Information processing device, inference device, machine learning device, information processing method, inference method, and machine learning method

Abstract

[Problem] To provide an information processing device capable of accurately generating processing information for a new processed product without relying on knowledge and experience of a skilled worker. [Solution] An information processing device 2 comprises: a feature amount acquisition unit 203B which inputs input information based on new processed product information to a feature amount inference model 220 to acquire a feature amount of the new processed product information; a similar processed product information extraction unit 203C which extracts, from a database 210 in which existing processed product information and processing information have been registered in association with each other, a plurality of pieces of existing processed product information similar to the new processed product information on the basis of the feature amount of the new processed product information; a processing information inference model learning unit 204 which performs machine learning of a processing information inference model 221 by using a plurality of sets of learning data composed of input information based on the existing processed product information and the processing information associated with the existing processed product information; and a processing information generation unit 205A which inputs the input information based on the new processed product information to the processing information inference model 221 to generate processing information for the new processed product information.

Description

Information processing device, inference device, machine learning device, information processing method, inference method, and machine learning method

[0001] The present invention relates to an information processing device, an inference device, a machine learning device, an information processing method, an inference method, and a machine learning method.

[0002] Processing companies that undertake requests to manufacture processed products are often given drawings of new processed products by the client in advance and are asked to submit estimates. In such cases, estimates have traditionally been created by relying on the knowledge and experience of experts, but in order to reduce the effort and prevent the work from becoming dependent on individuals, methods have recently been proposed that use inference models that utilize machine learning to support part of the work related to estimates.

[0003] For example, Patent Document 1 discloses a method in which a similarity determination model is generated in advance through machine learning using existing drawing data relating to existing processed products that is pre-stored in a database, and when new drawing data relating to a new processed product is input, the similarity determination model is used to output a score for each category of the new drawing data, and the existing drawing data belonging to the category with the highest score is extracted.

[0004] However, the above-mentioned method only extracts existing drawing data that is classified into a category that is predicted to contain new drawing data, but does not predict the processing information that will be obtained when a new processed product is processed according to the new drawing data. As a result, the creation of an estimate ultimately has to rely on the knowledge and experience of an expert.

[0005] Japanese Patent Application Laid-Open No. 2021-022330

[0006] The present invention has been made in response to the above-mentioned problems, and aims to provide an information processing device, an inference device, a machine learning device, an information processing method, an inference method, and a machine learning method that are capable of accurately generating processing information for new processed products without relying on the knowledge and experience of an expert.

[0007] In order to achieve the above object, an information processing device according to one aspect of the present invention includes: a new processed product information acquisition unit that acquires new processed product information including drawing data and additional information of a new processed product; a feature acquisition unit that acquires features of the new processed product information by inputting feature inference model input information based on at least the drawing data of the new processed product information acquired by the new processed product information acquisition unit into a feature inference model; and a similar processed product information extraction unit that extracts a plurality of existing processed product information similar to the new processed product information based on the features of the new processed product information acquired by the feature acquisition unit from a database in which a plurality of existing processed product information including drawing data and additional information of existing processed products and processing information when the existing processed product is processed in accordance with the existing processed product information are registered in association with each other. The method includes: a processing information inference model learning unit that uses multiple sets of learning data consisting of processing information inference model input information based on the existing processed product information extracted by the similar processed product information extraction unit and the processing information associated with the existing processed product information to have the processing information inference model learn the correlation between the processing information inference model input information and the processing information through machine learning; and a processing information generation unit that generates the processing information for the new processed product information by inputting the processing information inference model input information based on the new processed product information acquired by the new processed product information acquisition unit to the processing information inference model learned by the processing information inference model learning unit.

[0008] According to an information processing device of one aspect of the present invention, a similar processed product information extraction unit extracts existing processed product information similar to the new processed product information based on the feature quantities of the new processed product information, and machine learning of a processing information inference model is performed using learning data suitable for the new processed product. Therefore, compared to when the similarity between the new processed product and existing processed products is not taken into account, the processing information inference model is less affected by existing processed products whose characteristics are significantly different from those of the new processed product, thereby improving the inference accuracy. Therefore, processing information for the new processed product can be accurately generated without relying on the knowledge and experience of an expert.

[0009] Problems, configurations, and effects other than those described above will become apparent from the detailed description of the invention that follows.

[0010] FIG. 1 is an overall configuration diagram showing an example of an estimate creation support system 1 according to an embodiment. FIG. 2 is a diagram showing an example of various data related to a new processed product 10. FIG. 3 is a diagram showing an example of various data related to an existing processed product 11. FIG. 4 is a block diagram showing an example of an information processing device 2. FIG. 5 is a functional explanatory diagram showing an example of a feature inference model learning unit 202. FIG. 6 is a functional explanatory diagram showing an example of a similar processed product search processing unit 203. FIG. 7 is a functional explanatory diagram showing an example of a processing information inference model learning unit 204. FIG. 8 is a functional explanatory diagram showing an example of an estimate support processing unit 205. FIG. 9 is a hardware configuration diagram showing an example of a computer 900. FIG. 10 is a flowchart showing an example of the operation (information processing method) of the information processing device 2.

[0011] Hereinafter, an embodiment for carrying out the present invention will be described with reference to the drawings. The scope necessary for the explanation to achieve the object of the present invention will be schematically shown, and the scope necessary for explaining the relevant parts of the present invention will be mainly explained, and the parts that are omitted from the explanation will be based on publicly known techniques.

[0012] Fig. 1 is a diagram showing an overall configuration of an example of an estimate creation support system 1 according to an embodiment. Fig. 2 is a diagram showing an example of various data related to a new processed product 10. Fig. 3 is a diagram showing an example of various data related to an existing processed product 11.

[0013] The estimate creation support system 1 is a system that receives an estimate request from a user U and generates processing information D102 and estimated price information D103 as estimate results when a new processed product 10 is processed in accordance with drawing data D100 and additional information D101. When generating the processing information D102 and estimated price information D103 for the new processed product 10, the estimate creation support system 1 also refers to the processing information D112 and estimated price information D113 when an existing processed product 11 is processed in accordance with drawing data D110 and additional information D111, and therefore registers and manages various information in a database 210.

[0014] The processed products 10, 11 are manufactured by performing one or more processing steps on a material using various processing devices or by hand. Examples of the processed products 10, 11 include machine-processed products, buildings produced by architectural processing, and clothing produced by cutting and sewing. Note that the processed products 10, 11 are not limited to the above examples, as long as they are processed by processing steps in accordance with the drawing data D100, D110 and the accompanying information D101, D111. In this embodiment, the processed products 10, 11 will be mainly described as being machine-processed products.

[0015] User U may be, for example, a user (processing company) that receives an order for the processing of a new processed product 10, or a user (requesting company) that places an order for the processing of a new processed product 10.

[0016] The main data input to the estimate preparation support system 1 when requesting an estimate is new processed product information D10 including drawing data D100 of a new processed product 10 and additional information D101, as shown in FIG.

[0017] 3, the main data referenced by the estimate creation support system 1 when requesting an estimate are existing processed product information D11 including drawing data D110 and additional information D111 of the existing processed product 11, and processing information D112 when the existing processed product 11 is processed in accordance with the existing processed product information D11. These data are registered and stored in the database 210.

[0018] The main data output as the estimate result from the estimate creation support system 1 when requesting an estimate are, as shown in Figure 2, processing information D102 when a new processed product 10 is processed according to new processed product information D10, and estimated price information D103.

[0019] The drawing data D100, D110 is data recording design drawings and assembly drawings of the processed products 10, 11. The drawing data D100, D110 includes not only the outline lines of the processed products 10, 11 but also various lines such as center lines and dimension lines, as well as letters, numbers, symbols, notes, etc. (not shown) for indicating the processing steps.

[0020] The drawing data D100, D110 may be either 2D data (plan view, cross-sectional view, etc.) or 3D data (three-dimensional information), and may be either vector or raster format. The drawing data D100, D110 may be, for example, CAD data (an example of vector format) output by various CAD software, or image data (an example of raster format) output by scanning a drawing printed on paper using a scanner or the like. Note that, in the case of 3D data, the 3D data may be converted into 2D data and then handled as 2D data. In this embodiment, the drawing data D100, D110 will be mainly described as 2D data.

[0021] The incidental information D101, D111 is information that is created together with the drawing data D100, D110 and that supplements the information about the processed products 10, 11. The incidental information D101, D111 includes at least one of the shape category, dimensions, material, and processing quantity of the processed products 10, 11 as incidental items. Note that the incidental items of the incidental information D101, D111 are not limited to the above examples, as long as they indicate the characteristics of the processed products 10, 11 and details of the processing steps.

[0022] The shape category is a classification of the outer shape of the material before processing or the outer shape of the processed products 10, 11. Examples of shape categories include, but are not limited to, "plate," "rod," and "pipe."

[0023] Dimensions are expressions of the outer shape of a material before processing or the outer shape of the processed product 10, 11 after processing using two or more variables. Dimensions may be expressed using, for example, three variables (width x length x thickness), three variables (outer diameter x inner diameter x length), or two variables (outer diameter x length), but are not limited to these. Different dimension expressions may also be used depending on the shape category. For example, if the shape category is "plate," the dimension may be expressed using three variables (width x length x thickness), and if the shape category is "rod" or "tube," the dimension may be expressed using three variables (outer diameter x inner diameter x length).

[0024] The material indicates the material of the raw material before processing. The material may be specified by either the formal name or the abbreviated name.

[0025] The processing information D102, D112 includes at least one of a processing type that defines the type of processing step when the processed products 10, 11 are processed by one or more processing steps, and a processing time required for the processing step.

[0026] The processing type is defined as a classification of the combination and sequence of processing steps according to a pattern. For example, if there are two patterns of a series of processing steps, "setup → CAD → CAM → machining → finishing → inspection → packaging," and "setup → CAD → CAM → router processing → finishing → inspection → packaging," these can be defined as processing types "machining processing" and "router processing," respectively. Note that the classification of processing types may not only be defined based on the processing equipment or manual work used in the processing steps, but may also be defined by other methods. For example, processing types may be defined based on the processing method, such as "molding processing," "removal processing," and "additive processing."

[0027] The processing time may be a performance value when actually processed, or a simulation value when estimated by simulation. In addition, when the processed products 10 and 11 are processed through a plurality of processing steps, the processing time may be a processing time for each processing step.

[0028] As shown in Fig. 1, the estimate creation support system 1 includes an information processing device 2 and a user terminal device 3. The information processing device 2 and the user terminal device 3 are connected to a wired or wireless network 4 and configured to be able to send and receive various data to and from each other. Note that the number of information processing devices 2 and user terminal devices 3 and the connection configuration of the network 4 are not limited to the example in Fig. 1 and may be changed as appropriate.

[0029] The information processing device 2 is a server-type computer or a cloud-type computer, and is configured by a general-purpose or dedicated computer (see FIG. 9 described later), etc. The information processing device 2 performs machine learning of a feature amount inference model 220 and a processing information inference model 221. The information processing device 2 uses the trained feature amount inference model 220 and processing information inference model 221 to generate processing information D102 and estimated price information D103 as estimate results in response to an estimate request for a new processed product 10.

[0030] The user terminal device 3 is a client-type computer, and is configured as a general-purpose or dedicated computer (see FIG. 9 described later), etc. The user terminal device 3 accepts various input operations via a display screen such as an app or browser in order to input a request for quotation or check the quotation results, and outputs various information via the display screen or voice.

[0031] (Configuration of Information Processing Device 2) FIG. 4 is a block diagram showing an example of the information processing device 2. As shown in FIG.

[0032] The information processing device 2 includes a control unit 20 , a data storage unit 21 , a trained model storage unit 22 , a communication unit 23 , an input unit 24 , and an output unit 25 .

[0033] The communication unit 23 is connected to an external device (e.g., the user terminal device 3, etc.) via the network 4 and functions as a communication interface for transmitting and receiving various types of data. The input unit 24 accepts various input operations, and the output unit 25 functions as a user interface by outputting various types of information via a display screen or voice. Note that the input unit 24 and the output unit 25 may be omitted.

[0034] The data storage unit 21 stores a database 210 and an information processing program 211. As shown in Fig. 3, the database 210 stores a plurality of pieces of existing processed product information D11, including drawing data D110 and additional information D111 of an existing processed product 11, and processing information D112 obtained when the existing processed product 11 is processed in accordance with the existing processed product information D11, in association with each other. The specific configuration of the database 210 is not limited to the example shown in Fig. 3 and may be designed as appropriate.

[0035] The trained model storage unit 22 stores a trained feature quantity inference model 220 and a trained processed information inference model 221. The feature quantity inference model 220 and the processed information inference model 221 stored in the trained model storage unit 22 may be provided to another device via the network 4, a recording medium, or the like. Furthermore, the number of feature quantity inference models 220 and the processed information inference models 221 stored in the trained model storage unit 22 is not limited to one each, and multiple inference models with different conditions, such as differences in machine learning methods or data, may be stored and used selectively or in parallel.

[0036] 3, the data storage unit 21 and the trained model storage unit 22 are shown as two storage units, but they may be configured as a single storage unit or three or more storage units. Furthermore, at least one of the data storage unit 21 and the trained model storage unit 22 may be configured as a storage unit of an external computer (e.g., a server-type computer or a cloud-type computer).

[0037] The control unit 20 functions as a transmission / reception control unit 200, a database management unit 201, a feature inference model learning unit 202, a similar processed product search processing unit 203, a processing information inference model learning unit 204, and an estimate support processing unit 205 by executing an information processing program 211 recorded in the data storage unit 21.

[0038] (Transmission / reception control unit 200) The transmission / reception control unit 200 transmits and receives various types of data to and from an external device (e.g., the user terminal device 3, etc.). For example, the transmission / reception control unit 200 transmits display information to the user terminal device 3 for outputting various display screens on the user terminal device 3, and receives operation information from the user terminal device 3 for accepting input operations performed on the display screen of the user terminal device 3. The transmission / reception control unit 200 cooperates with the database management unit 201, the feature inference model learning unit 202, the similar processed product search processing unit 203, the processing information inference model learning unit 204, and the estimate support processing unit 205 to transmit display information and receive operation information.

[0039] (Database management unit 201) When the database management unit 201 receives existing processed product information D11 and processing information D112 for an existing processed product 11 from the user terminal device 3, the database management unit 201 associates the existing processed product information D11 with the processing information D112 and registers them in the database 210. Furthermore, when the estimation support processing unit 205 (details will be described later) generates processing information D102 and estimated price information D103 for a new processed product 10, the database management unit 201 associates the processing information D102 and estimated price information D103 with the new processed product information D10 and registers them in the database 210. As a result, the number of data registered in the database 210 increases with each estimate, thereby improving processing accuracy.

[0040] The various pieces of information registered in the database 210 may be made accessible to the user terminal device 3 by the transmission / reception control unit 200. In this case, editing operations such as adding, deleting, and modifying each piece of data may be performed on the display screen of the user terminal device 3. Furthermore, the information registered in the database 210 in association with the new processed product information D10 or the existing processed product information D11 is not limited to the processing information D102 or D112, and any other information may be associated as needed. For example, the database 210 may be configured to register, but is not limited to, estimated price information D103 or D113, feature values ​​(details of which will be described later) of the new processed product information D10 or the existing processed product information D11, the user name related to the estimate request, the user's affiliation (such as the name of a company), the estimate date and time, the estimated price, and the like.

[0041] 5 is a functional explanatory diagram showing an example of the feature amount inference model learning unit 202. The feature amount inference model learning unit 202 includes a learning data acquisition unit 202A and a machine learning unit 202B.

[0042] The learning data acquisition unit 202A refers to the database 210 and acquires learning data D12 made up of input data and output data.

[0043] The input data constituting the learning data D12 is feature quantity inference model input information based on at least the drawing data D110 of the existing processed product information D11. The feature quantity inference model input information may be based on the drawing data D110 and the incidental information D111 included in the existing processed product information D11, and may include, for example, the drawing data D110 and the incidental information D111. Furthermore, the feature quantity inference model input information may include at least outer shape data obtained by cutting out an area in which the outer shape of the existing processed product 11 is described.

[0044] The output data constituting the learning data D12 are feature quantities of the existing processed product information D11. The feature quantities are defined, for example, as vector-format data using a fixed-length numerical array. The feature quantities are not limited to the vector format and may be defined in other data formats.

[0045] The learning data D12 is data used as teacher data (training data), verification data, and test data in supervised learning. The output data constituting the learning data D12 is data used as a correct answer label in supervised learning.

[0046] The learning data acquisition unit 202A may acquire the learning data D12 from data that satisfies a predetermined condition among the data registered in the database 210. The learning data acquisition unit 202A may also acquire the learning data D12 by another method instead of or in addition to the database 210. The learning data acquisition unit 202A may acquire the learning data D12, for example, in cooperation with an external device connected via the network 4, or may acquire the learning data D12 by accepting an input operation via the input unit 24 and the output unit 25.

[0047] The machine learning unit 202B performs machine learning to make the feature inference model 220 learn the correlation between the input data and the output data using multiple sets of learning data D12 acquired by the learning data acquisition unit 202A.

[0048] The machine learning unit 202B causes the feature quantity inference model 220 to learn the correlation between the input data and the output data by performing distance learning so that the more similar the pieces of drawing data D110 are, the greater the distance when comparing the feature quantities, or the more dissimilar the pieces of drawing data D110 are, the smaller the distance when comparing the feature quantities. The trained feature quantity inference model 220 is stored in the trained model storage unit 22.

[0049] The feature inference model 220 may be, for example, a neural network, a vision transformer, a hash algorithm, or the like, but is not limited to these.

[0050] When the input data (feature inference model input information) is raster-format drawing data D110 (or external shape data), the feature inference model 220 may use an algorithm that acquires global features from the entire drawing data D110, or an algorithm that extracts local features from a portion of the drawing data D110 and acquires the overall features of the drawing data D110 using statistical information about the local features.

[0051] Furthermore, when the input data is drawing data D110 (which may be external shape data) and additional information D111, the feature inference model 220 may be composed of a single inference model that takes both of them as input data and outputs features, or may be composed of an inference model that takes drawing data D110 as input data and outputs features, and an inference model that takes additional information D111 as input data and outputs features, and the features output by the two inference models are combined to output a final feature.

[0052] Furthermore, the feature inference model 220 may be, for example, an inference model pre-trained using a dataset for general image recognition, or an inference model with randomly initialized parameters. In this case, the machine learning unit 202B may perform additional learning, such as fine tuning or transfer learning, on the feature inference model 220. As a method of additional learning, the machine learning unit 202B may perform distance learning on the feature inference model 220, for example, so that the more similar the pieces of drawing data D110 are, the greater the distance between the features when comparing them, or the more dissimilar the pieces of drawing data D110 are, the smaller the distance between the features when comparing them. This allows the feature inference model 220 to extract similar drawing data D110 by evaluating the distance between the features. The machine learning method is not limited to the above example and can be selected appropriately depending on the feature inference model 220.

[0053] The timing at which the machine learning unit 202B performs machine learning of the feature inference model learning unit 202 may be when the number of newly registered data in the database 210 exceeds a predetermined number, or when instructions or a request for a quote are received from the user U, but is not limited to these.

[0054] (Similar processed product search processing unit 203) Fig. 6 is a functional explanatory diagram showing an example of the similar processed product search processing unit 203. The similar processed product search processing unit 203 includes a new processed product information acquisition unit 203A, a feature amount acquisition unit 203B, and a similar processed product information extraction unit 203C.

[0055] The new processed product information acquisition unit 203A acquires new processed product information D10 including drawing data D100 and additional information D101 of the new processed product 10. The new processed product information acquisition unit 203A acquires the new processed product information D10 by receiving the new processed product information D10 from the user terminal device 3, for example, as a request for an estimate regarding the new processed product 10.

[0056] The new processed product information acquisition unit 203A may acquire some or all of the incidental information D101 from the drawing data D100 by reading characters, numbers, symbols, notes, etc. contained in the drawing data D100. For example, if text information is embedded in the drawing data D100, the incidental information D101 is acquired by reading the text information. Alternatively, the new processed product information acquisition unit 203A may perform optical character recognition (OCR) on the drawing data D100 to read the text information and acquire the incidental information D101. At this time, the acquired result of the incidental information D101 may be displayed on the display screen of the user terminal device 3, and the user U may perform an editing operation.

[0057] The feature acquisition unit 203B acquires the features of the new processed product information D10 by inputting feature inference model input information based on at least the drawing data D100 of the new processed product information D10 acquired by the new processed product information acquisition unit 203A into the feature inference model 220.

[0058] The feature quantity inference model 220 is obtained by machine learning performed by the feature quantity inference model learning unit 202 and stored in the learned model storage unit 22. The feature quantity acquisition unit 203B performs preprocessing on the new processed product information D10 in accordance with the definition of input data in the feature quantity inference model 220, and inputs the preprocessed information to the feature quantity inference model 220. The feature quantity acquisition unit 203B may input feature quantity inference model input information based on at least the drawing data D110 of the existing processed product information D11 to the feature quantity inference model 220, thereby acquiring the features of the existing processed product information D11 and registering them in the database 210.

[0059] The similar processed product information extraction unit 203C extracts multiple pieces of existing processed product information D11 similar to the new processed product information D10 based on the feature quantities of the new processed product information D10 acquired by the feature quantity acquisition unit 203B from the database 210 in which multiple pieces of existing processed product information D11, including drawing data D110 and accompanying information D111 of existing processed products 11, are registered. Here, the existing processed product information D11 similar to the new processed product information D10 among the multiple pieces of existing processed product information D11, is referred to as similar processed product information D11.

[0060] As a method of extracting similar processed product information D11, the similar processed product information extraction unit 203C extracts similar processed product information D11, for example, using the features of the existing processed product information D11, based on the similarity when comparing the features of the new processed product information D10 with the features of the existing processed product information D11 for each existing processed product 11.

[0061] If the feature amounts of the existing processed product information D11 are associated with the existing processed product information D11 and registered in the database 210, the similar processed product information extraction unit 203C may extract the similar processed product information D11 based on the similarity between the feature amounts of the new processed product information D10 and the feature amounts of the existing processed product information D11 registered in the database 210. If the feature amounts of the existing processed product information D11 are not associated with the existing processed product information D11 and registered in the database 210, the feature amount acquisition unit 203B may acquire the feature amounts of the existing processed product information D11 by inputting feature amount inference model input information based on at least the drawing data D110 of the existing processed product information D11 to the feature amount inference model 220, and the similar processed product information extraction unit 203C may extract the similar processed product information D11 based on the similarity between the feature amounts of the new processed product information D10 and the feature amounts of the existing processed product information D11 acquired by the feature amount acquisition unit 203B.

[0062] The similarity is defined, for example, as the distance between the features when comparing them. When the features are defined as vector data, the distance between the features may be determined using, but is not limited to, a distance index such as Euclidean distance, Manhattan distance, Chebyshev distance, or Mahalanobis distance, or a similarity index such as cosine similarity.

[0063] The extraction method and number of similar processed product information D11 extracted from the database 210 can be set as appropriate. For example, a predetermined number or a predetermined percentage of existing processed product information D11 may be extracted as similar processed product information D11 in descending order of similarity. Alternatively, a predetermined number or a predetermined percentage of existing processed product information D11 may be selected as candidates in descending order of similarity, and then a narrowing down or rearrangement of the similarity ranking may be performed based on the ancillary information D101 included in the new processed product information D10 and the ancillary information D111 included in the existing processed product information D11. Based on the results, a predetermined number or a predetermined percentage of existing processed product information D11 may be extracted as similar processed product information D11. The narrowing down method is not particularly limited, and examples include a method of narrowing down the candidates to only those that match a specific ancillary item (e.g., shape category) in the ancillary information D101, or a method of using a specific ancillary item (e.g., dimension) in the ancillary information D101 as a criterion and narrowing down the candidates to only those that fall within a predetermined range based on the criterion. The method for rearranging the similarity rankings is not particularly limited, but an example is a method of increasing the ranking of candidates that match specific additional items in the additional information D101.

[0064] In addition, the similar processed product information extraction unit 203C may limit the search range of the database 210 based on at least the additional information D101 of the new processed product information D10, and extract similar processed product information D11 from the database 210 limited by the search range based on the features of the new processed product information D10.

[0065] The method for limiting the search range of the database 210 is not particularly limited, but examples include a method for limiting the search range of the database 210 so that existing processed product information D11 that matches the additional items (e.g., shape category) of the additional information D101 is extracted. This makes it possible to remove from the search range existing processed products 11 that have significantly different characteristics from the new processed product 10, thereby making it possible to extract similar processed product information D11 with high accuracy.

[0066] (Processing Information Inference Model Learning Unit 204) Fig. 7 is a functional explanatory diagram showing an example of the processing information inference model learning unit 204. The processing information inference model learning unit 204 includes a learning data acquisition unit 204A and a machine learning unit 204B.

[0067] The learning data acquisition unit 204A refers to the database 210 and acquires learning data D13 made up of input data and output data.

[0068] The input data constituting the learning data D13 is processing information inference model input information based on the existing processed product information D11. The processing information inference model input information may be at least one of the drawing data D110 and the auxiliary information D111 included in the existing processed product information D11, or may be based on at least one of the drawing data D110 and the auxiliary information D111. In this case, only some of the auxiliary information D111 may be selected and used. For example, if the auxiliary information D111 includes a shape category, dimensions, material, and processing quantity as auxiliary items, only the shape category, dimensions, and material may be used as the processing information inference model input information. The processing information inference model input information may also be feature quantities of the existing processed product information D11. Furthermore, if the feature quantities of the existing processed product information D11 are vector-format data, the processing information inference model input information may be obtained by reducing the dimension of the feature quantities of the existing processed product information D11 using methods such as principal component analysis, independent principal component analysis, and nonnegative matrix factorization.

[0069] The output data constituting the learning data D13 is processing information D112 when the existing processed product 11 is processed in accordance with the existing processed product information D11.

[0070] The learning data D13 is data used as teacher data (training data), verification data, and test data in supervised learning. The output data constituting the learning data D13 is data used as a correct answer label in supervised learning.

[0071] When the similar processed product search processing unit 203 extracts multiple similar processed product information D11 for new processed product information D10 in a quotation request, the learning data acquisition unit 204A acquires multiple sets of learning data D13 by using, for each of the multiple similar processed product information D11, processing information inference model input information based on the similar processed product information D11 as input data and processing information D112 associated with the similar processed product information D11 as output data.

[0072] In this case, when the feature quantities of the existing processed product information D11 are used as input information for the processing information inference model, if the feature quantities of the existing processed product information D11 are associated with the existing processed product information D11 and registered in the database 210, the learning data acquisition unit 204A may acquire the learning data D13 by referring to the feature quantities of the existing processed product information D11 registered in the database 210. On the other hand, if the feature quantities of the existing processed product information D11 are not associated with the existing processed product information D11 and registered in the database 210, the feature quantity acquisition unit 203B may acquire the feature quantities of the existing processed product information D11 by inputting feature quantity inference model input information based on at least the drawing data D110 of the existing processed product information D11 into the feature quantity inference model 220, and the learning data acquisition unit 204A may acquire the learning data D13 based on the feature quantities of the existing processed product information D11 acquired by the feature quantity acquisition unit 203B.

[0073] The machine learning unit 204B performs machine learning to make the processing information inference model 221 learn the correlation between the input data and the output data, using the multiple sets of learning data D13 acquired by the learning data acquisition unit 204A. The learned processing information inference model 221 is stored in the learned model storage unit 22.

[0074] The multiple sets of training data D13 used in the machine learning of the processing information inference model 221 are acquired based on multiple sets of similar processed product information D11 similar to the new processed product information D10 in the quotation request. This eliminates the need to prepare a large amount of training data D13 in advance and perform machine learning of the processing information inference model 221 before starting use of the quotation preparation support system 1, and allows use of the quotation preparation support system 1 to begin with a small amount of training data D13. Furthermore, since the machine learning of the processing information inference model 221 is performed using similar processed product information D11 that is relatively similar to the new processed product information D10 in the quotation request, the processing information inference model 221 is less affected by existing processed products 11 whose characteristics are significantly different from those of the new processed product 10, compared to when the similarity between the new processed product information D10 and the existing processed product information D11 is not taken into consideration, and the inference accuracy of the processing information inference model 221 can be improved.

[0075] The processing information inference model 221 may be, for example, a linear model, a support vector machine, a decision tree model, a K-nearest neighbor algorithm, a neural network, or the like, but is not limited to these. The machine learning method can be appropriately selected in accordance with the feature quantity inference model 220. The processing information inference model 221 is constructed as a classification model when the processing information D102 outputs a categorical value such as the processing type, and is constructed as a regression model when the processing information D102 outputs a scalar value such as the processing time.

[0076] (Estimate Support Processing Unit 205) Fig. 8 is a functional explanatory diagram showing an example of the estimate support processing unit 205. The estimate support processing unit 205 includes a processing information generating unit 205A and an estimated price information generating unit 205B.

[0077] The processing information generation unit 205A generates processing information D102 for the new processed product information D10 by inputting processing information inference model input information based on the new processed product information D10 acquired by the new processed product information acquisition unit 203A into the processing information inference model 221 learned by the processing information inference model learning unit 204.

[0078] The processing information inference model 221 is obtained by machine learning performed by the processing information inference model learning unit 204 and stored in the learned model storage unit 22. That is, the processing information inference model 221 is obtained by machine learning performed using a plurality of similar processed product information D11 similar to the new processed product information D10 in the quotation request. The processing information generation unit 205A performs preprocessing on the new processed product information D10 in accordance with the definition of input data in the processing information inference model 221, and inputs the preprocessed information to the feature inference model 220, thereby generating processing information D102 for the new processed product information D10.

[0079] The estimated price information generation unit 205B generates estimated price information D103 for the new processed product information D10 based on the additional information D101 contained in the new processed product information D10 acquired by the new processed product information acquisition unit 203A and the processing information D102 generated by the processing information generation unit 205A.

[0080] The method by which the estimated price information generating unit 205B calculates the estimated price is not particularly limited, but may include, for example, a method of adding up the amounts related to each expense item such as material costs, processing costs, general administrative and sales expenses, profits, etc. In this case, the material costs reflect the material and processing quantity in the supplementary information D101, and the processing costs reflect the processing type and processing time in the processing information D102.

[0081] The processing information D102 generated by the processing information generation unit 205A and the estimated price information D103 generated by the estimated price information generation unit 205B are passed to, for example, the database management unit 201 and registered in the database 210 in association with the new processed product information D10.

[0082] Furthermore, the processing information D102 generated by the processing information generation unit 205A and the estimated price information D103 generated by the estimated price information generation unit 205B are displayed on the user terminal device 3 by the transmission / reception control unit 200. At this time, the processing information D102 and the estimated price information D103 may be edited, such as corrected, on the display screen of the user terminal device 3.

[0083] (Hardware Configuration of Each Device) Fig. 9 is a hardware configuration diagram showing an example of a computer 900. The information processing device 2 and the user terminal device 3 in the estimate creation support system 1 are configured by a general-purpose or dedicated computer 900.

[0084] 9, the computer 900 includes, as its main components, a bus 910, a processor 912, a memory 914, an input device 916, an output device 917, a display device 918, a storage device 920, a communication I / F (interface) unit 922, an external device I / F unit 924, an I / O (input / output) device I / F unit 926, and a media input / output unit 928. Note that the above components may be omitted as appropriate depending on the application of the computer 900.

[0085] The processor 912 is composed of one or more arithmetic processing devices (such as a CPU (Central Processing Unit), an MPU (Micro-processing unit), a DSP (Digital Signal Processor), or a GPU (Graphics Processing Unit)), and operates as a control unit that controls the entire computer 900. The memory 914 stores various data and programs 930, and is composed of, for example, a volatile memory (DRAM, SRAM, etc.) that functions as a main memory, a non-volatile memory (ROM), a flash memory, etc.

[0086] The input device 916 is composed of, for example, a keyboard, a mouse, a numeric keypad, an electronic pen, a microphone, etc., and functions as an input unit. The output device 917 is composed of, for example, a sound (audio) output device, a vibration device, etc., and functions as an output unit. The display device 918 is composed of, for example, a liquid crystal display, an organic EL display, electronic paper, a projector, etc., and functions as an output unit. The input device 916 and the display device 918 may be integrated, such as a touch panel display. The storage device 920 is composed of, for example, an HDD, an SSD, etc., and functions as a storage unit. The storage device 920 stores various data necessary for executing the operating system and the program 930.

[0087] The communication I / F unit 922 is connected to a network 940 such as the Internet or an intranet (which may be the same as network 4 in FIG. 1) via a wired or wireless connection and functions as a communication unit that transmits and receives data to and from other computers in accordance with a predetermined communication standard. The external device I / F unit 924 is connected to an external device 950 such as a camera, printer, scanner, or reader / writer via a wired or wireless connection and functions as a communication unit that transmits and receives data to and from the external device 950 in accordance with a predetermined communication standard. The I / O device I / F unit 926 is connected to I / O devices 960 such as various sensors and actuators and functions as a communication unit that transmits and receives various signals and data, such as detection signals from sensors and control signals to actuators, to and from the I / O devices 960. The media input / output unit 928 is composed of a drive device such as a DVD drive or a CD drive and reads and writes data from and to media (non-transitory storage media) 970 such as DVDs and CDs.

[0088] In the computer 900 having the above configuration, the processor 912 loads a program 930 stored in the storage device 920 into the memory 914, executes the program, and controls each unit of the computer 900 via the bus 910. The program 930 may be stored in the memory 914 instead of the storage device 920. The program 930 may be recorded on the medium 970 in an installable file format or an executable file format and provided to the computer 900 via the media input / output unit 928. The program 930 may be provided to the computer 900 by downloading it via the communication I / F unit 922 over the network 940. Furthermore, the computer 900 may implement various functions realized by the processor 912 executing the program 930 using hardware such as an FPGA or an ASIC.

[0089] The computer 900 is an electronic device of any type, such as a desktop computer or a portable computer. The computer 900 may be a client computer, a server computer, a cloud computer, or an embedded computer such as a control panel or a controller (including a microcomputer, a programmable logic controller, or a sequencer).

[0090] (Operation of Information Processing Device 2) Figure 10 is a flowchart showing an example of the operation (information processing method) of the information processing device 2. Note that the series of information processing methods by the information processing device 2 shown in Figure 10 will be described as being executed when the information processing device 2 receives a request for an estimate for a new processed product 10 from a user terminal device 3 operated by a user U. Also, the description will be given assuming that the database management unit 201 has registered existing processed product information D11 and processing information D112 for multiple existing processed products 11 in the database 210. Furthermore, the description will be given assuming that the feature inference model learning unit 202 has performed a learning data acquisition step, a machine learning step, and a learned model storage step as shown in Figure 5, thereby storing a learned feature inference model 220 in the learned model storage unit 22.

[0091] First, in step S100 (new processed product information acquisition step), the new processed product information acquisition unit 203A of the similar processed product search processing unit 203 acquires new processed product information D10 by receiving the new processed product information D10 from the user terminal device 3 as a request for an estimate for a new processed product 10. At this time, the new processed product information D10 includes drawing data D100 and accompanying information D101 of the new processed product 10.

[0092] In step S101 (feature acquisition process), the feature acquisition unit 203B acquires the features of the new processed product information D10 by inputting feature inference model input information based on at least the drawing data D100 of the new processed product information D10 acquired in step S100 into the feature inference model 220.

[0093] In step S102 (similar processed product information extraction process), the similar processed product information extraction unit 203C extracts multiple similar processed product information D11 (existing processed product information D11) that is similar to the new processed product information D10 based on the features of the new processed product information D10 acquired in step S101 from the database 210 in which multiple existing processed product information D11 is registered.

[0094] In step S110 (learning data acquisition process), the learning data acquisition unit 204A of the processing information inference model learning unit 204 acquires multiple sets of learning data D13 based on the multiple similar processed product information D11 extracted in step S102, by using the processing information inference model input information based on the similar processed product information D11 as input data and the processing information D112 associated with the similar processed product information D11 as output data.

[0095] In step S111 (machine learning process), the machine learning unit 204B performs machine learning using multiple sets of learning data D13 acquired in step S110 to have the processing information inference model 221 learn the correlation between the input data (processing information inference model input information) and the output data (processing information D112).

[0096] In step S120 (processing information generation process), the processing information generation unit 205A generates processing information D102 for the new processed product information D10 by inputting processing information inference model input information based on the new processed product information D10 acquired in step S100 into the processing information inference model 221 that underwent machine learning in step S111.

[0097] In step S121 (estimated price information generation process), the estimated price information generation unit 205B generates estimated price information D103 for the new processed product information D10 based on the additional information D101 contained in the new processed product information D10 acquired in step S100 and the processing information D102 generated in step S120.

[0098] In step S130 (quote result output step), the transmission / reception control unit 200 transmits the processing information D102 generated in step S120 and the estimated price information D103 generated in step S121 to the user terminal device 3. As a result, the processing information D102 and the estimated price information D103 are displayed on the display screen of the user terminal device 3 as the quotation result for the quotation request.

[0099] As described above, according to the information processing device 2 and information processing method of this embodiment, the similar processed product information extraction unit 203C extracts existing processed product information D11 similar to the new processed product information D10 based on the feature quantities of the new processed product information D10, and machine learning of the processing information inference model 221 is performed using the learning data D13 suitable for the new processed product 10. Therefore, compared to when the similarity between the new processed product 10 and the existing processed product 11 is not taken into consideration, the processing information inference model 221 is less affected by the existing processed product 11, whose features are significantly different from those of the new processed product 10, thereby improving the inference accuracy of the processing information inference model 221. Therefore, it is possible to accurately generate processing information D102 for the new processed product 10 without relying on the knowledge and experience of an expert.

[0100] (Other Embodiments) The present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the spirit and scope of the present invention, and all such modifications are included in the technical concept of the present invention.

[0101] In the above embodiment, the information processing device 2 has been described as being configured as a single device, but it may also be configured as a plurality of devices. For example, the units 202 to 205 of the information processing device 2 may be distributed across a plurality of devices, thereby resulting in a machine learning device including a feature quantity inference model learning unit 202 and performing a feature quantity inference model learning process, an information processing device including a similar processed product search processing unit 203 and performing a similar processed product search process, a machine learning device for the processing information inference model 221 including a processing information inference model learning unit 204 and performing a processing information inference model learning process, and an information processing device including an estimate support processing unit 205 and performing an estimate support process. In this case, each unit (each process) of each of the above devices may be realized by a program (an information processing program or a machine learning program) executable by the computer 900.

[0102] Each of the above devices can be configured, for example, as follows: The configuration and operation of each part of each device and the various types of data handled by each device are the same as those in the above embodiment, and therefore detailed description thereof will be omitted.

[0103] The machine learning device that performs the feature inference model learning process includes a learning data acquisition unit 202A that acquires multiple sets of learning data D12 each composed of input data and output data, a machine learning unit 202B that causes a feature inference model 220 to learn a correlation between the input data and the output data using the multiple sets of learning data D12 acquired by the learning data acquisition unit 202A, and a learned model storage unit 22 that stores the feature inference model 220 that has learned the correlation by the machine learning unit 202B. The input data is feature inference model input information that is based on at least the drawing data D110 of the existing processed product information D11. The output data is the feature of the existing processed product information D11.

[0104] The information processing device that performs the similar processed product search processing step further includes a new processed product information acquisition unit 203A that acquires new processed product information D10, a feature amount acquisition unit 203B that acquires feature amounts of the new processed product information D10 by inputting feature amount inference model input information based on at least drawing data D100 from the new processed product information D10 acquired by the new processed product information acquisition unit 203A to a feature amount inference model 220, and a similar processed product information extraction unit 203C that extracts one or more existing processed product information D11 similar to the new processed product information D10 from a database 210 in which a plurality of existing processed product information D11 is registered, based on the feature amounts of the new processed product information D10 acquired by the feature amount acquisition unit 203B. The similar processed product information extraction unit 203C may be omitted.

[0105] The machine learning device that performs the processing information inference model learning process includes a learning data acquisition unit 204A that acquires multiple sets of learning data D13 consisting of input data and output data, a machine learning unit 204B that causes a processing information inference model 221 to learn a correlation between the input data and the output data using the multiple sets of learning data D13 acquired by the learning data acquisition unit 204A, and a learned model storage unit 22 that stores the processing information inference model 221 that has learned the correlation by the machine learning unit 204B. The input data is processing information inference model input information based on existing processed product information D11. The output data is processing information D112 when a processed product 10 is processed according to the existing processed product information D11.

[0106] The information processing device that performs the estimate support processing step includes a new processed product information acquisition unit that acquires new processed product information D10, a processing information generation unit 205A that inputs processing information inference model input information based on the new processed product information D10 acquired by the new processed product information acquisition unit into a processing information inference model 221 to generate processing information D102 when a new processed product 10 is processed in accordance with the new processed product information D10, and an estimate price information generation unit 205B that generates estimate price information D103 for the new processed product information D10 based on incidental information D101 included in the new processed product information D10 acquired by the new processed product information acquisition unit and the processing information D102 generated by the processing information generation unit 205A. The estimate price information generation unit 205B may be omitted.

[0107] The transmission / reception control unit 200 and the database management unit 201 may be provided in each of the above devices. The database 210 may be configured to be accessible from each of the above devices, and may be stored in a storage unit of one of the devices or in a storage unit of an external computer. Furthermore, some of the above devices may be implemented by the user terminal device 3.

[0108] (Inference Device, Inference Method, and Inference Program) The present invention can be provided not only in the form of the information processing device 2 (information processing method or information processing program) according to the above embodiment, but also in the form of an inference device (inference method or inference program) used to infer the feature quantities of the new processed product information D10 and the processing information D102. In this case, the inference device (inference method or inference program) can include a memory and a processor, and the processor executes a series of processes. The series of processes includes an information acquisition process (information acquisition step) for acquiring the new processed product information D10, and, once the new processed product information D10 is acquired by the information acquisition process, inferring the feature quantities of the new processed product information D10 (inference process). The series of processes also includes an information acquisition process (information acquisition step) for acquiring the new processed product information D10, and, once the new processed product information D10 is acquired by the information acquisition process, inferring the processing information D102 that will be obtained when the new processed product 10 is processed in accordance with the new processed product information D10.

[0109] By providing it in the form of an inference device (inference method or inference program), it can be more easily applied to various devices than when an information processing device is implemented. It will be naturally understood by those skilled in the art that when the inference device (inference method or inference program) infers the feature quantities of the new processed product information D10 and the processing information D102, the inference method implemented by the information generation unit may be applied using a trained inference model generated by the machine learning device and machine learning method according to the above embodiments.

[0110] 1... Estimate creation support system, 2... Information processing device, 3... User terminal device, 10... New processed product, 11... Existing processed product, 20... Control unit, 21... Data storage unit, 22... Learned model storage unit, 23... Communication unit, 24... Input unit, 25... Output unit, 200... Transmission / reception control unit, 201... Database management unit, 202... Feature inference model learning unit, 202A... Learning data acquisition unit, 202B... Machine learning unit, 203... Similar processed product search processing unit, 203A... New processed product information acquisition unit, 203B... Feature acquisition unit, 203C... Similar processed product information extraction unit, 204... Processing information inference model learning unit, 204A... Learning data acquisition unit, 204B... Machine learning unit, 205... Support processing unit, 205A... Processing information generation unit, 205B... Price information generation unit, 210... Database, 211... Information processing program, 220...feature inference model, 221...processed information inference model

Claims

a new processed product information acquisition unit that acquires new processed product information including drawing data and additional information of a new processed product; a feature acquisition unit that acquires features of the new processed product information by inputting feature inference model input information based on at least the drawing data of the new processed product information acquired by the new processed product information acquisition unit into a feature inference model; a similar processed product information extraction unit that extracts a plurality of existing processed product information similar to the new processed product information based on the feature of the new processed product information acquired by the feature acquisition unit from a database in which existing processed product information including drawing data and additional information of existing processed products and processing information when the existing processed product is processed according to the existing processed product information are associated and registered in a plurality of databases; a processing information inference model learning unit that causes a processing information inference model to learn a correlation between the processing information inference model input information and the processing information by machine learning using a plurality of sets of learning data consisting of processing information inference model input information based on the existing processed product information extracted by the similar processed product information extraction unit and the processing information associated with the existing processed product information; a processing information generation unit that generates the processing information for the new processed product information by inputting processing information inference model input information based on the new processed product information acquired by the new processed product information acquisition unit to the processing information inference model trained by the processing information inference model learning unit.

2. The information processing device of claim 1, further comprising an estimated price information generation unit that generates estimated price information for the new processed product information based on the auxiliary information contained in the new processed product information acquired by the new processed product information acquisition unit and the processing information generated by the processing information generation unit.

3. The information processing device of claim 1, wherein the similar processed product information extraction unit extracts a plurality of pieces of existing processed product information similar to the new processed product information based on the similarity between the features of the new processed product information and the features of the existing processed product information associated with the existing processed product information and registered in the database, or based on the similarity between the features of the new processed product information and the features of the existing processed product information obtained by inputting feature inference model input information based on at least the drawing data among the existing processed product information into the feature inference model.

4. The information processing device of claim 1, wherein the similar processed product information extraction unit limits a search range of the database based on at least the ancillary information of the new processed product information acquired by the new processed product information acquisition unit, and extracts multiple pieces of existing processed product information similar to the new processed product information from the database limited by the search range based on the features of the new processed product information acquired by the feature acquisition unit.

5. The information processing device according to claim 1, wherein the feature inference model is a trained inference model that has been trained by machine learning to learn the correlation between feature inference model input information based on at least the drawing data of the existing processed product information and the features of the existing processed product information.

6. The information processing device of claim 1, wherein the feature inference model is a trained inference model that has learned by machine learning the correlation between feature inference model input information including at least outer shape data obtained by cutting out an area in which the outer shape of the existing processed product is described from the drawing data and the features of the existing processed product information, and the feature acquisition unit generates outer shape data obtained by cutting out an area in which the outer shape of the new processed product is described from the drawing data, and acquires the features of the new processed product information by inputting feature inference model input information including at least the outer shape data into the feature inference model.

7. An information processing device as described in claim 1, wherein feature quantity inference model input information based on at least the drawing data of the existing processed product information is based on the drawing data and the additional information included in the existing processed product information, and feature quantity inference model input information based on at least the drawing data of the new processed product information is based on the drawing data and the additional information included in the new processed product information.

8. An information processing device as described in claim 1, wherein input information of a processing information inference model based on the existing processed product information is a feature of the existing processed product information, and input information of a processing information inference model based on the new processed product information is a feature of the new processed product information.

9. The information processing device according to claim 1, wherein the additional information includes at least one of a shape category, a dimension, a material, and a processing quantity of the existing processed product or the new processed product.

10. The information processing device according to claim 1, wherein the processing information includes at least one of a processing type that defines the type of the processing process when the existing processed product or the new processed product is processed by one or more processing processes, and a processing time required for the processing process.

11. A new processed product information acquisition step of acquiring new processed product information including drawing data and additional information of a new processed product; a feature amount acquisition step of acquiring features of the new processed product information by inputting feature amount inference model input information based on at least the drawing data of the new processed product information acquired by the new processed product information acquisition step into a feature amount inference model; a similar processed product information extraction step of extracting a plurality of existing processed product information similar to the new processed product information based on the feature amount of the new processed product information acquired by the feature amount acquisition step from a database in which existing processed product information including drawing data and additional information of existing processed products and processing information when the existing processed product is processed according to the existing processed product information are associated and registered in a plurality of databases; a processing information inference model learning step of making a processing information inference model learn a correlation between the processing information inference model input information and the processing information by machine learning using a plurality of sets of learning data consisting of processing information inference model input information based on the existing processed product information extracted by the similar processed product information extraction step and the processing information associated with the existing processed product information; a processing information generation process for generating the processing information for the new processed product information by inputting processing information inference model input information based on the new processed product information acquired in the new processed product information acquisition process to the processing information inference model trained in the processing information inference model training process.

Images