JP2022081474A - Metallic material manufacturing specification determination method, manufacturing method, and manufacturing specification determination device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a manufacturing specification determination method which improves robustness against disturbances during manufacture of a metallic material, a manufacturing method, and a manufacturing specification determination device.

SOLUTION: A manufacturing specification determination device 10 optimizes various manufacturing specifications during manufacture of a thick steel plate. Manufacturing specification optimization includes: component composition adjustment by a converter 1; adjustment of a slub dimension, a casting speed, and secondary cooling by a continuous casting machine 2; adjustment of a slub heating temperature, a slub in-furnace time, and a slab extraction temperature by a heating furnace 3; adjustment of a product dimension, rolling condition and temperature condition by a rolling machine 5; and cooling adjustment by an accelerated cooling device 6. The manufacturing specification determination device acquires at least one or more actual data that has been confirmed after a prescribed process, performs inverse analysis on the basis of one or more actual data and a prediction model in which a manufacturing specification and a material property are combined, and optimizes the manufacturing specifications after the prescribed process by feed-forward processing during manufacture so that an estimated value of the material property becomes asymptotic to a desired value.

SELECTED DRAWING: Figure 1

COPYRIGHT: (C)2022,JPO&INPIT

Description

The present invention relates to a method for manufacturing a metal material such as a steel material, a method for determining a manufacturing specification of a metal material that makes desired properties suitable, a manufacturing method for manufacturing a metal material with the manufacturing specifications determined by the method, and a manufacturing specification determination. Regarding the device.

Steel materials used for structures such as ships, marine structures, bridges, buildings, tanks and construction machines are made stronger and thinner for the purpose of rationalizing the design of the structures and reducing the weight of steel materials used. Long materials are often applied. In such steel materials, in addition to being excellent in mechanical properties such as strength and toughness and weldability, they are also excellent in impact energy absorption ability in order to ensure the structural safety of the structure against external forces. May be required.

For example, in a ship, which is an example of a structure, such a demand is increasing. This is because if there is damage to the hull due to collisions between ships and grounding, cargo, fuel, etc. may flow out and cause damage such as marine pollution. As a technique for minimizing such damage, structural efforts such as double structuring of the hull are being carried out. However, it is not realistic to apply the structure to all the structures of the ship from the viewpoint of workability and manufacturing cost. Therefore, it is desired that the steel plate itself for the hull has an energy absorbing ability to prevent the hull from being destroyed in the event of a ship collision.

Here, in Patent Document 1, the volume fraction of the ferrite phase is 75% or more in the entire plate thickness direction, the hardness is Hv140 or more and 160 or less, the average crystal grain size is 2 μm or more and 40 μm or less, and the ferrite phase in the central portion in the plate thickness direction. Described is a steel material in which the ratio of the volume fraction of the ferrite phase in the surface layer portion in the plate thickness direction to the volume fraction in the plate thickness direction is 0.925 or more and 1.000 or less to increase the uniform elongation and improve the collision resistance. Has been done.

Further, in Patent Document 2, after rolling with a cumulative rolling reduction of 30 to 98% in the austenite single phase region, accelerated cooling is performed, and then air cooling or tempering is performed, so that the area ratio of ferrite is 85% or more, ferrite. Described are steel plates having an average crystal grain size of 5 to 40 μm and cementite particles in ferrite grains having a number density of 50,000 particles / mm2 or less and having an improved ability to absorb collision energy.

Further, in Patent Document 3, acceleration cooling is performed after final pass rolling (finish rolling), then reheating to a constant temperature and accelerated cooling are performed to obtain the space factor, average particle size and maximum particle size of ferrite. Further, a steel material having an improved impact absorption by controlling the size of the second phase is described.

It has also been proposed to automatically carry out material design. For example, Patent Document 4 proposes a method of designing a material using a prediction model and an optimization calculation in order to reduce the workload related to the design of a non-metal material.

Patent No. 5953952 Patent No. 4772932 Patent No. 4476923 Patent No. 4393586

However, the prior art described in Patent Documents 1 to 3 has the following problems. That is, in the methods described in Patent Documents 1 to 3, a steel sheet having desired crystal grains and hardness is produced by combining heating, hot rolling, accelerated cooling, and heat treatment. Since the guideline for structural design is constructed semi-empirically from the experimental results at the laboratory level and the actual data on the actual machine, the manufacturing method of these steel sheets has some disturbance (component composition, dimensions, and temperature) during manufacturing. If variations occur), the desired steel material cannot always be obtained. In addition, when manufacturing new high-grade steel materials with higher strength, higher ductility, thicker wall, etc., it is necessary to repeatedly carry out experiments at the laboratory level and trial production on actual machines in order to obtain the guidelines. should not. Further, in the method described in Patent Document 4, the production of a metallic material is not mentioned, and the resistance to disturbance during production is not considered at all.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a manufacturing specification determining method, a manufacturing method, and a manufacturing specification determining apparatus capable of enhancing robustness against disturbance during manufacturing of a metal material. do.

In order to solve the above problems, the manufacturing specification determination method according to the embodiment of the present invention is:
A step to acquire at least one or more performance data determined after a predetermined process during the manufacture of a metallic material, and
Inverse analysis is performed based on the one or more actual data and the prediction model that combines the manufacturing specifications and the material properties, and the manufacturing specifications related to the predetermined process are searched for so that the estimated value of the material properties approaches the desired value. Including steps to do.

Further, the manufacturing method according to the embodiment of the present invention manufactures a metal material according to the manufacturing specifications determined by using the manufacturing specification determining method.

Further, the manufacturing specification determination device according to the embodiment of the present invention is
A communication unit that acquires at least one or more performance data that has been determined after a predetermined process during the manufacture of a metal material.
Inverse analysis is performed based on the prediction model that combines the one or more actual data with the manufacturing specifications and material properties, and the manufacturing specifications related to the predetermined process are optimized so that the estimated value of the material properties approaches the desired value. It has a search processing unit and a search processing unit.

According to the method for determining the manufacturing specifications of the metal material, the manufacturing method, and the manufacturing specification determining device according to the embodiment of the present invention, it is possible to enhance the robustness against disturbance during manufacturing.

It is a schematic diagram which shows the whole outline of the system including the manufacturing specification determination apparatus which concerns on one Embodiment of this invention. It is a block diagram of the manufacturing specification determination apparatus which concerns on one Embodiment of this invention. It is a flowchart which shows the creation process of the prediction model which concerns on one Embodiment of this invention. It is a figure which shows the outline of the neural network model. It is a flowchart which shows the optimization process which concerns on one Embodiment of this invention. It is a graph which shows the prediction accuracy. An example of the prediction result obtained by the inverse analysis is shown. It is a graph which shows the comparison between an Example and a comparative example.

(Embodiment)
Hereinafter, an embodiment of the present invention will be described.

FIG. 1 is a schematic diagram showing an overall outline of a system including a manufacturing specification determination device 10 according to an embodiment of the present invention. Hereinafter, an example in which the metal material to be designed in the present embodiment is a thick steel plate will be described, but the metal material is not limited to the thick steel plate. As shown in FIG. 1, the system according to the present embodiment includes a converter 1, a continuous casting machine 2, a heating furnace 3, a thick steel plate 4, a rolling mill 5, an accelerating cooling device 6, and a product thick steel plate 7. And the manufacturing specification determination device 10. In the manufacturing process of thick steel sheets, the raw material iron ore is first charged into a blast furnace together with limestone and coke to produce molten pig iron. The components of carbon and the like are adjusted in the converter 1 for the pig iron produced in the blast furnace, and the final components are adjusted by secondary refining. In the continuous casting machine 2, refined steel is cast to produce an intermediate material called a slab. After that, the slab is heated by the heating step in the heating furnace 3, and the product thick steel plate 7 is produced through the hot rolling step by the rolling mill 5 and the cooling step by the accelerated cooling device 6. After the cooling step, a treatment step such as a pickling step, a cold rolling step, an annealing step, and a plating step may be appropriately performed.

As a general rule, the manufacturing specification determining device 10 according to this system optimizes various manufacturing specifications during the manufacturing of thick steel sheets. To optimize the manufacturing specifications, adjust the composition of the components by the converter 1, adjust the slab dimensions by the continuous casting machine 2, the casting speed, and the secondary cooling, adjust the slab heating temperature by the heating furnace 3, the slab residence time, and the slab extraction temperature. , Product dimensions, rolling conditions, temperature conditions adjustment by the rolling mill 5, and cooling adjustment by the acceleration cooling device 6. The manufacturing specification determination device 10 performs inverse analysis based on at least one or more actual data determined after a predetermined process and a prediction model, and derives an estimated value of the material characteristics of the product thick steel sheet 7 after cooling. Then, the manufacturing specification determination device 10 feed-forwards the required control amount so that the estimated value approaches the required material property (desired value), and the converter 1, the continuous casting machine 2, and the heating furnace 3 are used. , The rolling mill 5 and the acceleration cooling device 6 are given an indicated value. That is, the manufacturing specification determination device 10 according to the embodiment of the present invention acquires at least one or more actual data determined after a predetermined step during the production of the metal material, and is based on the one or more actual data and the prediction model. Inverse analysis is performed to optimize the manufacturing specifications after a predetermined step so that the estimated value of the material property approaches a desired value. For example, the predetermined step may be a secondary refining step, and at least one or more actual data may be actual data of component composition adjustment by the converter 1. Also, for example, material properties may include uniform elongation of the metallic material. By including the uniform elongation, the manufacturing specification determination device 10 can be used when the requirement of the impact energy absorbing ability for ensuring the structural safety of the structure against the external force is satisfied. Further, for example, the material property may include at least one selected from the group of yield stress, tensile stress, toughness value, surface hardness, and cross-sectional hardness so as to satisfy a desired thick steel sheet standard.

FIG. 2 shows a block diagram of the manufacturing specification determination device 10 according to the embodiment of the present invention. The manufacturing specification determining device 10 optimizes the manufacturing specifications by performing the material property estimation process of the thick steel sheet. The manufacturing specification determination device 10 includes a device main body 200, an input unit 300, a storage unit 400, an output unit 500, and a communication unit 600. The device main body 200 transmits and receives necessary information by communicating with the input unit 300, the storage unit 400, the output unit 500, and the communication unit 600 via the bus 205, and performs information processing. In FIG. 2, the device main body 200 and the input unit 300, the storage unit 400, the output unit 500, and the communication unit 600 are connected by wire via the bus 205, but the mode of connection is not limited to this, and the connection mode is wireless. It may be connected by, or it may be a connection mode that combines wired and wireless. Each configuration of the details of the apparatus main body 200 will be described later.

The input unit 300 includes any input interface that can detect the operation of the administrator of the system, such as a keyboard, a pen tablet, a touch pad, and a mouse. The input unit 300 receives an operation related to an instruction for various processes to the apparatus main body 200.

The storage unit 400 is, for example, a hard disk drive, a semiconductor drive, an optical disk drive, or the like, and is a device that stores information necessary for this system. For example, the storage unit 400 stores the actual value of the manufacturing specification (hereinafter, referred to as “manufacturing specification actual result”) related to the thick steel sheet manufactured in the past. Further, the storage unit 400 stores the actual value of the material characteristics of the cooled thick steel sheet corresponding to the actual manufacturing specifications (hereinafter referred to as "actual material characteristics").

The output unit 500 includes any display such as a liquid crystal display and an organic EL display. The output unit 500 can display a screen based on the output data and the signal.

The communication unit 600 receives the actual data transmitted from the converter 1, the continuous casting machine 2, the heating furnace 3, the rolling mill 5, and the accelerated cooling device 6, and outputs the data to the device main body 200. Further, the communication unit 600 transmits the data related to the optimized manufacturing specifications output from the apparatus main body 200 to the converter 1, the continuous casting machine 2, the heating furnace 3, the rolling mill 5, and the acceleration cooling device 6.

As shown in FIG. 2, the apparatus main body 200 includes an arithmetic processing unit 201, a ROM 202, and a RAM 204. The ROM 202 stores the program 203. Further, the arithmetic processing unit 201, the ROM 202, and the RAM 204 are connected by a bus 205, respectively.

The arithmetic processing unit 201 includes one or more processors such as a general-purpose processor and a dedicated processor specialized for a specific process. The arithmetic processing unit 201 reads the program 203 from the ROM 202 and realizes a specific function by using the RAM 204 which is a temporary storage unit. The arithmetic processing unit 201 controls the operation of the entire apparatus main body 200.

The arithmetic processing unit 201 includes an information reading unit 206, a preprocessing unit 207, a prediction model creation unit 208, a result storage unit 209, an information reading unit 210, a characteristic estimation unit 211, and an optimization processing unit 212. It is provided with a display / transmission unit 213. When the arithmetic processing unit 201 receives an instruction for predictive model creation processing based on the operation of the input unit 300, the calculation processing unit 201 includes an information reading unit 206, a preprocessing unit 207, a prediction model creation unit 208, and a result storage unit 209. Make it work and create a predictive model. Further, when the arithmetic processing unit 201 receives an instruction for estimation processing based on the operation of the input unit 300, the information reading unit 210, the characteristic estimation unit 211, the optimization processing unit 212, and the display / transmission unit 213 are combined. Make it work and perform manufacturing specification optimization processing. Here, the optimization processing unit is also referred to as a search processing unit that searches for a solution by the optimization processing.

Next, the information processing executed by the manufacturing specification determination device 10 according to the embodiment of the present invention will be described. In this system, in order to optimize the manufacturing specifications of thick steel sheets, we first create a predictive model that connects the material properties and manufacturing specifications of thick steel sheets. Here, in the present embodiment, it is assumed that a neural network model is created as a prediction model. FIG. 3 shows a flowchart related to the prediction model creation process. When the arithmetic processing unit 201 receives an instruction for the prediction model creation processing based on the operation of the input unit 300, the arithmetic processing unit 201 executes the processing according to the flowchart shown in FIG.

When receiving the prediction model creation instruction, the information reading unit 206 of the arithmetic processing unit 201 reads the manufacturing specification results from the storage unit 400. Further, the information reading unit 206 reads the material property results corresponding to the read manufacturing specification results from the storage unit 400. Specifically, the information reading unit 206 specifies various information related to the thick steel sheet based on the ID of the rolled material (step S201).

Next, the preprocessing unit 207 processes the manufacturing specification results input in step S201 for the prediction model creation process (step S202). Specifically, the preprocessing unit 207 normalizes the actual manufacturing specifications between 0 and 1, and removes noise from missing data and abnormal data.

Subsequently, the prediction model creation unit 208 creates a prediction model. Specifically, the prediction model creation unit 208 sets hyperparameters used for the neural network model (step S203), and performs learning by the neural network model using the hyperparameters (step S204).

In order to optimize the hyperparameters, the prediction model creation unit 208 first changes some of the hyperparameters step by step with respect to the training data (manufacturing specification results of about tens of thousands). To create. Then, the prediction model creation unit 208 sets hyperparameters so that the prediction accuracy for the verification data is the highest.

Hyperparameters include the number of hidden layers, the number of neurons in each hidden layer, the dropout rate in each hidden layer (which blocks the transmission of neurons with a certain probability), and the activation function in each hidden layer. , Not limited to this. Further, the hyperparameter optimization method is not particularly limited, and a grid search in which the parameters are changed stepwise, a random search in which the parameters are randomly selected, or a search by Bayesian optimization may be used.

FIG. 4 shows a processing flow diagram of the neural network model in this system. The neural network model according to the present embodiment includes an input layer 301, an intermediate layer 302, and an output layer 303 in order from the input side.

The input layer 301 stores the manufacturing specification results normalized between 0 and 1. It is desirable that variables related to the material properties of the thick steel sheet are selected as the explanatory variables of the stored manufacturing specifications, but the number and the degree of correlation with the material properties are arbitrary.

The intermediate layer 302 contains one or more hidden layers, and each hidden layer has a certain number of neurons arranged therein. The number of hidden layers formed in the intermediate layer 302 is not particularly limited, but empirically, if there are too many hidden layers, the prediction accuracy will decrease, so it is desirable that the number of hidden layers is 10 or less. Further, the number of neurons arranged in the hidden layer is preferably in the range of 1 to 20 times the input explanatory variable.

The transmission of neurons from one neuron to the hidden layer is performed via an activation function, along with variable weighting by a weighting factor. A sigmoid function, a hyperbolic tangent function, or a ramp function may be used as the activation function.

The output layer 303 is combined with the neuron information transmitted from the intermediate layer 302 and output as an estimated value of the final material properties. Learning is performed by gradually optimizing the weighting coefficient in the neural network model based on the estimated value output by such processing and the actual measured material property.

After the weighting coefficient of the neural network model is learned, the prediction model creation unit 208 inputs the evaluation data created in step S202 into the neural network model, and obtains an estimation result for the evaluation data.

Subsequently, the result storage unit 209 stores the training data, the evaluation data, the parameters of the neural network model, and the output result of the neural network model for the training data and the evaluation data in the storage unit 400. Further, the result storage unit 209 transmits the training data, the evaluation data, the parameters of the neural network model, and the output result of the neural network model for the training data and the evaluation data to the output unit 500, and causes the output unit 500 to display the output results. (Step S205). The output unit 500 outputs, for example, the estimation result in a tabular format.

When the arithmetic processing unit 201 receives an instruction of the manufacturing specification optimization processing based on the operation of the input unit 300, the arithmetic processing unit 201 executes the manufacturing specification optimization processing. The trigger for the arithmetic processing unit 201 to perform the optimization processing is not limited to the operation of the input unit 300. For example, the optimization process may be performed with the communication unit 600 receiving at least one or more actual data determined after a predetermined step as a trigger. FIG. 5 is a flowchart showing an optimization process of manufacturing specifications.

First, the information reading unit 210 reads the manufacturing specifications predetermined before the optimization process and at least one or more actual data determined after the predetermined process for the thick steel sheet to be estimated (step S601). Further, the information reading unit 210 acquires various data related to the neural network model stored in the storage unit 400.

Next, the characteristic estimation unit 211 performs inverse analysis using the manufacturing specifications read in step S601, at least one or more actual data determined after the predetermined process, and the prediction model, and optimizes the manufacturing specifications. Specifically, the characteristic estimation unit 211 estimates the material characteristics of the thick steel sheet after rolling (step S602). Subsequently, the optimization processing unit 212 compares the estimated value of the material property after rolling of the thick steel sheet estimated by the prediction model with the value of the target material property (desired value) (step S603). ). When the absolute value of the difference between the estimated value and the desired value is equal to or more than a certain threshold value or less than a certain number of convergences, the optimization processing unit 212 changes a part of the manufacturing specifications read in step S601. In step S602, the material properties are estimated again by the prediction model. The optimization process repeats these processes and searches for the manufacturing specifications to be optimized. The optimization method is not particularly limited, but for example, a constrained least squares method or the like can be used.
In this way, by re-performing and optimizing the inverse analysis using at least one actual data determined after the predetermined step, more appropriate manufacturing specifications can be obtained by readjusting the original manufacturing specifications.

The above-mentioned changes in the manufacturing specifications are changed within the preferable range described later, but in addition, process restrictions are taken into consideration. For example, slab thickness> product thickness, slab heating temperature> rolling inlet side steel plate temperature> rolling steel plate temperature> rolling finish temperature> cooling inlet side temperature> cooling exit side temperature, but with consistent constraint conditions in the manufacturing process. If there is, it is not particularly limited.

If the absolute value of the difference between the estimated value and the desired value is within a certain threshold value or reaches a certain number of convergences, step S603 is exited and the optimized manufacturing specifications are displayed by the output unit 500. (Step S604). The optimized manufacturing specifications are transmitted to the process after the above predetermined process among the converter 1, the continuous casting machine 2, the heating furnace 3, the rolling mill 5, and the accelerated cooling device 6 via the communication unit 600. Will be. Then, the thick steel plate is manufactured according to the optimized manufacturing specifications.

Next, the manufacturing specifications transmitted from the manufacturing specification determining device 10 in the material property estimation of the present invention will be specifically described. The following description shows one embodiment of the present invention, and the present invention is not limited to the following description.

[Ingredient composition]
The steel material used for producing the thick steel sheet of the present invention has, for example, the following composition.
By mass%,
C: 0.05 to 0.16%,
Si: 0.10 to 0.50%,
Mn: 0.80 to 2.50%,
P: 0.05% or less,
S: 0.02% or less,
Cu: 1.0% or less,
Ni: 2.0% or less,
Cr: 1.0% or less,
Mo: 1.0% or less,
Nb: 0.1% or less,
V: 0.1% or less,
Ti: 0.1% or less,
B: 0.005% or less,
Ca: 0.005% or less, and W: 0.05% or less.

The reasons for limiting the amount of each component in the above component composition will be described below. In addition, "%" in the following description shall represent "mass%" unless otherwise specified.

・ C: 0.05 to 0.16%
C is an element having an effect of increasing the hardness of the matrix phase and improving the strength. In order to obtain the above effect, it is necessary to set the C content to 0.05% or more. On the other hand, when the C content exceeds 0.16%, the hardness of the matrix phase increases excessively and the elongation deteriorates. Therefore, the C content is set to 0.16% or less. Preferably, the C content is 0.07 to 0.15%.

-Si: 0.10 to 0.50%
Si is an element that acts as a deoxidizing agent and dissolves in steel to increase the hardness of the matrix phase by solid solution strengthening. In order to obtain the above effect, the Si content needs to be 0.10% or more. On the other hand, when the Si content exceeds 0.50%, the hardness of the matrix phase increases excessively, the ductility and toughness decrease, and the amount of inclusions which are the starting points of voids due to local deformation increases. Therefore, the Si content is set to 0.50% or less. Preferably, the Si content is 0.20 to 0.40%.

-Mn: 0.80 to 2.50%
Mn is an element having the effect of increasing the hardness of the matrix phase and improving the strength. In order to obtain the above effect, the Mn content needs to be 0.80% or more. On the other hand, when the Mn content exceeds 2.50%, the weldability is lowered and the hardness of the matrix phase is excessively increased. Therefore, the Mn content is set to 2.50% or less. Preferably, the Mn content is 1.00 to 2.30%.

-P: 0.05% or less P is an element contained in steel as an unavoidable impurity. P is preferably reduced as much as possible because it segregates at the grain boundaries and has an adverse effect such as lowering the toughness of the base metal and the welded portion. However, a content of 0.05% or less is acceptable. Therefore, the P content is set to 0.05% or less. On the other hand, although the lower limit of the P content is not limited, it is preferable to set the P content to 0.001% or more because an excessive reduction causes an increase in the refining cost.

-S: 0.02% or less S is an element contained in steel as an unavoidable impurity. Since S is an element that exists in steel as a sulfide-based inclusion such as MnS and has an adverse effect such as being a starting point of fracture, it is preferable to reduce it as much as possible, but the content of 0.02% or less is preferable. acceptable. Therefore, the S content is 0.02% or less. The S content is preferably 0.01% or less. On the other hand, although the lower limit of the S content is not limited, it is preferable to set the S content to 0.0005% or more because an excessive reduction causes an increase in the refining cost.

-Cu: 1.0% or less Cu is an element having an effect of increasing the hardness of the matrix phase and improving the weather resistance of the steel sheet, and can be arbitrarily added according to desired characteristics. However, if the Cu content exceeds 1.0%, the weldability is impaired and defects are likely to occur during the production of steel materials. Therefore, when Cu is added, the content is 1.0% or less. More preferably, the Cu content is 0.01-0.8%.

-Ni: 2.0% or less Ni is an element that has the effect of improving low temperature toughness and weather resistance, and also improving the hot brittleness when Cu is added, and is arbitrarily added according to the desired properties. be able to. However, if the Ni content exceeds 2.0%, the weldability is impaired and the steel material cost increases. Therefore, when Ni is added, it should be 2.0% or less. More preferably, the Ni content is 0.01-1.5%.

-Cr: 1.0% or less Cr is an element having an effect of increasing the hardness of the matrix phase and improving the weather resistance, and can be arbitrarily added depending on the desired properties. However, if the Cr content exceeds 1.0%, weldability and toughness are impaired. Therefore, when Cr is added, it should be 1.0% or less. More preferably, the Cr content is 0.01-0.8%.

-Mo: 1.0% or less Mo is an element having an effect of increasing the hardness of the matrix phase, and can be arbitrarily added according to desired properties. However, if the Mo content exceeds 1.0%, weldability and toughness are impaired. Therefore, when Mo is added, the content is 1.0% or less. More preferably, the Mo content is 0.001 to 0.8%.

Nb: 0.1% or less Nb is an element that suppresses the recrystallization of austenite during hot rolling to make it finer, and has the effect of increasing the strength by precipitating in the air cooling process after hot rolling. And can be arbitrarily added depending on the desired properties. However, if the Nb content exceeds 0.1%, a large amount of NbC is deposited and the toughness is impaired. Therefore, when Nb is added, it should be 0.1% or less. More preferably, it is 0.001 to 0.08%.

-V: 0.1% or less V, like Nb, suppresses the recrystallization of austenite during hot rolling to make it finer, and increases the strength by precipitating in the air cooling process after hot rolling. It is an element having an effect and can be arbitrarily added depending on the desired properties. However, if the V content exceeds 0.1%, a large amount of VC is deposited and the toughness is impaired. Therefore, when V is added, it should be 0.1% or less. More preferably, the V content is 0.001 to 0.08%.

-Ti: 0.1% or less Ti has a strong tendency to form a nitride and fixes N to reduce solid solution N, so that it has an effect of improving the toughness of the base metal and the welded portion. Further, when B is added, by adding Ti together, it is possible to prevent Ti from fixing N and B from precipitating as BN. As a result, the hardenability improving effect of B can be promoted, and the strength can be further improved. Therefore, it can be arbitrarily added according to the desired properties. However, if the Ti content exceeds 0.1%, a large amount of TiC is deposited and the toughness is impaired. Therefore, when Ti is added, it should be 0.1% or less. More preferably, the Ti content is 0.001 to 0.08%.

B: 0.005% or less B is an element having an effect of significantly improving hardenability and increasing strength even when added in a small amount, and can be added according to desired properties. However, when the B content exceeds 0.005%, not only the effect is saturated but also the weldability is deteriorated. Therefore, when B is added, it should be 0.005% or less. More preferably, the B content is 0.0001 to 0.004%.

-Ca: 0.005% or less Ca binds to S, suppresses the formation of MnS and the like that extend long in the rolling direction, controls the shape so that the sulfide-based inclusions have a spherical shape, and toughness of the welded portion and the like. In order to contribute to the improvement, it can be added according to the desired properties. However, when the Ca content exceeds 0.005%, not only the effect is saturated, but also the cleanliness of the steel is lowered, surface defects occur frequently, and the surface texture is deteriorated. Therefore, when Ca is added, it should be 0.005% or less. More preferably, the Ca content is 0.0001 to 0.004%.

W: 0.05% or less W increases the hardness of the matrix phase and improves the weather resistance, so that it can be added according to the desired properties. However, if the W content exceeds 0.05%, the weldability is deteriorated or the alloy cost is increased. Therefore, when W is added, it should be 0.05% or less. More preferably, the W content is 0.0001 to 0.03%.

The component composition of the thick steel sheet rolled in the present invention contains the above components, and the balance is Fe and unavoidable impurities. When O (oxygen) and N are contained as unavoidable impurities, it is preferable to suppress the O content to 0.0050% or less and the N content to 0.0050% or less. That is, when the content of O exceeds 0.0050%, the abundance ratio of inclusions on the surface of the steel sheet becomes large, so that cracks starting from inclusions are likely to occur, and the elongation may decrease. .. Similarly, when the content of N exceeds 0.0050%, the abundance ratio of inclusions on the surface of the steel sheet becomes large, so that there is a possibility that cracks originating from the inclusions are likely to occur.

[Plate thickness]
The "steel plate" in the present embodiment refers to a steel plate having a thickness of 6 mm or more according to the usual definition in the present technical field, but the cross-sectional area and cross-sectional shape of the manufactured steel material are not particularly limited. A thin plate may be used as the manufactured steel material, or a shaped steel, a rod, or a pipe shape may be used.

[Production method]
In one embodiment of the present invention, a steel material having the above-mentioned composition is sequentially subjected to the following treatments to obtain a thick steel sheet.
(1) converter / smelting (2) continuous casting (3) heating (4) hot rolling (5) cooling

[Conversion / Smelting]
The composition described above is carried out by adjusting using a conventional method at the molten steel stage until the start of casting. For example, each alloying element can be contained in the steel by adding it to the molten steel in the converter step and / or the secondary refining step. At that time, pure metals and / or alloys can be used.

[Continuous casting]
The molten steel prepared in the converter step and / or the secondary refining step is formed into a slab by continuous casting using a vertical bending type or curved continuous casting machine. At that time, a slab having a desired temperature / shape is formed by changing the cooling conditions and the casting speed of the secondary cooling equipment.

[heating]
A slab having the above composition is heated to 900-1200 ° C. If the heating temperature is less than 900 ° C., the deformation resistance of the slab in the next hot rolling step increases, the load on the hot rolling mill increases, and hot rolling becomes difficult. Therefore, the heating temperature is set to 900 ° C. or higher. The heating temperature is preferably 950 ° C. or higher. On the other hand, when the heating temperature exceeds 1200 ° C., not only the crystal grains in the middle part of the steel sheet become coarse and the toughness decreases, but also the slab surface is significantly oxidized and the unevenness of the base iron-scale interface becomes sharp. Even after the product, surface irregularities tend to remain. Such surface irregularities may become the starting point of ductile fracture due to stress concentration. Therefore, the heating temperature is set to 1200 ° C. or lower. The temperature is preferably 1150 ° C. or lower. When the slab is manufactured by a method such as continuous casting, the slab may be directly subjected to a heating step without being cooled, or may be subjected to a heating step after being cooled. Further, the heating method is not particularly limited, but for example, heating can be performed in a heating furnace according to a conventional method.

[Hot rolling]
Next, the heated slab is hot-rolled to obtain a thick steel plate. At that time, in order to secure the toughness which is the basic performance of the product steel sheet, it is necessary to miniaturize the ferrite grains through the miniaturization of the austenite grains in the middle portion of the steel sheet. Therefore, the cumulative rolling reduction in hot rolling is set to 50% or more. That is, when the cumulative reduction rate is less than 50%, the ferrite grains in the middle portion of the steel sheet are not miniaturized, regions with low brittleness are locally generated, and brittle cracks are likely to occur. Other conditions relating to the hot rolling process are not particularly limited.

[cooling]
Next, the thick steel sheet after hot rolling is cooled. In the cooling step, it is preferable to cool to room temperature. The cooling can be performed by any method, for example, air cooling or accelerated cooling.

(Example)
Hereinafter, the effects of the present invention will be specifically described based on examples, but the present invention is not limited to these examples.

In the system according to this embodiment, the manufacturing specifications of a thick steel sheet having a product plate thickness of 15 mm and showing excellent uniform elongation were optimized. As pre-learning, first, the training data was learned by a neural network model, and the manufacturing specification results and the material property results were linked.

As a result of manufacturing specifications, component composition (C, Si, Mn, P, S, Ni, Cr, V, Ti, Nb), slab dimensions (thickness / width / length), rolling dimensions (thickness / width / length) , Slab heating temperature, rolling reduction during controlled rolling, controlled rolling temperature, finish rolling temperature, cooling rate information after rolling were taken into consideration. As a result of material characteristics, a tensile test was carried out using a full-thickness test piece of JIS Z 2201 1B collected so that the plate width direction coincided with the tensile direction from the center of the width of the thick steel plate after cooling, and the yield stress (YS). ) And the total thickness was sought and used for learning. For the number of training data, a total of 470 samples from which noise such as missing data was removed were used. The hyperparameters used for the neural network were searched by Bayesian optimization based on the Gaussian probability distribution and set as follows.
・ Number of epochs (number of repetitions of learning): 813
-Number of neurons in the hidden layer: 328
Dropout ratio (probability of blocking neuronal transmission): 0.3
・ Number of hidden layers: 8
・ Optimization method of learning weighting factor: Adam

A graph of prediction accuracy is shown in FIG. Verification of learning accuracy was performed by cross validation. The model prediction error was σ2.44%. As a target material property, a tensile strength of 440 MPa or more was accepted. The target elongation characteristic was 30%.

In this embodiment, the material properties of the steel slab whose component composition and slab size have already been determined are estimated by a neural network model, and the estimated material properties are continued so as to approach the target material properties. The required control amount was calculated for the process after the heating furnace. Specifically, the difference between the estimated value and the target value was obtained for each of the yield stress and the total thickness elongation, and the calculation was performed using the sequential least squares method so that the sum of the differences between the two was the minimum value. The number of repetitions until convergence was 500. Each explanatory variable obtained by calculation was given as an indicated value to the heating furnace 3, the rolling mill 5, and the accelerating cooling device 6.

In the comparative example, the product plate thickness was also set to 15 mm, and rolling was performed in the same manner as in the example except that the product was manufactured in a suitable range without using a neural network model.

FIG. 7 shows the upper and lower limits of the explanatory variables in the examples and the prediction results obtained by the inverse analysis. In FIG. 7, at least one or more slab thickness, slab width, slab length, and component composition (C, Si, Mn, P, S, Ni, Cr, V, Ti) are determined after a predetermined step during manufacturing. Corresponds to the actual data of. Based on the actual data, other manufacturing specifications, that is, rolling thickness, rolling width, rolling length, heating furnace temperature, rolling reduction during controlled rolling, controlled rolling temperature, finish rolling temperature, and cooling rate are optimized. Was readjusted. As a result of the optimization, the elongation characteristic that can be maximized within the upper and lower limit ranges of the explanatory variables was estimated to be 26.9%. FIG. 8 shows the relationship between the yield stress (YS) and the uniform elongation (EL) when actually manufacturing with the optimized manufacturing specifications obtained by the inverse analysis. By using the method of this example, it was possible to obtain a thick steel sheet having a higher uniform elongation while maintaining the yield stress as compared with the comparative example.

As shown in the above embodiment, according to the manufacturing specification determination method according to the embodiment of the present invention, the manufacturing guideline of a steel sheet having high uniform elongation and excellent impact resistance is accurately predicted, and the target material characteristics are predicted. Since it is possible to manufacture a thick steel sheet having the above, it can greatly contribute to the improvement of productivity and the speed of steel material development. In addition, in order to optimize the manufacturing specifications using at least one actual data determined after the predetermined process of the manufacturing process, even if a disturbance occurs during the manufacturing process, the manufacturing specifications take into consideration the change in the situation during the manufacturing process. Can be obtained. In other words, according to the manufacturing specification determination method according to the embodiment of the present invention, it is possible to enhance the robustness against disturbance during manufacturing of a metal material. Further, according to the manufacturing specification determination method according to the embodiment of the present invention, it is possible to manufacture a thick steel sheet having the target material properties without repeating experiments and trial production on an actual machine, and thus productivity. It can greatly contribute to the improvement of the steel material development speed. Therefore, the manufacturing specification determination method according to the embodiment of the present invention is structurally safe for, for example, steel plates having excellent impact resistance, particularly ships, marine structures, bridges, buildings, tanks and construction machines. It can be suitably used for welded structures for which properties are strongly required.

Although the present invention has been described with reference to the drawings and examples, it should be noted that those skilled in the art can easily make various modifications and modifications based on the present disclosure. Therefore, it should be noted that these modifications and modifications are within the scope of the present invention. For example, the functions included in each means, each step, etc. can be rearranged so as not to be logically inconsistent, and a plurality of means, steps, etc. can be combined or divided into one. ..

For example, the present invention can also be realized as a program describing the processing content that realizes each function of the above-mentioned manufacturing specification determination device 10 or as a storage medium on which the program is recorded. It should be understood that these are also included in the scope of the present invention.

For example, although an example of creating a prediction model in the manufacturing specification determination device 10 according to the present embodiment is shown, these may be realized by another information processing device. In this case, the manufacturing specification results and material property results required for the information processing apparatus to create a prediction model are aggregated to create a prediction model. Further, the information processing apparatus transmits the created mathematical model to the manufacturing specification determining apparatus 10.

Further, for example, the manufacturing specification determination device 10 according to the present embodiment applies a neural network as an algorithm for creating a prediction model, but the present invention is not limited to this, and any algorithm can be applied. For example, as an algorithm for creating a predictive model, statistical methods and machine learning methods such as local regression, support vector machine, neural network, or random forest can be applied. In other words, the prediction model may be a deep learning model learned based on actual data of manufacturing specifications and material properties, or a machine learning model including a statistical learning model.

1 converter 2 continuous casting machine 3 heating furnace 4 thick steel plate 5 rolling mill 6 acceleration cooling device 7 product thick steel plate 10 manufacturing specification determination device 200 device body 201 arithmetic processing unit 202 ROM
203 Program 204 RAM
205 Bus 206 Information reading unit 207 Preprocessing unit 208 Predictive model creation unit 209 Result storage unit 210 Information reading unit 211 Characteristic estimation unit 212 Optimization processing unit (search processing unit)
213 Display / transmission unit 300 Input unit 301 Input layer 302 Intermediate layer 303 Output layer 400 Storage unit 500 Output unit 600 Communication unit

Claims (7)

A step to acquire at least one or more performance data determined after a predetermined process during the manufacture of a metallic material, and
Inverse analysis is performed based on the one or more actual data and the prediction model that combines the manufacturing specifications and the material properties, and the manufacturing specifications related to the predetermined process are manufactured so that the estimated value of the material properties approaches the desired value. A method for determining manufacturing specifications of the metallic material, including a step of optimizing by a feed-forward process in the process. The manufacturing specification determination method according to claim 1, wherein the predetermined step is a secondary refining step, and the one or more actual data includes the actual data after adjusting the component composition. The manufacturing specification determination method according to claim 1 or 2, wherein the metal material is a thick steel plate. The manufacturing specification determination method according to any one of claims 1 to 3, wherein the material properties include uniform elongation. The prediction model according to any one of claims 1 to 4, wherein the prediction model is a deep learning model learned based on the manufacturing specifications and actual data of the material properties, or a machine learning model including a statistical learning model. Manufacturing specification determination method. A manufacturing method for manufacturing a metal material with manufacturing specifications determined by using the manufacturing specification determining method according to any one of claims 1 to 5. A communication unit that acquires at least one or more performance data that has been determined after a predetermined process during the manufacture of a metal material.
Inverse analysis is performed based on the prediction model that combines the one or more actual data with the manufacturing specifications and material properties, and the manufacturing specifications related to the predetermined process are being manufactured so that the estimated value of the material properties approaches the desired value. A manufacturing specification determination device having a search processing unit optimized by feed-forward processing.

Images