Setting calculation device for hot rolling production line

Abstract

The setpoint determination circuit obtains the measured inlet temperature, calculates the predicted inlet temperature of the finishing mill, uses the predicted inlet temperature and the target inlet temperature to calculate the final inlet temperature of the finishing mill, and uses the final inlet temperature to calculate the predicted outlet temperature of the finishing mill. The control setpoint is determined when the predicted outlet temperature matches the target outlet temperature. The learning calculation circuit calculates the actual recalculated outlet temperature of the finishing mill, uses the final inlet temperature to calculate the first instantaneous learning value and the first variance, uses the predicted inlet temperature to calculate the second instantaneous learning value and the second variance, uses the measured inlet temperature to calculate the third instantaneous learning value and the third variance, and uses the first variance, the second variance, and the third variance to calculate the influence of the predicted inlet temperature or the measured inlet temperature relative to the final inlet temperature. The final inlet temperature is calculated using both the predicted inlet temperature and the influence value.

Description

Technical Field

[0001] This disclosure relates to a setting calculation device for a hot rolling production line. Background Technology

[0002] A setting calculation device is installed in the hot rolling production line to control quality parameters such as temperature and thickness of the rolled material exiting the finishing mill within specified ranges. Such a setting calculation device typically includes a setting value determination device and a learning calculation device. The setting value determination device predicts the temperature of the rolled material exiting the finishing mill (hereinafter referred to as the "finishing mill exit temperature") and the rolling load required for rolling the material before rolling, and determines the finishing mill's roll speed, roll gap, etc., based on the predicted finishing mill exit temperature and rolling load values. The learning calculation device measures the temperature of the rolled material exiting the finishing mill after rolling and learns the measured finishing mill exit temperature value, thereby improving the prediction accuracy of the finishing mill exit temperature for the next material (the material to be rolled next).

[0003] When predicting the temperature on the finishing mill exit side, the temperature on the upstream side of the finishing mill, such as the temperature of the rolled material on the finishing mill inlet side (hereinafter also referred to as the "finishing mill inlet temperature"), is used as the input value.

[0004] In the technology described in Patent Document 1, the change in the temperature at the exit of the finishing roll, caused by a change in the temperature at the entry of the finishing roll, is used as an influence coefficient to change the speed of the rolled material. Patent Document 1 does not specify whether the temperature at the entry of the finishing roll is a measured value or a predicted value. That is, it is unclear how the measured temperature at the entry of the finishing roll is used to predict the temperature at the exit of the finishing roll.

[0005] In the technology described in Patent Document 2, a predicted temperature value for the finishing mill exit side is calculated, and the temperature of the rolled material is controlled based on this predicted value. In this case, the temperature distribution of the rolled material from the upstream to the downstream side of the rolling production line is predicted, and an organizational model composed of neural networks or the like is used to determine the target temperature value of the rolled material, thereby controlling it to achieve that target temperature value. Patent Document 2 describes changing the operating conditions of the rolling equipment on the downstream side, or changing the target temperature, based on the temperature of the rolled material measured on the upstream side of the rolling production line. On the other hand, it is unclear whether the measured value is used directly in Patent Document 2.

[0006] Existing technical documents Patent documents Patent Document 1: Japanese Patent Application Publication No. 2016-215237 Patent Document 2: Japanese Patent Application Publication No. 2022-522181 Summary of the Invention

[0007] The problem that the invention aims to solve Patent documents 1 and 2 do not disclose how the finishing mill inlet temperature is used during the rolling of the rolled material and when predicting the finishing mill exit temperature. That is, it is unclear whether the measured value obtained by the thermometer located at the finishing mill inlet is simply used as is, or whether the predicted value at the finishing mill inlet is corrected and used. If the finishing mill inlet temperature changes, the finishing mill exit temperature also changes, affecting the learning of the learning calculation device. Therefore, in order to improve the learning accuracy of the learning calculation device, it is important to determine the finishing mill inlet temperature based on the predicted and measured values. Furthermore, in a hot rolling production line, the rolled material is constantly exposed to the atmosphere, so the surface condition of the rolled material changes due to oxidation. Additionally, the ambient temperature and the temperature of the cooling water are not constant but constantly changing. Moreover, the average temperature of the rolled material is important when predicting the rolling load; not only the surface temperature but also the internal temperature distribution is important. Although the internal temperature distribution can be predicted and calculated, it cannot be actually measured by a thermometer. Therefore, there are problems with the inability to accurately predict the temperature of the rolled material during rolling and the inability to correctly predict the final finishing roll exit temperature.

[0008] This disclosure was made to solve the aforementioned problems. The purpose of this disclosure is to provide a setting calculation device for a hot rolling production line, which can improve the prediction accuracy of the finishing mill exit temperature and rolling load by determining the inlet temperature of the finishing mill to an appropriate value.

[0009] Methods for solving problems The first aspect of this disclosure relates to a setting calculation device for a hot rolling mill that calculates control setting values ​​for rolling a material in a finishing mill of a hot rolling production line. The setting calculation device includes: a setting value determination circuit that determines control setting values ​​before rolling the material; and a learning calculation circuit that calculates learning values ​​using actual values ​​from the hot rolling mill after rolling the material. The setting value determination circuit is configured to perform: acquiring an inlet temperature measurement value measured at the inlet side of the finishing mill; calculating a predicted inlet temperature value for the finishing mill; using the predicted inlet temperature value and the predicted inlet temperature value, calculating a determined inlet temperature value for the finishing mill; using the determined inlet temperature value, calculating a predicted outlet temperature value for the finishing mill; and determining a control setting value in a manner that makes the predicted outlet temperature value consistent with a target outlet temperature value. The learning calculation circuit is configured to perform the following: calculate the actual recalculated value of the exit side temperature of the finishing mill based on actual values; calculate a first learning instantaneous value using the determined value of the inlet side temperature in a manner that makes the actual recalculated value of the exit side temperature consistent with the measured value of the exit side temperature, and calculate a first variance value representing the variance of the first learning instantaneous value; calculate a second learning instantaneous value using the predicted value of the inlet side temperature in a manner that makes the actual recalculated value of the exit side temperature consistent with the measured value of the exit side temperature, and calculate a second variance value representing the variance of the second learning instantaneous value; calculate a third learning instantaneous value using the measured value of the inlet side temperature in a manner that makes the actual recalculated value of the exit side temperature consistent with the measured value of the exit side temperature, and calculate a third variance value representing the variance of the third learning instantaneous value; and calculate the influence degree of the predicted value of the inlet side temperature or the measured value of the inlet side temperature relative to the determined value of the inlet side temperature using the first variance value, the second variance value, and the third variance value. The determined value of the inlet side temperature is calculated using the predicted value of the inlet side temperature and the influence degree.

[0010] The second perspective, building upon the first, also possesses the following characteristics: The learning value and influence degree are at least stratified based on the steel grade of the rolled material. Calculating the learning value includes calculating the learning value for each stratum. Calculating the influence degree includes calculating the influence degree for each of the strata. Calculating the inlet temperature determination value includes using the influence degree calculated for each of the strata to calculate the inlet temperature determination value.

[0011] Invention Effects According to this disclosure, by calculating the influence of three variance values ​​on the temperature determination value of the finishing mill inlet side, the temperature determination value of the finishing mill inlet side can be determined to be an appropriate value. This improves learning stability and the prediction accuracy of the temperature and rolling load of the finishing mill outlet side. Attached Figure Description

[0012] Figure 1 This is a schematic diagram illustrating an example of a setting calculation device and a hot rolling production line using the setting calculation device in an embodiment.

[0013] Figure 2 This is a flowchart illustrating the process of determining the setpoints of the setting calculation device and the setting value determination process.

[0014] Figure 3 This is a flowchart illustrating the learning calculation process of the learning calculation device 12 of the setting calculation device 1.

[0015] Figure 4 It is a graph used to illustrate the changes in influence, variance, and learning instantaneous value.

[0016] Figure 5 This is a diagram illustrating the data stored in each layer of the learning value storage device, the learning instantaneous value storage device, and the influence storage device.

[0017] Figure 6 This is a diagram illustrating an example of the hardware structure of a computing device. Detailed Implementation

[0018] Hereinafter, the setting calculation device according to the embodiments of the present disclosure will be described with reference to the accompanying drawings. Furthermore, common elements in all figures are labeled with the same reference numerals, and repeated descriptions are omitted.

[0019] Figure 1 This is a schematic diagram illustrating an example of a setting calculation device 1 and a hot rolling production line (hereinafter also referred to as a "rolling production line") 2 using the setting calculation device 1. In the rolling production line 2, as the main rolling equipment, a heating furnace 21, a roughing mill 22, and a finishing mill 23 are provided, as well as a cooling device and a winding machine (not shown).

[0020] The heating furnace 21 is configured to heat the pre-rolled material (slab) 20 to a specified temperature (e.g., 1200°C). The roughing mill 22 typically has 1 to 3 units (in... Figure 1 In the example shown, there is one rolling mill stand (hereinafter referred to as "stand"). The roughing mill 22 is configured to reverse-roll the rolled material (slab) 20 drawn from the heating furnace 21. The finishing mill 23 consists of multiple mills arranged side by side along the rolling direction of the rolled material 20. Figure 1 The example shown is a tandem rolling mill with 7 stands. Each stand has two work rolls (upper and lower), a support roll, and a motor for roll rotation. A pressing device is installed on the support roll, allowing adjustment of the roll gap. The rolling load of each stand is measured by a rolling load sensor (not shown). The cooling system cools the rolled material 20 by injecting water into it using a cooling unit. The cooled rolled material 20 is then wound into a coil by a winding machine.

[0021] Multiple sensors, serving as measuring devices, are installed in key areas of the rolling production line 2. Figure 1 Examples of sensors include a roughing mill exit temperature gauge 31 installed on the exit side of the roughing mill 22, a finishing mill inlet temperature gauge 32 installed on the inlet side of the finishing mill 23, and a finishing mill exit temperature gauge 33 installed on the exit side of the finishing mill 23. The actual data measured by the various sensors are constantly transmitted to the setting calculation device 1 and the control computer (illustrated later).

[0022] The rolling production line 2 is operated using a computer control system with a hierarchical structure. Although not shown in the diagram, the control system includes a control computer and a host computer interconnected via a network. The control computer includes a setting calculation device 1 as described in this embodiment. If a rolling plan is input to the host computer, rolling information is sent from the host computer to the setting calculation device 1. The rolling information includes target plate thickness, target plate width, target temperature, etc. The setting calculation device 1 receives the rolling information input from the host computer, calculates the control setting values ​​for each rolling mill, and sends the calculated control design values ​​to the rolling production line 2, thereby executing the control of each rolling mill constituting the rolling production line 2. The structure and function of the setting calculation device 1 will be described below.

[0023] The setting calculation device 1 includes a setting value determination device 11 and a learning calculation device 12.

[0024] First, the structure of the setpoint determination device 11 will be described. The setpoint determination device 11 includes an input value prediction unit 111, an input value determination unit 112, and a setpoint calculation unit 113.

[0025] The input value prediction unit 111 is based on the roughing roll exit side temperature (measured value) T measured by the roughing roll exit side thermometer 31. RD The predicted temperature T on the finishing mill inlet side is calculated using the following formula (1). FE pre Alternatively, the input value prediction unit 111 predicts the roughing roll exit temperature (predicted value) T based on the time required for roughing, the rolling load required for roughing, and the rolling speed. RD The predicted temperature T on the finishing mill inlet side is calculated using the following formula (1). FE pre .

[0026] [Formula 1] In equation (1) above, T FE pre This represents the predicted temperature [degC] at the entry point of the finishing mill. RDTemperature at the roughing roll exit. T is the temperature of the rolled material at 20°C. σ is the Stefan-Boltmann coefficient. ε is the emissivity. ρ is the density [kg / m³]. 3 ]. Specific heat [kJ / (kg·degC)]. ht air It is the convective thermal conductivity with air [W / mm]. act is the actual speed of the rolled material 20 [m / s]. x is the position of the rolled material 20 in the hot rolling production line 2 [m].

[0027] The input value determination unit 112 uses the measured value T of the finishing mill entry temperature measured by the finishing mill entry temperature gauge 32. FE act The predicted temperature T on the finishing mill inlet side calculated by the input value prediction unit 111 FE pre The influence degree β, which will be discussed later, is calculated and determined by the following formula (2) to determine the temperature determination value T on the finishing mill inlet side. FE use The temperature T at the entry point of the finishing mill is determined by... FE use Also known as the input determination value for the finishing mill.

[0028] [Formula 2] In equation (2) above, T FE use This is the temperature determined by the entry temperature of the finishing mill [degC]. T FE pre This is the predicted temperature [degC] on the inlet side of the finishing mill. FE act This is the measured temperature on the entry side of the finishing mill. β is the influence (usual value).

[0029] The setpoint calculation unit 113 uses the finishing mill entry temperature determination value T calculated by the input value determination unit 112. FE use The predicted temperature T on the finishing mill exit side is calculated using the following formula (3). FD pre Predicted temperature T on the finishing mill exit side FD pre This is the predicted temperature value of the rolled material 20 at the location of the finishing mill exit temperature gauge 33. The predicted temperature T on the finishing mill exit side... FD preIn the calculation, from the finishing mill inlet thermometer 32 to the finishing mill outlet thermometer 33, the temperature T(x) of the rolled material 20 is calculated sequentially, taking into account the temperature drop caused by cooling between the stands, the processing heat caused by rolling, and the learning value. Here, the temperature drop caused by cooling is considered to be the temperature drop caused by the cooling device (not shown) installed between the stands. In addition, the processing heat based on rolling is considered to be the load prediction value and the temperature change based on the contact heat transfer with the work rolls.

[0030] [Formula 3] In equation (3) above, T FD pre This is the predicted temperature value on the finishing mill exit side. set p is the speed setting value [m / s] of the rolled material 20 at position x (hereinafter referred to as "rolled material position") on the rolling production line 2. set This is the rolling load setting [kN] at position x of the rolled material. set This is the setpoint for the cooling flow rate at position x of the rolled material [m]. 3 / hr].ht rl is the contact thermal conductivity of the rolled material at position x with the roll [W / mm]. z(s(x)) is the learning value at position x of the rolled material using the learning value of the number s(x) [degC].

[0031] Here, the learning value application number s(x) represents the number of the position where the temperature T(x) of the rolled material 20 is corrected using the learning value when the rolled material 20 is present at position x. For example, in the case of correcting the temperature T(x) of the rolled material 20 at the inlet side of the finishing mill 23, at the inlet side of each stand of the finishing mill 23 (in Figure 1 In the example shown, corrections are performed at 7 locations. Therefore, the learning value applied to the upstream rack inlet is numbered 1, and the learning value applied to the downstream rack inlet is numbered 7.

[0032] The setpoint calculation unit 113 calculates the predicted temperature T on the finishing mill exit side. FD pre The rolling speed and the flow rate of the cooling water in the cooling device (hereinafter also referred to as "cooling flow rate") are determined in a manner consistent with the target temperature value of the finishing mill exit side. Here, the setpoint calculation unit 113 calculates the predicted temperature value T of the finishing mill exit side. FD pre If the difference between the predicted temperature and the target temperature on the finishing mill exit side is large (e.g., exceeding a threshold), change the rolling speed and cooling flow rate, and then recalculate the predicted temperature T on the finishing mill exit side after the change. FD preThat is, the setpoint calculation unit 113 performs convergence calculations until the predicted temperature T on the finishing mill exit side is reached. FD pre It is consistent with the target temperature value of the finishing mill exit side.

[0033] The setpoint calculation unit 113 calculates the predicted temperature T on the finishing mill exit side. FD pre The set values ​​of rolling speed and cooling flow rate, which are consistent with the target value of the finishing roll exit temperature, are transmitted to the rolling production line 2.

[0034] Next, the structure of the learning calculation device 12 will be described. The learning calculation device 12 includes a performance recalculation unit 121, a learning value calculation unit 122, a learning value storage device 123, a learning instantaneous value storage device 124, an influence calculation unit 125, and an influence storage device 126.

[0035] The actual recalculation unit 121 calculates the actual recalculated value T of the finishing mill exit temperature based on the actual value of the rolling load, the actual value of the rolling speed, and the actual value of the cooling flow rate of the rolling material 20 obtained by the finishing mill 23, using the following formula (4). FD rep .

[0036] [Formula 4] In equation (4) above, T FD rep This is the actual recalculated value of the temperature on the finishing roll side. P act (x) is the actual value of the rolling load [kN] at position x of the rolled material. W act (x) is the actual cooling flow rate at position x of the rolled material [m 3 / hr]. v act (x) is the actual rolling speed [m / s] at position x of the rolled material.

[0037] The learning value calculation unit 122 uses the actual recalculation value T of the finishing mill exit temperature calculated by the actual recalculation unit 121. FD rep The instantaneous learning value is calculated using the following formula (5). The following will use the temperature determination value T at the mill inlet side. FE use The calculated instantaneous learning value is also called the "first instantaneous learning value".

[0038] [Formula 5] The calculation of the first instantaneous learning value uses the finishing mill inlet temperature determination value T calculated by the input value determination unit 112. FE useAs the temperature at the entry point of the finishing mill, the first instantaneous value Z is set. cur (s(x)) changes while calculating the actual recalculated value T of the finishing mill exit temperature. FD rep Until the calculated actual recalculated value T of the finishing roll exit temperature. FD rep Temperature measurement (actual value) T on the finishing mill exit side FD act The calculation is repeated until the difference becomes small enough.

[0039] The learning value calculation unit 122 recalculates the actual value T at the finishing mill exit temperature. FD rep Temperature measurement value T on the end of the finishing roll FD act When the difference is sufficiently small, specifically, when the difference is less than a threshold T TH In the case shown in equation (6), the first learning instantaneous value Z is... cur (s(x)) as Z cur (s(x))(use) is stored in the learning instant value storage device 124.

[0040] [Formula 6] The learning value calculation unit 122 calculates the value when the difference is greater than the threshold T. TH In the case of changing the instantaneous learning value Z as shown in equation (7), cur (s(x)), and then perform the calculation again.

[0041] [Formula 7] In equation (7) above, T FD act This is the actual temperature [degC] at the exit side of the finishing mill. cur (s(x))(use) is the first instantaneous learning value [degC]. d is the change adjustment coefficient [-]. N is the number of learning value application positions [-]. Here, since the learning value is applied on the inlet side of each stand of the finishing mill 23, the number of learning value application positions N is the number of stands of the finishing mill 23, i.e., 7.

[0042] Next, the learning value calculation unit 122 calculates the first instantaneous learning value Z based on the convergence calculation obtained by equations (5) to (7) above. cur (s(x))(use), update the learning value stored in the learning value storage device 123 by the following formula (8). old (s(x))(use).

[0043] [Formula 8] In equation (8) above, Z old (s(x))(use) is the learned value stored in [degC]. α is the adjustment gain [-].

[0044] Next, the learning value calculation unit 122 updates the learning value Z. new (s(x))(use) is stored in the learning value storage device 123.

[0045] Next, the learning value calculation unit 122 calculates the variance of the first learning instantaneous value (hereinafter also referred to as the "first variance value"). Here, the variance value, including the second variance value and the third variance value described later, represents the deviation (dispersion) of the learning instantaneous value. The variance value indicates how much it deviates from the average value of the learning instantaneous value. When calculating the variance value, the learning value calculation unit 122 traces back to the nd learning instantaneous values ​​that were previously acquired and stored in the learning instantaneous value storage device 124. The number nd acquired is defined as the adjustment coefficient.

[0046] The learning value calculation unit 122 calculates the nd first learning instantaneous values ​​Z obtained by the following formula (9). cur The average value Z of (s(x))(use) ave (s(x))(use).

[0047] [Formula 9] Next, the learning value calculation unit 122 calculates the average value Z from the first instantaneous learning value using the following formula (10). ave The first variance p of (s(x))(use) m cur (s(x))(use).

[0048] [Formula 10] The learning value calculation unit 122 uses the input value prediction unit 111 to calculate the predicted temperature T of the finishing mill entry side using the above formula (1). FE pre To replace the temperature determination value T on the finishing mill inlet side FE use The actual recalculated value T is calculated using the following formula (11). FD rep The predicted temperature T on the entry side of the finishing mill is calculated using the following formulas (12) to (14). FE pre The learning instantaneous value at time (hereinafter also referred to as the "second learning instantaneous value") Z cur (s(x))(pre) and the variance of the second learning instant (hereinafter also referred to as "second variance") pm cur (s(x))(pre). Using the predicted temperature T on the finishing mill inlet side. FE pre The second learning instantaneous value Z cur (s(x))(pre) is also stored in the learning instant value storage device 124.

[0049] [Formula 11] In addition, the learning value calculation unit 122 uses the finishing mill entry temperature measurement value T measured by the finishing mill entry temperature gauge 32. FE act Replaces the temperature determination value T of the finishing mill inlet side FE use The actual recalculated value T is calculated using the following formula (15). FD rep The measured temperature T on the end of the finishing mill is calculated using the following formulas (16) to (18). FE act The learning instantaneous value Z (hereinafter also referred to as the "third learning instantaneous value") cur The variance of (s(x))(act) and the variance of the third learning instant (hereinafter also referred to as "third difference") p m cur (s(x))(act). Using the temperature measurement value T on the finishing mill inlet side. FE act The third learning instantaneous value Z cur (s(x))(act) is also stored in the learning instant value storage device 124.

[0050] [Formula 12] The influence calculation unit 125 utilizes the temperature determination value T of the finishing mill entry side. FE use First variance p m cur (s(x))(use), using the predicted temperature T of the finishing mill inlet side. FE pre The second variance p m cur (s(x))(pre), and the temperature measurement value T on the finishing mill inlet side was used. FE act The third-party difference p m cur (s(x))(act), calculate the influence β of the next material (the next rolled material 20). The influence β represents the predicted temperature T on the finishing mill inlet side. FE preThe temperature determination value T relative to the entry temperature of the finishing mill FE use The numerical value of the proportion. The influence β is a value in the range of 0 ≤ β ≤ 1. At this time, the measured temperature T on the finishing mill inlet side. FE act The temperature determination value T relative to the entry temperature of the finishing mill FE use The ratio is "1-β". The influence β can also be described as the input value determining the gain. Furthermore, the influence β can also represent the measured temperature T on the finishing mill side. FE act The temperature determination value T relative to the entry temperature of the finishing mill FE use The proportion. Furthermore, when the number of rolled ends is small and the dispersion value is 0, it is possible to not update the influence β. Additionally, for example, when the variance of the learning instant is smaller than the variance using the predicted values ​​and the variance using the measured values, or when no significant difference is found (compared to the threshold d)... mar In the case of a small number of cases, the influence calculation unit 125 maintains the influence β (new) as the previous value β (old) as in the following formula (19).

[0051] [Formula 13] In equation (19) above, β(old) is the influence value stored in the influence value storage device 126. mar It is the threshold for judgment.

[0052] In other cases, i.e., in the case of the second variance p m cur (s(x))(pre) is greater than the third-party difference p m cur In the case of (s(x))(act), the influence calculation unit 125 uses the finishing mill entry temperature determination value T. FE use First variance p m cur (s(x))(use), using the predicted temperature T of the finishing mill inlet side. FE pre The second variance p m cur (s(x))(pre), using the temperature measurement value T on the finishing mill inlet side. FE act The third-party difference p m cur The influence β is calculated by formula (20) or (21) for (s(x))(act).

[0053] [Formula 14] The calculated influence β (new) is stored in the influence storage device 126.

[0054] Next, an example of the operation of the aforementioned setting calculation device 1 will be explained. Figure 2 This is a flowchart illustrating the process of setting value determination by the setting value determination device 11 of the setting calculation device 1.

[0055] The input value prediction unit 111 calculates the predicted temperature T of the finishing mill inlet side before the rolled material 20 reaches the finishing mill inlet side temperature gauge 32. FE pre (Step S1).

[0056] When the tip of the rolled material 20 reaches the finishing mill inlet temperature gauge 32, the finishing mill inlet temperature gauge 32 measures the temperature of the tip portion of the rolled material 20. The input value determination unit 112 collects the finishing mill inlet temperature measurement value T. FE act (Step S2).

[0057] Input value determination unit 112 calculates the temperature determination value T for the finishing mill inlet side. FE use (Step S3). Here, the influence β is the value stored in the influence storage device 126. As the initial value of the influence β, any value can be assigned in advance. For example, as in the following formula (22), the initial value of the influence β can be set to 0, but it is not limited to 0.

[0058] [Formula 15] If the initial value of the influence β is 0 as shown in equation (22) above, then the predicted temperature T on the finishing mill inlet side is... FE pre Directly used as the temperature determining value T for the finishing mill inlet side FE use The calculated temperature determination value T on the finishing mill inlet side is used. FE use The predicted temperature T on the finishing mill exit side is calculated using the above formula (3). FD pre (Step S4), determine the set values ​​for rolling speed, cooling flow rate, etc. (Step S5). The learning value z(x) used in the above formula (3) is stored in the learning value storage device 123 and is predetermined as the initial value. The initial value can be set to 0 as in the following formula (23).

[0059] [Formula 16] In addition, the set values ​​determined by the rolling speed, cooling flow rate, etc. are transmitted to the finishing mill 23 of the rolling production line 2, and the rolling material 20 is rolled based on these set values ​​(step S6).

[0060] Figure 3 This is a flowchart illustrating the learning calculation process of the learning calculation device 12 of the setting calculation device 1.

[0061] When the leading edge of the rolled material 20, rolled by the finishing mill 23, reaches the position of the finishing mill exit temperature gauge 33, the finishing mill exit temperature gauge 33 measures the finishing mill exit temperature of the rolled material 20. The performance recalculation unit 121 collects the measured value T of the finishing mill exit temperature. FD act (Step S11).

[0062] If the temperature measurement value T on the finishing mill exit side FD act Once the collection is complete, proceed to step S12. In step S12, the actual recalculation unit 121 uses the precision rolling entry temperature determination value T. FE use The actual values ​​of rolling speed, rolling load, and cooling flow rate are recalculated. Furthermore, in step S12, the learning value calculation unit 122 calculates the first learning instantaneous value and the variance of the first learning instantaneous value.

[0063] Next, in step S13, the learning value calculation unit 122 updates the learning value using equations (6) to (8) above, and updates the variance value using equations (9) to (10) above. These steps S12 and S13 are equivalent to normal learning calculations.

[0064] In this embodiment, in step S14, the performance recalculation unit 121 uses the predicted value T of the finishing mill entry temperature. FE pre The actual values, such as rolling speed, rolling load, and cooling flow rate, are recalculated using the above formula (11). The learning value calculation unit 122 uses the predicted value T of the finishing mill entry temperature. FE pre The second instantaneous learning value is calculated using the above formula (12). Furthermore, the learning value calculation unit 122 uses the predicted temperature T on the finishing mill entry side. FE pre The second variance value is calculated using the above formulas (13) to (14).

[0065] Next, in step S15, the performance recalculation unit 121 uses the finishing mill entry temperature measurement value T. FE act The actual values, including rolling speed, rolling load, and cooling flow rate, are recalculated using the above formula (15). The learning value calculation unit 122 uses the measured temperature T on the finishing mill inlet side.FE act The second instantaneous learning value is calculated using the above formula (16). Furthermore, the learning value calculation unit 122 uses the measured temperature T on the finishing mill entry side. FE act The third-party difference is calculated using the above formulas (17) to (18).

[0066] Next, in step S16, the influence calculation unit 125 calculates the influence degree β using the above formulas (19) to (21). When the number of rolled strands is the first strand, the variance is 0, and the influence degree β of the first strand remains 0.

[0067] Next, we will explain the changes in influence β, variances, and instantaneous values ​​of each learning step. Figure 4 It is a graph used to illustrate the changes in influence β, variance, and learning instantaneous value. Figure 4 (a) is a graph showing the change in influence β. Figure 4 (b) is a graph showing the change in variance. Figure 4 (c) is a graph showing the change in instantaneous learning values. The horizontal axis of each graph represents the rolling sequence of the rolled material 20. Figure 4 In the diagram, 131 represents the influence value used by the input value determination unit 112, and 132 represents the updated value of the influence value β calculated by the influence value calculation unit 125. 133 represents the first variance value of the first learning instantaneous value calculated by the learning value calculation unit 122, 134 represents the second variance value using the predicted value, and 135 represents the third variance value using the measured value. 136 represents the first learning instantaneous value, 137 represents the second learning instantaneous value using the predicted value, and 138 represents the third learning instantaneous value using the measured value. Furthermore, 139 represents the first learning instantaneous value of the first root, 140 represents the variance value and influence value (updated value) of the first root, 141 represents the influence value (used value) of the second root, 142 represents the variance value of the second root, 143 represents the influence value (updated value) of the second root, 144 represents the influence value (used value) of the third root, and 145 represents the learning instantaneous values ​​of the third to fifteenth roots.

[0068] At the moment when the learning calculation for rolling the first piece of rolled material 20 ends, a first learning instantaneous value, a second learning instantaneous value using measured values, and a third learning instantaneous value using predicted values ​​are calculated, and each learning instantaneous value is saved in the learning instantaneous value saving device 124. On the other hand, since there is only one learning instantaneous value, the variance is 0. The initial value of the influence β saved in the influence saving device 126 is 0, therefore the influence β saved in the influence saving device 126 is also 0.

[0069] During the rolling of the second rolled material 20, the influence degree β of the influence degree preservation device 126 is maintained at 0, therefore, the predicted temperature T on the finishing side is maintained in the same manner as the first material. FEpre Used as the temperature determining value T for the finishing rolling inlet side FE use .

[0070] In the second root, the learning value calculation unit 122 calculates the first variance value between the first and second roots (using the finishing rolling entry temperature determination value T). FE use The variance of the first instantaneous learning value, the second variance of the first and second roots (the variance of the second instantaneous learning value calculated using the predicted temperature of the finishing mill inlet side), and the third variance of the first and second roots (the variance of the third instantaneous learning value calculated using the measured temperature of the finishing mill inlet side) (refer to...) Figure 3 Steps S12, S14, and S15.

[0071] Since the variation of the third instantaneous value using the measured values ​​between the first and second roots is small, the third variance using the measured values ​​is smaller than the second variance using the predicted values ​​(see 142). The influence calculation unit 125 calculates the influence based on these variance values. Since the third variance is small when the measured values ​​are used, the influence β is close to 1.0 (see 143).

[0072] The influence β of the third rolled material 20 is 0.9. As the determining value for the finishing mill inlet temperature, the proportion of the measured finishing mill inlet temperature is used more frequently (see 144). As a result, it can be seen that the variation of the instantaneous learning value also decreases after the third material (see 145). Thereafter, the determining value T for the finishing mill inlet temperature... FE use The temperature measurement value T on the entry side of the finishing mill is mostly used. FE act The proportion.

[0073] Temperature measurement value T on the end of the finishing mill FE act The measurement conditions deteriorated, with significant fluctuations in the instantaneous values ​​learned in the sixteenth root (refer to 146). As a result, the temperature measurement value T on the finishing mill inlet side was used. FE act The third-party difference increases sharply (see 147). The influence calculation unit 125 uses the measured temperature T on the finishing mill entry side. FE act As the third-party difference increases, in order to stabilize the instantaneous learning value, the predicted temperature T on the finishing mill inlet side is increased. FE pre The ratio is used to determine the temperature determination value T on the finishing mill inlet side. FE use This reduces the influence β in a way that (see 148). As a result, after the seventeenth root, the change in the instantaneous value of the learning curve becomes smaller again (see 149).

[0074] As explained above, according to this embodiment, by calculating the influence β on the temperature determination value of the finishing mill inlet side using three variance values, the temperature determination value of the finishing mill inlet side can be determined to be an appropriate value. This improves learning stability and enhances the prediction accuracy of the finishing mill exit side temperature and rolling load.

[0075] However, sometimes the learning value is made to have at least a stratification based on the steel grade of the rolled material 20. In this case, it is preferable to determine the finishing mill inlet temperature determination value T by also making the influence degree stratified, thereby taking into account the different influence degrees of each stratum. FE use .

[0076] Figure 5 This diagram illustrates the data stored in the learning value storage device, the learning instantaneous value storage device, and the influence degree storage device. In the learning value storage device 123, a learning value Z is stored according to each steel grade number (sg). old (s(x))(use) (Ref. 151). In the learning instantaneous value storage device 124, the steel grade number (sg) and the first learning instantaneous value are stored as a pair of data according to the rolling sequence of the rolled material 20 (Ref. 152). In addition, the rolling sequence is set as a number, but it can also be an identification name or the rolling time, as long as the rolled material 20 can be identified. In addition, the second learning instantaneous value using the predicted value and the third learning instantaneous value using the measured value are also stored in the learning instantaneous value storage device 124. In the influence degree storage device 126, the influence degree β is stored for each steel grade number (sg) (Ref. 153).

[0077] After calculating the first instantaneous learning value using the above formulas (5) to (7), the learning value calculation unit 122 calculates the learning value for each layer. The learning value storage value of the same steel grade number (sg) as the rolled material 20 is extracted from the learning value storage device 123 and the learning value is updated as follows.

[0078] [Formula 17] The updated learning value of the steel grade number (sg) is re-saved in the learning value storage device 123. In addition, each instantaneous learning value is saved together with the steel grade number (sg) in the instantaneous learning value storage device 124.

[0079] Next, the variance of the learned instantaneous values ​​is calculated. In calculating the variance, the learned instantaneous values ​​of the same steel grade number (sg) stored in the learned instantaneous value storage device 124 are retrieved by tracing back to the past. After retrieval, the average value of the learned instantaneous values ​​of the steel grade number (sg) is calculated, and the variance of the learned instantaneous values ​​of the steel grade number (sg) is calculated based on this average value. The number of tracing back to the past, ndd, is predefined as an adjustment factor.

[0080] [Formula 18] Equations (11) and (12) above can be used in the calculation of the second learning instantaneous value using the predicted value, and equations (15) and (16) above can be used in the calculation of the third learning instantaneous value using the measured value. The calculated learning instantaneous values ​​are stored together with the steel grade number (sg) in the learning instantaneous value storage device 124. Then, similarly to the average and variance values ​​of the first learning instantaneous values, the average and variance values ​​of the second learning instantaneous values ​​using the predicted value of the steel grade number (sg) are calculated using equations (27) and (28), and the average and variance values ​​of the third learning instantaneous values ​​using the measured value of the steel grade number (sg) are calculated using equations (29) and (30).

[0081] [Formula 19] [Formula 20] Furthermore, similar to equations (19) to (21) above, the influence degree β is calculated by equation (31) using the first variance of the first instantaneous value of the first learning using the steel grade number (sg), the second variance of the predicted value using the steel grade number (sg), and the third variance of the measured value using the steel grade number (sg).

[0082] [Formula 21] Based on this structure, by calculating the influence degree β according to the steel grade, the accuracy of the influence degree β is improved, and the temperature of the finishing rolling side can be determined to a more appropriate value.

[0083] Figure 6 This diagram illustrates an example of the hardware structure of the aforementioned setting computing device 1. The functions of the setting computing device 1, namely the setting value determination device 11 and the learning computing device 12, can be achieved through… Figure 6 The processing circuit 10 shown is used to implement this. This processing circuit 10 can also be dedicated hardware 10a. This processing circuit 10 can also include a processor 10b and a memory 10c. A portion of this processing circuit 10 is formed as dedicated hardware 10a, and may also include a processor 10b and a memory 10c. Figure 6 In the example shown, a portion of the processing circuit 10 is formed as dedicated hardware 10a, and the processing circuit 10 also includes a processor 10b and a memory 10c. The processing circuit 10 can also be at least one dedicated hardware 10a. In this case, the processing circuit 10 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a circuit combining these. The processing circuit 10 may also include at least one processor 10b and at least one memory 10c. In this case, the functions of the setpoint determination device 11 and the learning computing device 12 are implemented through software, firmware, or a combination of software and firmware. The software and firmware are described as programs and stored in the memory 10c. The processor 10b implements the functions of the setpoint determination device 11 and the learning computing device 12 by reading and executing the programs stored in the memory 10c. The processor 10b is also referred to as a CPU (Central Processing Unit), processing device, arithmetic unit, microprocessor, microcomputer, or DSP. The memory 10c is, for example, a storage device equivalent to a non-volatile or volatile semiconductor memory such as RAM, ROM, flash memory, EPROM, or EEPROM. The memory 10c can also function as each of the storage devices 123, 124, and 126. Thus, the processing circuit 10 can implement the functions of the setting value determination device 11 and the learning calculation device 12 through hardware, software, firmware, or a combination thereof. That is, the setting value determination device 11 can be implemented as a setting value determination circuit, and the learning calculation device 12 can be implemented as a learning calculation circuit.

[0084] The embodiments of this disclosure have been described above, but this disclosure is not limited to the above-described embodiments and can be implemented in various modifications without departing from the spirit of this disclosure. In the above embodiments, when numerical values ​​such as the number, quantity, amount, and range of each element are mentioned, this invention is not limited to those mentioned values, except where specifically stated or clearly determined in principle. Furthermore, the structures described in the above embodiments are not necessarily essential in this disclosure, except where specifically stated or clearly determined in principle.

[0085] Explanation of reference numerals in the attached figures 1…Setting calculation device, 10…Processing circuit, 11…Setting value determination device, 111…Input value prediction unit, 112…Input value determination unit, 113…Setting value calculation unit, 12…Learning calculation device, 121…Performance recalculation unit, 122…Learning value calculation unit, 123…Learning value storage device, 124…Learning instantaneous value storage device, 125…Influence calculation unit, 126…Influence storage device, 2…Hot rolling production line, 20…Rolled material, 21…Heating furnace, 22…Roughing mill, 23…Finishing mill, 31…Roughing mill exit side thermometer, 32…Finishing mill inlet side thermometer, 33…Finishing mill exit side thermometer.

Claims

1. A setting calculation device of a hot rolling line that calculates a control set value of a finishing mill of the hot rolling line for rolling a rolling material, comprising: a set value decision circuit that decides the control set value before rolling of the rolling material; and a learning calculation circuit that calculates a learning value using an actual value of the hot rolling line after rolling of the rolling material, wherein the set value decision circuit is configured to: acquire an entry side temperature measurement value measured at an entry side of the finishing mill; calculate an entry side temperature prediction value of the finishing mill; calculate an entry side temperature decision value of the finishing mill using the entry side temperature prediction value and the entry side temperature prediction value; calculate an exit side temperature prediction value of the finishing mill using the entry side temperature decision value; and decide the control set value in such a manner that the exit side temperature prediction value coincides with an exit side temperature target value, and the learning calculation circuit is configured to: calculate an exit side temperature actual recalculation value of the finishing mill based on the actual value; calculate a first learning instantaneous value in such a manner that the exit side temperature actual recalculation value coincides with an exit side temperature measurement value using the entry side temperature decision value, and calculate a first variance value that represents a variance of the first learning instantaneous value; calculate a second learning instantaneous value in such a manner that the exit side temperature actual recalculation value coincides with the exit side temperature measurement value using the entry side temperature prediction value, and calculate a second variance value that represents a variance of the second learning instantaneous value; calculate a third learning instantaneous value in such a manner that the exit side temperature actual recalculation value coincides with the exit side temperature measurement value using the entry side temperature measurement value, and calculate a third variance value that represents a variance of the third learning instantaneous value; and calculate a degree of influence of the entry side temperature prediction value or the entry side temperature measurement value with respect to the entry side temperature decision value using the first variance value, the second variance value, and the third variance value, wherein the entry side temperature decision value is calculated using the degree of influence in addition to the entry side temperature prediction value and the entry side temperature prediction value.

2. The setting calculation device of the hot rolling line according to claim 1, wherein the learning value and the degree of influence have at least a stratification based on a steel grade of the rolling material, calculating the learning value includes calculating the learning value for each of the stratifications, calculating the degree of influence includes calculating the degree of influence for each of the stratifications, and calculating the entry side temperature decision value includes calculating the entry side temperature decision value using the degree of influence calculated for each of the stratifications. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​

Images