JP2017177203A - Method for detection of abnormality in rolling load measuring apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for detection of an abnormality in a rolling load measuring apparatus by which an error of a rolling load measuring apparatus in a rolling machine can be detected with good accuracy by use of a load calculation value which is calculated from a pressure in a hydraulic rolling reduction cylinder.SOLUTION: A method for detection of an abnormality in a rolling load measuring apparatus by which in a rolling machine having a hydraulic rolling reduction cylinder which presses a roll by a hydraulic system and the rolling load measuring apparatus for measuring a load of the rolling machine, a load calculation value, which is calculated based on a value of a pressure gauge provided on the hydraulic rolling reduction cylinder, and a load measurement value, which is measured by the rolling load measuring apparatus, are compared to each other over time, and a state that a difference between the load calculation value and the load measurement value becomes larger over time is determined as an abnormal state of the rolling load measuring apparatus.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an abnormality detection method for a rolling load measuring device that accurately detects an abnormality in a rolling load measuring device provided in a steel sheet rolling mill.

  In order to strictly control the thickness of the material to be rolled when rolling the steel plate, it is important to accurately measure the rolling load applied to the rolling mill (hereinafter sometimes simply referred to as “load”). . A load cell is mentioned as a specific example of the apparatus (rolling load measuring apparatus) for detecting a rolling load. The load cell is a member in which a strain gauge is attached to an elastic deformable body having a predetermined shape and size, and the rolling load can be calculated from the amount of strain generated in the elastic deformable body in the load cell. A load cell is widely used as an apparatus capable of measuring a rolling load with very high accuracy.

  In a rolling mill, a load cell is installed between a roll chock that supports a work roll, a backup roll, and the like, a housing, and the like. When the rolling is repeatedly performed by a rolling mill, a repeated rolling load is applied to the load cell, and an error occurs in the measured value of the rolling load by the load cell over time.

  While the error of the load measurement value by the load cell is small, the rolling (spinning rolling) can be continued primarily by adjusting various rolling conditions so as to correct such an error. However, in order to adjust the rolling conditions so as to correct the error, it is necessary to accurately detect how much error has occurred in the load measurement value by the load cell.

  Further, as the error of the load measurement value by the load cell increases, it becomes impossible to correct the error by adjusting the rolling conditions, and problems such as operation troubles occur. In such a case, the rolling process cannot be continued, and the load cell needs to be replaced. However, when replacing the load cell, it is necessary to stop the rolling line and sequentially remove the parts necessary for the replacement from the rolling mill, and then perform the load cell replacement operation. As described above, in order to replace the load cell, a work period of several days is required after the line is stopped. Therefore, if the load cell suddenly breaks down and must be replaced, the operation plan will be greatly adversely affected. In order to prevent such a problem, it is necessary to detect an error in the load measurement value by the load cell in advance and to establish a guideline for replacing the load cell in advance.

  Conventionally, Patent Documents 1 and 2 below are known as techniques for evaluating and calibrating an error in a measured value of a rolling load by a load cell.

  Patent Document 1 discloses a method of performing calibration by applying a roll bending force with the roll gap open at the time of non-rolling, comparing the roll bending force with the output of the rolling load measuring device.

  Patent Document 2 discloses a method for converting a pressure acting on a hydraulic reduction cylinder into a rolling load. Patent Document 2 discloses a method for measuring the pressure of hydraulic oil in a hydraulically-reduced cylinder in a static state with no inflow / outflow and converting the pressure into a rolling load.

JP-A-6-123667 JP 2002-35831 A

  However, while the roll bending force (when not rolled) in the open state of the roll gap is several tens of tons (hundreds of kN), the rolling load during rolling reaches several thousand tons (tens of thousands of kN). Thus, there is a large difference between the roll bending force during non-rolling and the rolling load during rolling. As disclosed in Patent Document 1, even when the rolling load measuring device is calibrated by using a relatively small non-rolling roll bending force, a relatively large rolling load is applied at the time of rolling in which a relatively large rolling load is applied. The error will be amplified, and it is difficult to perform calibration with high accuracy.

  In the invention disclosed in Patent Document 2, the relationship between the load converted from the pressure in the hydraulic pressure reduction cylinder and the measured value in the rolling load measuring device is evaluated based on the relationship at a certain point, not the trend over time. Is going. In order to obtain high measurement accuracy, it is necessary to suppress fluctuations in pressure in the hydraulic pressure reduction cylinder as much as possible, especially in a static state where there is no inflow / outflow of hydraulic oil. It is necessary to provide a restriction that

  Also, depending on the characteristics of the rolling mill and the like, even in the initial state (initial use), the load measurement value by the rolling load measuring device and the calculated load value in the hydraulic pressure reduction cylinder may not exactly match. As in Patent Document 2, in the method of using the calculated load value in the hydraulic reduction cylinder at a certain temporary point as the detected value of the rolling load (the method of Claim 1 of Patent Document 2), the rolling load measuring device in the initial state is used. There is also a problem that the difference between the measured load value and the calculated load value in the hydraulic cylinder cannot be taken into account.

  The present invention has been completed in view of the above problems, and it is possible to accurately detect an error in a rolling load measuring device in a rolling mill by using a load calculation value calculated from a pressure in a hydraulic reduction cylinder. An object of the present invention is to provide an abnormality detection method for a rolling load measuring device.

Means of the present invention are as follows.
[1] A load calculated based on a value of a pressure gauge provided in the hydraulic reduction cylinder in a rolling mill having a hydraulic reduction cylinder that presses a roll by a hydraulic method and a rolling load measuring device that measures a load of the rolling mill The calculated value and the load measurement value measured by the rolling load measuring device are compared with time, and the difference between the load calculation value and the load measurement value increases with time. An abnormality detection method for a rolling load measuring device for determining an abnormal state of the measuring device.
[2] Obtain the load calculation value and the load measurement value at a plurality of measurement points in a predetermined measurement span, and obtain a regression coefficient between the load calculation value and the load measurement value in the predetermined measurement span. When the absolute value of the difference between the regression coefficient in the first measurement span and the regression coefficient in the n-th measurement span (n is an integer of 2 or more) exceeds the threshold value, it is determined that the rolling load measuring device is in an abnormal state. The abnormality detection method of the rolling load measuring device according to [1].
[3] In a scatter diagram in which the calculated load value is on the horizontal axis and the measured load value is on the vertical axis, a regression coefficient between the calculated load value and the measured load value in the predetermined measurement span is obtained. When the absolute value of the difference between the regression coefficient in the span and the regression coefficient in the nth measurement span (n is an integer of 2 or more) exceeds 0.1, it is determined that the rolling load measuring apparatus is in an abnormal state [2 ] The abnormality detection method of the rolling load measuring apparatus as described in any one of.
[4] As described in any one of [1] to [3], an abnormal state of the rolling load measuring device is determined using a load calculation value and a load measurement value obtained in a non-rolling state of the rolling mill. An abnormality detection method for a rolling load measuring device.
[5] The load calculation value and the load measurement value are obtained in a rolling state and a non-rolling state of a rolling mill, and a difference between the load calculation value and the load measurement value obtained in the rolling state is determined over time. When the difference between the calculated load value obtained in the non-rolled state and the measured load value increases with time, it is determined that the rolling load measuring device is in an abnormal state [1] The abnormality detection method of the rolling load measuring device according to any one of [3].
[6] The abnormality detection method for a rolling load measuring device according to any one of [1] to [5], wherein the rolling load measuring device is a load cell.

  In the present invention, unlike the invention disclosed in Patent Document 1, the roll bending force at the time of non-rolling is not used, so that an error is not amplified due to the difference in load range between non-rolling and rolling. Thus, it is possible to detect an abnormality of the rolling load measuring apparatus with high accuracy.

  Further, in the present invention, the relationship between the two values at a plurality of time points is grasped over time, rather than comparing the calculated load value and the measured load value at the temporary point. In conventional methods that rely only on measurement results at a single point in time, various constraints to improve measurement accuracy (pressure measurement timing is limited to the state in which the roll gap is open, and static inflow and outflow of hydraulic oil are not required. However, in the present invention, the error of the rolling load measuring apparatus can be evaluated without providing these restrictions.

FIG. 1 is a schematic view of a rolling mill used in the present invention. FIG. 2 is a graph showing an example of measurement results during rolling. FIG. 3 is a graph showing an example of measurement results during non-rolling.

  The present invention will be specifically described with reference to FIG.

  FIG. 1 shows an example of a rolling mill 1 used in the present invention. The rolling mill 1 includes an upper work roll 12 and a lower work roll 22, and rolling is performed by passing a steel plate between these work rolls. The rolling mill 1 includes an upper backup roll 13 and a lower backup roll 23 that mechanically support these work rolls. Although not shown, the rolling mill 1 is provided with members such as a roll chock that supports the work roll and the backup roll, and a housing that houses the roll chock.

  The upper backup roll 13 includes hydraulic pressure reduction cylinders 4a and 4b on both end sides (op side and dr side) in the axial direction. Although not shown, the hydraulic pressure reduction cylinders 4 a and 4 b press both axial ends of the roll chock that supports the upper backup roll 13. Accordingly, a downward roll bending force is applied to the axial ends of the upper backup roll 13 and the upper work roll 12. The upper backup roll 13 and the upper work roll 12 are pressed downward at a predetermined pressure by the hydraulic pressure reduction cylinders 4a and 4b.

  Although not shown, each of the hydraulic pressure reduction cylinders 4a and 4b is provided with a pressure gauge for measuring the pressure due to the hydraulic pressure. These pressure gauges are attached to both the main pressure side and the back pressure side of the hydraulic circuit of the hydraulic pressure reduction cylinders 4a and 4b, and the pressure applied to the cylinder of the hydraulic circuit is calculated by calculating the difference between the two. . As a pressure measuring mechanism in the hydraulic pressure reduction cylinders 4a and 4b, a conventionally known mechanism can be used.

  Moreover, these pressure gauges can calculate the rolling load (load calculation value) in the hydraulic circuit by multiplying the measured pressure by the area (effective cross-sectional area) of the tip surface of the piston of the cylinder.

  It is preferable that at least one pressure gauge is provided in each of the hydraulic pressure reduction cylinders 4a and 4b. Thereby, it is possible to separately calculate the load applied to the op-side hydraulic pressure reduction cylinder 4a and the load applied to the dr-side hydraulic pressure reduction cylinder 4b.

  As shown in FIG. 1, load cells 5 a and 5 b, which are examples of a rolling load measuring device, are provided below the lower backup roll 23. The load cells 5a and 5b are members in which a strain gauge is attached to an elastic deformable body having a predetermined shape and size, and a load measurement value is obtained by converting the amount of strain generated in the elastic deformable body in the load cell into a rolling load. . Thus, the load measurement value can measure the rolling load applied to the rolling mill. Although not shown, the load cells 5a, 5b can be provided between the roll chock that supports the lower backup roll 23 and the housing.

  As shown in FIG. 1, it is preferable that at least one load cell 5a, 5b is also attached to the op side and the dr side. Thereby, it becomes possible to measure separately the rolling load concerning each edge part of the axial direction of a work roll and a backup roll.

  Moreover, contrary to the example of FIG. 1, a load cell can be provided in the upper part of a work roll, and a hydraulic pressure reduction cylinder can also be provided under a work roll.

  The load cells 5a and 5b can measure the rolling load during rolling with very high accuracy. Usually, at the time of rolling operation, various rolling conditions are adjusted so that rolling is performed under appropriate conditions while monitoring the measured values of the rolling load by the load cells 5a and 5b. For example, if the actual plate thickness is excessive with respect to the target plate thickness, the work roll is tightened. If the actual plate thickness is excessive with respect to the target plate thickness, the work roll is tightened. The measured value of the rolling load by the load cells 5a and 5b is used as a guide for the load in the thickness control.

  During repeated rolling, an error gradually occurs in the measured value of the rolling load by the load cells 5a and 5b. Such errors include the loss of linearity of the elastic region of the elastic body that supports the strain gauge due to repeated stress on the strain gauge, wear of the sensor on the surface of the strain gauge, and deviation due to displacement of the strain gauge. It is caused by load generation and abnormal output due to a decrease in insulation of the strain gauge. Therefore, when rolling is repeated, the error between the measurement value (load measurement value) of the rolling load by the load cells 5a and 5b and the true rolling load gradually increases.

  While the errors in the load measurement values by the load cells 5a and 5b are still small, the rolling can be performed by adjusting various rolling conditions so as to correct these errors. However, if the error of the load measurement value by the load cells 5a and 5b becomes too large, a trouble with a threading plate or the like is caused, so that the load cells 5a and 5b need to be replaced. The replacement of the load cells 5a and 5b requires a work period of several days after the rolling mill is stopped. Therefore, if the load cell needs to be replaced suddenly, the operation plan is adversely affected.

  On the other hand, the calculated value of the rolling load using the pressure gauges provided in the hydraulic reduction cylinders 4a and 4b has a large noise and error, and is difficult to use for strict control of rolling conditions. On the other hand, replacement and maintenance of the pressure gauges of the hydraulic reduction cylinders 4a and 4b can be performed without stopping the rolling mill, and can be performed easily and in a short period of time.

  In the present invention, the correlation between the rolling load value (load calculation value) calculated using a pressure gauge provided in the hydraulic cylinders 4a and 4b and the rolling load measurement value (load measurement value) by the load cells 5a and 5b is calculated as follows: By monitoring over time, measurement errors in the load cells 5a and 5b can be accurately evaluated.

  Specifically, the load calculation value in the hydraulic cylinder 4a during rolling is compared with the load measurement value in the load cell 5a at the same point. For example, as shown in FIG. 2, the horizontal axis indicates the load calculation value (PT load) in the hydraulic cylinder 4a, and the vertical axis indicates the load measurement value (LC load) by the load cell 5a−the load calculation value in the hydraulic cylinder 4a ( Data is plotted on a graph in which (PT load) is set. In one graph, a scatter diagram is obtained by plotting data obtained at a plurality of measurement points in a predetermined measurement span (for example, during one day). This scatter diagram shows the correlation between the load calculation value and the load measurement value in a predetermined measurement span.

  Further, when performing rolling a plurality of times within a predetermined measurement span, it is preferable to provide at least one measurement time point for each rolling and plot the load calculation value and the load measurement value on a graph. In addition, it is preferable to align the timing (longitudinal position of the steel plate to be rolled) for obtaining the load calculation value and the load measurement value at the time of each rolling. It is preferable to obtain the calculated load value and the measured load value at the same position.

  When all the data in a predetermined measurement span (for example, for one day) are collected, a scatter diagram is completed, and the correlation between the calculated load value and the load measurement value is extracted from this scatter diagram. Specifically, an approximate straight line (regression line) can be obtained by the least square method, and the correlation between the load calculation value and the load measurement value can be evaluated by the slope of the regression line (regression coefficient). In the example of FIG. 2A, the correlation between the load calculation value on the horizontal axis and (load measurement value−load calculation value) on the vertical axis is evaluated. In the drawing, the calculated load value is referred to as PT load or PT, and the measured load value is referred to as LC load or LC.

  Generally, at the same measurement time point, the load measurement value and the load calculation value are often the same value, and the value of (load measurement value−load calculation value) on the vertical axis in FIG. In many cases, it is close to 0 regardless of the load calculation value. However, depending on the standards of the pressure gauge and rolling load measuring device, there is already a difference between the load measurement value and the load calculation value at the time of the initial state (initial use) (load measurement value−load calculation value). ) And the calculated load value may be linear. In such a case, it is important to pay attention to the temporal variation of the slope of the approximate straight line obtained by the least square method. In other words, if the slope value gradually shifts with time from the slope of the straight line obtained by the least square method in the initial state, it can be evaluated that the measurement error due to the load cell has increased.

  In the example of FIG. 2A, data for one day is accumulated. By creating such a scatter diagram every day and comparing the slopes (regression coefficients) of the approximate lines obtained by the least square method in each scatter diagram, the correlation between the load measurement value and the load calculation value can be calculated. Change can be examined. Since the measurement error between the load measurement value by the load cell 5a and the load calculation value by the cylinder 4a gradually spreads, the vertical axis (load measurement value−load calculation value) as shown in FIG. ), The slope of the straight line obtained by the least square method increases or decreases over time.

  The absolute value of the difference between the slope of the approximate line (regression coefficient) obtained in the scatter diagram on the first day and the slope of the approximate line (regression coefficient) in the scatter chart obtained on and after the second day is gradually It gets bigger. When the absolute value of the difference between the regression coefficient on the nth day (n is an integer of 2 or more) and the regression coefficient on the first day exceeds a predetermined threshold, it is determined that an abnormality has occurred in the measurement value by the load cell. be able to. As an example, in FIG. 2A, when the absolute value of the difference between the regression coefficient on the first day and the regression coefficient on the nth day exceeds 0.1, it is determined that an abnormality has occurred in the measurement value by the load cell. Can do. Further, from the viewpoint of more strictly detecting the load cell abnormality, the threshold value may be set to 0.075 in the example of FIG.

  In the above description, the measurement span has been described as one day. However, the measurement span is not limited to this example. For example, the measurement span can be 12 hours or can be two days.

  The index used for the vertical axis or the horizontal axis of the graph is not limited to the example of FIG. 2A. For example, the rolling load measurement value (LC load) by the load cell 5a is used as the index of the vertical axis or the horizontal axis. It may be used. When the PT load is taken on the horizontal axis as shown in FIG. 2 (a), the difference in regression coefficient is the same regardless of whether (LC load−PT load) is set on the vertical axis or LC load is set. Similar value.

  In the example of FIG. 2A, the correlation between the op-side and dr-side loads is compared separately. Thus, by distinguishing and comparing the op side and the dr side, it is possible to distinguish and determine whether an abnormality has occurred in either the op side or the dr side.

  Note that the correlation of the sum of the respective loads may be compared without distinguishing the correlation between the loads on the op side and the dr side. In this case, the load cell abnormality detection is performed collectively without distinguishing the op side and the dr side.

  Although the method for comparing two load values during rolling has been described above, it is also possible to compare two load values during non-rolling. Specifically, the upper work roll 12 and the lower work roll 22 are brought into contact with each other in a state where the steel plate is not passed through the rolling mill 1, and the load measurement value and the load calculation value are compared. At this time, a plurality of measurement time points are set in the process of tightening the upper work roll 12 and the lower work roll 22, and load measurement values and load calculation values at the plurality of measurement time points are acquired. Approximate straight line is calculated. A similar experiment is performed on and after the second day, and the change in the slope of the approximate straight line over time is examined. If the difference between the regression coefficient of the first day (first time) and the regression coefficient of the nth time (n is an integer of 2 or more) exceeds 0.1, it is judged that an abnormality has occurred in the measurement value by the load cell. Can do. Although details will be described later, since it is difficult to perform load calculation and load measurement every day during non-rolling, the n-th measurement span is often not the n-th measurement during non-rolling.

  During rolling, a force called a thrust force in the width direction of the steel sheet (the axial direction of the roll) may be applied. When such a thrust force is applied, an excessive force is applied to one of the op side and the dr side, so that the load becomes large, and the load becomes small by that amount. At the time of rolling, such a thrust force causes an uneven load in the width direction, so that an error in measurement values by the load cell is likely to occur. Therefore, when it is determined that the load cell is abnormal from the change over time in the slope of the approximate straight line during rolling, there is a possibility of failure or deterioration of the load cell and a possibility of the influence of the thrust force. On the other hand, when the rolling line is operated, there is also an advantage that the calculated load value and the measured load value at the time of rolling can be obtained with high frequency.

  Since such a thrust force does not occur during non-rolling, there is a situation in which an error in a measurement value by the load cell hardly occurs. Therefore, at the time of non-rolling, only the measurement error of the load cell can be evaluated with high accuracy by eliminating the influence of the thrust force. On the other hand, there is a problem that the data at the time of non-rolling is limited in the timing at which the data can be acquired while the operation is continued, and it is difficult to acquire the data at a high frequency.

  In view of this, it is effective to temporarily determine the load cell error temporarily based on the rolling result and then determine the load cell error based on the non-rolling result. For example, data at the time of operation (during rolling) is acquired every day, and errors are evaluated. In parallel with this, data at the time of non-rolling is acquired intermittently (for example, about once every one to two weeks), and errors are similarly evaluated. The load cell is determined to be abnormal for the first time when the load cell is determined to be abnormal by evaluating the error during rolling and when the load cell is also evaluated to be abnormal by evaluating the error during non-rolling. By such a method, the load cell abnormality can be determined frequently based on the rolling data, and the load cell abnormality is finally determined based on the non-rolling data, so that the rolling cell abnormality can be determined. It is possible to prevent the load cell from being erroneously determined to be abnormal due to the generated thrust force.

  Next, an example in which the rolling mill is actually operated using the abnormality detection method of the rolling load measuring apparatus of the present invention will be described.

  First, at a predetermined measurement time, a load calculation value in a cylinder and a load measurement value in a load cell are obtained at a predetermined measurement time while performing rolling using a rolling mill at an arbitrary measurement start, preferably at the time of exchanging the load cell. Thereafter, for each rolling, a load calculation value and a load measurement value are obtained, and these data are sequentially plotted in a table as shown in FIG. When all the rolling operations on the first day have been completed, the plotting on the graph is terminated, and on the second day, the plotting on another new graph is started. The same experiment is performed on the second day and the third day, and the slope of the approximate line is obtained from each graph by the least square method. In addition, data is acquired in the same manner when not rolling.

  If the inclination of the approximate straight line at the time of rolling is followed with time, it will gradually deviate from the inclination of the graph of the first time (first day). While the error is small (when the difference from the slope of the approximate straight line on the first day is not more than a predetermined threshold value), various rolling conditions may be adjusted to correct this error.

  On the other hand, if the error becomes large and the slope of the approximate straight line becomes larger than a predetermined threshold value, it becomes impossible to adjust the error by adjusting the rolling conditions, and it is judged as abnormal.

  In this case, first, it is determined whether or not such an abnormality has been caused by the thrust force during rolling. Specifically, if the slope of the approximate straight line during non-rolling does not exceed the threshold value, it is determined that the measurement abnormality during rolling is due to the thrust force. In this case, the thrust force may be corrected by adjusting various rolling conditions. For example, the roll cross angle is adjusted, or various rolling reduction equipment is replaced. On the other hand, if the approximate straight line at the time of non-rolling exceeds the threshold as in the approximate straight line at the time of rolling, the abnormality is not due to the thrust force, but the failure of the instrument (for example, pressure gauge or load cell) It is judged to be due to deterioration.

  Next, if it is determined that the instrument is faulty or deteriorated, the pressure gauge installed in the hydraulic pressure reduction cylinder may have failed or deteriorated, the load cell may have failed or deteriorated, or the pressure gauge and load cell Either of them may have failed or deteriorated. Since the replacement work of the pressure gauge is easier than the replacement work of the load cell, the pressure gauge is first replaced. After replacing the pressure gauge, take similar data. If the inclination of the approximate straight line still does not become normal after replacement of the pressure gauge, it is determined that the load cell has failed or deteriorated, and therefore a load cell replacement work plan is made.

  Next, examples will be described. In the graph where the measured load value (LC load) and the calculated load value (PT load) of the rolling mill during rolling are obtained, the PT load is set on the horizontal axis, and (LC load−PT load) is set on the vertical axis. A scatter plot was created. A scatter diagram and an approximate straight line obtained on the first day are shown in FIG. The slope of the approximate straight line on the op side was 0.01. On the other hand, the slope of the approximate straight line on the dr side was 0.30.

  A similar experiment was performed about every third day on the third and fifth days, and an approximate straight line was calculated for each day. FIG. 2B is a graph in which differences (in units of “%”) between the slope of the approximate straight line on the first day and the slope of the approximate straight line on and after the second day are summarized in a graph. The number of measurements per day was about 400 to 600 times.

  Moreover, the same experiment was performed also at the time of non-rolling. The results are shown in FIGS. 3 (a) and (b). In addition, the measurement at the time of non-rolling was performed once every about 1 to 2 weeks. Further, in one measurement, in the process of tightening the upper work roll and the lower work roll, four measurement time points at which a load on one side becomes a predetermined magnitude between 1000 kN and 5000 kN are set, and measurement of these four points is performed. The measured load value and the calculated load value at the time were extracted.

  From FIG. 2B, it can be seen that the difference in slope on the op side has not changed greatly until the end, but the difference in slope on the dr side gradually increases. From this, it is presumed that there is no abnormality in the load cell on the op side, but there is an abnormality in the load cell on the dr side.

  Next, looking at FIG. 3B, the difference in inclination gradually increases on both the op side and the dr side. From this, it can be determined that a load error has actually occurred in both the op-side and dr-side load cells. In the rolling data, it is considered that the abnormality on the dr side was detected without detecting the abnormality on the op side because the measurement error of the load cell was canceled by the thrust force in the data on the op side.

  If the threshold value is 0.1, the difference in slope exceeds the threshold value on the 75th day both when rolling (FIG. 2 (b)) and when not rolling (FIG. 3 (b)). At this time, it is determined that the load cell is abnormal. On the other hand, the data on the dr side is not determined to be a load cell abnormality until the 90th day because the difference in slope does not exceed 0.1 both during rolling and during non-rolling.

DESCRIPTION OF SYMBOLS 1 Rolling machine 12 Upper work roll 22 Lower work roll 13 Upper support roll 23 Lower support roll 4a, 4b Hydraulic pressure reduction cylinder 5a, 5b Load cell

Claims (6)

In a rolling mill having a hydraulic reduction cylinder that presses a roll by a hydraulic method and a rolling load measuring device that measures a load of the rolling mill,
A load calculation value calculated based on a value of a pressure gauge provided in the hydraulic pressure reduction cylinder and a load measurement value measured by the rolling load measuring device are compared over time,
An abnormality detection method for a rolling load measuring apparatus, wherein a state where a difference between the calculated load value and the measured load value increases with time is determined as an abnormal state of the rolling load measuring apparatus. The load calculation value and the load measurement value are obtained at a plurality of measurement points in a predetermined measurement span,
Obtaining a regression coefficient between the load calculation value and the load measurement value in the predetermined measurement span;
When the absolute value of the difference between the regression coefficient in the first measurement span and the regression coefficient in the n-th measurement span (n is an integer of 2 or more) exceeds the threshold value, it is determined that the rolling load measuring device is in an abnormal state. The abnormality detection method of the rolling load measuring apparatus according to claim 1. In a scatter diagram with the load calculation value on the horizontal axis and the load measurement value on the vertical axis, a regression coefficient between the load calculation value and the load measurement value in the predetermined measurement span is obtained,
When the absolute value of the difference between the regression coefficient in the first measurement span and the regression coefficient in the n-th measurement span (n is an integer of 2 or more) exceeds 0.1, The abnormality detection method of the rolling load measuring device according to claim 2 for determination.   The rolling load measuring device according to any one of claims 1 to 3, wherein an abnormal state of the rolling load measuring device is determined using a load calculation value and a load measurement value obtained in a non-rolling state of the rolling mill. Anomaly detection method. In the rolling state and non-rolling state of the rolling mill, obtain the load calculation value and the load measurement value,
The difference between the calculated load value obtained in the rolling state and the measured load value increases over time, and the difference between the calculated load value obtained in the non-rolled state and the measured load value varies with time. The abnormality detection method of the rolling load measuring device according to any one of claims 1 to 3, wherein the abnormal state of the rolling load measuring device is determined when the rolling load measuring device is larger.   The abnormality detection method for a rolling load measuring device according to any one of claims 1 to 5, wherein the rolling load measuring device is a load cell.