An on-line detection feedback control method in strip rolling

Abstract

The application discloses a strip rolling online detection feedback control method, which detects the thickness or width of a strip through a sensor in an online high-precision mode, and controls the radial movement of a roller according to the detection result. The control criterion comprises controlling the target value a of the thickness or width of the strip within the error range c of the standard value b. The measurement precision of the sensor is better than 1 / 3 of the product error index. The feedback control method comprises a continuous filtering real-time control method or a cross-section detection control method. The cross-section detection control method is provided with a cross-repeating section between adjacent sections. The new method can reduce the influence of the detection noise of the high-precision sensor and the single-point measurement abnormal pulse on the feedback regulation of the roller, so that the feedback regulation is more stable, the product size is more stable, and the engineering practical significance is greater.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of metal rolling, and particularly relates to an online detection feedback control method in strip rolling. BACKGROUND

[0002] With the development of precision machining technology, in the rolling process of some metal strips, in addition to the requirements for the shape and structural strength, high requirements are also put forward for the dimensional accuracy, such as titanium alloy, aluminum alloy, stainless steel and other metal rolling for the machining and production of mobile phone frames, card holders, precision medical devices, precision automobile parts, smart home precision parts, and the dimensional accuracy of steel-aluminum, titanium-aluminum, magnesium-aluminum and other metal composite rolling strips is also required to be high.

[0003] For example, a well-known mobile phone brand requires that the rolling thickness error of titanium alloy mobile phone frame strips should not exceed 0.03 mm. For the traditional metal rolling machining process, this requirement is quite high. In particular, due to the movement and vibration of the driving system of the rolling equipment, the random error of dynamic detection may exceed the index requirement. If the detection result is used to directly control the roller, the roller will be continuously adjusted in the radial rolling distance, so that the actual rolling strip product thickness is always changing, which is difficult to meet the index requirement, and an effective detection and feedback control method is needed. SUMMARY

[0004] In view of the above problems, the application designs an online detection feedback control method in strip rolling, which can realize accurate control of the rolling process through filtering and removing outliers of the detection data, so as to ensure that the product quality meets the requirements.

[0005] The online detection feedback control method in strip rolling designed by the application realizes online high-precision detection of the thickness or width of the strip through a sensor, and controls the roller to make radial movement (changes the rolling distance of the roller, so as to change the thickness or width of the rolled strip) according to the detection result. The control criterion includes controlling the target value a of the thickness or width of the strip within the error range c of the standard value b, that is, a is within b±c, or b-c≦a≦b+c. The measurement accuracy of the sensor is better than 1 / 3 of the product error index (the width and thickness detection sensors are required to correspond to the respective index requirements, such as the requirement that the target product thickness error should not exceed 0.03 mm, and the measurement accuracy of the sensor should be better than 0.01 mm). The feedback control method includes continuous filtering real-time control method or cross-section detection control method (also known as sliding window detection control method). The cross-section detection control method is provided with a cross-repeating section between adjacent sections.

[0006] Further, the continuous filtering real-time control method comprises filtering the real-time thickness or width detection value by using a filtering model, and the target value a comprises the filtering result. If the filtering result a > b + c, the system controls the rolling distance to be reduced in the radial direction of the roller according to the deviation a - b. If the filtering result a < b - c, the system controls the rolling distance to be increased in the radial direction of the roller according to the deviation a - b.

[0007] Further, the continuous filtering real-time control method comprises the step of removing outliers of the actual detection value before filtering.

[0008] Further, the continuous filtering real-time control method comprises marking the strip point with a filtering deviation, and the marking method comprises marking physically, such as marking, or recording the distance (relative length) of the abnormal point relative to the starting or ending position of the strip by the system, so as to facilitate the subsequent process of the strip rolling to remove the size-unqualified segment.

[0009] Further, the cross-section detection control method comprises the following steps:

[0010] S1, setting the length of the detection section as L1 and the length of the cross-over between sections as L2, and the overlap L2 satisfies 0 < L2 < L1;

[0011] S2, setting the sampling interval AL or sampling period At of the strip thickness or width detection, the sampling interval AL < 1 / 10 L1, that is, at least 10 different detection points are sampled in each section, and a uniform sampling mode is generally used, and the sampling period At = AL / V, wherein V is the linear speed of the strip rolling output;

[0012] S3, continuously detecting and recording the detection results, and the detection results comprise real-time detection results of the strip thickness and / or width;

[0013] S4, sectionally counting and controlling the radial movement of the roller according to the counting results, calculating the statistical mean value of the detection results of the sections, and controlling the radial movement of the roller according to the control criterion.

[0014] Further, the outliers of the actual detection value are removed before step S4.

[0015] Further, the segments with a statistical mean value deviation are marked after step S4.

[0016] Further, the control criterion comprises calculating the control amount (such as voltage or current or frequency control pulse) according to the deviation a - b and the system transfer function (such as motor transfer function), and driving the roller to change the rolling distance.

[0017] Further, the error range c is not less than the radial distance change amount caused by a single control pulse of the roller precession control motor.

[0018] Further, the determination method of the segment length L1 includes, then the segment length range is 1 / 2L0≦L1≦10L0, wherein L0 is the minimum segment length cut in the next cutting process of the strip rolling process.

[0019] The advantages and beneficial effects of the present application are that the designed online detection feedback control method in the strip rolling can avoid the problem that the actual rolling strip product thickness is always changing and is difficult to meet the index requirements caused by the traditional use of detection value to directly control the roller, which always adjusts the radial rolling distance back and forth, by online high-precision detection of the strip thickness or width through the sensor, and feedback control of the roller radial movement according to the detection results using the continuous filtering real-time control method or the cross-section detection control method. Because the product processing precision requirement is high, the sensor measurement precision requirement is high, and the random error of dynamic detection may exceed the index requirement due to the movement and vibration of the rolling equipment driving system in the field production line, the measured value directly control is easy to cause oscillation phenomenon, that is, the rolling distance is always fluctuating. The present application reduces the influence of high-precision sensor detection noise and single-point measurement abnormal pulse on roller feedback adjustment, makes the feedback adjustment more stable, the product size more stable, and the engineering practical significance greater.

[0020] In addition, when the metal strip is high-precision hot rolled, the mold temperature rises and the rolling gap gradually decreases, and if it is not adjusted in time, the product thickness will become smaller and smaller, which is easy to lead to scrap, so it is particularly important to detect the product width and / or thickness in real time and feedback control the roller movement. Secondly, continuous real-time measurement can also find regular changes in product size, so as to feedback the dynamic round runout after the installation of the mold, which has important significance for rolling process improvement and equipment improvement and installation adjustment. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a schematic view of a strip thickness online detection device;

[0022] Figure 2 is Figure 1 side view.

[0023] Figure 3 is a principle step block diagram of the cross-section detection control method.

[0024] Markings in the figure:

[0025] 1. frame; 2. second driving cylinder; 3. distance sensor; 4. second guide rod; 5. second connecting plate; 6. sensor mounting bracket; 7. first slider bracket; 8. first linear guide rail; 9. second linear guide rail; 10. second spring; 11. measured workpiece; 12. second slider bracket; 13. second compression roller; 14. second guide rod; 15. first spring; 16. first guide rod; 17. first connecting plate; 18. first driving cylinder. DETAILED DESCRIPTION

[0026] The specific embodiments of the present application are further described below with reference to the accompanying drawings and examples. The following examples are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0027] Example 1

[0028] As shown in Figure 1 and Figure 2 , a titanium alloy strip thickness on-line detection device, a pair of bearings are provided on the frame 1, one of the bearings is rotatably provided on the frame 1, and the other bearing is rotatably provided on the first slider bracket 7, and the measured workpiece 11 of the metal strip is located between the two bearings; the frame 1 is also provided with a first linear guide rail 8, and the first slider bracket 7 is arranged on the first linear guide rail 8, so that the bearings are close to or away from the other bearing along the first linear guide rail 8, and the frame 1 on one side of the first slider bracket 7 is provided with a first driving cylinder 18, and the piston rod of the first driving cylinder 18 is provided with a first connecting plate 17, and the first connecting plate 17 is provided with a first guide rod 16 connected with the first slider bracket 7, and the first connecting plate 17 is provided with a guide hole to make the one end of the first guide rod 16 slide with the first connecting plate 17, and the other end of the first guide rod 16 is fixedly connected with the first slider bracket 7, and the first spring 15 is sleeved on the first guide rod 16 between the first connecting plate 17 and the first slider bracket 7; the frame 1 is provided with a sensor mounting bracket 6 close to the first slider bracket 7, and the sensor mounting bracket 6 is provided with a distance sensor 3, and the detection head of the distance sensor 3 is connected with the first slider bracket 7; the second slider bracket 12 is provided with a second guide rod 4, and the corresponding second driving cylinder 2 is provided with a second connecting plate 5, and the second driving cylinder 2 is arranged on the frame 1, one end of the second guide rod 4 is slidably connected with the second connecting plate 5, and the other end is fixedly connected with the second slider bracket 12, and the second guide rod 4 is sleeved with a second spring 10; the first compression roller 14 is rotatably arranged on the first slider bracket 7, the first slider bracket 7 is provided with a second linear guide rail 9, the second linear guide rail 9 is provided with a second slider bracket 12, and the second compression roller 13 is rotatably arranged on the second slider bracket 12.

[0029] The application discloses a strip rolling online detection feedback control method, which realizes online high-precision detection of strip thickness or width through a sensor, and controls the radial movement of a roller (changes the rolling distance of the roller, so as to change the thickness or width of the rolled strip) according to the detection result. The control criterion includes controlling the target value a of the strip thickness or width within the error range c of the standard value b, that is, a is within b±c, or b-c≦a≦b+c. The sensor has a measurement accuracy superior to 1 / 3 of a product error index (the width and thickness detection sensors are required to correspond to respective index requirements, for example, the sensor measurement accuracy should be superior to 0.01 mm for the requirement that the target product thickness error is not more than 0.03 mm). The feedback control method includes a continuous filtering real-time control method or a cross-section detection control method (also referred to as a sliding window detection control method). The cross-section detection control method is provided with a cross-repeated section between adjacent sections. The continuous filtering real-time control method is adopted in the embodiment.

[0030] The sensor realizes online high-precision detection, which can adopt photoelectric non-contact measurement, such as a bidirectional laser range finder for laser ranging on the upper and lower surfaces of the strip. Alternatively, a contact type pressure sensing range finder can be adopted. Since laser ranging is generally single-point ranging, for the strip with a certain surface area, there may be sampling deviation of the ranging point, which is not sufficient to represent the actual change of the strip thickness. In the embodiment, a high-precision digital contact sensor is adopted, which is provided with two rollers in one group on the upper and lower surfaces of the strip to sense the change of the strip thickness (or width). The local thickness change may not be within the detection range of the laser spot, but is within the sensing range of the roller. Figure 1 And Figure 2 The titanium alloy strip detection device shown in the embodiment adopts a high-precision digital contact sensor. The sensor senses the change of the strip thickness (or width) through the rolling of the two rollers in one group on the upper and lower surfaces of the strip. The local thickness change may not be within the detection range of the laser spot, but is within the sensing range of the roller. GT2-H12K is used for detection, and the measurement accuracy reaches 1 um. GT2-32 is used for width detection.

[0031] Traditional online monitoring feedback control is generally only direct feedback control of the measured value, which results in that the roller is continuously controlled and adjusted, and has no identification and processing capability for noise and single-point abnormality. Finally, the product size is also in an unstable state. The instability mainly depends on the sensor detection accuracy and the system control accuracy. The continuous filtering real-time control can filter part of the noise, especially the influence of single-point measurement abnormal pulse. The adjustment response probability and sensitivity of the system to noise and single-point measurement abnormal pulse are greatly reduced, so as to be beneficial to engineering practice. On the other hand, the cross-section detection control method is adopted, the roller rolling distance is controlled through the sectional analysis of the product, the influence of the sensor detection noise and single-point measurement abnormal pulse is further reduced, the feedback adjustment is more stable, the product size is more stable, and the engineering practical significance is greater.

[0032] Preferably, the continuous filtering real-time control method comprises filtering the real-time thickness or width detection value by using a filtering model, the target value a comprises the filtering result, if the filtering result a>b+c, the system controls the rolling distance to be reduced radially according to the deviation a-b, if the filtering result a<b-c, the system controls the rolling distance to be increased radially according to the deviation a-b, and the filtering model comprises any one of a constant speed model, a constant acceleration model, a Singer model, a current model, a polynomial model, a damping oscillation model, a trigonometric function model, a swing model, and a hybrid multi-model.

[0033] Preferably, the continuous filtering real-time control method comprises a step of eliminating outliers of actual detection values before filtering.

[0034] Preferably, the continuous filtering real-time control method comprises marking the strip points with filtering out-of-tolerance, the marking method comprises marking physically, such as marking, or recording the distance (relative length) of the abnormal points relative to the starting or ending position of the strip, so as to facilitate the subsequent process of the strip rolling to eliminate the size-unqualified paragraphs.

[0035] Preferably, the control criterion comprises calculating the control amount (such as voltage or current or frequency control pulse) according to the deviation a-b and the system transfer function (such as motor transfer function), and driving the roll to change the rolling distance.

[0036] Preferably, the error range c is not less than the radial distance change amount caused by a single control pulse of the roll precession control motor.

[0037] Embodiment 2

[0038] The difference from embodiment 1 is that the cross-section detection control method is used in this embodiment, which comprises the following steps:

[0039] S1, set the length of the detection section as L1 and the length of the cross-over between sections as L2, the overlap L2 satisfies 0≦L2<L1, and L2=1 / 2L1 is taken in this embodiment, that is, there is 1 / 2 overlap for each section;

[0040] S2, set the strip thickness or width detection sampling interval AL or sampling period At, AL≦1 / 10L1, that is, at least 10 different detection points are sampled in each segment, and generally uniform sampling is adopted, the sampling period At=AL / V, wherein V is the linear speed of the strip rolling output; the determination of the sampling period generally needs to consider the actual sampling capacity of the detection sensor, the computing capacity of data processing, and the minimum sampling point requirement of the segment, and generally the actual sampling capacity of the sensor is sufficient, and the computer or embedded CPU has sufficient computing capacity, and the key is the minimum sampling point requirement of the segment, that is, how many points are expected to be sampled in a segment to ensure that the obtained mean value is representative and can represent whether the segment actually has an anomaly or the roll actually needs to be adjusted; too few sampling points are easily affected by noise, and too many sampling points are not necessary on one hand, and affect the calculation efficiency on the other hand;

[0041] S3, continuously detect and record the detection results, and the detection results include the real-time detection results of the strip thickness and / or width;

[0042] S4, segment statistics and control of roll radial movement according to the statistical results, the statistical mean value is calculated according to the detection results of several segments, and the roll is controlled to move radially according to the control criterion; that is, the target value a includes the segment mean value, when a>b+c, the system controls the approach roll to move in the direction in which the radial distance from the counter-roller decreases, and the approach amount is determined according to the deviation a-b and the approach motor transfer function; when a<b-c, the system controls the approach roll to move in the direction in which the radial distance from the counter-roller increases, and the approach amount is determined according to the deviation b-a and the approach motor transfer function, the approach amount calculation technology is the existing technology of motor control, and generally the rolling force is inversely proportional to the distance between the upper and lower rolls of the rolling mill.

[0043] Compared with the continuous filtering real-time control method, the continuous filtering real-time control method is considered to be abnormal for a point or several continuous points, and thus the roll distance adjustment is started, the advantage of the scheme is that it is sensitive to fluctuations, but the disadvantage is that it is also sensitive to noise, which is not conducive to the stability of the actual rolling process, and the roll may be in dynamic adjustment all the time; and the cross-segment detection control only starts to control the roll adjustment when the mean value of a segment exceeds the index, is not sensitive to noise, is also not sensitive to actual fluctuations, but the control process is more stable, and is more practical in strip rolling, because the reasons that can cause the strip thickness or width to exceed the standard, such as obvious and continuous change of the strip raw material or change of the roll position, have long-term effects, and short-term effects are not required to respond in principle, so that the cross-segment detection control is more practical and reasonable, and the disadvantage of the scheme is that the roll adjustment response is slow, and there is a delay in response when an actual anomaly occurs, that is, there is a significant defective product, but compared with the continuous filtering real-time control method, the actual good rate of the product is higher.

[0044] The cross-section detection control method sets overlapping paragraphs between adjacent sections. On the one hand, it aims to appropriately improve the sensitivity and real-time performance of the system response without losing the smooth control performance (the section length remains unchanged, and the smoothness remains unchanged), which can reduce the defective rate when the actual abnormality occurs and implement the adjustment as soon as possible. The larger L2 is, the stronger the real-time performance of the control is. In an ideal case, taking the cumulative 10 sampling periods of the L1 paragraph as an example, the maximum L2 paragraph has 9 overlapping sampling periods. The real-time performance of this setting is the strongest, but because the mean value is taken for control adjustment, the real-time performance is still inferior to the adjustment strategy using the real-time detection value or the adjustment strategy using the measured value filtering.

[0045] On the other hand, the overlapping paragraphs between adjacent sections can accurately determine the specific position (or paragraph) of the strip thickness or width that does not meet the standard, facilitate the accurate determination of product quality during subsequent section cutting, and improve the real-time performance of the control. The longer the overlapping paragraph is, the greater the calculation amount is, and the shorter the control amount change period is. In the worst case, for the maximum L2 paragraph, the shortest period of the control amount change is the sampling period. For a high sampling rate, too frequent control is not necessarily a good thing. However, even in the above case, for the maximum L2 paragraph, the control smoothness is still better than the control strategy based on the single-point measured value or the continuous filtering result. Because for a single-point pulse abnormality, the measured value or the real-time filtering result may exceed the control threshold, but for the section mean value, a single-point abnormality is difficult to cause a significant change in the entire section mean value. That is, the sliding section statistical method has a significant inhibitory effect on the pulse value.

[0046] For the embodiment of L2=0, the paragraphs of each section are continuous, independent, and have no overlap. Although the real-time performance is insufficient, the positioning accuracy of defective products is poor, and a little more defective products (generally one section paragraph) may be caused, the control adjustment is more stable, and the control amount changes at most once a section paragraph. This is also feasible for actual engineering applications.

[0047] Preferably, the wild value rejection of the actual detection value is performed before step S4.

[0048] Preferably, the paragraphs with statistical mean value exceeding the standard are marked after step S4. The marking method includes marking physically, such as marking, or recording the serial number or the distance (relative length) from the starting position of the strip of the abnormal paragraph by the system, so as to facilitate the rejection of the paragraphs with size defects in the subsequent process of the strip rolling.

[0049] Preferably, the method for determining the segment length L1 comprises, then the segment length L1 is in the range of 1 / 2L0≦L1≦10L0, wherein L0 is the minimum segment length cut in the next cutting process of the strip rolling process, in order to ensure the reliability of the average detection result for adjusting the roll distance, the segment length can be slightly longer, so that small or short random fluctuations will be averaged out, the roll does not need to be adjusted, only when the longer segment average is out of tolerance, it is confirmed that the actual result is indeed abnormal, but the segment setting is too long, the response to the roll and the raw material change is not sensitive, which leads to the decrease of the product yield; if the segment setting is too short, the roll control will be too sensitive, and random noise may also respond, which will affect the actual product yield; the above setting optimization range is given according to experiments and analysis, but it is not necessary, and the titanium alloy strip for mobile phone frame in the present embodiment is taken as L1=L0.

[0050] In fact, in the cross-segment detection control method step S4, statistical analysis of the segment variance can also be performed, and a variance threshold is set. If the statistical variance exceeds the threshold, the segment strip should be marked as unqualified, and the equipment operation should be checked in serious cases. The average value mainly reflects the deviation of the average size (thickness or width) of the strip after rolling from the standard value, but the deviation of the actual size of the entire segment from the standard value cannot be known. Here, through statistical analysis of the segment variance, it can be known that the variance exceeding the standard means that the size unevenness in the segment is obvious. For example, even if the statistical average of a segment is within the error range, when the variance exceeds the standard, it indicates that the size fluctuation is large and / or the number of points is large, which is also a problem that needs attention.

[0051] The detection and feedback control method of width and thickness is similar, and is generally performed independently.

[0052] As shown in the titanium alloy strip thickness online detection device shown in Figure 1 and Figure 2 According to the traditional strip detection direct feedback control method, the actual rolling result data fluctuates greatly, the thickness fluctuation is about 0.08 mm, and the main reasons are: the up and down roll of the rolling mill jumps, the AGC measuring wheel jumps, and the incoming material thickness difference is large; the width fluctuation is about 0.2 mm (the width measurement device is not shown in the figure, but the structure principle is consistent with the thickness measurement), and the main reasons are: incoming material factors, width edge adjustment influence, etc. Overall, the length proportion of the local abnormal segment of the average 1500 m long strip per roll is higher than 5%.

[0053] By using the method of the present application, the width and thickness fluctuation can be controlled within the index requirement of 0.03 mm. Overall, the length proportion of the local abnormal segment of the average 1500 m long strip per roll is less than 1%.

[0054] Example 3

[0055] The difference from example 2 is that L2=0 is taken in this example, i.e. there is no intersection between adjacent segments, and each segment is independently sampled and analyzed.

[0056] Example 4

[0057] The difference from example 2 is that L1=1 / 2L0 is taken in this example, i.e. the segment length is 1 / 2 of the strip cutting length in the next process, which can be more accurate for positioning the thickness abnormality, but in order to ensure the statistical reliability, the number of sampling points in the segment range should be ensured, i.e. more than 10 sampling points are still required, so the overall data volume is large, and the requirement for the data processing system is slightly higher, but it is still achievable for the existing technology.

[0058] Example 5

[0059] The difference from example 2 is that L1=10L0 is taken in this example, i.e. the segment length is 10 times the strip cutting length in the next process, and the number of sampling points in the segment range of this example is set to 100, although the positioning accuracy of the thickness abnormality is poor, but the system control stability is higher, which is also desirable for large production lines.

[0060] The basic principle of the present application is that, in the processing of high-precision metal strip products, due to the high accuracy of the sensor, the random error of dynamic detection generally completely exceeds the sensor accuracy and exceeds the product performance index requirement due to the movement and vibration of the rolling equipment driving system in the field production line. The method of directly feeding back the measured value to control the rolling distance of the roller is prone to oscillation, i.e. the rolling distance is always fluctuating, and combined with the uneven size of the incoming material, it is particularly easy to cause the uneven width and thickness of the final strip product, which exceeds the index requirement. The present application analyzes and processes the real-time detection quantity, adopts continuous filtering real-time control method or cross-section detection control method, and continuously filters or sliding window averages (smooths) the measured size, reduces the detection noise, especially the influence of single abnormal detection pulse value on the control system, so that the roller control adjustment is more stable, the product size is more uniform, and the continuous rolling yield is higher.

[0061] The above only is part of the present application's more systematic comprehensive strip rolling on-line detection feedback control method embodiment, in fact, whether it is continuous filtering real-time control method or cross sectional detection control method, they have many preferred schemes, these preferred schemes can also be combined arbitrarily; on the other hand, other methods for reducing the influence of noise and vibration are adopted, for the present application, the segment length is controlled in the range of 1 / 2Lo≦L1≦10Lo, which is not absolute, only the present application considers that it is more economical, reliable and applicable in this range, if it is a little larger or a little smaller, although it can be implemented, but some possible adverse factors need to be tolerated, if the raw material size and the overall rolling environment are relatively stable, it can also be considered. These combinations or preferred schemes or schemes without substantial changes should also be considered as the protection scope of the present application, which will not be listed one by one here.

Claims

1. A method of on-line detection feedback control in strip rolling, in which the thickness or width of a strip is detected on-line by a sensor, and the roll is controlled in radial movement in feedback based on the detection result, the control criterion including controlling the target value a of the thickness or width of the strip within the error range c of the standard value b, characterized in that, The sensor measurement accuracy is better than 1 / 3 of the product error index, the feedback control method comprises a cross-section detection control method, the cross-section detection control method is provided with a cross-over repeated section between adjacent sections; ​ The cross-section detection control method comprises the following steps: S1, setting the length of the detection section as L1 and the length of the cross-over between sections as L2; S2, setting the sampling interval AL or sampling period At of the strip thickness or width detection; S3, continuously detecting and recording the detection results; S4, section statistics and controlling the roll radial movement according to the statistical results, calculating the statistical mean value of the detection results of the sections in turn, and controlling the roll radial movement according to the control criterion; The section length L1 meets the range 1 / 2L0≦L1≦10L0, wherein L0 is the minimum section length of the strip cutting process in the next cutting process; the cross-over length between sections is L2, and the overlap L2 meets 0≦L2<L1; The feedback control method further comprises statistical analysis of the section variance, setting a variance threshold, and if the statistical variance exceeds the threshold, the section strip should be marked as unqualified.

2. A strip rolling on-line detection feedback control method according to claim 1, characterized in that, The wild value elimination of the actual detection value is performed before step S4.

3. A strip rolling on-line detection feedback control method according to claim 1, characterized in that, The section with the statistical mean value exceeding the tolerance is marked after step S4.

4. A method of on-line detection feedback control in strip rolling according to any one of claims 1 to 3, characterized in that, The control criterion comprises calculating the control amount according to the deviation a-b and the system transfer function.

5. A method of on-line detection feedback control in strip rolling according to any one of claims 1 to 3, characterized in that, The error range c is not less than the radial distance change amount caused by a single control pulse of the roll advance control motor.

Images