A method for cooling hot-rolled steel sheets and a method for manufacturing hot-rolled steel sheets using the same.

Abstract

The cooling capacity of the cooling equipment can be easily controlled by incorporating the effect of the time spent in the cooling equipment on transformation heat into the learning coefficient. [Solution] In a method for cooling hot-rolled steel sheets, the cooling capacity of a cooling system that water-cools the hot-rolled steel sheet before coiling after hot rolling is corrected using a learning coefficient. The learning coefficient is managed using a stratified table with the manufacturing conditions of the hot-rolled steel sheet as factors, and is calculated at learning points taken at multiple positions in the longitudinal direction of the hot-rolled steel sheet.

Description

[Technical Field]

[0001] The present invention relates to a method for cooling hot-rolled steel sheets and a method for manufacturing hot-rolled steel sheets using the same, which controls water cooling on a runout table using a learning coefficient that includes the effect of transformation heat generation. [Background technology]

[0002] In the manufacturing process of hot-rolled steel sheets, a slab is heated to a predetermined temperature in a heating furnace, and a rough bar is produced by rolling the heated slab in a roughing mill. Next, this rough bar is rolled to a predetermined thickness in a continuous hot finishing rolling mill consisting of multiple rolling stands (hereinafter also simply referred to as a "finishing rolling mill") to produce a hot-rolled steel sheet (hereinafter also simply referred to as a "steel sheet" or "plate").

[0003] Then, cooling water is sprayed onto the top and bottom surfaces of the steel plate by a cooling system installed on the runout table to cool it, and then the steel plate is wound up by a winding machine. The temperature of the steel plate before being wound up by the winding machine is called the "winding temperature," and it has a significant impact on the material properties of the steel plate, making it an important item for quality control.

[0004] The winding temperature is controlled by the flow rate of cooling water injected into the cooling equipment (hereinafter also simply referred to as "flow rate") according to the speed at which the sheet metal passes on the runout table. However, due to changes in cooling capacity accompanying changes in the surface temperature of the steel sheet, changes in heat quantity accompanying the transformation of the steel sheet, and the characteristics of the cooling equipment, controlling the winding temperature becomes extremely difficult for certain materials.

[0005] In recent years, there has been an increasing trend towards steel sheets that are difficult to cool due to transformation heat generation, such as high-tensile steel sheets. As a result, steel sheet cooling technology is facing increasingly stringent requirements for target cooling temperature ranges, and the accuracy of model calculations for this cooling control is being demanded to improve more and more.

[0006] Generally, to ensure that the winding temperature is accurate according to the cooling conditions of the steel sheet, the cooling capacity is given by some function, or the winding temperature is controlled by multiplying the cooling capacity by a correction factor.

[0007] For example, Patent Document 1 discloses a technique for controlling the winding temperature by creating an equation for cooling capacity as a function of the cooling water temperature, the steel plate temperature, and the water density of the cooling water, and then multiplying it by a correction coefficient. The control method disclosed in Patent Document 1 is conventionally called learning control, and the correction coefficient is also called a learning coefficient. In Patent Document 1, the amount of heat generated due to transformation is considered by using temperature dependence data of the specific heat of the steel material, where the specific heat exhibits a peak in the transformation region.

[0008] The learning coefficient is calculated by matching the calculated and measured values ​​of the temperature drop from the start to the end of water cooling at a point designated for learning coefficient calculation at an appropriate position along the longitudinal direction of the steel plate. This point is called the learning point.

[0009] Furthermore, Patent Document 2 proposes a method for optimizing two variables, the transformation rate and the water cooling heat transfer coefficient, in the transformation heat generation model when reflecting the difference between the predicted temperature and the measured temperature in the temperature prediction model. [Prior art documents] [Patent Documents]

[0010] [Patent Document 1] Japanese Patent Application Publication No. 6-246320 [Patent Document 2] Japanese Patent Publication No. 2013-766 [Overview of the project] [Problems that the invention aims to solve]

[0011] However, while Patent Document 1 calculates the learning coefficient by considering the transformation heat of the steel plate based on the temperature dependence of its specific heat, the calculated learning coefficient can vary depending on the time spent in the cooling equipment, leaving room for further improvement in the control accuracy of the winding temperature.

[0012] Furthermore, while Patent Document 2 uses only the predicted temperature as the objective variable, optimizing two variables, the transformation rate and the heat transfer coefficient, increases the computational load required to find the optimal solution. In addition, since the specific heat that changes with the transformation rate differs depending on the material composition, temperature range, and elapsed time since the end of rolling, the burden of model construction and maintenance increases.

[0013] Therefore, the present invention aims to provide a method for cooling hot-rolled steel sheets and a method for manufacturing hot-rolled steel sheets that can easily control the cooling capacity of a cooling system by incorporating the effect of the time spent in the cooling system on transformation heat generation into the learning coefficient. [Means for solving the problem]

[0014] The inventors of this invention investigated the above problem and, as a result, focused on the fact that since the steel sheet is not passed through the cooling equipment at a constant speed during cooling, the time it spends in the cooling equipment changes, and the amount of heat generated due to transformation changes along the longitudinal direction of the steel sheet. In order to take into account this change in the amount of heat generated by transformation along the longitudinal direction of the steel sheet, they found that it is effective to set up multiple learning points where the time spent in the cooling equipment differs along the longitudinal direction of the steel sheet, and to calculate a learning coefficient for each manufacturing condition of the steel sheet at each learning point. In other words, by incorporating the effect of transformation heat generation into the learning coefficient that corrects the error of the temperature drop model due to external factors, the control accuracy of the winding temperature can be improved with a simple configuration without adjusting the temperature drop model itself. It should be noted that the learning coefficient inextricably incorporates not only the effect of transformation heat generation but also the effects of equipment malfunctions and other uncertain factors.

[0015] Conventionally, the learning coefficient was calculated at a single point at the leading edge where the winding of the steel plate in the longitudinal direction was stable, and the same learning coefficient was applied to the temperature drop calculation and cooling setting calculation for the entire length. Therefore, it was not possible to respond to changes in the effects of transformation heat generation, equipment malfunctions, or other uncertain factors in the longitudinal direction of the steel plate.

[0016] This invention was developed based on the above findings and further investigations, and its gist is as follows. 〔1〕In a method for cooling a hot-rolled steel sheet, which corrects the cooling capacity of a cooling facility that cools the hot-rolled steel sheet by water cooling before coiling after hot rolling, using a learning coefficient, the learning coefficient is managed in a stratified table based on the manufacturing conditions of the hot-rolled steel sheet, and is calculated at learning points taken at a plurality of positions in the longitudinal direction of the hot-rolled steel sheet. A method for cooling a hot-rolled steel sheet, characterized in that. 〔2〕In the above 〔1〕, the positions of the learning points taken at a plurality of positions in the longitudinal direction of the hot-rolled steel sheet are the inlet learning position and the outlet learning position that respectively coincide with the inlet side and the outlet side of the cooling facility for each change point of the speed pattern. A method for cooling a hot-rolled steel sheet, characterized in that. 〔3〕In the above 〔2〕, the speed pattern is a primary acceleration region, a secondary acceleration region, a maximum speed region, a deceleration region, and a creeping region. A method for cooling a hot-rolled steel sheet, characterized in that. 〔4〕In the above 〔2〕 or 〔3〕, further, in addition to the learning points at the inlet learning position and the outlet learning position based on the pattern, learning points are added at positions with a pitch of a predetermined length. A method for cooling a hot-rolled steel sheet, characterized in that. 〔5〕In any one of the above 〔2〕 to 〔4〕, further, for a learning point where the sheet passing speed at the maximum speed arrival point is less than the set maximum speed and the maximum speed arrival point becomes the deceleration start point, the learning point is corrected to correspond to the set maximum speed, and the learning coefficient calculated at the corrected learning point is saved, and the saved learning coefficient is corrected to correspond to the actual maximum speed and used. A method for cooling a hot-rolled steel sheet, characterized in that. 〔6〕A method for manufacturing a hot-rolled steel sheet, comprising a cooling step by the method for cooling a hot-rolled steel sheet according to any one of the above 〔1〕 to 〔5〕. A method for manufacturing a hot-rolled steel sheet, characterized in that.

Advantages of the Invention

[0017] According to the present invention, since the learning coefficient is managed in a stratified table based on the manufacturing conditions of the steel sheet and is calculated at learning points taken at a plurality of positions in the longitudinal direction of the steel sheet, the temperature prediction error due to the change in the amount of transformation heat generation in the longitudinal direction of the steel sheet can be reduced, and the control accuracy of the coiling temperature is improved.

Brief Description of the Drawings

[0018] [Figure 1] It is a schematic diagram showing an example of a cooling facility on a run-out table. [Figure 2] It is a schematic diagram showing the longitudinal position of a steel plate at a learning point corresponding to a general plate passing speed pattern and the plate passing speed. (a) shows the outlet speed, (b) shows the inlet speed and the average speed in addition to (a), and (c) shows a constant length learning point in addition to (b). [Figure 3] It is an explanatory diagram of a method for exception handling when storing and using learning points. [Figure 4] It is a graph showing the temperature dependence of specific heat. [Figure 5] It is a diagram showing an example of a chart of the coiling temperature of an example of the present invention and a comparative example.

Embodiments for Carrying Out the Invention

[0019] Hereinafter, embodiments of the present invention will be described. In the method for cooling a hot-rolled steel plate according to the present invention, the cooling capacity of a cooling facility for water-cooling a steel plate after hot rolling and before coiling is corrected using a learning coefficient. The learning coefficient is managed by a layer-by-layer table with the manufacturing conditions of the steel plate as a factor, and is calculated at learning points taken at a plurality of positions in the longitudinal direction of the steel plate.

[0020] 〔Cooling Facility and Setting of Its Cooling Capacity〕 As shown in FIG. 1, the cooling facility 4 used in the present invention water-cools a steel plate S passing on a run-out table (not shown) from both the upper and lower sides thereof from the final rolling stand 3 to the coiler 5. The cooling facility 4 has a plurality of cooling zones in the plate passing direction, and the cooling capacity of each cooling zone is set by selecting a target flow rate corresponding to a target cooling capacity acquired by a computer 40 using a learning coefficient and performing coincidence control to the selected target flow rate on a control panel 45. The computer 40 is composed of an ordinary online computer or a process computer. The above-mentioned coincidence control to the target flow rate is implemented by adjusting the valve ejection amount of a refrigerant (meaning cooling water. The same applies hereinafter). A plurality of on-off valves are arranged in each cooling zone, and the valve ejection amount is adjusted by the on-count rate of the valves.

[0021] On the inlet side and the outlet side of the cooling equipment 4, one thermometer 1 for measuring the water cooling start temperature and one thermometer 2 for measuring the water cooling end temperature are installed respectively. In addition to the thermometers 1 and 2, a separate thermometer (not shown) can be installed in the middle of the inlet side and the outlet side of the cooling equipment 4 for measuring the temperature of the steel plate during the period from the start of water cooling to the end of water cooling.

[0022] 〔Learning coefficient〕 In the present invention, the learning coefficient is a correction coefficient multiplied by the heat transfer coefficient of water cooling used in the temperature calculation of water cooling. In the temperature calculation of water cooling, the following temperature drop model formula (Formula (1)) considering water cooling and air cooling is used. Formula (1) is a simple one that does not consider the temperature distribution in the plate thickness direction because the hot-rolled steel plate is thin enough that there is no problem in ignoring the temperature distribution in the plate thickness direction. ρ×c×d×(∂T / ∂t)=-α×h×(T―T w )-h a ×(T-T a )‥‥(1) Here, ρ: density of the steel plate (kg / m 3 ), c: specific heat of the steel plate (J / kgK), d: thickness of the steel plate (m), T: temperature of the steel plate (K), t: time (s), h: heat transfer coefficient of water cooling (W / m 2 ), h a : equivalent heat transfer coefficient of air cooling (W / m 2 ), T w : temperature of the refrigerant (water) (K), T a : temperature of the air (K). α is the learning coefficient.

[0023] In the temperature calculation, numerical integration is performed using the following formula (2) which is the difference form of Formula (1). ρ×c×d×(ΔT / Δt)=-α×h×(T-T w )-h a ×(T-T a )‥‥(2) As the integration interval width Δt for the numerical integration in Formula (2), 0.3 to 1.0 s can be mentioned in consideration of the calculation time required and the calculation accuracy.

[0024] The heat transfer coefficient h for water cooling is predetermined as a function of flow rate, material surface temperature, and water temperature (not shown). Therefore, by obtaining the learning coefficient α, the cooling capacity can be corrected by the correction (α × h).

[0025] Note that the specific heat c in equations (1) and (2) is based on the specific heat data for general materials (low-carbon steel) shown in Figure 4, and is applicable to all steel materials, not just general materials. As shown in Figure 4, the specific heat c is temperature-dependent, so the specific heat value corresponding to the starting temperature of each integration interval of the numerical integral is used.

[0026] In the embodiment, the temperature drop model equations were in the form of equations (1) and (2) (heat transfer models that do not consider the temperature distribution in the thickness direction), but the model is not limited to this form. For example, a one-dimensional difference equation in the thickness direction that considers the temperature distribution in the thickness direction, or a two-dimensional difference equation that also considers the width direction, may be used. In addition, since the temperature calculation in the longitudinal direction is used for setting calculations and dynamic control, calculations are usually performed in units of bank length or 1 / 2 bank length.

[0027] [Method for obtaining the learning coefficient α] The learning coefficient α is managed using a stratified table with the steel plate manufacturing conditions as a factor. Preferably, the learning coefficient α is further managed using a stratified table with the water cooling time index as a factor.

[0028] Computer 40 is equipped with a stratification table (not shown), in which the learning coefficient α is associated with the steel plate manufacturing conditions, or further associated with an index of water cooling time. Computer 40 reads and obtains the learning coefficient α from the stratification table using information on the steel plate manufacturing conditions and the index of water cooling time provided by a higher-level computer (not shown). The method for calculating the learning rate α stored in the table will be described later.

[0029] [Manufacturing conditions for steel plates] Manufacturing conditions for steel sheets include steel type, chemical composition, sheet thickness, sheet width, and target temperature (target temperature for hot rolling and coiling, or an intermediate target temperature). By managing the learning coefficient α using a stratified table with steel plate manufacturing conditions as factors, the prediction error of the winding temperature due to differences in the manufacturing conditions of the steel plate before water cooling can be reduced.

[0030] [Indicators of water cooling time] Simply managing the learning coefficient α using a layered table with steel plate manufacturing conditions as factors may result in insufficient accuracy in predicting the winding temperature due to differences in transformation heat generation caused by varying water cooling times along the longitudinal direction of the steel plate. Therefore, in this invention, it is preferable to further adopt an index of water cooling time as a factor in the layered table.

[0031] As the aforementioned water cooling time, in addition to the time spent in the cooling equipment (hereinafter also simply referred to as "stay time"), a plate speed proportional to the reciprocal of the stay time can be used. When using stay time, for example, a continuous range of stay time from 6 to 20 seconds divided into 20 equal parts (referred to as "sectional stay time") can be used as an indicator.

[0032] By stratifying the data by the water cooling time index and expressing the learning coefficient α in this way, it is possible to reduce temperature prediction errors due to differences in the amount of heat generated during transformation in the longitudinal direction of the steel plate during water cooling, and further improve the accuracy of the winding temperature prediction.

[0033] [How to earn learning points] The aforementioned dwell time for each section can be correlated with the longitudinal section of the steel plate through analysis of operational data. Therefore, a point within the longitudinal section of the steel plate, such as the midpoint, which is correlated with the dwell time for each section, is taken as a learning point.

[0034] [Calculation of the learning coefficient α] The learning coefficient α, which is stored in the stratified table, is calculated in advance for each manufacturing condition of the steel plate and for each water-cooling time index, at multiple learning points taken at positions along the longitudinal direction of the steel plate in correspondence with the water-cooling time index.

[0035] The learning rate at each learning point is calculated using the following procedure. (Step 1) The learning coefficient α when water-cooling a steel plate S from a measured starting temperature T0 at a thermometer on the water-cooling section entrance side (e.g., thermometer 1) to a measured ending temperature T1 at a thermometer on the water-cooling section exit side (e.g., thermometer 2) in τ seconds with a predetermined cooling capacity is calculated by the following procedure. (Step 2) Equation (2) is used to calculate the temperature drop, and the learning coefficient α is initially set to α=1. The calculation is performed by dividing the process into each bank and air-cooled section with different boundary conditions between the start and end points. Actual values ​​related to the cooling capacity (flow rate, passage time, steel plate surface temperature, passage speed) as the learning point passes through each section are collected. From these actual values, the water cooling heat transfer coefficient h and cooling time τ are calculated, and the calculation endpoint temperature T1C is determined by numerical integration with α=1 in equation (2). (Step 3) The temperature calculation for water cooling the steel plate S with the predetermined cooling capacity for τ seconds from the starting temperature T0 is performed by numerical integration with α=1 in equation (2), and the calculation endpoint temperature T1C is obtained. (Step 4) Measured temperature drop ΔT 実測 (=T0-T1) and calculated temperature drop ΔT 計算 The process repeats step 2, adjusting α according to the error in the endpoint, until (=T0-T1C) is equal and the endpoint is T1. The process ends when the error in the endpoint is within an acceptable range, and the value of α at that point is taken as the learning rate.

[0036] [Pattern speed patterns (speed patterns)] While the aforementioned dwell time in each section can be used as an indicator of water cooling time, the longitudinal section of the steel plate corresponding to the dwell time may differ depending on the steel plate. Considering this case would result in a complex structure for the layered table. Therefore, it is preferable that the longitudinal section of the steel plate corresponding to the water cooling time is at least partially common among individual steel plates. One such indicator of water cooling time is the plate speed pattern (hereinafter also referred to as the "speed pattern").

[0037] A speed pattern refers to a pattern (type) of acceleration region, maximum speed region, deceleration region, etc., that is unique to the rolling equipment. These speed patterns are common to all steel plates. Therefore, it is preferable that the indicator of the water cooling time is the speed pattern, and that the multiple positions in the longitudinal direction of the steel plate where the learning points are taken coincide with the inlet and outlet of the cooling equipment, respectively, for each change in the speed pattern, and are inlet-side learning positions and outlet-side learning positions based on the speed pattern.

[0038] By adopting water cooling time indicators and speed patterns, the process of determining learning points in the longitudinal direction of the steel plate can be standardized for any steel plate, and the area of ​​the stratified table where the learning coefficient is stored can also be standardized.

[0039] Furthermore, the multiple positions along the longitudinal direction of the steel plate where the learning points are taken are assumed to be inlet and outlet learning positions based on the speed pattern, coinciding with the inlet and outlet of the cooling equipment, respectively, for each change in the speed pattern. The advantage of providing inlet and outlet learning positions is that the average speed in that section does not contain a mixture of acceleration and deceleration, making it easier to linearly interpolate the learning coefficient.

[0040] [Saving the calculated learning rate] As described above, the learning rate is calculated at both the input learning position and the output learning position for each change in the velocity pattern, and both are stored in the stratification table.

[0041] [Addition of learning points] The multiple locations along the longitudinal direction of the steel plate where the learning points are taken are entry-side learning positions and exit-side learning positions based on the velocity pattern. However, the length of the acceleration region and the maximum velocity region within the plate can become very long depending on the total length of the steel plate and the acceleration rate in the acceleration region. In such regions, the distance between learning points becomes long, which can result in insufficient correction of equipment malfunctions and other uncertain factors using the learning coefficient. Therefore, it is preferable to add learning points at predetermined fixed intervals in addition to the entry-side and exit-side learning points based on the velocity pattern. This limits the length of regions without learning points, improving the effectiveness of correcting for equipment malfunctions and other uncertain factors. The fixed interval is preferably 150 to 250 m.

[0042] Furthermore, in cases where the total length of the plate is short, there may be steel plates that do not correspond to the learning points at the aforementioned fixed length pitch. If the learning coefficient is not calculated at such learning points, when updating the learning coefficient, the old and new learning coefficients will be mixed in the longitudinal direction of the plate within the same speed pattern region. Since the learning coefficient corrects for the effects of equipment malfunctions and other uncertain factors, it is desirable to use the most recent learning coefficient possible. Therefore, even for learning points that do not correspond to the learning points, it is preferable to update the learning coefficient by linear interpolation or extrapolation from learning points for which the learning coefficients on both sides or one side of the learning point have already been calculated.

[0043] [Examples of learning points based on speed patterns] The method for selecting learning points for the input and output learning positions based on the speed patterns described above is explained below.

[0044] In the case of a runout table, the patterns of the sheet metal speed are more specifically as follows: (a) The primary acceleration region from when the leading edge of the plate leaves the final stand 3 of the finishing rolling mill until it reaches the winding machine 5. (b) The secondary acceleration region from the start of winding until the set maximum speed is reached. (c) The range of maximum speed that is maintained after reaching the maximum speed. (d) A deceleration region in which the tail end slows down to the speed at which it passes the final stand 3. (e) The creeping region from when the tail end leaves the final stand 3 until it is wound up by the winding machine 5.

[0045] Note that the threading speed, which is the speed at which the leading edge of the plate exits the final stand of the finishing rolling mill, is assumed to be constant and is not included as a factor in the stratification table for calculating the learning coefficient, but it may be included if necessary.

[0046] Figure 2(a) shows examples of the longitudinal position of the steel plate at the learning point and the plate speed corresponding to the plate speed patterns described above, for the exit speed.

[0047] In Figure 2(a), the horizontal axis, the "longitudinal" axis, represents the longitudinal distance (in meters) from the leading edge of the steel plate, and the direction of plate passage is the opposite of that in Figure 1 (from right to left). The vertical axis, the "speed" axis, represents the speed of the steel plate passage (in m / s).

[0048] In Figure 2(a), the exit speed 6 (dashed line) is the measured feed speed, defined as the exit speed of the final stand with the advance rate taken into consideration until the start of winding at the tip of the board, and then by the winding speed after the tip is wound. The exit speed 6 is measured from the tip of the board, which is the starting point of the measurement (the intersection of the vertical and horizontal axes), to the tail end of the board as it progresses.

[0049] The exit velocity 6 (dashed line) exhibits a bell-shaped curve as shown in the figure, due to the velocity pattern sequentially passing through the regions of primary acceleration, secondary acceleration, maximum speed, and creeping. Therefore, the longitudinal positions of the steel plate at the exit position secondary acceleration start point 10, the exit position maximum speed arrival point 13, the exit position deceleration start point 16, and the creeping start point 18 on the exit velocity 6 (dashed line) can be determined as follows. That is, the distances from the leading edge of the steel plate to each of these points are represented by the symbols L10, L13, L16, and L18, respectively, and are as follows. L10 = Distance from the cooling equipment outlet to the winding machine L13 = L10 + (Set maximum speed squared - Speed ​​at the start of secondary acceleration squared) ÷ (2 × Acceleration rate in the secondary acceleration region) L16 = L13 + Set maximum speed 9 × Duration of maximum speed L18 = L16 + (Set maximum speed squared - Speed ​​at which the tail end passes the final stand 3 squared) ÷ (2 × Deceleration rate in the creeping region) In this way, for each change in the velocity pattern, a longitudinal position of the steel plate that coincides with the outlet of the cooling equipment (an outlet learning position based on the velocity pattern) can be determined, and a learning point can be taken at that position.

[0050] On the other hand, Figure 2(b) is a diagram that further adds entry speed and average speed to Figure 2(a). In Figure 2(b), the entry speed 8 (dotted line) is the speed calculated by considering the advance rate of the rolling mill's final stand speed before winding, and the winding speed after winding. It is measured from the leading edge of the plate, which is the starting point for measuring the entry speed 8 (intersection of the vertical and horizontal axes), to the trailing edge of the plate as the plate progresses. This entry speed 8 (dotted line) exhibits a mountain shape as shown in the figure, as the speed pattern sequentially passes through the regions of secondary acceleration, maximum speed, and deceleration. Therefore, the longitudinal positions of the steel plate at the entry position secondary acceleration start point 11, the entry position maximum speed arrival point 14, and the entry position deceleration start point 17 can be determined as follows. That is, the distances from the leading edge of the steel plate to each of these points are represented by the symbols L11, L14, and L17, respectively, and are as follows. Here, ΔL is a symbol representing the distance from the entry side to the exit side of the cooling equipment. L11 = L10 + ΔL L14 = L13 + ΔL L17 = L16 + ΔL In this way, for each change in the velocity pattern, a longitudinal position of the steel plate that coincides with the cooling equipment inlet (inlet-side learning position based on the pattern) can be determined, and a learning point can be taken at that position.

[0051] Furthermore, in Figure 2(b), the average speed 7 (solid line) represents the average of the exit speed 6 and entry speed 8 at the same position along the longitudinal direction of the steel plate, over the longitudinal direction of the steel plate. By using this average speed 7, the points of change between the two speeds are eliminated, preventing acceleration and deceleration from appearing between the learning points, thus facilitating linear interpolation of the learning coefficient.

[0052] Furthermore, Figure 2(c) is a diagram showing Figure 2(b) with additional learning points of a fixed length. In Figure (c), the fixed-length learning point 12 in the secondary acceleration region and the fixed-length learning point 15 in the maximum speed region are examples showing two of the multiple learning points added at the fixed-length pitch positions. In this example, the fixed length is set to 200m, so the distance between learning point 12 and learning point 15 is a multiple of 200m.

[0053] [Comparison of calculated temperature and measured temperature at the learning point] This section describes a method for comparing the calculated temperature and the measured temperature at a learning point using a measured cooling curve taken over the entire length of each steel plate. The explanation focuses on the case where the temperature measurement means is the outlet thermometer 2, but the same method applies when the input thermometer 1 is used. Furthermore, this section assumes that no learning points (learning points 12 and 15) are added at fixed length intervals, but the same method applies when they are added.

[0054] The relationship between the elapsed time from the start of temperature measurement and the longitudinal distance of the steel plate from the leading edge of the steel plate can be determined based on the velocity pattern in the exit-side measured cooling curve (not shown) measured by thermometer 2. Therefore, the exit-side secondary acceleration start point 10, the exit-side maximum speed reach point 13, the exit-side deceleration start point 16, and the creeping start point 18 can be plotted on the exit-side measured cooling curve. Then, for each velocity pattern in the secondary acceleration region between points 10 and 13, the maximum speed region between points 13 and 16, and the deceleration region between points 16 and 18, the calculated temperature and the measured temperature are compared, and the learning coefficient is calculated.

[0055] The calculated learning coefficient is stored in a layered table, corresponding to the steel plate manufacturing conditions and speed pattern at the time of calculation. However, filtering, such as exponential smoothing as shown in equation (3), is performed during storage. Learning rate update value = β × current learning rate + (1-β) × previous learning rate ... (3) (β: Exponential smoothing gain 0~1.0)

[0056] [Exceptional handling when saving and using the learning rate] In cases where the board length is short, for example, an exceptional speed pattern may occur where the board decelerates without reaching the set maximum speed, unlike the usual speed pattern that reaches the set maximum speed. If the learning coefficient is calculated and saved at the actual point where the maximum speed is reached in this case, the board speed when saved in the same storage area in the layer table will differ from the usual value. Furthermore, when using the learning coefficient in the aforementioned exceptional speed pattern, the saved learning coefficient corresponds to the set maximum speed in the usual speed pattern, and therefore does not represent the maximum speed actually reached.

[0057] Even in such exceptional speed patterns, it is desirable that the learning coefficient during saving corresponds to the set maximum speed, and the learning coefficient during use corresponds to the actual speed of the sheet metal. To this end, the present invention prefers to perform the following exceptional processing for saving and using the learning coefficient, as this does not require modification of the stratified table. That is, during saving, for learning points where the sheet metal speed at the point of reaching the maximum speed is less than the set maximum speed and the point of reaching the maximum speed is the point where deceleration begins, the learning point is corrected to correspond to the set maximum speed, and the learning coefficient calculated for the corrected learning point is saved. During use, the saved learning coefficient is corrected to correspond to the actual maximum speed and used.

[0058] Figure 3 is an explanatory diagram of this exceptional processing method. In Figure 3, (a) and (b) show the relationship between the position of the learning point in the longitudinal direction of the steel plate, the plate speed, and the learning coefficient, respectively. In Figure 3, the maximum speed actually reached at learning point 24, the starting point of deceleration on the exit side on the dashed line with an exit speed of 6, and learning point 27, the starting point of deceleration on the entry side on the dashed line with an entry speed of 8, both fall short of the set maximum speed of 19.

[0059] At this time, the learning point 24 at the exit deceleration start point is extrapolated to the set maximum speed 19 using the acceleration side extrapolation line 20A, and this is set as the acceleration side learning point 26 corresponding to the set maximum speed 19, and the acceleration side learning coefficient 22A calculated at this learning point 26 is saved. On the other hand, the learning point 27 at the entry deceleration start point is extrapolated to the set maximum speed 19 using the deceleration side extrapolation line 20B, and this is set as the deceleration side learning point 25 corresponding to the set maximum speed 19, and the deceleration side learning coefficient 22B calculated at this learning point 25 is saved.

[0060] The saved learning coefficients have been corrected to correspond to the set maximum speed, so when using them, they should be corrected to the learning coefficient for the actual maximum speed reached. For example, in Figure 3(b), when using the acceleration-side learning coefficient 22A, the acceleration-side learning coefficient 22A is plotted on a straight line connecting the learning coefficients of the acceleration-side learning point 23, which is one step before the actual maximum speed reached, and the learning point 24, which is the starting point of deceleration on the exit side. Then, the straight line connecting the learning coefficient of learning point 24 and the learning coefficient 22A of learning point 26 is used as the learning coefficient of the learning point to be used that does not reach the set maximum speed.

[0061] Furthermore, when using the deceleration learning coefficient 22B, the deceleration learning coefficient 22B is plotted on a straight line connecting the learning coefficients of the acceleration learning point 28 (the next highest speed reached) and the learning point 27 (the starting point of deceleration on the entry side). Then, the learning coefficient at the intersection of the straight line connecting the learning coefficient of learning point 27 and the learning coefficient 22B of learning point 25, and the position of the desired learning point where the set maximum speed is not reached, is taken as the learning coefficient of the desired learning point.

[0062] In this embodiment, the learning method involves using a temperature drop model to calculate a correction coefficient for the water cooling heat transfer coefficient from actual data as the learning coefficient, exponentially smoothing it with the previous learning coefficient value, and then managing it in a stratified table with manufacturing conditions as factors. However, the learning method is not limited to this. For example, a combination of short-term learning and stratified table learning (long-term learning) may be used, or machine learning may be applied. Furthermore, simultaneous learning of multiple items, including learning related to transformations, as in the prior art (Patent Document 2), may also be used.

[0063] [Method of manufacturing hot-rolled steel sheets] The method for manufacturing hot-rolled steel sheets according to the present invention includes a cooling step according to the hot-rolled steel sheet cooling method described above. This cooling step is applied to the water cooling of the steel sheets being transported on the runout table from the final stand of the finishing rolling mill to the winding machine. This ensures accurate winding temperature and improves the yield and quality stability of the hot-rolled steel sheets. [Examples]

[0064] The effects of the present invention will be specifically explained below by comparing examples of the present invention with comparative examples.

[0065] In this example of the present invention, the cooling equipment shown in Figure 1 was used. The target material was a general material (low carbon steel) with a plate thickness of 2.02 mm, a plate width of 937 mm, a target finishing temperature of 870°C, and a target winding temperature of 640°C.

[0066] As described in the embodiment, the learning coefficient for general materials is in the form of a layered table of steel plate manufacturing conditions, and the learning points are set at multiple positions in the longitudinal direction of the plate that coincide with the inlet and outlet of the cooling equipment, respectively, for each change in the speed pattern. That is, in the present invention example, the learning coefficient calculated for each longitudinal region of the steel plate corresponding to each speed pattern is applied to that longitudinal region. On the other hand, in the comparative example, the learning coefficient calculated at a single point at the leading edge of the plate is uniformly applied to the entire length of the plate.

[0067] Figure 5 shows a comparison of charts of winding temperature results over the entire length (vertical axis: winding temperature results [°C], horizontal axis: position in the longitudinal direction of the steel plate (left end is the tip)). From Figure 5, in the comparative example (thin line), the variation in the deviation of the actual temperature from the target temperature in the longitudinal direction of the steel plate is 1σ = 9.1°C. In contrast, in the example of the present invention (thick line), 1σ = 6.3°C, and the variation in the deviation between the target temperature and the actual temperature in the longitudinal direction has decreased, demonstrating the superiority of the example of the present invention over the comparative example. [Explanation of symbols]

[0068] 1. Thermometer (Thermometer at the entrance of the cooling equipment) 2. Thermometer (Thermometer at the outlet of the cooling equipment) 3. Final rolling stand 4 Cooling equipment 5 Winder 6 Output speed 7 Average speed 8 Entry speed 9 Setting maximum speed 10 Exit position secondary acceleration start point 11 Entry side position secondary acceleration start point 12. Secondary acceleration region constant length learning point 13 Output position maximum speed reached point 14 Entry side position maximum speed reached point 15. Maximum speed range constant length learning points 16 Exit position deceleration start point 17 Entry side position deceleration start point 18. Creeping start point 19 Maximum setting speed 20A Acceleration side extrapolation line 20B Deceleration side extrapolation straight line 21. Learning rate when calculating the learning rate 22A Acceleration learning coefficient corresponding to the set maximum speed 22B Learning coefficient for deceleration corresponding to the set maximum speed 23 The learning point on the acceleration side just before the actual maximum speed reached. 24 Learning point of the starting point of deceleration on the exit side 25 Learning points for deceleration corresponding to the set maximum speed 26 Acceleration learning points corresponding to the set maximum speed 27 Learning point of entry-side deceleration start point 28 The learning point for the next deceleration after the actual maximum speed reached 40 calculator 45 Control Panel S steel plate

Claims

1. In a method for cooling hot-rolled steel sheets, in which the cooling capacity of a cooling system that water-cools the hot-rolled steel sheet before coiling after hot rolling is corrected using a learning coefficient, A method for cooling a hot-rolled steel sheet, characterized in that the learning coefficient is managed using a stratified table with the manufacturing conditions of the hot-rolled steel sheet as factors, and is calculated at learning points taken at multiple positions in the longitudinal direction of the hot-rolled steel sheet.

2. The method for cooling a hot-rolled steel sheet according to claim 1, characterized in that the positions of the learning points taken at multiple locations in the longitudinal direction of the hot-rolled steel sheet are inlet learning positions and outlet learning positions that coincide with the inlet and outlet of the cooling equipment, respectively, for each change in the speed pattern.

3. The method for cooling a hot-rolled steel sheet according to claim 2, characterized in that the speed pattern comprises a primary acceleration region, a secondary acceleration region, a maximum speed region, a deceleration region, and a creeping region.

4. Furthermore, the method for cooling a hot-rolled steel sheet according to claim 2 or 3, characterized in that learning points are added at positions with a predetermined length pitch in addition to the learning points at the input and output learning positions.

5. Furthermore, the method for cooling a hot-rolled steel sheet according to claim 2 or 3, characterized in that, for learning points where the sheet speed at the point of reaching the maximum speed is less than the set maximum speed and the point of reaching the maximum speed is the point where deceleration begins, the learning point is corrected to correspond to the set maximum speed, the learning coefficient calculated at the corrected learning point is saved, and the saved learning coefficient is corrected to correspond to the actual maximum speed and used.

6. A method for manufacturing a hot-rolled steel sheet, characterized by comprising a cooling step by a method for cooling a hot-rolled steel sheet as described in any one of claims 1 to 3.

Images