Material quality estimation model creation method, material quality estimation method, material quality estimation model creation apparatus, material quality estimation apparatus and material quality estimation model creation system

Abstract

To provide a material quality estimation model creation method, a material quality estimation method, a material quality estimation model creation apparatus, a material quality estimation apparatus and a material quality estimation model creation system which can create a model for estimating the quality of a new manufacturing material at high precision.SOLUTION: A material quality estimation model creation method comprises: a material quality estimation base model creation step in which a model creation unit of a computer executes machine learning with existing manufacturing condition data as input and existing quality data linked to the existing manufacturing condition data as output to create a material quality estimation base model; and a material quality estimation model relearning step in which the model creation unit executes the relearning of the material quality estimation base model with additional manufacturing condition data as input and additional quality data liked to the additional manufacturing condition data as output to create a material quality estimation model.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to a method for creating a material quality estimation model, a method for estimating material quality, an apparatus for creating a material quality estimation model, an apparatus for estimating material quality, and a system for creating a material quality estimation model.

Background Art

[0002] Patent Document 1 discloses a quality prediction system that predicts the quality of a target product based on the similarity between a learning prediction model and a quality prediction model derived from the operating conditions of an actual target product. In this system, a machine learning algorithm is applied to construct the prediction model, and the occurrence probability of defects is estimated with high accuracy.

[0003] Further, Patent Document 2 discloses a prediction system including a prediction model that inputs input data including the output factors of a metal plate manufacturing facility and the component values of the metal plate during manufacturing, and predicts the material characteristic values of the manufactured metal plate. In this system, a machine learning algorithm is applied, and by inputting the input data to output the manufacturing condition factors and using the manufacturing condition factors as inputs to output the material characteristic values, the material characteristic values are estimated with high accuracy.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, the material quality estimation models disclosed in Patent Documents 1 and 2 are models applied to materials manufactured under manufacturing conditions that are the same as or similar to those of the manufactured materials in actual manufacturing facilities. That is, the material quality estimation models disclosed in Patent Documents 1 and 2 are generally created on the premise that a large number of manufacturing condition data and quality data exist.

[0006] On the other hand, in the development of new manufacturing materials, it is common to verify whether the materials manufactured under newly designed manufacturing conditions satisfy the desired quality by conducting preliminary experiments in experimental facilities. In the above preliminary experiments, it is usually the case that only some types of manufacturing conditions that are assumed to have a large contribution to the determination of the material quality are carefully designed and controlled. Therefore, in the preliminary experiments, other manufacturing conditions including environmental conditions and disturbance conditions that only exist when manufacturing in actual manufacturing facilities are rarely considered. Also, due to the nature of the preliminary experiments, the number of times the experiments are tried is usually small, so the amount of experimental result data that can be obtained is also reduced.

[0007] When new manufacturing is carried out in actual manufacturing facilities based on the manufacturing conditions designed in the preliminary experiments as described above, there is a problem that in many cases, the desired material quality cannot be obtained due to the differences from actual manufacturing that could not be fully considered in the preliminary experiments. Also, when starting the manufacturing of new materials, since there is not enough experimental condition data and quality data during the preliminary experiments as described above, it has not been easy to create a highly accurate material quality estimation model.

[0008] The present invention has been made in view of the above, and an object thereof is to provide a method for creating a material quality estimation model, a material quality estimation method, a device for creating a material quality estimation model, a material quality estimation device, and a system for creating a material quality estimation model that can create a model for accurately estimating the quality of new manufacturing materials.

Means for Solving the Problems

[0009] In order to solve the above-described problems and achieve the object, a method for creating a material quality estimation model according to the present invention includes: a material quality estimation base model creation step in which a model creation unit provided in a computer performs machine learning with existing manufacturing condition data as an input and existing quality data associated with the existing manufacturing condition data as an output, and creates a material quality estimation base model; and a material quality estimation model re-learning step in which the model creation unit performs re-learning of the material quality estimation base model with additional manufacturing condition data as an input and additional quality data associated with the additional manufacturing condition data as an output, and creates a material quality estimation model.

[0010] In the method for creating a material quality estimation model according to the present invention, in the above invention, the existing manufacturing condition data is data related to existing manufacturing materials, and the additional manufacturing condition data is data related to new manufacturing materials.

[0011] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, when the types of the additional manufacturing condition data are insufficient with respect to the types of the existing manufacturing condition data that are inputs to the material quality estimation base model, the model creation unit adds the types of the additional manufacturing condition data so as to match the types of the existing manufacturing condition data, and performs re-learning of the material quality estimation base model.

[0012] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, for the types of the additional manufacturing condition data that are insufficient with respect to the existing manufacturing condition data, for each insufficient type, the model creation unit adds the data by complementing it using the statistic of the corresponding type of the existing manufacturing condition data.

[0013] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, for the types of the additional manufacturing condition data that are lacking in the existing manufacturing condition data, the model creation unit adds them by complementing each lacking type with sample values randomly extracted from the corresponding types of the existing manufacturing condition data.

[0014] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, in the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, the model creation unit locally extracts only the manufacturing condition data similar to the additional manufacturing condition data from the set of the existing manufacturing condition data, and for the types of the additional manufacturing condition data that are lacking in the existing manufacturing condition data, adds them by complementing each lacking type with the statistic of the corresponding type of the existing manufacturing condition data.

[0015] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, in the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, the model creation unit locally extracts only the manufacturing condition data similar to the additional manufacturing condition data from the set of the existing manufacturing condition data, and for the types of the additional manufacturing condition data that are lacking in the existing manufacturing condition data, adds them by complementing each lacking type with sample values randomly extracted from the corresponding types of the existing manufacturing condition data.

[0016] In the method for creating a material quality estimation model according to the present invention, in the above invention, in the material quality estimation model re-learning step, when the types of the additional manufacturing condition data are lacking with respect to the types of the existing manufacturing condition data that are inputs to the material quality estimation base model, the material quality estimation base model is created by reducing the types of the existing manufacturing condition data so as to match the types of the existing manufacturing condition data.

[0017] In the method for creating a material quality estimation model according to the present invention, in the above invention, the method for re-learning the material quality estimation base model is transfer learning.

[0018] In the method for creating a material quality estimation model according to the present invention, in the above invention, the method for re-learning the material quality estimation base model is fine-tuning.

[0019] In order to solve the above-described problems and achieve the object, the material quality estimation method according to the present invention includes a quality estimation step in which a quality estimation unit provided in a computer estimates the quality of a manufactured product based on a material quality estimation model created by the method for creating a material quality estimation model described above.

[0020] In order to solve the above-described problems and achieve the object, a device for creating a material quality estimation model according to the present invention includes a model creation unit that performs machine learning with existing manufacturing condition data as an input and existing quality data associated with the existing manufacturing condition data as an output to create a material quality estimation base model, and the model creation unit inputs additional manufacturing condition data and performs re-learning of the material quality estimation base model with additional quality data associated with the additional manufacturing condition data as an output to create a material quality estimation model.

[0021] In order to solve the above-described problems and achieve the object, a material quality estimation device according to the present invention includes a quality estimation unit that estimates the quality of a manufactured product based on a material quality estimation model created by the device for creating a material quality estimation model described above.

[0022] In order to solve the above problems and achieve the object, a system for creating a material quality estimation model according to the present invention uses existing manufacturing condition data as input, performs machine learning with existing quality data associated with the existing manufacturing condition data as output, and includes a model creation device for creating a material quality estimation base model. The model creation device uses additional manufacturing condition data as input, performs retraining of the material quality estimation base model with additional quality data associated with the additional manufacturing condition data as output, and creates a material quality estimation model.

Advantages of the Invention

[0023] In the method for creating a material quality estimation model, the device for creating a material quality estimation model, and the creation of a material quality estimation model according to the present invention, a machine learning model based on existing manufacturing condition data is used as a base model, and retraining is performed using preliminary experiment data (additional manufacturing condition data) of a new manufacturing material. Thereby, even when there is not enough manufacturing data when actually manufacturing a new manufacturing material, it is possible to create a model capable of estimating the material quality with high accuracy.

[0024] In the material quality estimation method and the material quality estimation device according to the present invention, by using the model created as described above, it is possible to estimate the variation in material quality when assuming variations in manufacturing conditions in a new manufacturing material in advance. Further, in the material quality estimation method and the material quality estimation device according to the present invention, it is possible to perform material design of manufacturing conditions that do not deviate from the allowable range of quality variation. Thereby, material development can be efficiently carried out.

Brief Description of the Drawings

[0025]

Figure 1

Figure 2

Figure 3

Figure 4

Embodiments for Carrying Out the Invention

[0026] A method for creating a material quality estimation model according to an embodiment of the present invention (hereinafter referred to as "model creation method"), a material quality estimation method, an apparatus for creating a material quality estimation model (hereinafter referred to as "model creation apparatus"), and a material quality estimation apparatus will be described with reference to the drawings. Note that the constituent elements in the following embodiments include those that can be replaced and are easy for those skilled in the art, or those that are substantially the same.

[0027] (Device Configuration) FIG. 1 shows an example of the configuration of an information processing apparatus 1 for realizing a model creation apparatus and a material quality estimation apparatus according to an embodiment. The information processing apparatus 1 includes an input unit 10, a storage unit 20, an arithmetic unit 30, and a display unit 40. The model creation apparatus according to the embodiment is realized by the constituent elements of the information processing apparatus 1 excluding the quality estimation unit 33 of the arithmetic unit 30. Further, the material quality estimation apparatus according to the embodiment is realized by all the constituent elements of the information processing apparatus 1.

[0028] The input unit 10 is an input means for the arithmetic unit 30 and is realized by an input device such as a keyboard, a mouse pointer, a numeric keypad, or the like.

[0029] The storage unit 20 is composed of recording media such as an EPROM (Erasable Programmable ROM), a hard disk drive (HDD), and a removable media. Examples of the removable media include disk recording media such as a USB (Universal Serial Bus) memory, a CD (Compact Disc), a DVD (Digital Versatile Disc), and a BD (Blu-ray (registered trademark) Disc). The storage unit 20 can store an operating system (OS), various programs, various tables, various databases, and the like. An operation DB (database) 21 is stored in the storage unit 20.

[0030] The operation DB 21 stores existing manufacturing condition data, existing quality data, additional manufacturing condition data, and additional quality data. The existing manufacturing condition data and the existing quality data are data used in actual equipment and there is a large amount of data. On the other hand, the additional manufacturing condition data and the additional quality data are data used in experimental equipment, for example, when developing a new manufacturing material, and generally the number of data is limited.

[0031] The existing manufacturing condition data indicates existing data obtained by measuring or estimating the manufacturing state during product manufacturing by some means. This existing manufacturing condition data is preferably data that varies according to the position within the product.

[0032] The existing quality data indicates existing data obtained by measuring or estimating the quality of the product by some means. Although the method of measurement or estimation varies depending on the type of the quality, this existing quality data is preferably data at a specific position within the product.

[0033] The additional manufacturing condition data indicates additional data obtained by measuring or estimating the manufacturing state during product manufacturing by some means. This additional manufacturing condition data is preferably data that varies according to the position within the product.

[0034] The additional quality data indicates additional data obtained by measuring or estimating the quality of the product by some means. Although the method of measurement or estimation varies depending on the type of quality, it is preferably data at a specific position within the product.

[0035] The arithmetic unit 30 is realized by a processor such as a CPU (Central Processing Unit) and a memory (main storage unit) such as a RAM (Random Access Memory) and a ROM (Read Only Memory).

[0036] The arithmetic unit 30 loads a program into the working area of the main storage unit and executes it, and controls each component etc. through the execution of the program, thereby realizing a function that meets a predetermined purpose. The arithmetic unit 30 functions as a data creation unit 31, a model creation unit 32, and a quality estimation unit 33 through the execution of the above-described program. In FIG. 1, an example is shown in which the functions of each unit are realized by a single computer, but the specific method of realizing the functions of each unit is not particularly limited, and for example, the functions of each unit may be realized by a plurality of computers.

[0037] The data creation unit 31 creates data necessary for creating a material quality estimation base model described later. That is, the data creation unit 31 collects existing manufacturing condition data and existing quality data from the operation DB 21. Subsequently, the data creation unit 31 associates the existing manufacturing condition data with the existing quality data to create existing manufacturing data. This existing manufacturing data is data regarding existing manufacturing materials as described above.

[0038] Specifically, the data creation unit 31 aggregates the collected existing manufacturing condition data and existing quality data and combines them with each other. That is, the data creation unit 31 associates the data obtained by measuring or estimating the physical state of the product with the data obtained by measuring or estimating the quality of the product. Here, in the creation of the existing manufacturing data, it is desirable to associate a specific position of the product where the existing quality data was measured or estimated with the existing manufacturing condition data measured or estimated at the same position.

[0039] In addition, the data creation unit 31 creates data necessary for creating a material quality estimation model described later. That is, the data creation unit 31 collects additional manufacturing condition data and additional quality data from the operation DB 21. Subsequently, the data creation unit 31 associates the additional manufacturing condition data with the additional quality data to create additional manufacturing data. This additional manufacturing data is data regarding a new manufacturing material as described above.

[0040] Specifically, the data creation unit 31 aggregates the collected additional manufacturing condition data and additional quality data and combines them with each other. That is, the data creation unit 31 associates the data obtained by measuring or estimating the physical state of the product with the data obtained by measuring or estimating the quality of the product. Here, in the creation of the additional manufacturing data, it is desirable to associate a specific position of the product where the additional quality data was measured or estimated with the additional manufacturing condition data measured or estimated at the same position.

[0041] The model creation unit 32 creates a material quality estimation base model using the existing manufacturing data. That is, the model creation unit 32 uses the existing manufacturing condition data as input, performs machine learning with the existing quality data associated with the existing manufacturing condition data as output, and creates a material quality estimation base model representing the relationship between the manufacturing conditions and the quality. A machine learning algorithm having a neural network structure is used for the learning of this material quality estimation base model.

[0042] If the existing manufacturing data is taken as the manufacturing data of the existing manufactured materials in the actual manufacturing equipment, it is often possible to secure a sufficient number of data for machine learning, and thus create a model with high estimation accuracy, that is, high generalization performance, even for unknown data.

[0043] In addition, the model creation unit 32 creates a material quality estimation model using the material quality estimation base model and the additional manufacturing data. That is, the model creation unit 32 inputs the additional manufacturing condition data and retrains the material quality estimation base model with the additional quality data associated with the additional manufacturing condition data as the output, and creates a material quality estimation model for the additional manufacturing data.

[0044] For retraining, for example, transfer learning is used. FIG. 2 shows an overview of transfer learning. In transfer learning, generally, when a model M m ,…,X M that accurately estimates the output Y A has been learned and created, the middle layer is diverted from the input layer of the model M A , and only the linear layer A close to the output layer is replaced with another linear layer B. Then, for the input X = [X1, X2,…, X A ,…,X m ,…,X M , it is learned with data of another output Y B , and a model M B that accurately estimates the output Y B is created.

[0045] Here, the linear layer is also called the fully connected layer. Also, in transfer learning, since the input layer and the middle layer of the already learned model M A are diverted, the model M BIt is known that even when the data for learning is smaller in number than that in general supervised learning, a highly accurate model can be created. In the present embodiment, when additional manufacturing data is used as manufacturing data in a preliminary experiment of a new manufacturing material, there may be a case where the number of data sufficient for performing general machine learning cannot be secured. Even when the additional manufacturing data is limited and small in number in this way, by using transfer learning, a highly accurate material quality estimation model for the additional manufacturing data can be created.

[0046] Furthermore, when the types of additional manufacturing condition data are insufficient compared to the types of existing manufacturing condition data that are input to the material quality estimation base model, the model creation unit 32 adds the types of additional manufacturing condition data so as to match the types of existing manufacturing condition data, and performs relearning of the material quality estimation base model. In this case, the model creation unit 32 adds the types of additional manufacturing condition data, for example, as in the following (1) to (4).

[0047] (1) For the types of additional manufacturing condition data that are insufficient with respect to the existing manufacturing condition data, for each insufficient type, it is added by complementing using the statistic (for example, average value, etc.) of the corresponding type of the existing manufacturing condition data. (2) For the types of additional manufacturing condition data that are insufficient with respect to the existing manufacturing condition data, for each insufficient type, it is added by complementing using the sample value randomly extracted from the corresponding type of the existing manufacturing condition data. (3) In the manufacturing condition data of the types that already match between the existing manufacturing condition data and the additional manufacturing condition data, only the manufacturing condition data similar to the additional manufacturing condition data is locally extracted from the set of the existing manufacturing condition data. Then, for the types of additional manufacturing condition data that are insufficient with respect to the existing manufacturing condition data, for each insufficient type, it is added by complementing using the statistic of the corresponding type of the locally extracted existing manufacturing condition data. (4) In the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, only the manufacturing condition data similar to the additional manufacturing condition data is locally extracted from the set of existing manufacturing condition data. Then, for the types of additional manufacturing condition data that are lacking in the existing manufacturing condition data, for each lacking type, it is added by complementing with sample values randomly extracted from the corresponding types of the locally extracted existing manufacturing condition data.

[0048] Here, in the above (3) and (4), as a method of locally extracting only the manufacturing condition data similar to the additional manufacturing condition data from the set of existing manufacturing condition data, for example, the Just-In-Time method or the like can be used. In this Just-In-Time method, in the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, only the data in the vicinity of the additional manufacturing condition data that is the required point is locally extracted from the set of existing manufacturing condition data.

[0049] Figure 3 shows a method of complementing the types of additional manufacturing condition data. For example, let the existing manufacturing data be the manufacturing data of the existing manufacturing materials in the actual manufacturing equipment, and the additional manufacturing data be the manufacturing data in the preliminary experiment of the new manufacturing materials. In this case, for example, re-learning is performed using some types of manufacturing conditions designed in the preliminary experiment of the new manufacturing materials from the material quality estimation base model learned using the manufacturing data of the existing manufacturing materials. As a result, re-learning considering other manufacturing conditions including environmental conditions and disturbance conditions that exist only when manufacturing in the actual manufacturing equipment becomes possible.

[0050] Note that in the above (1) to (4), the types of additional manufacturing condition data that are lacking in the existing manufacturing condition data are added, but conversely, the types of existing manufacturing condition data may be reduced. That is, when it has already been determined at the stage of creating the material quality estimation base model that the types of additional manufacturing condition data are lacking compared to the types of existing manufacturing condition data, the types of existing manufacturing condition data that are inputs to the material quality estimation base model may be reduced.

[0051] When the types of additional manufacturing condition data are insufficient compared to the types of existing manufacturing condition data that serve as the input for the material quality estimation base model, the model creation unit 32 reduces the types of existing manufacturing condition data to match the types of existing manufacturing condition data, and creates a material quality estimation base model. In this way, by pre-reducing the types of existing manufacturing condition data that serve as the input for the material quality estimation base model, it becomes possible to align the types of additional manufacturing condition data with the types of existing manufacturing condition data during the relearning of the material quality estimation base model.

[0052] Also, in this embodiment, a method using a general neural network algorithm is exemplified, but the specific algorithm is not limited as long as it is a machine learning algorithm capable of performing transfer learning. Further, in this embodiment, a method using transfer learning is exemplified, but fine-tuning may be used instead of transfer learning.

[0053] The quality estimation unit 33 estimates the quality of the manufactured product based on the material quality estimation model created by the model creation unit 32. That is, the quality estimation unit 33 inputs the manufacturing condition data of the product for which the quality is estimated to the material quality estimation model, and outputs a quality estimation result.

[0054] The display unit 40 is realized by a display device such as an LCD display or a CRT display. The display unit 40 displays, in the form of characters, graphics, etc., for example, the quality estimation result by the quality estimation unit 33, based on the display signal input from the arithmetic unit 30.

[0055] (Model Creation Method) A model creation method by a model creation device according to an embodiment will be described with reference to FIG. 4. The model creation method includes an existing manufacturing data collection step (step S1), an existing manufacturing data creation step (step S2), and a material quality estimation base model creation step (step S3). Further, the model creation method further includes an additional manufacturing data collection step (step S4), an additional manufacturing data creation step (step S5), and a material quality estimation model re-learning step (step S6). Note that the timing of executing steps S4 and S5 is not particularly limited, and they may be executed in parallel with steps S1 and S2.

[0056] In the existing manufacturing data collection step, the data creation unit 31 collects existing manufacturing condition data and existing quality data (step S1).

[0057] In the existing manufacturing data creation step, the data creation unit 31 associates the existing manufacturing condition data and the existing quality data to create existing manufacturing data (step S2).

[0058] In the material quality estimation base model creation step, the model creation unit 32 performs machine learning with the existing manufacturing condition data as input and the existing quality data as output to create a material quality estimation base model (step S3).

[0059] In the additional manufacturing data collection step, the data creation unit 31 collects additional manufacturing condition data and additional quality data (step S4).

[0060] In the additional manufacturing data creation step, the data creation unit 31 associates the additional manufacturing condition data and the additional quality data to create additional manufacturing data (step S5).

[0061] In the material quality estimation model re-learning step, the model creation unit 32 performs re-learning of the material quality estimation base model with the additional manufacturing condition data as input and the additional quality data as output to create a material quality estimation model (step S6).

[0062] (Material Quality Estimation Method) In the agent surface quality estimation method according to the embodiment, the quality estimation unit 33 performs a quality estimation step of estimating the quality of the manufactured product based on the material quality estimation model created by the model creation unit 32.

[0063] (Examples) Examples of the method for creating a material quality estimation model according to the embodiment will be described. In this example, the method for creating a material quality estimation model according to the embodiment is applied to the manufacture of a specific product.

[0064] Specifically, in the development of a new manufacturing material for steel products, machine learning was performed with the existing manufacturing condition data in the actual manufacturing equipment of the existing manufacturing material as the input and the existing quality data in the actual manufacturing equipment of the existing manufacturing material as the output, and a material quality estimation base model was created. Subsequently, transfer learning from the base model was performed with the additional manufacturing condition data in the preliminary experiment of the new manufacturing material as the input and the additional quality data in the preliminary experiment of the new manufacturing material as the output, and a material quality estimation model for the new manufacturing material data was created.

[0065] In the existing manufacturing data collection step, the manufacturing condition data (components in steel) in the steelmaking process in the actual manufacturing equipment of the existing manufacturing material and the manufacturing condition data (hot rolling temperature, hot rolling speed) in the hot rolling process were collected. Also, in the existing manufacturing data collection step, the manufacturing condition data (cold rolling speed) in the cold rolling process and the manufacturing condition data (annealing temperature, annealing time) in the annealing process were collected. Also, in the existing manufacturing data collection step, the quality data (tensile strength) of the steel plate inspected after the annealing process in the actual manufacturing equipment of the existing manufacturing material was collected.

[0066] In the existing manufacturing data creation step, the manufacturing condition data of the existing manufacturing material and the quality data of the existing manufacturing material were combined to create existing manufacturing data. In this step, the tip position and the tail end position of the product where the quality data of the existing manufacturing material was measured were associated with the measurement at the same position or the manufacturing condition data of the existing manufacturing material.

[0067] In the step of creating the material quality estimation base model, using the manufacturing condition data (chemical composition in steel, hot rolling temperature, hot rolling speed, cold rolling speed, annealing temperature, annealing time) of each process of the off-the-shelf material as input and the quality data (tensile strength) of the off-the-shelf material as output, learning of a neural network having an input layer, an intermediate layer, a linear layer, and an output layer was performed. Thereby, a material quality estimation base model was created.

[0068] In the step of collecting and creating additional manufacturing data, the manufacturing condition data (chemical composition in steel) in the steelmaking process, the manufacturing condition data (hot rolling temperature) in the hot rolling process, and the manufacturing condition data (annealing temperature, annealing time) in the annealing process in the preliminary experiment of the new manufacturing material were collected. Also, in the step of collecting and creating additional manufacturing data, the quality data (tensile strength) of the steel plate inspected after the annealing process in the preliminary experiment of the new manufacturing material was collected.

[0069] In the step of creating additional manufacturing data, the manufacturing condition data of the new manufacturing material and the quality data of the new manufacturing material were combined to create additional manufacturing data. In this step, the position of the product where the quality data of the new manufacturing material was measured was associated with the measurement at the same position or the manufacturing condition data of the new manufacturing material.

[0070] In the step of re-learning the material quality estimation model, using the various manufacturing condition data (chemical composition in steel, hot rolling temperature, annealing temperature, annealing time) of each process of the new manufacturing material as input and the quality data (tensile strength) of the new manufacturing material as output, transfer learning from the material quality estimation base model was performed. Thereby, a material quality estimation model for the new manufacturing material production data was created.

[0071] Here, in this embodiment, the types of the manufacturing condition data of the new manufacturing material for performing transfer learning were insufficient compared to the types of the off-the-shelf manufacturing condition data that are input to the material quality estimation base model. Therefore, in order to match the types of the conditions of the manufacturing condition data of the off-the-shelf material, for the insufficient hot rolling speed and cold rolling speed, for each type, the average value of that type of the off-the-shelf material was used to perform complementation and addition.

[0072] Thus, in this embodiment, a material quality estimation base model is created using manufacturing data from the actual manufacturing equipment of off-the-shelf materials, and transfer learning from the material quality estimation base model is performed using manufacturing data from a preliminary experiment of a new manufacturing material. As a result, a material quality estimation model capable of accurately estimating the material quality of a product could be created.

[0073] In the method for creating a material quality estimation model and the apparatus for creating a material quality estimation model according to the embodiment described above, a machine learning model based on existing manufacturing data is used as a base model, and relearning is performed using preliminary experiment data (additional manufacturing data) of a new manufacturing material. As a result, even when there is not enough manufacturing data when actually manufacturing a new manufacturing material, a model capable of accurately estimating the material quality can be created.

[0074] Also, in the material quality estimation method and the material quality estimation apparatus according to the embodiment, by using the model created as described above, it is possible to estimate the variation in material quality when the variation in manufacturing conditions in a new manufacturing material is assumed in advance. Further, in the material quality estimation method and the material quality estimation apparatus according to the present invention, it is possible to perform a material design of manufacturing conditions that does not deviate from the allowable range of quality variation. As a result, material development can be efficiently carried out.

[0075] As described above, the method for creating a material quality estimation model, the material quality estimation method, the apparatus for creating a material quality estimation model, the material quality estimation apparatus, and the system for creating a material quality estimation model according to the present invention have been specifically described by the forms and examples for carrying out the invention. However, the gist of the present invention is not limited to these descriptions, and should be broadly interpreted based on the description of the claims. Also, it goes without saying that various changes, modifications, etc. based on these descriptions are also included in the gist of the present invention.

[0076] For example, by configuring each part of the material quality estimation model creation device in different devices in the embodiment, a material quality estimation model creation system may be constructed. In this case, the material quality estimation model creation system includes a data creation device having the same function as the data creation unit 31 and a model creation device having the same function as the model creation unit 32.

Explanation of Signs

[0077] 1 Information processing device 10 Input unit 20 Storage unit 21 Operation DB 30 Arithmetic unit 31 Data creation unit 32 Model creation unit 33 Quality estimation unit 40 Display unit

Claims

1. A material quality estimation base model creation step in which a model creation unit included in a computer performs machine learning with existing manufacturing condition data as input and existing quality data associated with the existing manufacturing condition data as output, and creates a material quality estimation base model; A material quality estimation model re-learning step in which the model creation unit performs re-learning of the material quality estimation base model with additional manufacturing condition data as input and additional quality data associated with the additional manufacturing condition data as output, and creates a material quality estimation model; A method for creating a material quality estimation model including the above.

2. The existing manufacturing condition data is data related to existing manufacturing materials, The additional manufacturing condition data is data related to new manufacturing materials, The method for creating a material quality estimation model according to Claim 1.

3. In the material quality estimation model re-learning step, the model creation unit When the types of the additional manufacturing condition data are insufficient compared to the types of the existing manufacturing condition data that serve as the input of the material quality estimation base model, adds the types of the additional manufacturing condition data so as to match the types of the existing manufacturing condition data, and performs re-learning of the material quality estimation base model. The method for creating a material quality estimation model according to Claim 1.

4. In the material quality estimation model re-learning step, the model creation unit For the types of the additional manufacturing condition data that are insufficient for the existing manufacturing condition data, adds them by complementing each insufficient type using the statistic of the corresponding type of the existing manufacturing condition data. The method for creating a material quality estimation model according to Claim 3.

5. In the material quality estimation model re-learning step, the model creation unit For the types of the additional manufacturing condition data that are insufficient for the existing manufacturing condition data, adds them by complementing each insufficient type using sample values randomly extracted from the corresponding types of the existing manufacturing condition data. The method for creating a material quality estimation model according to Claim 3.

6. In the material quality estimation model re-learning step, the model creation unit In the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, locally extracts only the manufacturing condition data similar to the additional manufacturing condition data from the set of the existing manufacturing condition data. For the types of the additional manufacturing condition data that are lacking in the existing manufacturing condition data, for each lacking type, it is added by complementing using the statistic of the corresponding type of the existing manufacturing condition data. The method for creating a material quality estimation model according to claim 3.

7. In the material quality estimation model re-learning step, the model creation unit In the types of manufacturing condition data that already match between the existing manufacturing condition data and the additional manufacturing condition data, only the manufacturing condition data similar to the additional manufacturing condition data is locally extracted from the set of the existing manufacturing condition data. For the types of the additional manufacturing condition data that are lacking in the existing manufacturing condition data, for each lacking type, it is added by complementing using the sample values randomly extracted from the corresponding types of the existing manufacturing condition data. The method for creating a material quality estimation model according to claim 3.

8. In the material quality estimation model re-learning step, the model creation unit When the types of the additional manufacturing condition data are lacking with respect to the types of the existing manufacturing condition data that are inputs to the material quality estimation base model, the types of the existing manufacturing condition data are reduced so as to match the types of the existing manufacturing condition data, and the material quality estimation base model is created. The method for creating a material quality estimation model according to claim 1.

9. The method for re-learning the material quality estimation base model is transfer learning. The method for creating a material quality estimation model according to any one of claims 1 to 8.

10. The method for re-learning the material quality estimation base model is fine-tuning. The method for creating a material quality estimation model according to any one of claims 1 to 8.

11. A quality estimation step of estimating the quality of a manufactured product based on a material quality estimation model created by the method for creating a material quality estimation model according to any one of claims 1 to 8, which is included in a quality estimation unit provided in a computer. A material quality estimation method.

12. A model creation unit that performs machine learning with existing manufacturing condition data as an input and existing quality data associated with the existing manufacturing condition data as an output to create a material quality estimation base model. The model creation unit uses the additive manufacturing condition data as input and the additive quality data associated with the additive manufacturing condition data as output to retrain the material quality estimation base model and create a material quality estimation model. An apparatus for creating a material quality estimation model.

13. A material quality estimation apparatus comprising a quality estimation unit that estimates the quality of a manufactured product based on the material quality estimation model created by the apparatus for creating a material quality estimation model according to Claim 12.

14. It includes a model creation apparatus that performs machine learning with existing manufacturing condition data as input and existing quality data associated with the existing manufacturing condition data as output to create a material quality estimation base model. The model creation apparatus uses the additive manufacturing condition data as input and the additive quality data associated with the additive manufacturing condition data as output to retrain the material quality estimation base model and create a material quality estimation model. A system for creating a material quality estimation model.

Images