JP2657444B2 - Learning method for changing the running gauge of a cold rolling mill - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a method for changing a running gauge of a cold rolling mill in which sheet materials having different specifications are joined and continuously rolled, and actual rolling data of a preceding rolling material is fed back to a setup calculation of a next rolling material. The present invention relates to a setup learning method and a setup method in the case of performing.

[0002]

2. Description of the Related Art Conventionally known adaptive correction methods in cold rolling mills are described, for example, in Chapter 11.4 of "Theory and Practice of Sheet Rolling (published by The Iron and Steel Institute of Japan, September 1984)". As has been performed, at a certain timing during rolling, the measured values of the thickness, speed, tension, etc. of each stand are input, and a rolling load formula, a rolling position formula, and a rolling torque formula are calculated using the measured values, The respective calculated values (rolling load Pc, rolling position Sc, rolling torque GRC) are obtained. On the other hand, rolling load (PA), rolling position (SA), rolling torque G of each stand
RA) is input, and the ratio or difference between PC and PA, SC and SA, GRC and GRA is calculated as the following adaptive correction coefficient.

ZP: Load adaptive correction coefficient ZP = PA / PC (1) ΔS: Rolling position adaptive correction coefficient (zero point error) ΔS = SA−SC (2) ZM: Motor load adaptive correction coefficient ZM = GRA / GRC (3) This adaptive correction coefficient is multiplied or added to the setup calculation result of the high-speed rolling section and low-speed rolling section (rolled material tip) of the next rolled material, and the high-speed rolling section and low-speed rolling section (rolled material tip) are set. The value is estimated and calculated. FIG. 1 shows a conventional setup method.

[0004]

The purpose of calculating the set value for a cold rolling mill is roughly to improve the rolling efficiency and shorten the length of the off gauge. Improvements in rolling efficiency are achieved by setting a maximum rolling speed that uses 100% of the capacity of the motor, but for this purpose the motor load must be accurately predicted. Therefore, although the motor load calculation adaptive correction coefficient is important, this motor load prediction calculation currently has sufficient accuracy. On the other hand, the reduction of the length of the off-gauge has improved the control ability of the automatic thickness control at present, and the occurrence of off-gauge at the center (high-speed rolling section) of the product is almost eliminated. However, a satisfactory value has not yet been achieved for the off-gauge length at the leading end (low-speed rolling section) of the material. This is because, in the continuous rolling, even if different plates such as base material plate thickness, steel type, and plate width are rolled, the rolling is performed without stopping the rolling mill,
This is because the plates are welded to form a series of continuous long plates, and the long plates are continuously rolled as one plate. For example, even if the product thickness (target thickness) is the same, the required rolling load changes if the base material thickness differs. Therefore, it is necessary to change the setting of each stand near the point where the base materials are welded (welding point) while the rolling mill is running (during rolling) (change of running gauge).

[0005] The setting change is performed by the roll-down control device moving the roll position from the current roll-down position to the roll-down position of the leading end of the next material given by the setup calculation. During this time, it is inevitable that the exit side plate thickness becomes an off-gauge, so the rolling speed is reduced at a low speed before the welding point enters the rolling mill, and the roll rotation is started from the point when the welding point leaves the mill. Rolling is performed so as to accelerate and reach the set maximum rolling speed. The problem is that when the rolling position accuracy given by the setup calculation is poor, an off-gauge portion longer than the inevitable off-gauge length caused by changing the running gauge is generated. There is no other way to shorten the off-gauge length except by increasing the prediction accuracy of the rolling position (load) immediately after the welding point (the tip of the material), and it is necessary to obtain the correct values of the adaptive correction factors for the load and rolling position. It becomes important.

In the conventional learning control method, when the rolling reaches the maximum rolling speed, the actual measured values at the same time at each position from the entrance to the exit of the rolling mill are input at once, and the high-speed part actual measurement is performed. The rolling load, motor load, roll peripheral speed, and rolling position are calculated using the values. An error (adaptive correction coefficient) between the calculated value and the actually measured value is calculated, the result is stored in a correction coefficient feedback table, and a setup calculation process for the next rolled material is started. In the setup calculation process,
Using the adaptive correction coefficient stored in the correction coefficient feedback table, the rolling load at the front end of the next material, the rolling position, and the maximum rolling speed at the high-speed part are predicted. However, the rolling state when the actual measurement value for calculating the adaptive correction coefficient is input is high-speed rolling, and the rolling state to be predicted and calculated using the calculated adaptive correction coefficient is low-speed rolling, and the rolling state is Are different. In a general rolling schedule, the rolling load required to obtain the same rolling reduction is as follows:
00 mpm) and the low-speed part (200 mpm), the low-speed part requires a larger value than 10%.

[0007] The rolling load formula includes a rolling speed as an input parameter. If there is an error between the rolling load formula and the actual rolling phenomenon, an adaptive value calculated from the measured value of the high-speed rolling section and the rolling load formula is used. The correction coefficient should be different from the adaptive correction coefficient calculated by the measured value of the low-speed rolling section and the rolling load equation, and the conventional method is not optimal. Furthermore, since the conventional method of collecting measured values is a method of inputting measured values from the entrance to the exit of the rolling mill at one time, the collected measured values are measured at different points of the material to be rolled at the same time. It can be said that the error has already been included in the input data of the learning calculation. The rolling position adaptive correction coefficient (zero error) is
This value is caused by the change in roll diameter due to the heat generated by the friction between the material and the roll during rolling, and changes during rolling.In the high-speed rolling section and the low-speed rolling section, the high-speed rolling section is larger by several tens of microns. Become. Therefore, the method using the zero point error obtained in the high speed rolling section for calculating the rolling position of the leading end of the next material ignores the amount of change in the zero point error that changes from the high speed rolling section to the leading end rolling of the next material. This is one factor that does not increase the accuracy of the rolling position set value.

An object of the present invention is to reduce the length of an off-gauge in the vicinity of a welded joint of a material which is joined by welding and continuously rolled.

[0009]

[Means for Solving the Problems]

(1) The setup learning calculation, which was conventionally performed at the same timing, is performed twice for the high-speed part and the low-speed part.
The learning of the rolling torque and the advance rate is performed, and the learning of the rolling load is performed in the low-speed part.

[0010] PC = fP (HA, hA, TBA, TFA, VRA, R, b) ... (4) PC: Calculated rolling load value HA: Actual measured value of incoming sheet thickness hA: Actual measured value of outgoing sheet thickness TBA: Back tension TFA: Forward tension VRA: Roll speed R: Roll diameter b: Sheet width ZP = PA / PC ... (5) ZP: Rolling load adaptive correction coefficient PA: Actual measured value of rolling load First, material tip (low speed) Part), input the measured value, and (4)
The rolling load calculation value PC is determined using the equation. The ratio ZP between the calculated rolling load value PC and the measured rolling load value PA is determined using equation (5), and this is stored in a table as a feedback value to the calculation of the rolling load at the leading end of the next material. Next, after the rolling shifts to high-speed rolling, the actual measured values are input, and the rolling torque calculation value GRC is obtained by using the rolling torque calculation formula in the same manner as the calculation of the rolling load adaptive correction coefficient, and the ratio ZM between the calculated value and the actually measured value is calculated. The calculated value is stored in the table as a feedback value, and the next material setup calculation is started. In the setup calculation, the rolling load, roll speed, and roll speed of the high-speed portion are calculated using the feedback values ZP and ZM and the respective calculation formulas.

(2) Rolling is not stable in the low-speed section (the leading end of the material to be rolled) immediately after the change of the running gauge. When the actual measured values of the thickness of the entrance side of each stand, the rolling load, the rolling position, the roll speed, and the thickness of the exit side of each stand are collected at the same timing, when the entrance side thickness (HA) actually reaches just below the roll, , Rolling load, rolling position, roll speed, etc. do not match the rolling load (PA), rolling position (SA), roll speed (VRA) collected at the same time when the entry side plate thickness (HA) is collected. Therefore, in order to guarantee the consistency of this data, data tracking from the entrance thickness gauge to the exit thickness gauge is performed for each stand, and the entrance thickness (H
Each measured value when the point where A) is measured comes to each measuring point of the rolling load, the rolling position, the roll speed, and the thickness at the side of each stand is treated as one data used for learning. As a result, more accurate calculation of the adaptive correction coefficient in a state where the rolling state of the tip or the like is unstable can be performed.

[0012]

[Action]

(1) By performing data tracking, the consistency of the measured values of the entry side plate thickness, rolling load, rolling position, roll speed, and the stand exit side plate thickness at the front end of the material is guaranteed. Based on this, a rolling load adaptive correction coefficient is calculated. Since the result is used for the setup calculation of the leading end of the next material, the adaptive correction caused by the difference in the rolling speed is compared with the prediction calculation that was conventionally performed using the adaptive correction coefficient calculated from the actual measurement value of the high-speed rolling section. The coefficient mismatch is eliminated, and more accurate prediction of the rolling load and the rolling position can be performed. Also, in the setup learning calculation in the high-speed rolling section, it is possible to eliminate an error based on the difference in the measurement position from the input value.

(2) By changing the setup method from the setup at the rolling position to the setup at the rolling load, it is possible to ignore the amount of change of the zero point error which mainly changes due to the thermal expansion of the roll during rolling.

[0014]

An embodiment of the present invention will now be described with reference to FIGS.
This will be described in detail with reference to FIG.

FIG. 2 shows a configuration of one rolling stand 1 in a tandem cold rolling mill including a plurality of rolling stands to which the present invention is applied. The illustrated stand 1 includes a pair of work rolls 1A for rolling the material 5 to be rolled, a pair of intermediate rolls 1B, and a pair of backup rolls 1C, and is mounted on the stand 1 to measure a rolling load. A work roll based on the output of the vessel 2, a stand thickness gauge 3 on the entrance side of the stand 1, a thickness gauge 4 on the exit side of the stand 1 disposed on the exit side of the stand 1, and the output of the rolling load measuring device 2. And a rolling-down control device 12 for controlling the rolling-down position of the vehicle. The material 5 to be rolled is conveyed to the entrance side of the present plant (tandem cold rolling mill) in the form of a coil wound on the exit side of the hot finishing mill, and the tip of the coil is rolled before the coil. Is welded to the end of the machine.

The data tracking process 6 starts collecting the actual rolling data of the coil at the timing when this welding point has passed through the thickness gauge on the stand entrance side, and continues the material processing until the next welding point passes through the thickness gauge on the stand exit side. Tens in the longitudinal direction of
Collect rolling performance data for each cm pitch. The concept of data tracking processing will be described with reference to the tracking table shown in FIG. First, when the welding point on the tip side of the coil reaches the entry-side thickness gauge, the actual value of the entry-side thickness is collected from the thickness gauge, and Hi of the first case of the tracking table (the left end case of the table in the figure) is used. Store in the column. Next, the material is several tens of cm (this distance is 0.2 m in advance in this embodiment).
After the table is moved, the contents of the tracking table are shifted to the right by one case, and the next entry side sheet thickness result is collected and stored in the Hi column of the first case after the table is moved. The above processing is performed every time the material moves one pitch,
When the position of the material to be rolled at which the actual entry side plate thickness is measured reaches just below the stand, the backward tension (tb), the rolling load (P),
Rolling position (S), roll speed (VR), forward tension (t
f) The material speed (vo) is taken from each measuring device, and tbi, Pi, Si, VRi, tfi, tfi,
Stored in each column of voi. Data tracking is continued, and when the material to be rolled that has finished collecting data under the entrance thickness gauge and the stand reaches the exit thickness gauge, the actual thickness value is taken in from the exit thickness gauge, and All the actual data of a certain point on the material are stored in the column.

By repeating this process every several tens of centimeters in the longitudinal direction of the material, actual rolling data is collected over the entire length of the rolled material in the longitudinal direction, and it is possible to execute learning calculation at any point. . Tip learning process (7)
The number of cases in the data tracking table required to perform the above is the value obtained by dividing the distance from the thickness gauge on the entrance side of the i-stand to the thickness gauge on the exit side of the i-stand by the data collection pitch (length) + the number of cases. is necessary. The hardness of the material is increased in the vicinity of the welding point by welding, and even if a correction calculation is performed using the data of this portion, it is not preferable to use the result for the tip portion setup calculation. Therefore, the timing of executing the tip learning processing is set to the time when the welding point passes through the next stand, and the actual data to be used is the actual data of the case corresponding to the position of the exit thickness gauge at this time. The exit side thickness gauge position and the next stand roll position are separated from each other by a distance enough to avoid the effect of increasing the hardness of the material by welding, and the speed of the rolling roll is accelerated. Since the welding point is after being wound on the tension reel, the rolling speed is low at the stage when the welding point is at the next stand position, and there is no hindrance to using the obtained data for the tip setup calculation. Absent.

In the high-speed part learning calculation, the rolling speed is sampled and monitored every several hundred milliseconds as in the conventional method, and when the detected rolling speed is equal to or more than a certain value and does not change from the previous sampling value (2). (When the same value is detected consecutively), and at this time, the actual data (data of the data tracking table) of the rolled material portion at the exit side thickness gauge position is used.

FIG. 4 shows a flowchart of the setup learning calculation. First, the actual data is read from the case corresponding to the exit thickness gauge position of the data tracking table, and at the tip learning calculation timing, the rolling load and the roll speed are calculated using the respective model formulas used in the setup calculation. The rolling load value and the roll speed value obtained by these model formulas are compared with these actual values stored in the tracking table, and a correction coefficient is calculated. At the high-speed part learning calculation timing, the actual load data read from the case corresponding to the outlet thickness gauge position of the data tracking table as in the case of the tip part is used, and the rolling load and the roll speed are calculated by the following equations (6) and (7). Calculate and calculate each correction factor. Further, a rolling torque calculation value GRC is calculated using the read data, and an actual rolling torque measurement value GRA is calculated from the current and voltage measurement values.
The equations (1) and (3) and the following equation (8) are used for calculating the correction coefficient.

Rolling load PC = b · k · κ · dp · R ′ (H−h) (6) P: rolling load b: plate width k: deformation resistance κ: tension correction term dp: friction coefficient correction Item R ': Flat work roll radius H: Inlet plate thickness h: Outlet plate thickness The roll speed VRi' is calculated by the following equation (7).

[0021]

(Equation 1)

I: stand number, 1 stand to final stand F: final stand ri: reduction rate (calculated setup value) hi: exit side plate thickness (calculated setup value) VRi: roll speed (calculated setup value, corrected by ZVR) HAi: Actual value of exit side plate thickness rAi: Actual value of rolling reduction (= (HAi-hi) / HAi) fi: Advanced rate (calculated setup value) Roll speed adaptive correction coefficient ZVRi is calculated by the following equation (8). calculate.

[0023]

(Equation 2)

VRAi: i actual stand roll speed value VRAF: final stand roll speed actual value Each correction coefficient is stored in the correction coefficient feedback table shown in FIG. Since the correction coefficients have different values depending on the steel type, the sheet thickness, and the sheet width, the correction coefficient feedback table has a classification by layer of the steel type, the sheet width, and the product sheet thickness, and stores the average value of each correction coefficient. The learning calculation process determines a corresponding case in the correction coefficient feedback table from the product information of the corresponding coil, and calculates and stores the average value of the correction coefficient storage value of the tip (or high-speed part) and the current calculation value.
The correction coefficient feedback table includes an inter-coil learning table in addition to the long-term learning table described above.
In this table, the correction coefficients calculated this time are stored as they are. After the high-speed part learning calculation is completed, the setup calculation is started and the set values of the high-speed part and the tip part, such as the rolling load and the roll speed, are calculated.If the product specifications of the next coil and the previous coil are different, the next coil The corresponding case of the long-term learning table is obtained from the product information of the above, and the correction coefficients of the tip part and the high-speed part stored therein are used.

The following formulas (9) and (10) are used to calculate the set values of the rolling load, the roll speed, and the motor load.
(11) is used respectively.

Rolling load equation P = b · k · κ · dp · R ′ (H−h) Zp (9) Zp: Load adaptation correction coefficient Roll speed

[0027]

(Equation 3)

VRi: Roll speed VOi: i-stand exit side plate speed fi: i-stand advanced ratio ZVRi: Roll speed adaptive correction coefficient Motor load HP = 0.0002192 × VR × GM / R (11) HP: Motor load VR: Roll speed R: Work roll radius GM: Motor torque = GR · ZM + GT + GL GR: Rolling torque GT: Tension torque GL: Loss torque ZM: Motor load adaptation correction coefficient When the previous coil and product specifications are equal, the coil-to-coil learning table is used. The stored contents are used. The setup calculation result is transmitted to the reduction control device of FIG. 2, and the reduction control device obtains the rolling load given by the setup calculation at the timing when the coil change point (weld point) enters the corresponding stand. Operate the rolling position of the rolling mill.

According to the above embodiment, the load prediction accuracy is improved as compared with the conventional method in which the rolling load at the leading end of the material is predicted by using the adaptive correction coefficient calculated based on the actually measured values in the different rolling states. Can be done. Also, by setting the setup value to the rolling load instead of the rolling position, the influence of the prediction error of the zero point error can be eliminated. As described above, the off-gauge length at the material tip (near the welding point) is reduced,
This has the effect of improving the yield. Comparing the rolling conditions of the high-speed portion and the low-speed portion, the load adaptation correction coefficient often differs by about ± 10% due to changes in the roll speed, tension, and frictional force between the roll and the material to be rolled. Therefore, in the conventional method of predicting the rolling load at the tip using the load adaptive correction coefficient of the high-speed portion, a load error of ± 10% occurs.

FIG. 6 shows the effect of the rolling load deviation on the exit side plate thickness. It is assumed that the load required to roll the input side plate thickness of 0.700 mm to the output side plate thickness of 0.550 mm at a certain stand is 600 tons. On the other hand, when the load prediction calculation is performed, if there is an error of -5%, the calculated load becomes 570 tons. When this is converted into the change amount Δh of the exit side plate thickness, Δh = ΔP / M Δh: deviation of the exit side plate thickness (mm) M: plasticity coefficient (ton / mm) ΔP: load deviation (ton) Is 800 ton / mm,
Δh = (600−570) /800=0.0375 mm
Becomes

As described above, the delivery side sheet thickness target value is 0.55 m.
An error of about 7% with respect to m results in an off-gauge portion. Further, when comparing the zero point error between the high-speed rolling and the leading end rolling section, considering that the error changes by about 0.050 mm, the error further increases due to the prediction error of the zero point.

[0032]

According to the present invention, since the rolling load at the leading end of the material is predicted using the adaptive correction coefficient calculated based on the data in the low-speed rolling state, the load prediction accuracy is improved.
Since the adaptive correction coefficient is calculated based on the data collected for the same point of the material to be rolled, the accuracy of the adaptive correction coefficient is improved, and by setting the setup value to the rolling load instead of the rolling position, The influence of the prediction error of the zero point error can be eliminated. By improving the load prediction accuracy and eliminating the zero point prediction error, there is an effect of reducing the length of the off-gauge.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram for explaining a conventional setup calculation method.

FIG. 2 is a block diagram illustrating a main configuration of an embodiment of the present invention.

FIG. 3 is a conceptual diagram for explaining data tracking.

FIG. 4 is a procedure diagram illustrating an example of a calculation flow of a setup learning calculation.

FIG. 5 is a diagram showing a configuration example of a correction coefficient feedback table applied to the present invention.

FIG. 6 is a conceptual diagram showing the effect of rolling load fluctuation on sheet thickness.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 i stand 2 Load cell 3 Entry side thickness gauge 4 Exit side thickness gauge 5 Rolled material 6 Data tracking processing 7 Tip learning processing 8 High speed part learning processing 9 Feedback table 10 Setup processing 11 Reduction control device ZP Load adaptation correction coefficient ZM Motor load adaptive correction coefficient ΔS Zero point error

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Takashige 5-2-1, Omika-cho, Hitachi City, Ibaraki Prefecture Within Hitachi Information & Control Systems Co., Ltd. (72) Inventor Masaaki Nakajima 5-chome, Omika-cho, Hitachi City, Ibaraki Prefecture No. 1 Hitachi, Ltd. Omika Factory (56) References JP-A-59-174207 (JP, A)

Claims (9)

(57) [Claims] 1. A learning item in a running gauge change setup learning calculation of a cold rolling mill for joining and continuously rolling plate materials having different specifications and feeding back actual rolling data of a preceding rolling material to a setup calculation of a next rolling material. A method for changing the running gauge of a cold rolling mill, which separately selects the optimal data collection timing and performs learning calculation. 2. A learning calculation for a change set-up of a running gauge of a cold rolling mill in which sheet materials having different specifications are joined and continuously rolled, and actual rolling data of a preceding rolling material is fed back to a setup calculation of a next rolling material. A cold rolling mill characterized in that a learning calculation for a setup targeting a high speed rolling state of a rolling mill and a learning calculation for a setup targeting a low speed rolling state of a rolling mill are separately performed. To learn how to change the gauge between runs. 3. The low-speed rolling state is a rolling state from a point at which the leading end of the material to be rolled to the preceding rolled material passes through the stand to the start of acceleration of the rolling roll. 2. The learning method for changing a running gauge of a cold rolling mill according to item 2. 4. The learning calculation in a low-speed rolling state is performed at least for a rolling load and a roll speed, and the learning calculation in a high-speed rolling state is performed for at least a rolling load, a roll speed, and a rolling torque. The learning method for changing a running gauge of a cold rolling mill according to claim 2 or 3. 5. A learning calculation for a change set-up of a running gauge of a cold rolling mill, in which sheet materials having different specifications are joined and continuously rolled, and actual rolling data of a preceding rolling material is fed back to a setup calculation of a next rolling material. The actual rolling data used for learning is that each time a certain portion of the material to be rolled passes through the data sampling point of the rolling mill, the data is collected for the relevant portion, and that the data is collected for the same portion of the material to be rolled. A learning method for changing a running gauge of a cold rolling mill, which is a common feature. 6. The learning calculation for the setup calculation for the low-speed rolling state is performed based on the actual rolling data collected in the low-speed rolling state, and the learning calculation for the setup calculation for the high-speed rolling state is performed in the high-speed rolling state. The running gauge change setup learning method for a cold rolling mill according to any one of claims 2 to 4, wherein the learning is performed based on the actual rolling data collected in (1). 7. The actual rolling data is obtained every time a certain portion of the material to be rolled passes through the data sampling point of the rolling mill, and is obtained for the same portion of the material to be rolled. 7. The learning method for changing the running gauge of a cold rolling mill according to claim 6, wherein the learning method is a common feature. 8. The learning calculation for the setup calculation includes:
The correction coefficient is calculated from at least the calculated rolling load and the actual value, the calculated roll speed and the actual value, and the calculated rolling torque and the actual value, respectively. How to learn the setup for changing the running gauge of a cold rolling mill. 9. A setup method for changing a running gauge of a cold rolling mill, in which sheet materials of different specifications are joined and continuously rolled, and actual rolling data of a preceding rolled material is fed back to a setup calculation of a next rolled material. A setup method for changing a running gauge of a cold rolling mill for setting up a rolling load value for a draft control device.