A method for width self-adaptive learning using dynamic acquisition function

Abstract

The present application relates to a kind of methods for width self-adaptive learning using dynamic acquisition function, the method comprises the following steps: S1, dynamic beat and measure width, S2, dynamic steel and measure width, S3, width feedforward of intermediate pass.The technical scheme improves the precision of width control, starts from the data acquisition of site, 100% acquisition is carried out to the data of width meter, and 100% trust is trusted.Using the redundant backup mode of width measuring equipment, ensure that data can be collected, through the design of dynamic acquisition mode, the accuracy of data is realized, and the control mode of model itself is designed, to ensure that model can efficiently use site data in width setting to improve width.

Description

TECHNICAL FIELD

[0001] The present application relates to a learning method, in particular to a method for width self-adaptive learning using dynamic acquisition function, and belongs to the technical field of improving the width accuracy of rough rolling in hot rolling line. BACKGROUND

[0002] There are two width gauges between the racks of Meisteel 1422 hot rolling production line, which are respectively located after the reversible rolling mills R1 and R2. Due to the water control between the racks of the reversible rolling mill and the influence of the production environment in different seasons, the width gauge needs to be installed far away from the rack position to ensure that the width gauge will not be affected by the water mist in the field and cannot work. However, there is a problem that the width gauge cannot collect the data of the middle of the strip due to its distance, which leads to the opposite control of the width control affected by dirty data, and the width is too wide or too narrow. This technology is developed to solve the contradiction in the field, that is, a method for collecting and controlling the width on site.

[0003] After the novelty search, the method for setting the rough rolling width by using the width gauge between the racks of the hot rolling mill mainly introduces how to apply and set the data collected by the width gauge in the model. This method can only collect the last width of the rack for setting, while the present method is aimed at the problem that the correct data cannot be collected due to the remote position of the width gauge caused by the field, and optimizes the collection and transmits the data to the model setting. The important point of this method is to ensure that the strip can receive correct data in all rough rolling passes and optimize the model setting. The two methods are completely different.

[0004] The computer model control method for the width of hot rolled products is a method for classifying reading and storing the rough rolling width expansion learning value according to the steel grade, product width, rough rolling mill, width reduction amount and inlet thickness. The factors in this classification method are obviously related to the width expansion change amount, and are not affected by the rough rolling pass combination mode and product thickness, which can significantly narrow the fluctuation range of the rough rolling width expansion learning value, and is suitable for the production of products with various incoming slab thicknesses. It can effectively improve the product width control precision, reduce the scrap cutting rate and cost loss caused thereby, while the present method accurately and effectively utilizes the data of the on-site width gauge, thereby improving the width accuracy, which is obviously different from the content of the above-mentioned invention patent and paper. SUMMARY

[0005] The present application is just for the problems existing in the prior art, providing a method for width self-adaptive learning by dynamic acquisition function, which improves the precision of width control, starts from the data acquisition of the field, and collects 100% of the data of the width meter, and is 100% trusted. The redundant backup mode of the width measuring equipment is used to ensure that the data can be collected, the design of the dynamic acquisition mode realizes the accuracy of the data, and the control mode of the model itself is designed to ensure that the model can efficiently use the field data for width improvement in the width setting.

[0006] In order to achieve the above-mentioned purpose, the technical scheme of the present application is as follows: a method for width self-adaptive learning by dynamic acquisition function, comprising the following steps:

[0007] S1, dynamic tapping width measurement,

[0008] S2, dynamic steel throwing width measurement,

[0009] S3, width feedforward of intermediate pass.

[0010] As an improvement of the present application, S1, dynamic tapping width measurement, specifically as follows:

[0011] S11, there are two tapping width measurement devices before R1 in the field, respectively located at the descaling box position (device one) and the position before R1 rack (device two), and the tapping 2 device close to the R1 rack position is the main one in the field control, because the tapping device two is closer to the R1 rolling mill, that is, the inlet device of width control, the width loss caused by the temperature drop will be minimum, and the width data of the inlet is more accurate;

[0012] S12, when the following conditions occur in the field, switch to tapping device one for tapping:

[0013] Tapping device two fails, and the field cannot collect data, so tapping device one is switched;

[0014] Tapping device two produces more dirty data due to the influence of descaling water between racks, so tapping device one is switched;

[0015] When the production of short material, the strip steel cannot stop at the tapping 2 position, and it is easy to rush into the R1 rack to cause equipment and production scrap, so tapping device one is switched at this time;

[0016] When it is necessary to check the two tapping devices, tapping device one and tapping device two are used at the same time;

[0017] In summary, the combination of the two tapping devices has the following modes:

[0018] Tapping 1 and tapping device two are used at the same time;

[0019] Beat device one is put into use and beat device two is not put into use;

[0020] Beat device two is put into use and beat device one is not put into use;

[0021] S13, for the use and switching mode of the device, the model is designed correspondingly, when beat 1 and beat device two are used simultaneously, the model remembers the values of the two beat modes, and when the deviation of the two values is within plus or minus 3mm, the width of beat 2 is used for width setting of the model, when the deviation of the two values exceeds plus or minus 3mm, the original width entry data needs to be introduced, the width of beat is compared with the original width data, when the deviation is within the range of plus or minus 5, according to the control loss on site, the smaller one of the beat values is used for width setting of the model,

[0022] When beat device one is put into use + beat device two is not put into use, the actual value collected by beat device one is directly used for control;

[0023] When beat device two is put into use + beat device one is not put into use, the actual value collected by beat device two is directly used for control.

[0024] As an improvement of the present application, S2, dynamic steel throwing width measurement, specifically as follows:

[0025] S21, according to the detector and speed on site, the tail position TailPos when the strip steel is forward is tracked, and the length Len of the strip steel at each pass is calculated by using the equal volume method,

[0026] S22, the relative position SGPos of the width gauge and the deviation position DeltaPos when the strip steel leaves the rolling mill and is thrown are calculated,

[0027] S23, in order to ensure that the middle width of the strip steel can be detected, it is necessary to ensure that the middle position exceeds the width gauge position, and at the same time, in order to save the production rhythm of the rolling line, the tail does not need to be detected, so the rolling mill throwing position is dynamically calculated for different strip steels.

[0028] As an improvement of the present application, S3, width feedforward of the intermediate pass, specifically as follows:

[0029] S31, according to the rolling strategy on site, the corresponding feedforward learning mode is configured, there are two reversible rolling mills on site, so there are multiple combination modes of the corresponding feedforward learning mode. When the feedforward learning is configured, since the reverse pass collection function is not started, it is only opened in the forward direction.

[0030] S32, according to the situation on site, the feedforward learning confidence of different regions is configured, since the number and quality of the width gauge on-site sampling data are greatly different in different regions, different feedforward learning confidences need to be configured in different regions, and the range is 0-1;

[0031] S33, according to the final feedforward learning result, as the input of the reverse pass, the width model calculation between passes is carried out. Through the dog bone and the horizontal spread, combined with the input of the feedforward learning, the output width of each pass is finally calculated.

[0032] Compared with the prior art, the present application has the following advantages: the present application uploads the accurate width data of the middle part of the strip steel to the model control during the dynamic casting function of width data acquisition, which not only maximizes the use of the equipment on site, but also ensures the accuracy and availability of the width data, improves the precision of the model setting, and through the development of the corresponding width feedforward learning function of the model, the production demand can be better met, each width gauge is controlled and adjusted individually, the adjustability and adaptive learning function are more flexible, the accuracy of the hot rolling strip steel width self-learning is ensured, the width control precision of hot rolling is greatly improved, and the yield of the strip steel during the production of the downstream process is increased. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 It is a layout of the on-site beating equipment;

[0034] Figure 2 It is a width gauge distribution diagram between stands;

[0035] Figure 3 It is a dynamic data acquisition process. DETAILED DESCRIPTION

[0036] In order to deepen the understanding of the present application, the present embodiment will be described in detail below with reference to the accompanying drawings.

[0037] Embodiment 1: see Figure 2 、 Figure 3 A method for width adaptive learning using dynamic acquisition function, the method comprising the following steps:

[0038] In order to improve the precision of width control, the present method starts from the data acquisition on site, and the data of the width gauge is 100% collected and trusted. The redundant backup mode of the width measuring equipment is used to ensure that the data can be collected, the design of the dynamic acquisition mode is used to realize the accuracy of the data, and the control mode of the model itself is designed to ensure that the model can efficiently use the on-site data to improve the width in the width setting.

[0039] S1, dynamic beating width measurement function

[0040] S2, dynamic casting width measurement function

[0041] S3, width feedforward function of intermediate pass

[0042] Among them, S1, dynamic beating width measurement function:

[0043] The width of the hot-rolled strip remains substantially unchanged from the time the strip is extracted from the heating furnace to the time the strip is gripped by the rough rolling R1 rolling mill. The two side guide plate devices on site are used to measure the width by means of redundant backup before the descaling box and the R1 rack. The two tapping devices are separated by a descaling device, and when in use, the strip is tapped at different positions according to the device layout diagram. Figure 1 : On-site tapping device layout diagram

[0044] The specific implementation of the dynamic tapping width measurement function is as follows:

[0045] S11, the on-site tapping device two is mainly used. The reason is that the tapping device two is closer to the R1 rolling mill, that is, the inlet device of the width control, and the width loss caused by the temperature drop will be minimal, and the width data at the inlet will be more accurate;

[0046] S12, when the following conditions occur on site, switch to the tapping device one for tapping:

[0047] The tapping device two fails, and the data cannot be collected on site, so the tapping device one is switched;

[0048] The tapping device two produces more dirty data due to the influence of the descaling water between the racks, so the tapping device one is switched;

[0049] When the strip cannot be stopped at the position of the tapping device two during short material production, it is easy to rush into the R1 rack, causing equipment and production scrap, so the tapping device one is switched;

[0050] When the two tapping devices need to be checked, the tapping device one and the tapping device two need to be used at the same time;

[0051] In summary, the two tapping devices can be combined in the following ways:

[0052] The tapping device one and the tapping device two are used at the same time; the tapping device one is used + the tapping device two is not used; the tapping device two is used + the tapping device one is not used;

[0053] S13, the model needs to be designed accordingly according to the use and switching mode of the device.

[0054] When the tapping device one and the tapping device two are used at the same time, the model remembers the values of the two tapping methods, and when the deviation of the two values is within ±3 mm, the width of the tapping device two is used for the width setting of the model. When the deviation of the two values exceeds ±3 mm, the original width inlet data needs to be introduced, the tapped width and the original width data are compared, and when the deviation is within ±5, the smaller value of the tapping is used for the width setting of the model according to the control loss on site.

[0055] When the first beating device is used and the second beating device is not used, the actual value collected by the first beating device is directly used for control.

[0056] When the second beating device is used and the first beating device is not used, the actual value collected by the second beating device is directly used for control.

[0057] The two rough rolling mills basically meet the function of reversible rolling. The width meter between the stands can be used most effectively when the reversible rolling mill is produced, so the width meter between the stands plays a key role in the width control during the production process. It is understood that the width meter between the stands of other domestic steel plants can only be used in the last pass width calculation of each stand, and the use efficiency is not high. For width control, the thickness is thinned at the last pass, and the roll force of the vertical roll limits the correction range of the width, greatly affecting the width control efficiency.

[0058] One of the important features of the method is to make full use of the width meter in all vertical roll involved pass calculation, which improves the efficiency by 50% compared to before.

[0059] Due to the change of the environment temperature during hot rolling production, the water between the stands will appear mist, which affects the authenticity of the width meter sampling data, so in order to avoid the influence of the mist, the width meter position is slightly moved back. Figure 2 Width meter distribution between stands

[0060] Although this device distribution can avoid the influence of the mist, the width sampling length is too small, and the reliability is not high. When producing short materials, the collected data is even 0. To solve this contradiction, a dynamic width meter model is designed.

[0061] When the strip width is controlled, the head and tail of the strip length direction will appear width loss phenomenon, so when collecting the width, this area needs to be avoided, and the width of the middle position of the strip is collected as much as possible. Figure 3 Dynamic data collection process

[0062] S21, according to the detector and speed on site, track the tail position TailPos when the strip is forward, and calculate the length Len of the strip at each pass by using the equal volume method

[0063] S22, calculate the relative position SGPos of the width meter and the deviation position DeltaPos when the strip leaves the mill

[0064] S23, in order to ensure that the middle width of the strip can be detected, it is necessary to ensure that the middle position exceeds the width meter position, and at the same time, in order to save the production rhythm of the rolling line, the tail does not need to be detected, so the rolling mill casting position is dynamically calculated for different strips

[0065] S3, width feedforward function of intermediate pass

[0066] Because the hot rolling rough rolling area is adopted by reversible rolling, the model needs to be calculated frequently when rolling between passes. This algorithm is a function of self-correction of the model. At present, most of the hot rolling calculation, the input data are the data pre-calculated by the model, which has little effect on the improvement of the width on site. Through the data collection, screening, cleaning and other work in the early stage, the pass data is fully utilized, and the data is substituted into the calculation between passes through the model feedforward learning mode, so as to achieve the purpose of correction.

[0067] S31, according to the rolling strategy on site, configure the corresponding feedforward learning mode of the strategy. There are two reversible rolling mills on site, so there are many combination ways of the corresponding feedforward learning mode. When configuring the feedforward learning, because the reverse pass acquisition function is not opened, it is only opened in the forward direction.

[0068] S32, configure the feedforward learning confidence of different areas according to the situation on site. Because the quantity and quality of the sampling data of the width meter on site are different in different areas, different feedforward learning confidences need to be configured in different areas, ranging from 0 to 1.

[0069] S33, according to the final feedforward learning result, as the input of the reverse pass, the width model calculation between passes is carried out. Through the dog bone and horizontal spread, combined with the input of the feedforward learning, the output width of each pass is finally calculated.

[0070] Example 2:

[0071] 1422 hot rolling width double tapping function, one tapping before R1 descaling and one before R1 rolling mill. The two tapping devices can be switched at will or used at the same time, generally for short material, using R1 descaling tapping, to prevent the strip positioning from being inaccurate, causing the strip to be measured only at the head or the strip to be punched into the rack. The control process is as follows:

[0072] 1, the initial calculation of the model of the outlet width of each pass

[0073] ps obj outlet width calculated by the model 1 e1 1101.9 1 r1 1147 2 r1 1157.4 2 e1 1157.4 3 e1 1100.8 3 r1 1142.3 4 e2 1110.9 4 r2 1136.6 5 r2 1141.3 5 e2 1141.3 6 e2 1139 6 r2 1143.1

[0074] 2, the width deviation of each pass measured on site through dynamic measurement;

[0075]

[0076] Through the maximum deviation and the minimum deviation, it can be seen that the deviation of the data after dynamic measurement and cleaning and screening is reasonable, which can truly reflect the actual measurement width on site.

[0077] 3, the inlet width setting of each pass after feedback calculation;

[0078] ps obj inlet width calculated by the width gauge feedback 1 e1 1180.6 1 r1 1102.3 2 r1 1146.1 2 e1 1157.5 3 e1 1157.4 3 r1 1101.1 4 e2 1136.9 4 r2 1110.9 5 r2 1136.3 5 e2 1141.2 6 e2 1140.9 6 r2 1138.8 .

[0079] It should be noted that the above examples are not intended to limit the scope of the present application, and any equivalent variations or substitutions made on the basis of the above technical solutions fall within the scope of the claims of the present application.

Claims

1. A method for width adaptive learning using dynamic acquisition function, characterized in that, The method includes the following steps: S1. Dynamic tapping width measurement S2, Dynamic steel throwing for width measurement S3, width feedforward for intermediate passes; S1, dynamic tapping width measurement, is detailed below: S11. There are two tapping width measuring devices in front of R1 on site. Device 1 is located at the descaling box in front of R1 and Device 2 is located in front of the R1 stand. During on-site control, the tapping device 2, which is closer to the R1 stand, is the main device. The reason is that the tapping device 2 is closer to the R1 mill, which is the entrance device for width control. Its width loss caused by temperature drop will be the smallest, and the width data at the entrance will be more accurate. S12. When the following situations occur on site, switch to the first beating device for beating: The second beater device malfunctioned and could not collect data on site. Switch to beater device one. Due to the influence of the descaling water between the racks, the second beater device generated a lot of dirty data, so the beater device was switched to the first beater device. When producing shorter strips, the strip may not stop at position 2 of the tapping machine and may rush into the R1 frame, causing equipment damage and scrap steel. In this case, switch to tapping equipment 1. When two beating devices need to be calibrated, both beating device one and beating device two must be put into operation simultaneously. In summary, the two tapping devices can be combined in the following ways: Simultaneously, tapping equipment 1 and tapping equipment 2 were put into operation; The first tapping device is put into use, while the second tapping device is not. The second slapping device is put into operation, while the first slapping device is not in operation; S13. The model must be designed accordingly for the deployment and switching methods of the equipment. When both beater 1 and beater 2 are deployed simultaneously, the model remembers the values ​​of both beater methods. If the deviation between the two values ​​is within ±3mm, the width of beater 2 is used to set the model's width. If the deviation exceeds ±3mm, the original width input data needs to be introduced, and the width of the beater is compared with the original width data. If the deviation does not exceed ±5mm, the model width is set based on the smaller beater value, according to the control losses on site. When tapping device 1 is in operation and tapping device 2 is not in operation, the actual value collected by tapping device 1 is used directly for control. When tapping device 2 is in operation and tapping device 1 is not in operation, the actual value collected by tapping device 2 is used directly for control. S2. Dynamic steel throwing width measurement, details are as follows: S21. Based on the on-site detectors and speed, track the tail position TailPos of the strip when it is in the forward direction, and simultaneously calculate the length Len of the strip for each pass using the equal volume method. S22. Calculate the relative position SGPos of the width gauge and the deviation position DeltaPos when the strip leaves the rolling mill and is thrown. S23. To ensure that the width of the middle strip can be detected, it is necessary to ensure that the middle position exceeds the width gauge position. At the same time, to save the production speed of the rolling line, the tail end does not need to be detected. Therefore, the rolling mill discard position is dynamically calculated for different strips. S3, the width feedforward of the intermediate passes, is as follows: S31. Based on the on-site rolling strategy, configure the corresponding feedforward learning mode. S32. Configure the feedforward learning confidence level for different areas according to the on-site conditions. Since the quantity and quality of the width meter on-site sampling data vary greatly in different areas, it is necessary to configure different feedforward learning confidence levels for different areas, with a range of 0 to 1. S33. Based on the final feedforward learning result, as the input for the reverse path, the width model between paths is calculated. By combining the dog bone and horizontal width with the input of the feedforward learning, the output width of each path is finally calculated.

Images