Method and system for predicting ballistic performance of martensitic steel

Abstract

The present invention provides a method and a system for predicting ballistic performance of martensitic steel by using a microstructure image to predict hardness of martensitic steel. The method for predicting ballistic performance of martensitic steel, according to one embodiment of the present invention, comprises the steps of: acquiring a martensite microstructure image of martensitic steel; classifying a martensite phase from the martensite microstructure image; calculating hardness of the martensitic steel on the basis of the content of the classified martensite phase; and calculating ballistic performance from the hardness of the martensitic steel.

Description

Method and System for Predicting Ballistic Performance of Martensitic Steel

[0001] The technical concept of the present invention relates to a method and system for predicting the ballistic performance of martensitic steel.

[0002] Martensitic steel forms a martensitic phase through rapid cooling and possesses characteristics of high strength and hardness. Such martensitic steel consists of a martensite structure and has a surface hardness ranging from HB 350 to HB 550. However, for the user to actually verify the surface hardness values ​​presented by the supplier, physical tests must be performed. For example, one method to verify martensitic steel is through Electron Backscatter Diffraction (EBSD) analysis, which checks for the presence of large-angle grain boundaries with an orientation difference of 15 degrees or more within the martensitic structure. However, this method has the disadvantage of requiring cumbersome preliminary preparation work and being difficult for the supplier to operate. Additionally, analysis methods using a transmission electron microscope are cumbersome and pose significant difficulties for the user to perform. Furthermore, physical destructive tests such as hardness testing and ballistic testing require specific measurement equipment or cannot be performed without the cooperation of government agencies.

[0003] The technical problem that the technical concept of the present invention aims to solve is to provide a method and system for predicting the ballistic performance of martensitic steel using microstructure images.

[0004] In addition, the technical problem that the technical concept of the present invention aims to solve is to provide a deep learning-based ballistic performance prediction method and system for martensitic steel that predicts the ballistic performance of martensitic steel using microstructure images.

[0005] However, these tasks are exemplary, and the technical concept of the present invention is not limited thereto.

[0006] According to one aspect of the present invention, a method and system for predicting the ballistic performance of martensitic steel are provided, which predicts the ballistic performance of martensitic steel using microstructure images.

[0007] According to one embodiment of the present invention, a method for predicting the ballistic performance of the martensitic steel may include: a step of acquiring a martensitic microstructure image of the martensitic steel; a step of classifying martensitic phases from the martensitic microstructure image; a step of calculating the hardness of the martensitic steel based on the content of the classified martensitic phases; a step of establishing a correlation between the hardness of the martensitic steel and the ballistic performance; and a step of calculating the ballistic performance from the hardness of the martensitic steel using the correlation.

[0008] According to one embodiment of the present invention, in the step of establishing the correlation, the correlation may use the following formula.

[0009] Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661

[0010] According to one embodiment of the present invention, the step of classifying the martensite phase may classify the martensite phase according to the width of the lath of the martensite phase.

[0011] According to one embodiment of the present invention, the step of classifying the martensite phase may be performed by classifying the martensite phase into a first martensite phase including a first lath having a width of more than 1 μm and less than or equal to 20 μm; a second martensite phase including a second lath having a width of more than 0.1 μm and less than or equal to 1 μm; and a third martensite phase not including lath.

[0012] According to one embodiment of the present invention, the step of classifying the martensite phase may classify the martensite phase according to at least one of the width of the lath of the martensite phase, the length of the lath, the ratio of the width to the length of the lath, and whether the lath contains a cementite phase.

[0013] According to one embodiment of the present invention, the step of calculating the hardness of the martensite steel can calculate the hardness of the martensite steel according to the respective contents of the first martensite phase, the second martensite phase, and the third martensite phase.

[0014] According to one embodiment of the present invention, in the step of calculating the hardness of the martensitic steel, the hardness may be at least one of Brinell hardness, Vickers hardness, Rockwell hardness, and Shore hardness.

[0015] According to one embodiment of the present invention, a ballistic performance prediction system for the martensitic steel may include: an image acquisition unit for acquiring an image of the martensitic microstructure of the martensitic steel; a martensitic phase classification unit for classifying the martensitic phase from the image of the martensitic microstructure; a hardness calculation unit for calculating the hardness of the martensitic steel based on the content of the classified martensitic phase; and a ballistic performance calculation unit for calculating the ballistic performance from the hardness of the martensitic steel using correlation information between the hardness of the martensitic steel and the ballistic performance.

[0016] According to one embodiment of the present invention, the correlation information may use the following formula.

[0017] Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661

[0018] According to one embodiment of the present invention, the martensite phase classification unit can classify the martensite phase according to the width of the lath of the martensite phase.

[0019] According to one embodiment of the present invention, the martensite phase classification unit may classify the martensite phase into a first martensite phase including a first lath having a width of more than 1 μm and less than or equal to 20 μm; a second martensite phase including a second lath having a width of more than 0.1 μm and less than or equal to 1 μm; and a third martensite phase not including lath.

[0020] According to one embodiment of the present invention, the martensite phase classification unit includes a first database having martensite classification information according to at least one of the width of the lath of the martensite phase, the length of the lath, the ratio of the width to the length of the lath, and whether the lath includes a cementite phase, and can classify the martensite phase according to the martensite classification information provided in the first database.

[0021] According to one embodiment of the present invention, the hardness calculation unit can calculate the hardness of the martensite steel according to the respective contents of the first martensite phase, the second martensite phase, and the third martensite phase.

[0022] According to one embodiment of the present invention, the hardness calculation unit includes a second database having hardness information of the martensite steel according to the content of the classified martensite phase, and can calculate the hardness of the martensite steel according to the hardness information of the martensite provided in the second database.

[0023] According to one embodiment of the present invention, the ballistic performance calculation unit includes a third database having correlation information between the hardness of the martensitic steel and ballistic performance, and can calculate the ballistic performance of the martensitic steel according to the correlation information between the hardness and ballistic performance provided in the third database.

[0024] According to one embodiment of the present invention, a method for predicting the ballistic performance of a martensitic steel may include: a step of preparing a plurality of data sets consisting of training martensitic microstructure image data, training hardness data, and training ballistic performance data of a training martensitic steel; a step of having a computer system perform deep learning using the training martensitic microstructure image data as an input value and the training hardness data and the training ballistic performance data as output values; a step of providing actual martensitic microstructure image data of a martensitic steel to be predicted; and a step of inputting the actual martensitic microstructure image data into the deep-learned computer system to predict the hardness data and ballistic performance data of the material to be predicted.

[0025] According to the technical concept of the present invention, a method for predicting the ballistic performance of martensitic steel classifies the martensitic phase according to the Ras structure of the microstructure observed with a scanning electron microscope, and through this, the hardness and ballistic performance of the martensitic steel can be easily predicted.

[0026] The effects of the present invention described above are illustrative and the scope of the present invention is not limited by these effects.

[0027] FIG. 1 is a flowchart illustrating a method for predicting the bulletproof performance of martensitic steel according to an embodiment of the present invention.

[0028] FIG. 2 is a block diagram illustrating a bulletproof performance prediction system for martensitic steel according to an embodiment of the present invention.

[0029] FIG. 3 is a schematic diagram illustrating a method for classifying the microstructure of martensitic steel used in a method for predicting the bulletproof performance of martensitic steel according to an embodiment of the present invention.

[0030] FIG. 4 is a scanning electron microscope image showing hardness values ​​calculated using the microstructure of a martensitic steel used in the method for predicting the bulletproof performance of a martensitic steel according to an embodiment of the present invention.

[0031] FIG. 5 is a flowchart illustrating a method for predicting the bulletproof performance of martensitic steel based on deep learning according to an embodiment of the present invention.

[0032] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The embodiments of the present invention are provided to more completely explain the technical concept of the present invention to those skilled in the art, and the following embodiments may be modified in various different forms, and the scope of the technical concept of the present invention is not limited to the following embodiments. Rather, these embodiments are provided to make the present disclosure more faithful and complete and to fully convey the technical concept of the present invention to those skilled in the art. In this specification, the same reference numerals denote the same elements throughout. Furthermore, various elements and areas in the drawings are depicted schematically. Accordingly, the technical concept of the present invention is not limited by the relative sizes or spacing depicted in the attached drawings.

[0033] FIG. 1 is a flowchart illustrating a method (S100) for predicting the bulletproof performance of martensitic steel according to an embodiment of the present invention.

[0034] Referring to FIG. 1, a method for predicting the ballistic performance of martensitic steel (S100) comprises the steps of: acquiring a martensitic microstructure image of martensitic steel (S110); classifying martensitic phases from the martensitic microstructure image (S120); calculating the hardness of the martensitic steel based on the content of the classified martensitic phases (S130); establishing a correlation between the hardness of the martensitic steel and the ballistic performance (S140); and calculating the ballistic performance from the hardness of the martensitic steel using the correlation (S150).

[0035] The step (S110) of acquiring a martensite microstructure image of the above-mentioned martensite steel can be performed by acquiring a microstructure image obtained using a scanning electron microscope. For example, the step (S110) can be performed by directly receiving the microstructure image from the scanning electron microscope. Alternatively, it can be performed by acquiring image data by scanning a printed photograph or a displayed screen of the microstructure image observed using the scanning electron microscope, or by receiving digital data regarding the microstructure image from the scanning electron microscope.

[0036] The step (S120) of classifying the martensite phase from the above martensite microstructure image may classify the martensite phase according to the width of the lath of the martensite phase. For example, the step (S120) may be performed by classifying the martensite phase into a first martensite phase including a first lath having a width of more than 1 μm and less than or equal to 20 μm; a second martensite phase including a second lath having a width of more than 0.1 μm and less than or equal to 1 μm; and a third martensite phase not including lath. Since the first martensite phase, the second martensite phase, and the third martensite phase have very different shape characteristics, it may be easy to determine the microstructure.

[0037] The first martensite phase is in the form of having a relatively large first lath. The first lath may have a width of, for example, greater than 1 μm and less than or equal to 20 μm, and a length of, for example, 5 μm to 50 μm. The length is larger than the width. The first martensite phase may be composed of an initial martensite structure formed by an initial transformation from the austenite interface immediately below the Ms temperature. The first lath has cementite precipitated therein, for example, with a size of 20 nm or less, for example, 5 nm to 20 nm. The cementite may be composed of hexagonal cementite. Additionally, the first lath does not contain Fe-carbide.

[0038] The second martensite phase is in the form of having a second lath that is relatively smaller than the first martensite phase. The second lath may have a width of, for example, 0.1 μm or more and 1 μm or less, and a length of, for example, 5 μm to 50 μm. The length is greater than the width. In addition, the ratio of length (L) to width (T) (T / L) may be 0.5 to 0.7. The second lath may be positioned surrounding the first lath. The second lath has a very thin shape and may have a width of 1 / 10 to 1 / 20 of that of the first lath. The second lath may exhibit a shape in which the length is significantly longer than the thickness. The second lath may not contain cementite precipitates inside.

[0039] The third martensite phase may include all cases where Ras is absent and may have a length of 1 μm to 15 μm. The third martensite phase may include a region that is not etched. The third martensite phase may include a region where conventional etching does not occur, and accordingly, Ras does not appear, so it can be distinguished from the first and second martensite phases. The second martensite phase may not contain cementite precipitates internally.

[0040] The distinction between the first martensite phase, the second martensite phase, and the third martensite phase is exemplary, and the present invention is not limited thereto. That is, cases in which the martensite phases are distinguished and applied in various ways are also included within the technical scope of the present invention.

[0041] For example, the step (S120) of classifying the martensite phase from the above martensite microstructure image can classify the martensite phase according to at least one of the width of the lath of the martensite phase, the length of the lath, the ratio of the width to the length of the lath, and whether the lath contains a cementite phase.

[0042] The step (S130) of calculating the hardness of the martensite steel based on the content of the classified martensite phase can calculate the hardness of the martensite steel according to the respective contents of the first martensite phase, the second martensite phase, and the third martensite phase. The martensite steel can be classified based on the hardness of the martensite steel calculated in this way.

[0043] The above hardness may be a value measured using various hardness measurement methods. The above hardness may be, for example, at least one of Brinell hardness, Vickers hardness, Rockwell hardness, and Shore hardness.

[0044] The above method for predicting the ballistic performance of martensitic steel can be applied to martensitic steel having a martensitic transformation start temperature of 400°C or higher. That is, it can be applied when the reustenizing temperature is 400°C or higher. However, these temperatures are exemplary, and the technical concept of the present invention is not limited thereto.

[0045] The step of establishing the correlation (S140) can be performed by establishing a correlation between the hardness and ballistic performance of the martensitic steel. For example, the correlation can be derived by directly measuring the hardness and ballistic performance of the martensitic steel and mathematically analyzing them. Alternatively, the correlation can be established based on various reported data regarding the hardness and ballistic performance of the martensitic steel.

[0046] Alternatively, in the step of establishing the correlation (S140), the correlation may use the following Equation 1.

[0047] <Equation 1>

[0048] Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661

[0049] For example, the Brinell hardness may be in the range of 350 HB to 550 HB. The ballistic performance may be V50 ballistic performance based on US military standard (MIL) 46100 and may be in the range of 380 m / sec to 860 m / sec. This may be for a 7.62 mm armor-piercing (Armor-Piercing M2) bullet incident at an angle of 30 degrees.

[0050] Here, the V50 ballistic performance mentioned above is a figure obtained through tests specified in the United States and Korea to numerically express ballistic performance, for example, by firing at least 6 rounds at various bullet velocities from a certain distance to secure 3 rounds of complete penetration and 3 rounds of partial penetration velocities, and then deriving a 50% penetration probability based on this using a statistical formula.

[0051] The step of calculating the ballistic performance (S150) can be performed by calculating the ballistic performance from the hardness of the martensitic steel using the correlation.

[0052] However, this is exemplary, and the step (S150) of calculating the ballistic performance may be performed by calculating the ballistic performance using various correlations between hardness and ballistic performance, such as the correlation between hardness and ballistic performance established based on data in literature.

[0053] FIG. 2 is a block diagram illustrating a bulletproof performance prediction system (100) for martensitic steel according to an embodiment of the present invention.

[0054] Referring to FIG. 2, a ballistic performance prediction system (100) for martensitic steel comprises: an image acquisition unit (110) for acquiring an image of the martensitic microstructure of the martensitic steel; a martensitic phase classification unit (120) for classifying the martensitic phase from the image of the martensitic microstructure; a hardness calculation unit (130) for calculating the hardness of the martensitic steel based on the content of the classified martensitic phase; and a ballistic performance calculation unit (140) for calculating the ballistic performance from the hardness of the martensitic steel using correlation information between the hardness of the martensitic steel and the ballistic performance.

[0055] The image acquisition unit (110) can acquire an image of the martensite microstructure of the martensite steel.

[0056] The image acquisition unit (110) can acquire a microstructure image of the martensitic steel obtained, for example, using a scanning electron microscope (112).

[0057] The image acquisition unit (110) is configured to include, for example, a scanning electron microscope (112), so that it can be directly connected to the scanning electron microscope and receive microstructure images directly.

[0058] The image acquisition unit (110) may acquire image data by, for example, scanning a printed photograph or a displayed screen of the microstructure image observed using a separately installed scanning electron microscope, or by receiving digital data regarding the microstructure image from the scanning electron microscope. However, this is exemplary, and cases where various devices capable of providing microstructure images in place of the scanning electron microscope are applied are also included within the technical scope of the present invention.

[0059] The martensite phase classification unit (120) can classify the martensite phase from the martensite microstructure image.

[0060] The martensite phase classification unit (120) can classify the martensite phase according to, for example, the width of the lath of the martensite phase.

[0061] The martensite phase classification unit (120) may classify, for example, the martensite phase into a first martensite phase including a first lath having a width of more than 1 μm and less than or equal to 20 μm; a second martensite phase including a second lath having a width of more than 0.1 μm and less than or equal to 1 μm; and a third martensite phase not including lath. However, such classification is exemplary and the technical concept of the present invention is not limited thereto.

[0062] The martensite phase classification unit (120) may include a first database (122) having martensite classification information according to at least one of, for example, the width of the lath, the length of the lath, the ratio of the width to the length of the lath, and whether the lath includes a cementite phase. Accordingly, the martensite phase classification unit (120) can classify the martensite phase according to the martensite classification information provided in the first database (122).

[0063] For example, the first database (122) may provide martensite classification information such that, for example, a phase having a first lath with a width greater than 1 μm and less than or equal to 20 μm is corresponded to a first martensite phase, a phase having a second lath with a width greater than 0.1 μm and less than or equal to 1 μm is corresponded to a second martensite phase, and a phase not containing lath is corresponded to a third martensite phase.

[0064] The hardness calculation unit (130) can calculate the hardness of the martensite steel based on the content of the classified martensite phase.

[0065] The hardness calculation unit (130) may include a second database (132) having hardness information of the martensite steel according to the content of the classified martensite phase. Accordingly, the hardness calculation unit (130) can calculate the hardness of the martensite steel according to the hardness information of the martensite provided in the second database (132).

[0066] For example, the second database (132) may provide information on the hardness of the martensite, such that when the content of the first martensite phase is 30%, the content of the second martensite phase is 50%, and the content of the third martensite phase is 20%, the hardness of the martensite steel is corresponded to HB400. Alternatively, when the content of the first martensite phase is 0%, the content of the second martensite phase is 70%, and the content of the third martensite phase is 30%, the hardness of the martensite steel is corresponded to HB360.

[0067] The bulletproof performance calculation unit (140) can calculate the bulletproof performance from the hardness of the martensite steel by using the correlation between the hardness of the martensite steel and the bulletproof performance.

[0068] The ballistic performance calculation unit (140) may include a third database (142) having correlation information between the hardness and ballistic performance of the martensite steel. Accordingly, the ballistic performance calculation unit (140) can calculate the ballistic performance of the martensite steel based on the correlation information between the hardness and ballistic performance provided in the third database (142).

[0069] The above correlation information can use the following Equation 1.

[0070] <Equation 1>

[0071] Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661

[0072] However, this is exemplary and the technical concept of the present invention is not limited thereto.

[0073] The first database (122), the second database (132), and the third database (142) may be separate databases or integrated databases.

[0074] Even in the case of martensitic steel manufactured with the same compositional system, material variations may occur due to process conditions such as cooling rate and cooling time. That is, changes may occur in the presence or content of the first to third martensitic phases. Manufacturers of martensitic steel supply materials with a hardness difference of ± 30 to 40, but it is difficult for users to verify this.

[0075] For example, in HB 400, all of the first to third martensite phases may appear, but in HB 360, the first martensite phase may not appear, and only the second and third martensite phases may appear. In HB 520, the first martensite phase and the second martensite phase may appear, and the third martensite phase may not appear.

[0076] Through the technical concept of this invention, a method is presented that enables martensitic steel manufacturers and users to predict the physical properties of materials by providing a microstructure analysis method in addition to physical testing for steel development and use. By inferring the expected degradation of physical properties when the phase essential for the low-carbon martensitic microstructure composition is not identified, a more precise method for using the steel can be presented.

[0077] FIG. 3 is a schematic diagram illustrating a method for classifying the microstructure of martensitic steel used in a method for predicting the bulletproof performance of martensitic steel according to an embodiment of the present invention.

[0078] Referring to FIG. 3, a microstructure image of a martensitic steel is shown. After extracting the red region from the microstructure image, the martensitic phase is classified according to the width of the lath. Cases where the lath width is relatively large, for example, having a width greater than 1 μm and less than or equal to 20 μm, are classified as the first martensitic phase and designated as region "A". Region A is an area without precipitates, for example, appearing white. Subsequently, cases where the lath width is relatively small, for example, having a width greater than 0.1 μm and less than or equal to 1 μm, are classified as the second martensitic phase and designated as region "B". Subsequently, the remaining area, or the area where lath is not observed, is designated as region "C". Region "C" may not appear.

[0079] For example, as the carbon content increases, the above-mentioned region A may increase, and the above-mentioned region B may decrease. Brinell hardness can be calculated by considering the size of the above-mentioned region A, the size of the above-mentioned region B, and the carbon content. Therefore, the consistency of the manufacturing process can be verified from the analysis of carbon content and the balance between the above-mentioned region A and the above-mentioned region B, and hardness can be predicted without performing physical tests.

[0080] Accordingly, the classification of the martensite phases in the above martensite microstructure image can be completed. In the case of the above martensite steel, it may be formed to include a first martensite phase, a second martensite phase, and a third martensite phase.

[0081] The hardness and ballistic performance of the martensitic steel can be calculated based on the content of the martensitic phase classified in this manner. The martensitic steel can be classified based on the hardness and ballistic performance of the martensitic steel calculated in this manner.

[0082] In addition, in the method for predicting the bulletproof performance of a martensitic steel according to one embodiment of the present invention, the hardness can be calculated based on the carbon content. For example, the hardness can be calculated by the following Equation 2. The hardness below may be Brinell hardness.

[0083] <Equation 2>

[0084] Brinell Hardness (HBW) = 330.08 + 527.3x[C] + 336.62x[C] 2 - 2.705 x[C] 3 - 107.02 x[C] 4 + 43.523 x[C] 5

[0085] (Here, [C] is the carbon content in the steel, in weight percent)

[0086] For reference, HBW stands for Brinell tungsten hardness and indicates the Brinell hardness of a tungsten indenter.

[0087] For example, a method for predicting the ballistic performance of a martensitic steel according to one embodiment of the present invention may include the step of calculating hardness based on the carbon content of the martensitic steel; the step of establishing a correlation between the hardness of the martensitic steel and the ballistic performance; and the step of calculating the ballistic performance from the hardness of the martensitic steel using the correlation.

[0088] In this case, information regarding the martensite microstructure can be predicted in advance based on the carbon content and the manufacturing process conditions below. Therefore, it is possible to calculate the hardness without performing a step of classifying the martensite phase.

[0089] A method for predicting the bulletproof performance of martensitic steel according to one embodiment of the present invention can be performed based on deep learning.

[0090] Experimental Example

[0091] Preferred experimental examples are presented below to aid in understanding the present invention. However, the following experimental examples are intended only to aid in understanding the present invention and do not limit the present invention. Details not described herein can be sufficiently technically inferred by those skilled in the art, so their description is omitted.

[0092] Table 1 shows the composition of the martensitic steel used in the method for predicting the bulletproof performance of martensitic steel according to an embodiment of the present invention. The content unit of each component is weight%. In Table 1, the remainder consists of iron (Fe) and impurities inevitably contained in the steelmaking process, etc.

[0093] Classification CsiMnAlPSFe Experimental Example 1 0.26 0.28 0.7 20.0 30.0 1 20.00 21 Residue Experimental Example 2 0.27 0.3 20.76 0.0 20.0 230.00 22 Residue Experimental Example 3 0.29 0.3 30.66 0.0 30.0 220.00 18 Residue

[0094] The above-mentioned martensite steel was manufactured by performing reheating, hot rolling, normalizing, and quenching on the above-mentioned steel under process conditions as shown in Table 2. The above-mentioned martensite steel was formed into a scanning electron microscope specimen, and after etching by the Nital etching method, the microstructure was observed through a scanning electron microscope.

[0095] Table 2 shows the process conditions for forming martensitic steel.

[0096] Separate reheating temperature ( o C) Rolling end temperature ( o C) Normalizing( o C) Quenching cooling rate ( o C / sec) Experimental Example 1 1 5094590551 Experimental Example 2 1 1 7094490550 Experimental Example 3 1 1 9093590550

[0097] Referring to Table 2, the process conditions for forming martensitic steel are shown. However, this is exemplary, and the process conditions for forming martensitic steel can be varied. The present invention, for example, 1000 o C ~ 1250 o Reheat temperature in the C range, 800 o C ~ 1000 o Rolling finish temperature in the C range, 850 o C ~ 950 o Normalizing temperature in the C range, and 30 o C / sec ~ 100 o Process conditions may include quenching cooling rates in the range of C / second.

[0098] Table 3 shows the Brinell hardness and ballistic performance of martensitic steel.

[0099] Classification Brinell Hardness Calculated Value Ballistic Performance Predicted Value Ballistic Performance Measured Value Measurement Error (%) Experiment Example 1 4897137393.5 Experiment Example 2 4967297583.8 Experiment Example 3 5037467632.2

[0100] Referring to Table 3, Brinell hardness was calculated based on the microstructure of the martensitic steel, and a predicted ballistic performance value was calculated from the Brinell hardness using the correlation of Equation 1. In addition, the ballistic performance of the martensitic steel was directly measured using the V50 ballistic performance test method.

[0101] The measurement error was calculated using Equation 3 below.

[0102] <Equation 3>

[0103] Measurement Error (%) = ((Measured Ballistics - Predicted Ballistics) x 100) / Measured Ballistics

[0104] The above measurement error was found to be less than 4%, which indicates that the predicted ballistic performance value and the measured ballistic performance value match well.

[0105] Therefore, the method for predicting the ballistic performance of martensitic steel according to the present invention can predict ballistic performance with high reliability based on the martensitic microstructure without directly measuring the ballistic performance.

[0106] For reference, it was confirmed that the Brinell hardness calculated based on the carbon content according to Equation 2 above matches well with the Brinell hardness calculated based on the microstructure of the martensitic steel.

[0107] FIG. 4 is a scanning electron microscope image showing hardness values ​​calculated using the microstructure of a martensitic steel used in the method for predicting the bulletproof performance of a martensitic steel according to an embodiment of the present invention.

[0108] Referring to FIG. 4, when the martensitic steel contains all of the first martensitic phase (indicated by A), the second martensitic phase (indicated by B), and the third martensitic phase (indicated by C), the hardness of the martensitic steel can be calculated as HB400. On the other hand, when the martensitic steel does not contain the first martensitic phase but contains the second martensitic phase (indicated by B) and the third martensitic phase (indicated by C), the hardness of the martensitic steel can be calculated as HB360.

[0109] Deep learning-based

[0110] FIG. 5 is a flowchart illustrating a method for predicting the bulletproof performance of martensitic steel based on deep learning according to an embodiment of the present invention.

[0111] Referring to FIG. 5, a method (S200) for predicting the ballistic performance of martensitic steel based on deep learning comprises: a step (S210) of preparing a plurality of data sets consisting of training martensitic microstructure image data, training hardness data, and training ballistic performance data of training martensitic steel; a step (S220) of having a computer system perform deep learning using the training martensitic microstructure image data as an input value and the training hardness data and the training ballistic performance data as output values; a step (S230) of providing actual martensitic microstructure image data of the martensitic steel to be predicted (S240); and a step (S240) of inputting the actual martensitic microstructure image data into the deep-learned computer system to predict the hardness data and ballistic performance data of the material to be predicted.

[0112] In this specification, when martensite microstructure image data, hardness data, and ballistic performance data are included in a dataset for deep learning, the term "for training" is added to the name, and when measured by actual experiments, the term "actual" is added to the name. Of course, to increase learning accuracy, actual martensite microstructure image data, actual hardness data, and actual ballistic performance data may be used in the deep learning of an artificial neural network.

[0113] The aforementioned deep learning is a term widely used in the field of artificial intelligence, and the present invention does not specifically limit the method thereof. According to one embodiment of the present invention, by inputting the data set into an artificial neural network (or artificial intelligence) and performing deep learning (or machine learning), an artificial neural network can be constructed that predicts an output value when an input value is input. The principle by which such an artificial neural network is constructed is similar to a regression problem for a general linear function. That is, it involves creating a function that indicates the relationship between the input and output values ​​of a given data set. The deep learning method of the artificial neural network is a widely known method, such as general deep learning, and the present invention may adopt any method among the already known methods.

[0114] The deep learning described above may be performed using a computer system. The computer system referred to in this specification may include an artificial intelligence program, an artificial neural network program, or any system equipped with such a program.

[0115] The training martensite microstructure image data, training hardness data, and training ballistic performance data included in the above data set may, for example, be 100 sets or more to ensure the accuracy of the output value, and to further increase accuracy, may, for example, be 1,000 sets or more, or for example, 2,000 sets or more. However, the technical concept of the present invention is not limited thereto. Although the accuracy of the output value increases as the number of the above data sets increases, they can be appropriately selected considering time and cost. The above data sets may be provided through actual experimental results, or by repeatedly performing operations while changing input values ​​using finite element simulation, etc.

[0116] According to the technical concept of the present invention, a deep artificial neural network (NN) model capable of identifying non-linear correlations between input data and output data can be applied. To improve the performance of the artificial neural network model, the structure of the artificial neural network model can be optimized by tuning basic hyperparameters, namely the number of hidden layers and the number of neurons per hidden layer, using a Bayesian optimization algorithm. The prepared dataset can be randomly assigned to deep learning, validation, and test sets. During the deep learning process, the Mean Squared Error (MSE) of the validation set can be monitored to determine the completion of deep learning while avoiding overfitting. The performance of the trained artificial neural network can be evaluated using a test dataset excluded during deep learning. New information not used in deep learning can be provided to the deep-learned artificial neural network, and the calculated and measured parameters can be compared. Since the initial weights, biases, and dataset partitioning for deep learning can be configured differently, deep learning can be performed independently multiple times, for example, 5 times, and the mean squared error and output can be calculated to ensure sufficient generalization. The maximum epoch of deep learning can be set to, for example, 30 or more, for example, 3000, which is large enough to stabilize the artificial neural network system and allow the early stopping method to operate.

[0117] Through the deep learning process of artificial intelligence as described above, unlike conventional methods which required separate work by the actual user, the present invention allows for immediate acquisition of predicted values ​​simply by inputting measured input values ​​into a deep-learned computer system without any separate work by the actual user. Therefore, since it can be easily used in combination with existing experimental equipment, a significant reduction in time and cost is possible.

[0118] A computing system according to the technical concept of the present invention may include a central processing unit (154) and a database unit (156) as a computing system for performing a method to predict the ballistic performance of martensitic steel. The central processing unit (154) may perform the following steps: preparing a plurality of data sets consisting of training martensitic microstructure image data, training hardness data, and training ballistic performance data of training martensitic steel; performing deep learning on a computer system using the training martensitic microstructure image data as an input value and the training hardness data and training ballistic performance data as output values; providing actual martensitic microstructure image data of the martensitic steel to be predicted; and inputting the actual martensitic microstructure image data into the deep-learned computer system to predict the hardness data and ballistic performance data of the material to be predicted.

[0119] The method according to the present invention may be implemented as computer-readable code on a computer-readable storage medium. The computer-readable storage medium may include any type of recording device in which data that can be read by a computer system is stored. Examples of the computer-readable storage medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc., and also include implementation in the form of a carrier wave (e.g., transmission over the Internet). Additionally, the computer-readable storage medium may store computer-readable code that can be executed in a distributed manner by a networked distributed computer system. The code may include a computer program.

[0120] It will be obvious to those skilled in the art that the technical concept of the present invention described above is not limited to the aforementioned embodiments and attached drawings, and that various substitutions, modifications, and changes are possible within the scope of the technical concept of the present invention.

Claims

1. A step of acquiring an image of the martensite microstructure of martensite steel; A step of classifying the martensite phase from the above martensite microstructure image; A step of calculating the hardness of the martensite steel based on the content of the classified martensite phase; A step of establishing a correlation between the hardness and bulletproof performance of the above-mentioned martensitic steel; and A step comprising calculating bulletproof performance from the hardness of the martensitic steel using the above correlation, Method for predicting the ballistic performance of martensitic steel.

2. In Paragraph 1, In the step of establishing the above correlation, The above correlation uses the following formula, Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661 Method for predicting the ballistic performance of martensitic steel.

3. In Paragraph 1, The step of classifying the above martensite phase is, Classifying the martensite phase according to the width of the lath of the martensite phase, Method for predicting the ballistic performance of martensitic steel.

4. In Paragraph 1, The step of classifying the above martensite phase is, The above martensite phase, A first martensite phase comprising a first lath having a width of more than 1 μm and less than or equal to 20 μm; A second martensite phase comprising a second lath having a width of 0.1 μm or more and 1 μm or less; and Performed by classifying into a third martensite phase that does not contain lath, Method for predicting the ballistic performance of martensitic steel.

5. In Paragraph 1, The step of classifying the above martensite phase is, Classifying the martensite phase according to at least one of the width of the lath of the martensite phase, the length of the lath, the ratio of the width to the length of the lath, and whether the lath contains a cementite phase. Method for predicting the ballistic performance of martensitic steel.

6. In Paragraph 4, The step of calculating the hardness of the above martensitic steel is, Calculating the hardness of the martensite steel according to the respective contents of the first martensite phase, the second martensite phase, and the third martensite phase. Method for predicting the ballistic performance of martensitic steel.

7. In Paragraph 1, In the step of calculating the hardness of the above martensitic steel, The above hardness is at least one of Brinell hardness, Vickers hardness, Rockwell hardness, and Shore hardness, Method for predicting the ballistic performance of martensitic steel.

8. An image acquisition unit for acquiring an image of the martensite microstructure of martensite steel; A martensite phase classification unit that classifies the martensite phase from the above martensite microstructure image; A hardness calculation unit that calculates the hardness of the martensite steel based on the content of the classified martensite phase; and A ballistic performance calculation unit comprising a ballistic performance calculation unit that calculates ballistic performance from the hardness of the martensite steel using correlation information between the hardness and ballistic performance of the martensite steel, Ballistic performance prediction system for martensitic steel.

9. In Paragraph 8, The above correlation information uses the following formula, Ballistic performance (m / s) = 6.632 x (Brinell hardness - 467) x 0.3557 + 661 Ballistic performance prediction system for martensitic steel.

10. In Paragraph 8, The above martensite phase classification section is, It includes a first database having martensite classification information according to at least one of the width of the lath on the martensite phase, the length of the lath, the ratio of the width to the length of the lath, and whether the lath includes a cementite phase, and Classifying the martensite phase according to the martensite classification information provided in the first database, Ballistic performance prediction system for martensitic steel.

11. In Paragraph 8, The above hardness calculation unit is, It includes a second database having hardness information of the martensite steel according to the content of the classified martensite phase, and Calculating the hardness of the martensite steel according to the hardness information of the martensite provided in the second database, Ballistic performance prediction system for martensitic steel.

12. In Paragraph 8, The above ballistic performance calculation unit is, It includes a third database having correlation information between the hardness and bulletproof performance of the above-mentioned martensitic steel, and Calculating the ballistic performance of the martensitic steel according to the correlation information between the hardness and ballistic performance provided in the third database above. Ballistic performance prediction system for martensitic steel.

13. A step of preparing a plurality of data sets consisting of training martensite microstructure image data, training hardness data, and training ballistic performance data of training martensite steel; A step of having a computer system perform deep learning using the above-mentioned training martensite microstructure image data as input values, the above-mentioned training hardness data, and the above-mentioned training ballistic performance data as output values; A step of providing actual martensite microstructure image data of a predicted martensite steel; and A method comprising the step of inputting the actual martensite microstructure image data into the deep-learned computer system to predict the hardness data and ballistic performance data of the predicted target material. Method for predicting the ballistic performance of martensitic steel.

Images