JP7835178B2 - Saw cutting load estimation device, saw cutting load estimation method, saw cutting speed determination method, metal material manufacturing method, and method for generating a saw cutting load estimation model. - Google Patents

Description

This invention relates to a sawing load estimation device, a sawing load estimation method, a method for generating a sawing load estimation model used in the estimation, and a method for manufacturing a metal material, all for estimating the sawing load when sawing a rolled metal material using a saw blade in a hot rolling line for metal materials.

Structural steel sections with webs and flanges, such as H-beams and steel sheet piles, are manufactured by hot-rolling steel billets to form rolled material with the desired cross-sectional shape, and then cutting this rolled material to a predetermined length using a sawing device. In this sawing device, sawing generates a sawing load on the electric motor that drives the saw blade. If this sawing load becomes too large and exceeds the motor's allowable limit, fatigue damage accumulates in the motor, shortening its lifespan. Therefore, the sawing device is controlled to ensure that the sawing load remains within the device's allowable limit.

As a technique for controlling the saw cutting load, Patent Document 1 discloses a saw cutting device that suppresses the surge in the saw cutting load. According to Patent Document 1, the relationship between the position of the shift cart that moves the saw blade and the load current value is determined, and the surge in the load current value is suppressed by sufficiently slowing down the speed of the shift cart from a position before the position where a surge in the load current value occurs.

Furthermore, Patent Document 2 discloses a machine learning model capable of determining a suitable threshold for determining whether or not a machine tool is experiencing an abnormal load. According to Patent Document 2, the threshold is determined using a machine learning model that takes at least one of the following as input: information about the machine tool, the material of the workpiece, the direction of tool movement, the cutting speed, the spindle speed, the depth of cut, and the coolant amount, and outputs a load current value.

Japanese Patent Publication No. 2008-296346 Japanese Patent Publication No. 2017-220111

Patent Document 1 discloses a technique for suppressing the surge in sawing load, but not a technique for estimating sawing load. Furthermore, Patent Document 2 estimates the load current value using a machine learning model that only takes machine tool information as input, and this estimation method has the problem of not considering the rolling conditions of the rolled material being sawed. This invention addresses these problems, and its objective is to provide a sawing load estimation device and method that can estimate the sawing load considering the hot rolling conditions of the rolled metal material being sawed, as well as a method for generating a sawing load estimation model used to estimate the sawing load, and a method for manufacturing metal materials.

The means to solve the above problems are as follows:
[1] A sawing load estimation device for estimating the sawing load generated when sawing rolled metal material having the same cross-sectional shape rolled in a hot rolling process, wherein the target cross-sectional shape classification of the rolled metal material for which the sawing load is estimated is the same, and the sawing load estimation unit inputs the operating parameters of the hot rolling process and the sawing speed into a sawing load estimation model and outputs the sawing load of the rolled metal material.
[2] The sawing load estimation device according to [1], wherein the sawing load estimation unit further inputs the sawing position to the sawing load estimation model and outputs the sawing load of the rolled metal material.
[3] The sawing load estimation device according to [1] or [2], wherein the sawing load estimation unit further inputs at least one of the cooling parameters of the hot rolling process, the sawing condition parameters, and the attribute parameters of the rolled metal material to the sawing load estimation model and outputs the sawing load of the rolled metal material.
[4] The saw cutting load estimation device according to any one of [1] to [3], further comprising a saw cutting speed determination unit that determines a saw cutting speed such that the saw cutting load estimated by the saw cutting load estimation unit falls within a predetermined target load range.
[5] A method for estimating the sawing load generated when sawing rolled metal materials with the same cross-sectional shape rolled in a hot rolling process, wherein the target cross-sectional shape classification of the rolled metal materials for which the sawing load is estimated is the same, and the operating parameters of the hot rolling process and the sawing speed are input to a sawing load estimation model, and the sawing load of the rolled metal material is output to estimate the sawing load.
[6] The sawing load estimation method according to [5], further comprising inputting the sawing position into the sawing load estimation model and outputting the sawing load of the rolled metal material to estimate the sawing load.
[7] The sawing load estimation method according to [5] or [6], further comprising inputting at least one of the cooling parameters of the hot rolling process, the sawing condition parameters, and the attribute parameters of the rolled metal material into the sawing load estimation model and outputting the sawing load of the rolled metal material to estimate the sawing load.
[8] A method for estimating saw cutting load according to any one of [5] to [7], which determines a saw cutting speed such that the saw cutting load output from the saw cutting load estimation model falls within a predetermined target load range.
A method for manufacturing a metal material, comprising cutting the rolled metal material at a cutting speed determined by the cutting load estimation method described in [9] and [8] to produce the metal material.
[10] A method for generating a sawing load estimation model for estimating the sawing load generated when a rolled metal material rolled in a hot rolling process is sawed, wherein the target cross-sectional shape classification of the rolled metal material for which the sawing load is estimated is the same, and a machine learning model is trained using multiple datasets, each of which consists of actual values of the operating parameters of the hot rolling process, the sawing speed, and the sawing load of the rolled metal material that has been sawed in the past, as training data, and a sawing load estimation model is generated that takes the operating parameters and the sawing speed as input and outputs the sawing load.

According to the present invention, the sawing load is estimated using a sawing load estimation model that takes input data including the operating parameters of the hot rolling process as input and outputs the sawing load of the rolled metal material. Therefore, the sawing load can be estimated while considering the hot rolling conditions of the rolled metal material to be sawed.

Figure 1 is a schematic cross-sectional view of a steel sheet pile. Figure 2 is a schematic top view showing an example of a hot rolling mill used for manufacturing steel sheet piles. Figure 3 is a schematic side view of the saw cutting device. Figure 4 illustrates the sawing load when rolling steel sheet pile material is sawn at a constant sawing speed. Figure 5 illustrates the sawing load when cutting rolled metal material of steel sheet piles at an increased sawing speed. Figure 6 is a functional block diagram of the saw break load estimation device. Figure 7 is a flowchart showing the process of determining the saw cutting speed by the saw cutting speed determination unit. Figure 8 illustrates the sawing load when a rolled metal material of round steel is cut with a saw. Figure 9 illustrates the sawing load when a rolled round steel is sawn at an increased sawing speed. Figure 10 is a graph showing the correspondence between estimated current values and measured current values.

The present invention will be described below through embodiments of the present invention. The sawing load estimation device and sawing load estimation method according to this embodiment estimate the sawing load when manufacturing a metal material that is a structural steel by sawing multiple rolled metal materials, each having the same target cross-sectional shape classification, which have been rolled in a hot-rolling process, using a sawing device. The structural steel to be manufactured is, for example, an H-beam or a hat-shaped steel sheet pile (hereinafter referred to as "steel sheet pile"), but the following embodiments will use an example of sawing rolled metal material of a steel sheet pile. Furthermore, the material of the metal material is not limited to iron; it may also be aluminum or copper. Note that "multiple materials with the same target cross-sectional shape classification" here refers to multiple metal materials manufactured using the same roll set in a rolling mill, as will be described later. For steel sheet piles, each type of steel sheet pile (e.g., 10H, 3W, 4 type, etc.) will have the "same target cross-sectional shape classification." For H-beams, series classified by web height and flange width will have the "same target cross-sectional shape classification."

Figure 1 is a schematic cross-sectional view of a steel sheet pile 10. The steel sheet pile 10 has a web 12, two flanges 14, two arm portions 16, and two joint portions 18. The two flanges 14 are provided at an angle, extending downwards continuously from both sides of the web 12. The two arm portions 16 are provided extending horizontally continuously from the flanges 14. The two joint portions 18 are provided at the ends of the two arm portions 16, respectively.

Figure 2 is a schematic top view showing an example of a hot rolling mill 20 for manufacturing steel sheet piles 10. The hot rolling mill 20 includes, for example, a heating furnace 22, a roughing mill 24, a transfer bed 26, an intermediate rolling mill 28, a finishing mill 30, a cooling device 32, a table roller 34, a sawing device 36, a process computer 38, and a sawing load estimation device 40. Temperature sensors 25, 29, 31, and 33 are provided downstream of the roughing mill 24, the intermediate rolling mill 28, and the finishing mill 30, and upstream of the sawing device 36, respectively, to measure the surface temperature of the rolled metal material 11. These temperature sensors measure the rolling temperature of each mill and the temperature of the rolled metal material 11 before sawing. Furthermore, sensors (not shown) are provided to measure the length and mass of the rolled metal material 11 being conveyed on the table roller 34. These sensors measure the mass and length of the rolled metal material 11 in the rolling direction.

The slabs, blooms, and beam blanks (hereinafter referred to as "slabs, etc.") that will be used as the material for steel sheet piles are heated to 1100-1350°C in a heating furnace 22. The slabs, etc. heated in the heating furnace 22 are then roughly rolled in a roughing mill 24. The roughing mill 24 has a pair of upper and lower rolls with multiple "hole molds" engraved on them. By rolling with these rolls, the slabs, etc. are rolled into rough steel billets with the general shape of the steel sheet piles formed. The rough steel billets are then moved laterally from the roughing line to the intermediate and finishing rolling lines by a transfer bed 26.

In the intermediate rolling mill 28, the rough steel billet is further intermediate-rolled so that its cross-sectional shape closely resembles that of a steel sheet pile. In the finishing rolling mill 30, the material is finish-rolled so that its cross-sectional shape resembles that of a steel sheet pile. The rolled metal material 11, having been rolled in the finishing rolling mill 30 and now having a cross-sectional shape resembling a steel sheet pile, is water-cooled in the cooling device 32 and then transported to the table rollers 34. The rolled metal material 11, transported on the table rollers 34, is then sawn by the saw-cutting device 36 to produce steel sheet piles 10 of a predetermined length.

The process computer 38 is connected to each device constituting the hot rolling mill 20 by wired or wireless means and oversees the manufacturing process of the steel sheet piles 10. The process computer 38 sets the manufacturing parameters of each device constituting the hot rolling mill 20 based on the manufacturing conditions of the steel sheet piles 10, and collects and stores manufacturing performance data from each device. Furthermore, the process computer 38 is connected to the saw-cutting load estimation device 40 by wired or wireless means, and transmits and receives the setting values of the manufacturing parameters and manufacturing performance data necessary for the saw-cutting load estimation device 40 to estimate the saw-cutting load. Note that each device constituting the hot rolling mill 20 may have a control device for controlling it. In this case, the process computer 38 connects to the control device to set manufacturing parameters and collect manufacturing performance data.

The sawing load estimation device 40 estimates the sawing load when the rolled metal material 11 is sawn by the sawing device 36. The sawing load estimation device 40 also acquires the setting values of the manufacturing parameters used for estimating the sawing load and actual manufacturing data from the process computer 38.

The sawing load estimation device 40, when estimating the sawing load, obtains the operating parameters of the hot rolling process of the rolled metal material 11 for which the sawing load is to be estimated, and the set value of the sawing speed at a specific sawing position, from the process computer 38, and then estimates the sawing load. Here, the operating parameters of the hot rolling process are at least one of the following: the heating temperature in the heating furnace 22, the rolling temperature and rolling time of the roughing mill 24, the rolling temperature and rolling time of the intermediate rolling mill 28, and the rolling temperature and rolling time of the finishing rolling mill 30.

Figure 3 is a schematic side view of the saw cutting device 36. The saw cutting device 36 comprises a drive unit 42, a trolley 44, an electric motor 46, a drive transmission belt 48, and a saw blade 50. The electric motor 46 rotates the saw blade 50 by transmitting rotational driving force to it via the drive transmission belt 48. The electric motor 46, drive transmission belt 48, and saw blade 50 are mounted on the trolley 44. The drive unit 42 causes the trolley 44 to reciprocate toward the rolled metal material 11. This causes the rolled metal material 11 to be cut by the saw blade 50.

Furthermore, the drive unit 42 outputs the position of the trolley 44 to the process computer 38, associating it with time. The electric motor 46 also outputs its load current value to the process computer 38, associating it with time. Upon acquiring the position of the trolley 44, the process computer 38 refers to the time associated with that position to determine the sawing position and the sawing speed at that position. Furthermore, upon acquiring the load current value, the process computer 38 refers to the time associated with that load current value and the information acquired from the drive unit 42 to determine the sawing load at each sawing position.

Next, the sawing load when the rolled metal material 11 is sawn by the sawing device 36 will be explained. Figure 4 illustrates the sawing load when the rolled metal material 11 of the steel sheet pile is sawn at a constant sawing speed. Figure 4(a) shows the sawing position of the rolled metal material 11 of the steel sheet pile. The dashed line in Figure 4(a) shows the outer shape of the saw blade 50. Figure 4(b) is a graph showing the relationship between the sawing position and the sawing speed, and Figure 4(c) is a graph showing the relationship between the sawing position and the sawing load. The sawing positions (1) to (6) in Figure 4(a) correspond to the sawing positions (1) to (6) in Figures 4(b) and (c).

As shown in Figure 4(a), the saw blade 50 moves from left to right, approaching the rolled metal material 11 of the steel sheet pile. The saw blade 50 contacts the rolled material of the steel sheet pile at position (1) in Figure 4 and begins cutting. The saw blade 50 continues to move to the right, cutting the steel sheet pile, and upon reaching position (6), the rolled metal material 11 of the steel sheet pile is cut, and the steel sheet pile 10 is manufactured.

Figure 4(b) shows the sawing speed at each sawing position. As shown in Figure 4(b), the saw blade 50 moves at high speed until just before contact with the rolled metal material 11. Just before the saw blade 50 contacts the rolled metal material 11 at position (1), the sawing speed is reduced to the reference speed. After that, the saw blade 50 moves at a constant speed at the reference speed to position (6), and the rolled metal material 11 is sawed.

Figure 4(c) shows the sawing load (current value) at each sawing position. As shown in Figure 4(c), the sawing load increased when the saw blade 50 contacted the rolled metal material 11 at position (1), and increased further when it moved to position (2) and the saw blade 50 contacted the flange 14. As the saw blade 50 passed position (2) and moved to position (3), the sawing load decreased, and decreased even further at position (4). Subsequently, the sawing load increased slightly at positions (5) and (6), and decreased significantly when the sawing of the rolled metal material 11 was completed.

From the sawing load profile shown in Figure 4(c), it can be seen that the sawing load increases as the contact length between the saw blade 50 and the rolled metal material 11 increases, and decreases as the contact length decreases. That is, as shown in Figure 4(a), at sawing position (2), the contact length between the saw blade 50 and the flange 14 is the longest, so the largest sawing load occurred at sawing position (2). Furthermore, at sawing positions (1), (5), and (6), the contact length with the saw blade 50 is the next longest, and the contact lengths are similar on both sides, so a sawing load of roughly the same magnitude occurs immediately after the start of sawing and near the end of sawing. On the other hand, at sawing positions (3) and (4), the contact length between the saw blade 50 and the rolled metal material 11 is shorter, so the sawing load was smaller at sawing positions (3) and (4).

Furthermore, it is known that in the sawing device 36, increasing the sawing speed increases the sawing load, while decreasing the sawing speed decreases the sawing load. As shown in Figure 4(c), the sawing load changes depending on the contact length between the saw blade 50 and the rolled metal material 11. Therefore, at sawing positions (3) and (4), where the sawing load is small, it may be possible to increase the sawing speed to improve sawing productivity.

Figure 5 illustrates the sawing load when the rolled metal material 11 of the steel sheet pile is sawn at an increased sawing speed. Figure 5(a) shows the sawing position of the rolled metal material 11 of the steel sheet pile, and is the same figure as Figure 4(a). Figure 5(b) is a graph showing the relationship between the sawing position and the sawing speed, and Figure 5(c) is a graph showing the relationship between the sawing position and the sawing load. Note that in Figure 5, the sawing positions (1) to (6) in Figure 5(a) correspond to the sawing positions (1) to (6) in Figures 5(b) and (c).

As shown in Figure 5(b), in the example shown in Figure 5, the sawing speed is increased at sawing positions (3) and (4), where the sawing load is small. Therefore, as shown in Figure 5(c), the sawing load at positions (3) and (4) is high. However, since the sawing load at these positions is lower than the equipment upper limit of the sawing device 36, this increase in sawing load does not cause any problems. Note that keeping the sawing load below the equipment upper limit is one example of keeping the sawing load within a predetermined target load range.

Thus, at cutting positions where the sawing load is low, the sawing speed can be increased while keeping the sawing load within a predetermined target load range, thereby improving sawing productivity. Furthermore, if the sawing load at increased sawing speed can be estimated with high accuracy, the safety factor required to ensure the upper limit is not exceeded can be reduced, allowing for even faster sawing speeds and further increasing sawing productivity.

Next, the saw-cutting load estimation device 40, which estimates the saw-cutting load, will be described. Figure 6 is a functional block diagram of the saw-cutting load estimation device. The saw-cutting load estimation device 40 is, for example, a general-purpose computer such as a workstation or personal computer. The saw-cutting load estimation device 40 has a control unit 52, an input unit 54, an output unit 56, and a storage unit 58. The control unit 52 is, for example, a CPU, and by executing the program stored in the storage unit 58, it functions as a data acquisition unit 60, a saw-cutting load estimation unit 62, a saw-cutting speed determination unit 64, and a saw-cutting load estimation model generation unit 66.

The input unit 54 is, for example, a keyboard, a touch panel integrated with a display, etc. The output unit 56 is, for example, an LCD or CRT display, etc. The storage unit 58 is, for example, an information recording medium such as a flash memory capable of updating data, a hard disk (built-in or connected via a data communication terminal), or a memory card, and its reading and writing device. The storage unit 58 stores the database 68 and the sawing load estimation model 70. The database 68 stores more than 100 data sets, each containing operating parameters for the hot rolling process of rolled metal material 11 that has been rolled in the past using the same hot rolling equipment 20 (with the same roll set for the roughing mill 24, intermediate mill 28, and finishing mill 30) and sawn by the sawing device 36, with the same target cross-sectional shape classification. The specific sawing position may be, for example, the sawing position (3) or (4) in Figure 4(a).

The saw-cutting load estimation model 70 is a pre-trained machine learning model that has been trained using the dataset stored in the database 68 as training data. The saw-cutting load estimation model 70 according to this embodiment is a pre-trained machine learning model that takes input data including the operating parameters of the hot rolling process and the saw-cutting speed at a specific saw-cutting position as input, and outputs the saw-cutting load.

Next, the processes performed by the data acquisition unit 60 and the sawing load estimation unit 62 will be described. The data acquisition unit 60 acquires the operating parameters of the hot rolling process and the set value of the sawing speed at a specific sawing position as input data from the process computer 38. Alternatively, the data acquisition unit 60 may acquire the above data from the input unit 54. In this case, the data is input by the operator from the input unit 54. The data acquisition unit 60 outputs the acquired data to the sawing load estimation unit 62.

The sawing load estimation unit 62, upon acquiring input data from the data acquisition unit 60, reads the sawing load estimation model 70 from the storage unit 58, inputs the input data into the sawing load estimation model 70, and outputs the sawing load. By outputting the sawing load from the sawing load estimation model 70 in this way, the sawing load estimation unit 62 can estimate the sawing load at a specific sawing position. The sawing load estimation unit 62 may also output the output sawing load to the output unit 56 and display it on the output unit 56. This allows the operator to confirm the estimated sawing load value by visually checking the output unit 56.

Thus, in the saw-cutting load estimation device and saw-cutting load estimation method according to this embodiment, the saw-cutting load is estimated using a saw-cutting load estimation model 70 that takes input data including the operating parameters of the hot rolling process as input and outputs the saw-cutting load. Therefore, the saw-cutting load can be estimated while considering the hot rolling conditions of the rolled metal material to be sawed.

The heating temperature in the heating furnace 22, which is an operating parameter of the hot rolling process, and the rolling temperature and rolling time in the roughing mill 24, intermediate rolling mill 28, and finishing rolling mill 30, which are operating parameters of the rolling mill, affect the deformation resistance of the rolled metal material 11 due to its temperature. This deformation resistance affects the sawing load of the rolled metal material 11.

The rolling load changes due to changes in the deformation resistance of the rolled metal material 11. Therefore, the heating temperature in the heating furnace 22, which is an operating parameter of the hot rolling process, and the rolling temperature and rolling time in the roughing mill 24, intermediate rolling mill 28, and finishing rolling mill 30, which are operating parameters of the rolling mill, also affect the cross-sectional shape of the rolled metal material 11. The cross-sectional shape of the rolled metal material 11 affects the contact length between the rolled metal material 11 and the saw blade 50. As explained in Figures 4(a) and (c), the contact length with the saw blade 50 correlates with the saw cutting load; therefore, these parameters also correlate with the saw cutting load through changes in the cross-sectional shape. Consequently, it can be seen that using the saw cutting load estimation model 70, which uses input data including the operating parameters of the hot rolling process, allows for a higher accuracy prediction of the saw cutting load than using a saw cutting load estimation model that does not include these operating parameters in its input data.

Furthermore, it is preferable that the input data for the sawing load estimation model 70 includes at least one of the following: sawing position, cooling parameters for the hot rolling process, sawing condition parameters, and attribute parameters of the rolled metal material. If these data are included in the input data, they will also be included in the dataset stored in the database 68.

It is preferable that the sawing position is included in the input data of the sawing load estimation model 70. As shown in Figure 4(c), the sawing position significantly affects the sawing load. By generating a sawing load estimation model 70 limited to specific sawing positions, it is possible to create a model that can estimate the sawing load even without including the sawing position in the input data. However, by including the sawing position in the input data, a sawing load estimation model that can estimate the sawing load without being limited to specific sawing positions is obtained. By estimating the sawing load using such a sawing load estimation model, estimated values of the sawing load corresponding to each sawing position, as shown in Figure 4(c), can be obtained.

Furthermore, it is preferable to include at least one of the following parameters in the input data of the sawing load estimation model 70: the cooling parameter for the hot rolling process, the sawing condition parameter, and the attribute parameter of the rolled metal material. Here, the cooling parameter for the hot rolling process is at least one of the cooling water volume and cooling time of the cooling device 32. The sawing condition parameter is at least one of the number of sawings of the rolled metal material 11, the cumulative number of sawings of the saw blade 50, and the temperature of the rolled metal material 11 before sawing. The number of sawings represents the number of sawings performed on a given rolled metal material 11. The attribute parameter of the rolled metal material is at least one of the length in the rolling direction of the rolled metal material 11 and the mass of the rolled metal material 11.

The cooling parameters of the cooling device 32, namely the amount of cooling water and the cooling time, affect the deformation resistance caused by the temperature of the rolled metal material 11, as described above. Therefore, these parameters also affect the sawing load.

Furthermore, as the number of sawing cycles of the rolled metal material 11 increases, the temperature of the rolled metal material 11 decreases, and therefore, the sawing load tends to increase with increasing sawing cycles. The cumulative number of sawing cycles of the saw blade 50 correlates with the wear of the saw blade 50, and as wear progresses, the sawing load tends to increase. A higher temperature of the rolled metal material 11 before sawing tends to decrease the sawing load. Therefore, sawing condition parameters also affect the sawing load.

Furthermore, since the length of the rolled metal material 11 in the rolling direction and the mass of the rolled metal material 11 also affect the sawing load, the attribute parameters of the rolled metal material also affect the sawing load. Therefore, by including at least one of the cooling parameters of the hot rolling process, the sawing condition parameters, and the attribute parameters of the rolled metal material in the input data of the sawing load estimation model 70, the accuracy of the sawing load estimation is improved.

The chemical composition, steel grade classification, and target strength (strength classification) of the rolled metal material 11 may also be added as attribute parameters for the rolled metal material. By adding these to the attribute parameters of the rolled metal material, it becomes possible to estimate the sawing load of rolled metal materials with different chemical compositions, steel grade classifications, and strength classifications, but rolled using the same roll set.

Next, we will explain the processes performed by the saw cutting speed determination unit 64 and the saw cutting load estimation model generation unit 66. The saw cutting speed determination unit 64 identifies a saw cutting speed at which the saw cutting load estimated by the saw cutting load estimation unit 62 falls within a predetermined target load range. The process of identifying the saw cutting speed by the saw cutting speed determination unit 64 will be explained using Figure 7.

Figure 7 is a flowchart showing the process of determining the saw cutting speed by the saw cutting speed determination unit 64. The flow shown in Figure 7 is initiated, for example, by an instruction from the operator to start the process and input of a target saw cutting speed.

First, the data acquisition unit 60 acquires the target sawing speed set as the reference upper limit, or the target sawing speed input by the operator from the input unit 54 (step S101). The data acquisition unit 60 also acquires other input data, such as the operating parameters of the hot rolling process, from the process computer 38 (step S102). The data acquisition unit 60 outputs this data to the sawing load estimation unit 62.

The sawing load estimation unit 62 reads the sawing load estimation model 70 from the storage unit 58, inputs the sawing speed and hot rolling temperature parameters into the sawing load estimation model 70, outputs the sawing load, and estimates the sawing load (step S103). The sawing load estimation unit 62 outputs the outputted sawing load to the sawing speed determination unit 64.

The saw cutting speed determination unit 64 reads the upper limit value of the saw cutting load of the saw cutting device 36, which is stored in the storage unit 58, and determines whether the saw cutting load obtained from the saw cutting load estimation unit 62 is less than or equal to this upper limit (step S104). If the saw cutting load exceeds the upper limit, the saw cutting speed determination unit 64 determines that the saw cutting load obtained from the saw cutting load estimation unit 62 is not within the target load range (step S104: No), and changes the saw cutting speed so that the saw cutting speed obtained in step S101 becomes slower (step S105). The saw cutting speed determination unit 64 outputs the changed saw cutting speed to the saw cutting load estimation unit 62 and causes the process in step S103 to be executed again. The processes in steps S105 and S103 are repeated until the saw cutting load estimated in step S103 is determined to be within the target load range in step S104. Note that one unit of slowing down the saw cutting speed in step S105 is, for example, 20 mm/sec.

On the other hand, if the sawing load is less than or equal to the upper limit of the sawing load of the sawing device 36, the sawing speed determination unit 64 determines that the sawing load obtained from the sawing load estimation unit 62 is within the range of the target load (step S104: Yes), determines the sawing speed used for sawing load estimation as the sawing speed of the sawing device 36 (step S105), outputs the determined sawing speed to the process computer 38, and terminates the process of determining the sawing speed. When the process computer 38 obtains the sawing speed from the sawing speed determination unit 64, it controls the drive unit 42 of the sawing device 36 so that the speed of the trolley 44 becomes the determined sawing speed. In this way, by adjusting the sawing speed of the sawing device 36, the sawing speed can be increased while keeping the sawing load within the range of the target load, thereby improving the productivity of sawing. The sawing speed determination unit 64 may also display the determined sawing speed on the output unit 56. This allows the operator to visually monitor the output unit 56 and adjust the sawing speed of the sawing device 36.

The above describes a method for determining the saw cutting speed using the upper limit of the reference speed as the initial value, but this is not the only method. For example, the saw cutting speed may be determined by setting the lower limit of the reference speed as the initial value and gradually increasing the saw cutting speed until the saw cutting load does not exceed the equipment's upper limit. Alternatively, the saw cutting speed may be determined by setting the standard saw cutting speed as the initial value and gradually increasing or decreasing the saw cutting speed so that the saw cutting load falls within the target range.

Next, the method for generating the sawing load estimation model used to estimate the sawing load will be described. The data acquisition unit 60 acquires actual values of the hot rolling process's operating parameters, actual values of the sawing speed at a specific sawing position, and actual values of the sawing load from the process computer 38. These are then stored in the database 68 of the storage unit 58 as a set of datasets. The number of datasets stored in the database 68 is at least 100, preferably 500 or more, and more preferably 1000 or more.

The saw-cut load estimation model generation unit 66 reads a pre-stored machine learning model from the storage unit 58, uses the dataset stored in the database 68 as training data to train the machine learning model, and generates a trained machine learning model. This trained machine learning model becomes the saw-cut load estimation model. The machine learning model used in the saw-cut load estimation method and saw-cut load estimation device according to this embodiment may be, for example, a neural network, decision tree learning, random forest, support vector regression, Gaussian process, or k-nearest neighbor method.

As explained above, the saw-cutting load estimation model used in the saw-cutting load estimation method and saw-cutting load estimation device according to this embodiment includes the operating parameters of the hot rolling process as input. As mentioned above, the operating parameters of the hot rolling process have a correlation with the saw-cutting load. Therefore, by including the operating parameters of the hot rolling process that manufactures rolled metal materials as input data for the saw-cutting load estimation model, the saw-cutting load estimation model becomes an estimation model that takes into account the hot rolling conditions of the rolled metal material, thus increasing the accuracy of the saw-cutting load estimation. If the saw-cutting load can be estimated with high accuracy in this way, the occurrence of overload exceeding the equipment limit during saw-cutting can be suppressed, and the occurrence of "tripping" of the electric motor 46 due to overload and "sticking" of the saw blade 50 to the rolled metal material 11 can be suppressed.

Furthermore, by determining a sawing speed such that the sawing load estimated by the sawing load estimation model falls within a predetermined target load range, and then sawing the rolled metal material at that speed to manufacture the metal material, it becomes possible to manufacture the metal material while maintaining the sawing load within the target range.

Note that while Figure 2 shows an example where the hot rolling mill 20 has a process computer 38 and a sawing load estimation device 40, it is not limited to this configuration. For example, the process computer 38 may have the functions of the sawing load estimation device 40, and these may be configured as a single device. Also, while Figure 6 shows an example where the sawing load estimation device 40 has a control unit 52 with a sawing speed determination unit 64 and a sawing load estimation model generation unit 66, it is not limited to this configuration. If the sawing load estimation is performed by the sawing load estimation device 40, it is not necessarily required to have a sawing speed determination unit 64. Furthermore, if the sawing load estimation model 70 is generated externally and stored in the storage unit 58 via the input unit 54, the sawing load estimation device 40 does not necessarily need to have a sawing load estimation model generation unit 66.

Next, we will explain the sawing load when cutting round steel, which has a different cross-sectional shape from steel sheet piles. Figure 8 illustrates the sawing load when cutting rolled round steel. Figure 8(a) shows the sawing position of the round steel. Figure 8(b) is a graph showing the relationship between the sawing position and the sawing speed, and Figure 8(c) is a graph showing the relationship between the sawing position and the sawing load. Note that the sawing positions (1) to (3) in Figure 8(a) correspond to the sawing positions (1) to (3) in Figures 8(b) and (c).

As shown in Figure 8(a), the saw blade 50 moves from left to right, approaching the rolled metal material of the round steel. Then, at position (1) in Figure 8, the saw blade 50 contacts the rolled metal material and begins cutting. The saw blade 50 continues to move to the right, cutting the rolled metal material, and when it reaches position (3), the rolled metal material is cut and a round steel is manufactured.

Figure 8(b) shows the sawing speed at each sawing position. As shown in Figure 8(b), in the production of round steel, the sawing speed is reduced immediately before contact with the rolled metal material (1) to prevent a sudden increase in sawing load. Afterward, the sawing speed is increased to the standard speed. As the sawing approaches (3), the end of sawing, the contact length with the saw blade 50 shortens. Therefore, the sawing speed is increased above the standard speed to move to position (3), and the rolled metal material is sawed.

Figure 8(c) shows the sawing load (current value) at each sawing position. As shown in Figure 8(c), the sawing load increases when the saw blade 50 contacts the rolled round steel at position (1), but a sudden increase in the sawing load is prevented because the sawing speed has been reduced beforehand. Subsequently, the sawing speed is controlled to keep the sawing load constant, and upon reaching position (3) and completing the sawing, the sawing load decreases. As shown in Figure 8(c), the sawing load at position (2) is lower than the equipment's upper limit, suggesting that increasing the sawing speed at position (2) could potentially improve sawing productivity.

Figure 9 illustrates the sawing load when a rolled round steel is sawn at an increased sawing speed. Figure 9(a) shows the sawing position of the round steel, and is the same as Figure 8(a). Figure 9(b) is a graph showing the relationship between the sawing position and the sawing speed, and Figure 9(c) is a graph showing the relationship between the sawing position and the sawing load. Note that in Figure 9, the sawing positions (1) to (3) in Figure 9(a) correspond to the sawing positions (1) to (3) in Figures 9(b) and (c).

As shown in Figure 9(b), in the example shown in Figure 9, the sawing speed is increased at position (2), where the sawing load is small. Therefore, as shown in Figure 9(c), the sawing load at position (2) is high, but since it is lower than the upper limit of the sawing device 36, this increase in sawing load does not cause any problems.

Thus, even with round steel, at cutting positions where the sawing load is low, the sawing speed can be increased while keeping the sawing load within a predetermined target load range, thereby improving sawing productivity. Furthermore, if the sawing load estimation device and method according to this embodiment can estimate the sawing load when the sawing speed is increased with high accuracy, it becomes possible to further increase the sawing speed while maintaining the target load range, thereby increasing productivity.

Furthermore, in the case of round bars, it is sometimes possible to produce products with diameters varying by several millimeters to over ten millimeters using the same roll set by adjusting the rolling conditions. In such cases, by adding the target product diameter to the attribute parameters of the rolled metal material, it becomes possible to estimate the sawing load for round bars of different diameters using the same sawing load estimation model.

Furthermore, the saw-cut load estimation device and saw-cut load estimation method according to this embodiment can also be applied to H-shaped steel beams with an H-shaped cross-section, and to square steel bars and flat steel bars with a rectangular cross-section, all manufactured using the same roll set. When applied to H-shaped steel beams, adding the target web thickness and target flange thickness to the attribute parameters of the rolled metal material allows for the estimation of saw-cut loads for H-shaped steel beams with different target web thicknesses and target flange thicknesses using the same saw-cut load estimation model. Additionally, when applied to square steel bars and flat steel bars, adding the target wall thickness and target width to the attribute parameters of the rolled metal material allows for the estimation of saw-cut loads for square steel bars and flat steel bars with different target wall thicknesses and target widths using the same saw-cut load estimation model.

[Example 1]
This section describes an example of generating a sawing load estimation model and estimating the sawing load. In this example, a sawing load estimation model was generated for each sawing position (3), (4), and (5) shown in Figure 4 to estimate the sawing load when sawing rolled metal material of steel sheet pile (type: 10H). The correlation between the estimated sawing load value estimated by each sawing load estimation model and the actual sawing load value was confirmed. The input data for the sawing load estimation model is shown in Table 1 below. The output of the sawing load estimation model is the maximum sawing load current generated in the motor when sawing was performed at each position (3) to (5).

Using a dataset of 1100 data points, each consisting of actual input data and actual saw-break load current values, a machine learning model was trained to generate a pre-trained machine learning model. A decision tree model was used as the machine learning model. This pre-trained machine learning model was used as a saw-break load prediction model to predict saw-break loads, and the correlation with the actual measured saw-break load current values generated by the motor was confirmed.

Figure 10 is a graph showing the correspondence between estimated and measured current values. In Figure 10, the horizontal axis represents the estimated current value (A), and the vertical axis represents the measured current value (A). As shown in Figure 10, the plot showing the correspondence between estimated and measured current values lies on the 45° line, confirming that the saw break load can be estimated with high accuracy.

Furthermore, the standard deviation of the difference between the estimated current value and the measured current value (predicted current value - measured current value) shown in Figure 10 was 3.46 A, and the standard deviation of the ratio of the predicted current value to the measured current value (predicted current value / measured current value) was 4.72%. In contrast, when the load was estimated using only the saw cutting condition parameters, the standard deviation of the difference between the estimated current value and the measured current value was 5.98 A, and the standard deviation of the ratio of the predicted current value to the measured current value was 8.13%. From these results, it can be seen that the saw cutting load can be estimated with high accuracy by using a saw cutting load estimation model that includes the operating parameters of the hot rolling process as input data.

[Example 2]
Next, we will describe Example 2, in which the sawing load at sawing positions (3) and (4) in Figure 5 was estimated, and the sawing speed was increased to a sawing speed that did not exceed the equipment's upper limit to saw the rolled metal material. In Example 2, the sawing load estimation model created in Example 1 was used. In Invention Example 1, the sawing load was estimated at sawing positions (3) and (4) in Figure 5, and the sawing speed was increased (up to a maximum of 450 mm/second) to a range that did not exceed the equipment's upper limit to saw the rolled metal material. The sawed rolled metal material was steel sheet pile (model: 10H), and 100 steel sheet piles (80 of 295 steel and 20 of 390 steel) were sawn 12 times per steel sheet pile (including crop cuts).

In Comparative Example 1, the sawing load at cutting positions (3) and (4) in Figure 5 was not estimated, and the steel sheet piles were sawn at a standard sawing speed of 200 mm/second. The type, number, and number of saw cuts per sheet pile were the same as in the Inventive Example. In Comparative Example 3, the sawing load at cutting positions (3) and (4) in Figure 5 was not estimated, and the steel sheet piles were sawn at a sawing speed of 400 mm/second for all cuts. The type, number, and number of saw cuts per sheet pile were the same as in the Inventive Example. The sawing results for the Inventive Example, Comparative Example 1, and Comparative Example 2 are shown in Table 2 below.

As shown in Table 2, in the inventive example, there were no problems with the saw blade jamming, and the sawing time was shorter than in Comparative Examples 1 and 2, resulting in improved sawing productivity. In Comparative Example 1, although there were no problems with the saw blade jamming, the sawing time was longer because the sawing speed was slower than in Inventive Example 1, and improved sawing productivity was not achieved. In Comparative Example 2, because the sawing speed was increased, the saw blade jamming problem occurred twice, and the processing time for this was long, resulting in a longer total time than in Comparative Example 1. From these results, it was confirmed that it is preferable to estimate the sawing load using the sawing load estimation device and sawing load estimation method according to this embodiment, and to increase the sawing speed so that the estimated sawing load does not exceed the equipment upper limit. This allows for an increase in sawing speed while suppressing problems such as jamming, and thus improves sawing productivity.

10 Steel sheet pile 11 Rolled metal material 12 Web 14 Flange 16 Arm section 18 Joint section 20 Hot rolling equipment 22 Heating furnace 24 Roughing mill 25 Temperature sensor 26 Transfer bed 28 Intermediate rolling mill 29 Temperature sensor 30 Finishing mill 31 Temperature sensor 32 Cooling device 33 Temperature sensor 34 Table roller 36 Saw cutting device 38 Process computer 40 Saw cutting load estimation device 42 Drive device 44 Trolley 46 Electric motor 48 Drive transmission belt 50 Saw blade 52 Control unit 54 Input unit 56 Output unit 58 Storage unit 60 Acquisition unit 62 Saw cutting load estimation unit 64 Saw cutting speed determination unit 66 Saw cutting load estimation model generation unit 68 Database 70 Saw cutting load estimation model 80 Round steel

Claims (13)

A sawing load estimation device for estimating the sawing load generated when a rolled metal material, rolled in a hot rolling process, is sawn at a specific sawing position,
The system includes a sawing load estimation unit that inputs the operating parameters of the hot rolling process, including at least one of the heating temperature in the heating furnace, the rolling temperature of the roughing mill, the rolling time of the roughing mill, the rolling temperature of the intermediate mill, the rolling time of the intermediate mill, the rolling temperature of the finishing mill, and the rolling time of the finishing mill, as well as the sawing speed, into a sawing load estimation model to output the sawing load of the rolled metal material.
The sawing load estimation device is a pre-trained machine learning model that has been trained using multiple datasets, each dataset consisting of the operating parameters of the hot rolling process, the sawing speed, and the sawing load. A sawing load estimation device for estimating the sawing load generated when sawing a rolled metal material that has been rolled in a hot rolling process,
The system includes a sawing load estimation unit that inputs the operating parameters of the hot rolling process, including at least one of the heating temperature in the heating furnace, the rolling temperature of the roughing mill, the rolling time of the roughing mill, the rolling temperature of the intermediate mill, the rolling time of the intermediate mill, the rolling temperature of the finishing mill, and the rolling time of the finishing mill, as well as the sawing speed and sawing position, into a sawing load estimation model to output the sawing load of the rolled metal material.
The sawing load estimation device is a pre-trained machine learning model that has been trained using multiple datasets, each dataset consisting of the operating parameters of the hot rolling process, the sawing speed, the sawing position, and the sawing load. The saw-cutting load estimation device according to claim 1 or 2, further comprising the saw-cutting load estimation unit inputting at least one of the cooling parameters of the hot rolling process, the saw-cutting condition parameters, and the attribute parameters of the rolled metal material into the saw-cutting load estimation model to output the saw-cutting load of the rolled metal material. The saw cutting load estimation device according to claim 1 or 2, further comprising a saw cutting speed determination unit that determines a saw cutting speed such that the saw cutting load estimated by the saw cutting load estimation unit falls within a predetermined target load range. The saw cutting load estimation device according to claim 3, further comprising a saw cutting speed determination unit that determines a saw cutting speed such that the saw cutting load estimated by the saw cutting load estimation unit falls within a predetermined target load range. A method for estimating the sawing load generated when a rolled metal material, rolled in a hot rolling process, is sawn at a specific sawing position,
The operating parameters of the hot rolling process, including at least one of the heating temperature in the heating furnace, the rolling temperature of the roughing mill, the rolling time of the roughing mill, the rolling temperature of the intermediate mill, the rolling time of the intermediate mill, the rolling temperature of the finish mill, and the rolling time of the finish mill, as well as the sawing speed, are input into a sawing load estimation model, and the sawing load of the rolled metal material is output to estimate the sawing load.
The sawing load estimation method is a pre-trained machine learning model that has been machine-trained using multiple datasets, each dataset consisting of the operating parameters of the hot rolling process, the sawing speed, and the sawing load. A method for estimating the sawing load generated when sawing a rolled metal material that has been rolled in a hot rolling process,
The operating parameters of the hot rolling process, including at least one of the heating temperature in the heating furnace, the rolling temperature of the roughing mill, the rolling time of the roughing mill, the rolling temperature of the intermediate mill, the rolling time of the intermediate mill, the rolling temperature of the finishing mill, and the rolling time of the finishing mill, along with the sawing speed and sawing position, are input into a sawing load estimation model, and the sawing load of the rolled metal material is output to estimate the sawing load.
The sawing load estimation method is a pre-trained machine learning model that has been machine-trained using multiple datasets, each dataset consisting of the operating parameters of the hot rolling process, the sawing speed, the sawing position, and the sawing load. Furthermore, the sawing load estimation method according to claim 6 or 7, wherein at least one of the cooling parameters of the hot rolling process, the sawing condition parameters, and the attribute parameters of the rolled metal material is input to the sawing load estimation model, and the sawing load of the rolled metal material is output to estimate the sawing load. A method for determining a saw cutting speed , which determines a saw cutting speed such that the saw cutting load output from the saw cutting load estimation model of the saw cutting load estimation method according to claim 6 or claim 7 falls within a predetermined target load range. A method for determining a saw cutting speed , which determines a saw cutting speed such that the saw cutting load output from the saw cutting load estimation model of the saw cutting load estimation method according to claim 8 falls within a predetermined target load range. A method for manufacturing a metal material, comprising cutting the rolled metal material with a saw at a saw speed determined by the saw cutting speed determination method described in claim 9. A method for manufacturing a metal material, comprising cutting the rolled metal material with a saw at a saw speed determined by the saw cutting speed determination method described in claim 10. A method for generating a sawing load estimation model for estimating the sawing load generated when a rolled metal material rolled in a hot rolling process is sawn at a specific sawing position,
A machine learning model is trained using multiple datasets as training data, each dataset consisting of a set of operational parameters for a hot rolling process, including at least one of the following: heating temperature in the heating furnace of the previously sawed rolled metal material, rolling temperature of the roughing mill, rolling time of the roughing mill, rolling temperature of the intermediate rolling mill, rolling time of the intermediate rolling mill, rolling temperature of the finishing mill, and rolling time of the finishing mill, along with the sawing speed and sawing load.
A method for generating a sawing load estimation model, which takes the operating parameters of the hot rolling process and the sawing speed as inputs and generates a sawing load estimation model that outputs the sawing load.