​of each characteristic item, by referring to the relationship between each combination of target values ​​and the overall achievement probability. Therefore, it becomes easier to set appropriate target values.

[0101] Next, a design support program for causing the computer to function as the design support device 10 of this embodiment will be described. Figure 11 is a diagram showing the configuration of the design support program.

[0102] The design support program P1 is comprised of a main module m10 that comprehensively controls the design support processing in the design support device 10, a data acquisition module m11, a model construction module m12, a target value acquisition module m13, an optimization range acquisition module m14, a score function construction module m15, a target achievement probability acquisition module m16, and an output module m17. Each of the modules m11 to m17 realizes the respective functions for the data acquisition unit 11, the model construction unit 12, the target value acquisition unit 13, the optimization range acquisition unit 14, the score function construction unit 15, the target achievement probability acquisition unit 16, and the output unit 17.

[0103] The design support program P1 may be transmitted via a transmission medium such as a communication line, or it may be stored on a recording medium M1, as shown in Figure 11.

[0104] According to the design support device 10, design support method, and design support program P1 of this embodiment described above, a prediction model is constructed based on actual data, target values ​​for each characteristic item are obtained, and a score function including a target achievement probability term is constructed based on the obtained target values ​​and the prediction model. Then, the overall achievement probability is obtained and output by optimizing the score function. Therefore, by referring to the outputted overall achievement probability, it is possible to determine whether the target value is appropriate or not, making it easy to set an appropriate target value.

[0105] The present invention has been described in detail above based on its embodiments. However, the present invention is not limited to the above embodiments. The present invention can be modified in various ways without departing from its spirit. [Explanation of Symbols]

[0106] 10...Design support device, 11...Data acquisition unit, 12...Model construction unit, 13...Target value acquisition unit, 14...Optimization range acquisition unit, 15...Score function construction unit, 16...Target achievement probability acquisition unit, 17...Output unit, 21...Design parameter storage unit, 22...Observed value storage unit, 23...Target value storage unit, 24...Optimization range storage unit, M1...Recording medium, m10...Main module, m11...Data acquisition module, m12...Model construction module, m13...Target value acquisition module, m14...Optimization range acquisition module, m15...Score function construction module, m16...Target achievement probability acquisition module, m17...Output module, P1...Design support program.

Claims

1. A design support device that assists in setting target values ​​for observed values ​​of multiple characteristic items that indicate the characteristics of a product, work in progress, semi-finished product, part, or prototype manufactured based on a set of design parameters consisting of multiple design parameters, A data acquisition unit that acquires multiple data sets consisting of the design parameter group and the observed values ​​of each of the multiple characteristic items for the manufactured product, work in progress, semi-finished product, part, or prototype, A model building unit constructs a prediction model based on the actual data, which predicts the observed values ​​of the characteristic items as a probability distribution or an approximation or surrogate index based on the aforementioned design parameter group, A target value acquisition unit that acquires target values ​​for the observed values ​​of each characteristic item, A score function construction unit constructs a score function that is calculated using the design parameter group as variables based on the prediction model and includes at least a target achievement probability term that represents the overall achievement probability, which is the probability that the target values ​​of all characteristic items are achieved. A target achievement probability acquisition unit that acquires the overall achievement probability by optimizing the score function, An output unit that outputs the overall achievement probability, A design support device equipped with the following features.

2. The target value acquisition unit acquires the target value for at least one of the multiple characteristic items based on user input. The design support device according to claim 1.

3. The target value acquisition unit acquires a target value for at least one of the multiple characteristic items from a predetermined database that stores the target values ​​of the characteristic item in advance. The design support device according to claim 1.

4. The output unit outputs the target value obtained by the target value acquisition unit together with the overall achievement probability. The design support device according to claim 1.

5. The target value acquisition unit acquires the range of the target value with respect to the observed value of at least one characteristic item among the plurality of characteristic items, The score function construction unit constructs the score function corresponding to each target value within the range of the target value, The target achievement probability acquisition unit acquires the overall achievement probability corresponding to each target value within the range of acquired target values ​​by performing optimization of each score function. The output unit outputs the overall achievement probability corresponding to each target value within the set range of target values, in association with the set target value. The design support device according to claim 4.

6. The aforementioned target achievement probability acquisition unit acquires the probability of achieving the target value for each characteristic item, The output unit outputs the probability of achieving the target value of each characteristic item together with the overall achievement probability. The design support device according to claim 1.

7. An optimization range acquisition unit that acquires an optimization range, which is the search range in the optimization of the score function, for at least one of the design parameters included in the group of design parameters, wherein the optimization range of the design parameter is defined by at least one of an upper limit, lower limit, upper and lower limits, and a fixed value of the value of the design parameter, further comprising: The unit that acquires the probability of achieving the target performs optimization of the score function based on the acquired optimization range. The design support device according to claim 1.

8. The model building unit constructs a prediction model for at least one of the multiple characteristic items based on the input of parameters that constitute the prediction model. The design support device according to claim 1.

9. The unit for acquiring the probability of achieving the target acquires the set of design parameters that maximize the overall probability of achieving the target by optimizing the score function. The output unit outputs the set of design parameters that maximize the overall success probability. The design support device according to claim 1.

10. A design support method in a design support device that supports setting target values ​​for observed values ​​of multiple characteristic items that indicate the characteristics of a product, work in progress, semi-finished product, part, or prototype manufactured based on a set of design parameters consisting of multiple design parameters, A data acquisition step for acquiring multiple actual data sets consisting of the design parameter group and the observed values ​​of each of the multiple characteristic items for the manufactured product, work in progress, semi-finished product, part, or prototype; A model construction step in which a prediction model is constructed based on the actual data, which predicts the observed values ​​of the characteristic items as a probability distribution or an approximation or surrogate index based on the aforementioned design parameter set, A target value acquisition step to obtain target values ​​for the observed values ​​of each characteristic item, A score function construction step involves constructing a score function that includes at least a target achievement probability term, which is the probability that the target values ​​of all characteristic items are achieved, calculated using the design parameter group as variables based on the prediction model; A step to obtain the overall probability of achievement by optimizing the score function, An output step which outputs the overall achievement probability, A design support method having [a certain feature].

11. A design support program for causing a computer to function as a design support device that assists in setting target values ​​for observed values ​​of multiple characteristic items that indicate the characteristics of a product, work in progress, semi-finished product, part, or prototype manufactured based on a set of design parameters consisting of multiple design parameters, To the aforementioned computer, A data acquisition function that acquires multiple data sets consisting of the design parameter group and the observed values ​​of each of the multiple characteristic items for the manufactured product, work in progress, semi-finished product, part, or prototype, A model building function that constructs a prediction model based on the actual data, which predicts the observed values ​​of the characteristic items as a probability distribution or an approximation or surrogate index based on the aforementioned design parameter set, A target value acquisition function that obtains target values ​​for the observed values ​​of each characteristic item, A score function construction function that constructs a score function that includes at least a target achievement probability term, which is the probability that the target values ​​of all characteristic items are achieved, calculated using the design parameter group as variables based on the prediction model, A target achievement probability acquisition function that obtains the overall achievement probability by optimizing the score function, An output function that outputs the overall achievement probability, A design support program to make this a reality.

Images