An online identification and early warning method for abnormal stability state of PC rolling mill

Abstract

The present invention discloses an online identification and early warning method for abnormal stability state of a PC rolling mill. First, the PC angle, rolling force data of the transmission side and the working side of the strip steel within a period of time are obtained from a basic automation computer. After invalid data is eliminated, linear decomposition is performed with the PC angle as the independent variable and the rolling force deviation as the dependent variable. The dynamic change index of the stability state obtained after the decomposition is subjected to high-pass and low-pass filtering, and then abnormal discrimination is performed. Finally, the early warning type is judged according to the early warning threshold, and an early warning signal is output. The early warning signal is used for the alarm of abnormal stability state of the rolling mill, prompting the technician to handle it in time. The present invention can conveniently and timely realize the identification of abnormal defects of the stability of the PC rolling mill, and realize long-term online monitoring and early warning.

Description

Technical Field

[0001] The present invention relates to the field of hot rolling technology, and more specifically to an online identification and early warning method for abnormal finishing rolling stability state of a PC rolling mill. Background Art

[0002] During the hot-rolled strip production process, the stability of the finishing mill is a key production concern. When the stability problem of the rolling mill is relatively minor, it will affect production efficiency and increase the incidence of product defects; when the stability problem is serious, it will affect production stability and cause scrap steel accidents.

[0003] Since PC rolling mills have a very significant effect on controlling the convexity of strip steel, they are widely used in hot rolling and cold rolling. The characteristic of PC rolling mills is that the upper and lower rollers are in a crossed state, and the maximum crossing angle reaches 1.5 tesla. Due to the crossing between the rollers, a large axial force will be generated during the rolling process, which places high demands on the accuracy of the rolling mill equipment system. When the accuracy of the crossing equipment deviates, it is easy to cause unstable problems such as strip deviation, which will have a very serious impact on the stability of the finishing mill.

[0004] When stability problems occur, the mill equipment causing the problem needs to be dealt with in a timely manner to avoid further deterioration of quality. However, since the hot rolling finishing mill generally has 7 racks and the racks equipped with PC equipment generally have 3 to 7 racks, it is necessary to accurately locate the specific rack with the problem in order to solve the problem in a timely manner.

[0005] At present, the stability state of the PC rolling mill is mainly judged by measuring the deviation of the stiffness on both sides of the rolling mill when it is stopped. When the deviation exceeds a certain range, it is considered that there is a problem with the stability state of the frame. There are two problems with this method: one is that the machine needs to be stopped for measurement, and the other is that the PC angle is 0 during measurement, that is, the rollers do not cross, which is inconsistent with the constantly changing state of the PC angle during actual production, and cannot fully reflect the impact of the dynamic changes of the PC rolling mill on the stability state. Therefore, it is easy to make wrong judgments, and the judgment conclusions are lagging and prone to large losses. Therefore, how to conveniently and timely realize the monitoring of stability anomalies and accurately warn is a key issue that needs to be solved.

[0006] In the existing patent applications, such as patent (application) No. 201810322829.1, a finishing rolling stability control method is disclosed, which belongs to the field of rolling process technology. The finishing rolling stability control method includes the following steps: controlling the load distribution of the finishing mill; controlling the tension distribution of the finishing mill; controlling the convexity ratio of the plate and strip; the control of the convexity ratio of the plate and strip includes: according to the plate type straightness control theory between frames, the formula Ci / Hi=Ci+1 / Hi+1 is adopted, wherein Ci is the convexity of the upstream frame outlet, Hi is the upstream frame outlet thickness, i is the i-th frame number, and the convexity distribution between frames should be set to convexity 1; when the frame outlet thickness is greater than 7mm, the formula Ci / Hi=(Ci+1 / Hi+1)*is adopted, and the δ is the lateral uneven extension coefficient. The finishing rolling stability control method of the present invention ensures the stability during the finishing rolling process, solves the problems of rolling mill vibration and surface quality, centerline control and plate type. This patent belongs to the use of model control methods and is not related to the technical field and implementation technology methods involved in the present invention.

[0007] For example, patent (application) No. 201610044456.7 discloses an online monitoring method for the cross-roll system of a PC mill, which relates to a control device or method specifically suitable for a metal mill or its processed products, and in particular to a method for monitoring the cross-roll system of a PC mill using zero adjustment deviation data of a hot rolling production roll change process, comprising the following steps: collecting the static pressure oil column deviation and the transfer oil column deviation of each frame; calculating and judging whether the cross-accuracy of the frame exceeds the limit by judging the oil column deviation of the frame, so as to evaluate the cross-accuracy of the cross-roll system of the mill. The method of the present invention can realize online monitoring of the cross-accuracy of the mill by analyzing the data of the usual equipment action process without stopping the system; by analyzing the trend of the oil column deviation of the frame after multiple zero adjustments, it can timely find the frame with poor cross-accuracy and the need to stop the machine for measurement and calibration. It can improve the stability of the PC mill rolling, ensure the stability of the entire rolling, and ensure that the quality of the final product is stable and controlled. This patent belongs to the measurement of the cross-accuracy of the offline PC mill, which is not related to the stability abnormality recognition technology implemented by the present invention. Summary of the invention

[0008] In view of the above-mentioned defects existing in the prior art, the purpose of the present invention is to provide an online identification and early warning method for abnormal stability state of PC rolling mill finishing, which can conveniently and timely realize the identification of abnormal defects of PC rolling mill finishing stability, and realize long-term online monitoring and early warning.

[0009] To achieve the above object, the present invention adopts the following technical solution:

[0010] An online identification and early warning method for abnormal stability state of PC rolling mill finishing:

[0011] Firstly, the PC angle, rolling force data of the transmission side and the working side of the strip within a period of time are obtained from the basic automation computer. After the invalid data is eliminated, linear decomposition is performed with the PC angle as the independent variable and the rolling force deviation as the dependent variable. The dynamic change index of the stability state obtained after the decomposition is subjected to high-pass and low-pass filtering, and then abnormality discrimination is performed. Finally, the warning type is judged according to the warning threshold, and a warning signal is output. The warning signal is used to alarm the abnormal stability state of the rolling mill, prompting the technical personnel to deal with it in time.

[0012] Preferably, the PC angle, rolling force data of the transmission side and the working side of the strip steel are obtained from the basic automation computer over a period of time, as follows:

[0013] Obtaining a data set of a length of the head of each stand of each strip steel from the basic automation computer, and calculating the rolling force deviation percentage DiffP_j between the transmission side and the working side according to formula (1);

[0014] DiffP_j=(Pdj-Pwj) / Pwj*100(1)

[0015] In formula (1), Pdj represents the rolling force data of the transmission side, in KN; Pwj represents the rolling force data of the working side, in KN; and j represents the rack number.

[0016] Preferably, the basic automation computer acquires a data set of 10 to 20 meters of the head of each rack of each strip steel.

[0017] Preferably, the invalid data removal is specifically as follows:

[0018] The data with PC angle less than 0 and greater than 1.2, and the data with rolling force less than 4000 KN and greater than 40000 KN are eliminated.

[0019] Preferably, the strain variable is linearly decomposed as follows:

[0020] The actual test data obtained are grouped according to the production time, with days as intervals. The least square method is used to perform linear fitting on the daily data group according to formula (2) to calculate the dynamic change index of the stability state Dynlndex;

[0021] DiffP_ij=Dynlndex_ij×Pcact_ij+Sta_ij (2)

[0022] In formula (2), i represents the serial number grouped at daily intervals; j represents the rack number; Dynlndex_ij represents the dynamic change index of the stability state of rack j on day i; Pcact_ij represents the PC angle data of rack j on day i; Sta_ij represents the static change index of the stability state of rack j on day i.

[0023] Preferably, the stability state dynamic change index is subjected to high-pass and low-pass filtering as follows:

[0024] Perform high-pass and low-pass filtering on Dynlndex_ij. The data after high-pass filtering is High_DiffP_ij, and the data after low-pass filtering is Low_DiffP_ij.

[0025] Preferably, the high-pass and low-pass filtering are performed by a third-order Butterworth filter, the cut-off frequency range of the low-pass filtering is 0.06-0.08, and the cut-off frequency range of the high-pass filtering is 0.09-0.11.

[0026] Preferably, the warning types include trend warning judgment and volatility warning judgment;

[0027] The trend warning judgment is as follows:

[0028] According to formula (3), the average value Ave_Low_DiffP_ij of Low_DiffP_ij in the n days before the warning evaluation day is calculated, and the deviation percentage Diff_Low_ij is calculated according to formula (4);

[0029] If the absolute value of Diff_Low_ij is greater than the trend warning threshold G_LOW, it is determined that the rolling mill stability trend is abnormal and a warning signal is issued;

[0030]

[0031] Diff_Low_ij=(Low_DiffP_ij-Ave_Low_DiffP_ij) / Ave_Low_DiffP_ij*100(4)

[0032] In formula (3), n is 5 to 10; G_LOW is 6 to 10;

[0033] The volatility early warning judgment is as follows:

[0034] According to formulas (5) and (6), calculate the mean value Ave_High_DiffP_ij and standard deviation Std_High_DiffP_ij of High_DiffP_ij in the n days before the warning evaluation day;

[0035] If the absolute value of High_DiffP_ij is greater than the volatility warning threshold G_HIGH_ij, it is determined that the rolling mill stability fluctuation is abnormal, and a warning observation signal is issued;

[0036]

[0037] In formulas (5) and (6), n is 30 to 60.

[0038] Preferably, the volatility warning threshold G_HIGH_ij is calculated by formula (7):

[0039] G_HIGH_ij=Ave_High_DiffP_ij+2×Std_High_DiffP_ij(7).

[0040] Preferably, the early warning signal is used to warn of abnormal stability of the rolling mill, prompting the operator to handle it in time as follows:

[0041] Outputting the trend warning judgment and the fluctuation warning judgment to the technical personnel, taking shutdown measures for the output of the trend warning judgment, and repairing the rack that issued the warning;

[0042] For the output of the fluctuation warning judgment, the corresponding rack is observed. If the warning is issued for N consecutive days, the rack that issues the warning is shut down for repair;

[0043] The value of N is ≥ days.

[0044] The present invention provides an online identification and early warning method for abnormal stability status of PC mill finishing rolling, which utilizes the existing detection equipment and detection data of the production plant without adding detection equipment. Under normal production conditions, by analyzing the changing trend of rolling force deviation in the PC angle change process, online monitoring of PC mill stability can be achieved, and frames with poor stability that need to be shut down for repair can be discovered in time, thereby improving the production stability of the PC mill finishing unit without increasing costs and ensuring that the quality of the final product is stable and controlled. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a structural schematic diagram of the online identification and early warning method of the present invention;

[0046] Figure 2 It is a curve diagram of the settlement result of the stability state dynamic change index in the embodiment of the online identification and early warning method of the present invention;

[0047] Figure 3 It is a curve diagram of the output result after low-pass filtering in the embodiment of the online identification and early warning method of the present invention;

[0048] Figure 4is a scatter diagram of the output result after high-pass filtering in the embodiment of the online identification and early warning method of the present invention;

[0049] Figure 5 It is a curve diagram of the deviation percentage output result in the embodiment of the online identification and early warning method of the present invention. DETAILED DESCRIPTION

[0050] In order to better understand the above technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0051] Combination Figure 1 The present invention provides an online identification and early warning method for abnormal stability state of a PC rolling mill, comprising the following steps:

[0052] 1) Obtaining actual test data: Obtain a data set of a length (10 to 20 meters) of each stand head of each strip steel from the process control computer, including the strip steel number, production time (Prod_date), PC angle (Pcact_j, unit degree), working side rolling force (Pwj, unit KN), and transmission side rolling actual data (Pdj, unit KN), and calculate the rolling force deviation percentage DiffP_j between the transmission side and the working side according to formula (1);

[0053] DiffP_j=(Pdj-Pwj) / Pwj*100(1)

[0054] In formula (1), Pdj represents the rolling force data of the transmission side, in KN; Pwj represents the rolling force data of the working side, in KN; and j represents the rack number.

[0055] 2) Elimination of invalid test data: Eliminate data with PC angle less than 0 and greater than 1.2, and rolling force less than 4000KN and greater than 40000KN.

[0056] 3) Decomposition of the dynamic change index of stability state: The actual test data obtained are grouped according to the production time, with days as the interval. If the number of strips produced per day is less than 50, this group of data will be eliminated. The least squares method is used to perform linear fitting on the daily data group according to formula (2) to calculate the dynamic change index of stability state Dynlndex;

[0057] DiffP_ij=Dynlndex_ij×Pcact_ij+Sta_ij (2)

[0058] In formula (2), i represents the serial number grouped by day; j represents the rack number; DynIndex_ij represents the dynamic change index of the stability state of the j rack on the i day; Pcact_ij represents the PC angle data of the j rack on the i day; Sta_ij represents the static change index of the stability state of the j rack on the i day. The dynamic change index of the stability state DynIndex reflects the change characteristics of the rolling force deviation with the change of the PC angle, which can characterize the change of the stability state of the PC rolling mill, and its size can measure the degree of stability.

[0059] 4) High-pass and low-pass filtering of the stability state dynamic change index: Dynlndex_ij is subjected to high-pass and low-pass filtering, and the data after high-pass filtering is High_DiffP_ij, and the data after low-pass filtering is Low_DiffP_ij. High-pass and low-pass filtering are completed by a third-order Butterworth filter, the cutoff frequency range of the low-pass filter is 0.06-0.08, and the cutoff frequency range of the high-pass filter is 0.09-0.11.

[0060] 5) Trend warning judgment: Calculate the average value Ave_Low_DiffP_ij of the n (5 to 10) days before the warning evaluation day according to formula (3), and calculate the deviation percentage Diff_Low_ij according to formula (4);

[0061] If the absolute value of Diff_Low_ij is greater than the trend warning threshold G_LOW (value ranges from 6 to 10), it is determined that the rolling mill stability trend is abnormal and a warning signal is issued;

[0062]

[0063] Diff_Low_ij=(Low_DiffP_ij-Ave_Low_DiffP_ij) / Ave_Low_DiffP_ij*100(4).

[0064] 6) Volatility early warning judgment: According to formulas (5) and (6), calculate the mean value Ave_High_DiffP_ij and standard deviation Std_High_DiffP_ij of the n (range 30 to 60) days before the early warning evaluation day;

[0065] If the absolute value of High_DiffP_ij is greater than the volatility warning threshold G_HIGH_ij, it is determined that the rolling mill stability fluctuation is abnormal, and a warning observation signal is issued;

[0066]

[0067] The volatility warning threshold G_HIGH_ij is calculated by formula (7):

[0068] G_HIGH_ij=Ave_High_DiffP_ij+2×Std_High_DiffP_ij(7).

[0069] 7) Outputting the early warning signal processing: outputting the early warning judgment results obtained in steps 5 and 6) to the technical personnel, and taking shutdown processing measures for the output of the trend early warning judgment, and repairing the rack that issued the early warning;

[0070] The output of the fluctuation warning judgment is observed on the corresponding rack. If it is issued for N consecutive days (value ≥ value), the rack that issues the warning is shut down for repair.

[0071] Example

[0072] The online identification and early warning method of the present invention is further illustrated by taking the identification and early warning embodiment of the abnormal finishing stability state of the finishing rolling F4 stand of a hot rolling production line as an example, and taking the early warning on March 15 of a certain year as an example.

[0073] 1) Obtain the actual measured data (in KN) of the rolling force and PC angle of the F4 stand from February 1 of a certain year from the basic automation computer, and calculate the rolling force deviation percentage, and obtain a total of 112,678 data. Some of the data are shown in the following table.

[0074] F4 stand rolling force measured data

[0075] STRIP_NO Pdj PcqI Pcact Prod_date DiffPj 123147152800 7273.209 7127.577 0.749 xxxx0201 2.04 123147454400 7351.342 7548.947 0.549 xxxx0201 -2.62 123147454300 6901.794 6861.661 0.549 xxxx0201 0.58 123147454500 7208.435 7419.605 0.549 xxxx0201 -2.85 123147454600 7141.322 7167.056 0.549 xxxx0201 -0.36 123147454700 7383.279 7543.960 0.549 xxxx0201 -2.13 123147454800 6996.353 7044.060 0.549 xxxx0201 -0.68 123147454900 6688.334 6726.168 0.549 xxxx0201 -0.56 123147455000 6545.596 6536.228 0.549 xxxx0201 0.14 123147455100 6817.173 6755.909 0.549 xxxx0201 0.91 123147455200 6685.328 6687.954 0.500 xxxx0201 -0.04 123147455300 6739.371 6738.809 0.549 xxxx0201 0.01 123147455400 6382.666 6462.721 0.500 xxxx0201 -1.24 123147455500 6643.800 6652.278 0.500 xxxx0201 -0.13 123147455600 6294.891 6243.692 0.549 xxxx0201 0.82 123147455700 7226.027 7304.578 0.500 xxxx0201 -1.08 123147455800 6697.727 6534.369 0.549 xxxx0201 2.50 123147455900 6591.687 6569.127 0.500 xxxx0201 0.34 123147456000 6396.691 6280.024 0.549 xxxx0201 1.86 ……

[0076] 2) The data with PC angle less than 0 degree and greater than 1.2 degree, and rolling force less than 4000 KN and greater than 40000 KN were eliminated, and 112673 data remained after elimination.

[0077] 3) The data processed in step 2) are grouped by day. If the number of strip steel produced per day is less than 50, this group of data is eliminated, and a total of 43 groups are divided. Then, linear fitting is performed on each group of data according to formula (2) to calculate the stability state dynamic change index DynIndex_4, such as Figure 2 shown.

[0078] 4) Perform high-pass and low-pass filtering on DynIndex_4 obtained in step 3). The cutoff frequency of the low-pass filter is 0.08, and the cutoff frequency range of the high-pass filter is 0.10. The output result is as follows: Figure 3 and Figure 4 shown.

[0079] 5) Calculate Ave_Low_DiffP_4 according to formula (3), and calculate the deviation percentage Diff_Low_4 according to formula (4). The calculation results are as follows: Figure 5 As shown in the figure, the absolute value of Diff_Low_4 on March 13 of that year was 7.09, which was greater than the trend warning threshold G_LOW (value 6), and it was determined that the rolling mill stability trend was abnormal, and a warning signal was issued.

[0080] 6) Since the stability trend abnormal warning was triggered, a trend warning signal was output to the technical staff. On-site inspection found that there was a problem with the mortar pad of the PC cross mechanism of the rack, and it was replaced during the scheduled maintenance downtime.

[0081] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.

Claims

1. An online identification and early warning method for abnormal stability state of PC rolling mill, characterized by: Firstly, the PC angle, rolling force data of the transmission side and the working side of the strip steel within a period of time are obtained from the basic automation computer. After the invalid data is eliminated, the PC angle is used as the independent variable and the rolling force deviation is used as the dependent variable for linear decomposition. The dynamic change index of the stability state obtained after the decomposition is processed by high-pass and low-pass filtering, and then abnormality discrimination is performed. Finally, the warning type is judged according to the warning threshold, and a warning signal is output. The warning signal is used for the alarm of abnormal stability state of the rolling mill to prompt the technical personnel to deal with it in time. The PC angle, rolling force data of the transmission side and the working side of the strip steel are obtained from the basic automation computer within a period of time, as follows: Obtaining a data set of a length of the head of each stand of each strip steel from the basic automation computer, and calculating the rolling force deviation percentage DiffP_j between the transmission side and the working side according to formula (1); DiffP_j=(Pdj-Pwj) / Pwj*100 (1) In formula (1), Pdj represents the rolling force data of the transmission side, in KN; Pwj represents the rolling force data of the working side, in KN; j represents the frame number, The specific details of invalid data removal are as follows: Eliminate the data with PC angle less than 0 degree and greater than 1.2 degree, and rolling force less than 4000 KN and greater than 40000 KN. The linear decomposition of the dependent variable is as follows: The actual test data obtained are grouped according to the production time, with days as intervals. The least square method is used to perform linear fitting on the daily data group according to formula (2) to calculate the dynamic change index of the stability state Dynlndex; DiffP_ij=Dynlndex_ij×Pcact_ij+Sta_ij (2) In formula (2), i represents the serial number grouped by day; j represents the rack number; Dynlndex_ij represents the dynamic change index of the stability state of rack j on day i; Pcact_ij represents the PC angle data of rack j on day i; Sta_ij represents the static change index of the stability state of rack j on day i, The high-pass and low-pass filtering of the stability state dynamic change index are specifically performed as follows: Perform high-pass and low-pass filtering on Dynlndex_ij. The data after high-pass filtering is High_DiffP_ij, and the data after low-pass filtering is Low_DiffP_ij. The high-pass and low-pass filtering are completed by a third-order Butterworth filter. The warning types include trend warning judgment and volatility warning judgment; The trend warning judgment is as follows: According to formula (3), the average value Ave_Low_DiffP_ij of Low_DiffP_ij in the n days before the warning evaluation day is calculated, and the deviation percentage Diff_Low_ij is calculated according to formula (4); If the absolute value of Diff_Low_ij is greater than the trend warning threshold G_LOW, it is determined that the rolling mill stability trend is abnormal and a warning signal is issued; Diff_Low_ij=(Low_DiffP_ij-Ave_Low_DiffP_ij) / Ave_Low_DiffP_ij*100(4) In formula (3), n is 5 to 10; G_LOW is 6 to 10; The volatility early warning judgment is as follows: According to formulas (5) and (6), calculate the mean value Ave_High_DiffP_ij and standard deviation Std_High_DiffP_ij of High_DiffP_ij in the n days before the warning evaluation day; If the absolute value of High_DiffP_ij is greater than the volatility warning threshold G_HIGH_ij, it is determined that the rolling mill stability fluctuation is abnormal, and a warning observation signal is issued; In formulas (5) and (6), n is 30 to 60.

2. The method for online identification and early warning of abnormal stability state of PC rolling mill according to claim 1, characterized in that: The length of the section is 10 to 20 meters.

3. The method for online identification and early warning of abnormal stability state of PC rolling mill according to claim 1, characterized in that: The cut-off frequency range of the low-pass filter is 0.06 to 0.08, and the cut-off frequency range of the high-pass filter is 0.09 to 0.

11.

4. The method for online identification and early warning of abnormal stability state of PC rolling mill according to claim 1, characterized in that: The volatility warning threshold G_HIGH_ij is calculated by formula (7): G_HIGH_ij=Ave_High_DiffP_ij+2×Std_High_DiffP_ij(7).

5. The method for online identification and early warning of abnormal stability state of PC rolling mill according to claim 1, characterized in that: The warning signal is used to warn of abnormal stability of the rolling mill, prompting the operator to handle it in time as follows: Outputting the trend warning judgment and the fluctuation warning judgment to the technical personnel, taking shutdown measures for the output of the trend warning judgment, and repairing the rack that issued the warning; For the output of the fluctuation warning judgment, the corresponding rack is observed. If the warning is issued for N consecutive days, the rack that issues the warning is shut down for repair; The value of N is ≥ days.

Images