Working roller thermal wear and thermal expansion control method for improving plate shape quality

Abstract

The invention belongs to the technical field of hot rolling, and particularly relates to a working roll thermal wear and thermal expansion control method for improving plate shape quality, which comprises the following steps: collecting hot rolling historical production data; preprocessing the hot rolling historical production data, and dividing the data into a training set and a test set; a working roll thermal wear and thermal expansion prediction model Inform-RF based on the combination of Inform and a random forest is constructed; inputting the training set into an Informer-RF (Radio Frequency) model for training; acquiring field production data, inputting the field production data into the trained Informer-RF model to obtain a thermal wear predicted value of the working roll, adjusting the roll shifting position according to a circulating roll shifting strategy, simultaneously obtaining a thermal expansion predicted value of the working roll, and adjusting the thermal expansion amount of the working roll according to a flow strategy of cooling water of the working roll; and the roll bending force is adjusted according to the adjusted roll shifting position, the working roll thermal expansion amount and the difference value between the actually-measured strip steel convexity and the target convexity.

Description

Technical Field

[0001] The invention relates to the technical field of hot rolling, and in particular to a method for controlling thermal wear and thermal expansion of a working roll to improve plate shape quality. Background Art

[0002] Hot-rolled strip is an important raw material for industrial manufacturing and is widely used in vehicle manufacturing, oil pipelines, infrastructure, transportation and other fields. With the advancement of science and technology and industrial upgrading, downstream customers have put forward higher and higher requirements on the plate quality of hot-rolled strip. According to research, the shifting and bending rolls of the rolling mill can effectively eliminate the plate defects such as middle waves and edge waves of hot-rolled strip, but they are unable to eliminate the plate defects such as rib waves, edge thinning, and "cat ear"-shaped local high points. The main reason is that the wear and thermal expansion of the working rolls are difficult to measure and the corresponding closed-loop control strategy. Therefore, a method for controlling the thermal wear and thermal expansion of the working rolls is urgently needed to improve the plate quality of hot-rolled strip. Summary of the invention

[0003] In order to overcome the problems of hot-rolled strip edge defects, local high points, and large roll consumption caused by uneven wear and thermal expansion of working rolls, the purpose of the present invention is to provide a method for controlling thermal wear and thermal expansion of working rolls to improve the quality of plate shape.

[0004] The present invention provides a method for controlling thermal wear and thermal expansion of a working roll for improving plate shape quality, comprising the following steps: Step 1: Collect historical hot rolling production data, including working roll parameters, rolled product parameters and process parameters; Step 2: Preprocess the hot rolling historical production data and divide it into training set and test set; Step 3: Construct the work roll thermal wear and thermal expansion prediction model Informer-RF which combines Informer and random forest; Step 4: Input the training set into the Informer-RF model for training to obtain the trained Informer-RF model; Step 5: Input the test set into the trained Informer-RF model to evaluate the prediction accuracy of the Informer-RF model; Step 6: Collect on-site production data and input it into the trained Informer-RF model to obtain the predicted value of thermal wear of the working roll, and adjust the roll shifting position according to the cyclic roll shifting strategy; Step 7: Collect on-site production data and input it into the trained Informer-RF model to obtain the predicted value of thermal expansion of the work roll, and adjust the thermal expansion of the work roll according to the flow strategy of the work roll cooling water; Step 8: The bending roll force is adjusted according to the shifting roll position adjusted in step 6, the thermal expansion of the working roll adjusted in step 7, and the difference between the measured strip convexity and the target convexity.

[0005] The method for controlling thermal wear and thermal expansion of a working roll for improving plate shape quality of the present invention has the following beneficial effects: The present invention establishes a prediction model for working roll wear and thermal expansion based on Informer-RF, which effectively improves the prediction accuracy of working roll wear and thermal expansion. At the same time, the present invention provides a circulating roll shifting strategy and a flow strategy for working roll cooling water based on the predicted values ​​of working roll wear and thermal expansion based on the working roll thermal wear and thermal expansion prediction model Informer-RF. Circulating roll shifting can even out roll wear, reduce strip shape defects such as edge waves and local high points, and extend the rolling mileage within the rolling cycle. The flow strategy for working roll cooling water can effectively compensate for the thermal expansion of the roll to adjust the high-order strip shape defects. Finally, the bending roll force is adjusted based on the plate shape deviation to achieve feedback control of the plate shape, which can effectively improve the dimensional accuracy of the strip thickness and convexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 It is a flow chart of a method for controlling thermal wear and thermal expansion of a working roll for improving plate shape quality according to the present invention; Figure 2 Schematic diagram of the work roll thermal wear and thermal expansion prediction model Informer-RF of the present invention; Figure 3a The comparison chart of the working roll thermal wear value of the F1 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3b It is a comparison chart of the working roll thermal wear value of the F2 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3c It is a comparison chart of the working roll thermal wear value of the F3 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3d The comparison chart of the working roll thermal wear value of the F4 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3e The comparison chart of the working roll thermal wear value of the F5 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3f The comparison chart of the working roll thermal wear value of the F6 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 3g The comparison chart of the working roll thermal wear value of the F7 stand of the finishing mill predicted by the Informer-RF model and the measured working roll thermal wear value; Figure 4a This is a comparison chart of the thermal expansion value of the working roll of the F1 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 4b It is a comparison chart of the working roll thermal expansion value of the F2 stand of the finishing mill predicted by the Informer-RF model and the measured seal roll thermal expansion value; Figure 4c This is a comparison chart of the thermal expansion value of the working roll of the F3 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 4d This is a comparison chart of the thermal expansion value of the working roll of the F4 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 4e This is a comparison chart of the thermal expansion value of the working roll of the F5 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 4f This is a comparison chart of the thermal expansion value of the working roll of the F6 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 4g This is a comparison chart of the thermal expansion value of the working roll of the F7 stand of the finishing mill predicted by the Informer-RF model and the measured thermal expansion value of the seal roll; Figure 5a A comparison of the mean absolute errors of work roll thermal wear prediction is shown; Figure 5b A comparison of the root mean square error of the work roll thermal wear prediction is shown; Figure 6a A comparison of the mean absolute errors of the prediction of thermal expansion of the work rolls is shown; Figure 6b A comparison of the root mean square error of the work roll thermal expansion prediction is shown. DETAILED DESCRIPTION

[0007] The present invention is further described below in conjunction with specific embodiments and drawings. In this embodiment, the working rolls of the F1 to F7 stands in the hot rolling finishing mill are taken as an example, and the parameters of the working rolls are shown in Table 1.

[0008] Table 1 Roller parameters of F1~F7 stands

[0009] like Figure 1 As shown, a method for controlling thermal wear and thermal expansion of a working roll to improve plate quality of the present invention specifically comprises the following steps: Step 1: Collect historical hot rolling production data, including working roll parameters, rolled product parameters and process parameters.

[0010] Roll parameters include: rolling mileage, working roll diameter, working roll hardness, working roll surface roughness, working roll thermal wear, and working roll thermal expansion.

[0011] The rolled product parameters include: strip inlet thickness, strip outlet thickness, strip width, strip inlet temperature, strip outlet temperature, and strip hardness.

[0012] The process parameters include: strip inlet speed, strip outlet speed, rolling force, working roll cooling water flow, and inter-stand cooling water flow.

[0013] Step 2: preprocessing the hot rolling historical production data and dividing it into a training set and a test set, said step 2 comprising: Step 2.1: Divide the collected hot rolling historical production data according to the rolling cycle, and convert the rolling kilometers into a time series.

[0014] Step 2.2: Use the 3σ principle to detect outliers in the historical production data of each rolling cycle and remove outliers.

[0015] Step 2.3: Use linear interpolation to fill in the missing data after removing missing values ​​and outliers.

[0016] Step 2.4: Perform Z-score standardization on the data. The calculation formula is as follows: ; in, x and x * They are original historical production data and standardized historical production data, and ρ are the mean and standard deviation of the original data, respectively.

[0017] Step 2.5: Divide the standardized historical production data into a training set and a test set.

[0018] Step 3: Construct the work roll thermal wear and thermal expansion prediction model Informer-RF combining Informer and random forest; the Informer model includes an encoder, a decoder, and a fully connected layer. Figure 2 As shown, step 3 includes: Step 3.1: Assume that the input data is a two-dimensional matrix X, the expression is as follows: ; In the formula, x i Indicates i input samples, one sample consists of a set of historical hot rolling production data. n represents the number of samples, d represents the feature dimension of the input sample, R Represents the set of real numbers.

[0019] Step 3.2: Encode the input data and map it to a uniform dimension through CNN. The calculation formula is as follows: ; In the formula, X input Represents the encoded input information. , and They represent feature coding, location information coding and time information coding respectively.

[0020] Step 3.3: Construct the encoder in Informer, which includes the ProbSparse self-attention mechanism, convolution, activation and pooling operations; The calculation formula of the ProbSparse self-attention mechanism is as follows: ; In the formula, Indicates input information X input The coefficient query matrix after sparsity measurement; K and V Respectively represent the input information X input The key matrix and value matrix extracted from ; is the weight matrix of the coefficient query matrix, W K is the weight matrix of the key matrix, W V is the weight matrix of the value matrix; Soft max() represents a normalization function.

[0021] The calculation formulas for the convolution, activation and pooling operations are as follows: ; In the formula, Indicates j The feature matrix of the layer through the ProbSparse self-attention mechanism, Convld () represents a one-dimensional convolution operation, ELU() is the activation function, Maxpool () represents the maximum pooling operation.

[0022] Step 3.4: Construct the decoder in Informer and input information X input The output of the encoder after the masked ProbSparse self-attention mechanism Perform the attention mechanism calculation together to get the output of the decoder .

[0023] Step 3.5: Construct the fully connected layer in Informer Input to the fully connected layer to get the output of the fully connected layer .

[0024] Step 3.6: Build a random forest and transform the encoder output And the output of the fully connected layer Connect them together as the input of random forest, build multiple decision trees according to the number of input samples and feature dimensions, perform feature selection and division on each decision tree, and obtain the prediction results of each decision tree; average or weighted average the prediction results of each decision tree to obtain the final output result The output results include the predicted values ​​of thermal wear of the working roll and the predicted values ​​of thermal expansion of the working roll. w i Indicates that according to i The predicted value of work roll thermal wear obtained by input samples, Δ E i Indicates that according to i The predicted value of the thermal expansion of the work roll is obtained by using 10 input samples.

[0025] Step 3.7: Construct the working roll thermal wear and thermal expansion prediction model Informer-RF which is a combination of Informer and random forest.

[0026] Step 4: Input the training set into the Informer-RF model for training to obtain a trained Informer-RF model. Step 4 includes: Step 4.1: Set the hyperparameters in the Informer-RF model, including the number of cycles, learning rate, optimizer, batch size, dropout, number of attention heads in Informer, output sequence length in Informer, start token length of decoder in Informer, prediction sequence length of decoder in Informer, number of trees in RF, and maximum depth allowed in RF. The detailed settings are shown in Table 2.

[0027] Table 2 Hyperparameters in the Informer-RF model

[0028]

[0029] Step 4.2: Establish a loss function, the expression of the loss function is: ; In the formula, Indicates i The predicted value of samples, Indicates i The measured value of a sample.

[0030] Step 4.3: Input the training set data into the Informer-RF model in batches for training until the maximum number of cycles is reached.

[0031] Step 5: Input the test set into the trained Informer-RF model to evaluate the prediction accuracy of the Informer-RF model.

[0032] In the specific implementation, the test set data is input into the trained Informer-RF model to obtain the predicted values ​​of working roll thermal wear and working roll thermal expansion, and the mean absolute error (MAE) and root mean square error (RMSE) are calculated with the corresponding true values ​​to evaluate the prediction accuracy of the Informer-RF model.

[0033] Figure 3a-3g The comparison chart of the working roll thermal wear values ​​of the F1-F7 stands of the finishing mill predicted by the Informer-RF model and the measured roll wear values ​​is shown. Figure 4a-4g The comparison chart of the work roll thermal expansion values ​​of the F1-F7 stands of the finishing mill predicted by the Informer-RF model and the measured roll thermal expansion values ​​is shown. It can be seen from the figure that the work roll thermal wear value and work roll thermal expansion value on the test set have a good fit with the corresponding measured values.

[0034] In order to evaluate the Informer-RF prediction model of the present invention, it is compared with the Informer, Transformer, and LSTM of the prior art. Figure 5a The mean absolute error comparison chart of work roll thermal wear prediction is shown. Figure 5b A comparison of the root mean square error of the work roll thermal wear prediction is shown. Figure 6a The mean absolute error comparison chart of the work roll thermal expansion prediction is shown. Figure 6bThe root mean square error comparison diagram of the working roll thermal expansion prediction is shown. It can be seen from the figure that compared with the prior art, the mean absolute error and root mean square error of the Informer-RF prediction model of the present invention are smaller, indicating that the Informer-RF prediction model has higher prediction accuracy in roll wear and thermal expansion and better generalization performance.

[0035] Step 6: Collect on-site production data and input it into the trained Informer-RF model to obtain the predicted value of thermal wear of the working roll, and adjust the roll shifting position according to the cyclic roll shifting strategy. The cyclic roll shifting strategy is: When the working roll thermal wear prediction value w satisfy: w When ≤100um, the roller shifting stroke is -150mm~150mm, and the roller shifting step length is 15mm; When the working roll thermal wear prediction value w Meet: 100um< w When ≤200um, the roller shifting stroke is -120mm~120mm, and the roller shifting step length is 12mm; When the working roll thermal wear prediction value w Meet: 200um< w When ≤300um, the roller shifting stroke is -80mm~80mm, and the roller shifting step length is 9mm; When the working roll thermal wear prediction value w satisfy: w When the diameter is >300um, the roller shifting stroke is -50mm~50mm, and the roller shifting step length is 5mm.

[0036] Step 7: Collect on-site production data and input it into the trained Informer-RF model to obtain the predicted value of thermal expansion of the working roll, and adjust the thermal expansion of the working roll according to the flow strategy of the working roll cooling water. The flow strategy of the working roll cooling water is: When the predicted value of thermal expansion of the working roll Δ E Satisfaction: Δ E When the pressure is less than 10um, close the cooling water nozzles on both sides of the working roll, and set the flow rate of the cooling water nozzle in the middle of the working roll to 536.51m 3 / h; When the predicted value of thermal expansion of the working roll Δ E Satisfy: 100um<Δ E When the pressure is ≤20um, close the cooling water nozzles on both sides of the working roll, and set the flow rate of the cooling water nozzle in the middle of the working roll to 554.16m 3 / h; When the predicted value of thermal expansion of the working roll Δ E Satisfy: 100um<Δ EWhen the diameter is ≤20um, the flow rate of the cooling water nozzles on both sides of the working roll is set to 447.15m 3 / h; the flow rate of the working roll intermediate cooling water nozzle is set to 599.25 m 3 / h.

[0037] Step 8: The bending roll force is adjusted according to the roll shifting position adjusted in step 6, the thermal expansion of the working roll adjusted in step 7, and the difference between the measured strip convexity and the target convexity. Step 8 is specifically as follows: Step 8.1: Collect the measured convexity of the strip at the exit of the finishing mill and calculate the difference Δ between it and the target convexity C : ; in, C mea is the measured convexity of the strip at the exit of the finishing mill. C tar Target convexity.

[0038] Step 8.2: Calculate the adjustment amount of the bending roll force according to the real-time production data; ; ; in, m is the number of stands of the finishing mill, Δ C j For the j The convexity adjustment of the rack is in μm; Δ F j For the j The adjustment amount of the frame bending roll force, in KN; W j For the j The strip steel width of the rack, in mm; For the j The efficiency factor of the strip width of the rack, in μm / mm; P j For the j The rolling force of the stand, in kN; For the j The efficiency coefficient of the rolling force of the stand, in μm / kN; S j For the j The roller shifting amount of the frame, in mm; For the j The efficiency coefficient of the roll shifting of the frame, in μm / mm; For the j The entrance convexity of the rack, in μm; For the jThe efficiency coefficient of the rack inlet convexity; T j For the j The inlet temperature of the rack, in °C; For the j The efficiency factor of the rack inlet temperature; H j For the j The inlet thickness of the rack, in mm; For the j The efficiency coefficient of the rack inlet thickness, in μm / mm; w j For the j Thermal wear of the working roll of the stand, in μm; For the j Efficiency coefficient of thermal wear of working roll of the stand; Δ E j For the j Thermal expansion of the working roll of the frame, in μm; For the j Efficiency coefficient of thermal expansion of the working rolls of the frame.

[0039] The method for controlling thermal wear and thermal expansion of working rolls for improving plate shape quality of the present invention was applied to the production line of a hot rolling mill, and the convexity and thickness results of the produced strip steel are shown in Tables 3 and 4; it can be seen from the hit rate of the embodiments that the method of the present invention can improve the strip steel thickness and plate shape indicators.

[0040] Table 3 Convexity hit rate statistics

[0041]

[0042] Table 4 Thickness hit rate statistics

[0043]

[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the concept of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for controlling thermal wear and thermal expansion of a work roll to improve plate quality, characterized in that: The following steps are involved: Step 1: Collect historical hot rolling production data, including working roll parameters, rolled product parameters and process parameters; Step 2: Preprocess the hot rolling historical production data and divide it into training set and test set; Step 3: Construct the work roll thermal wear and thermal expansion prediction model Informer-RF which combines Informer and random forest; Step 4: Input the training set into the Informer-RF prediction model for training to obtain the trained Informer-RF model; Step 5: Input the test set into the trained Informer-RF prediction model to evaluate the prediction accuracy of the Informer-RF prediction model; Step 6: Collect on-site production data and input it into the trained Informer-RF prediction model to obtain the predicted value of thermal wear of the working roll, and adjust the roll shifting position according to the cyclic roll shifting strategy; Step 7: Collect on-site production data and input it into the trained Informer-RF prediction model to obtain the predicted value of thermal expansion of the work roll, and adjust the thermal expansion of the work roll according to the flow strategy of the work roll cooling water; Step 8: The bending roll force is adjusted according to the shifting roll position adjusted in step 6, the thermal expansion of the working roll adjusted in step 7, and the difference between the measured strip convexity and the target convexity.

2. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 1, characterized in that: Roll parameters include: rolling mileage, working roll diameter, working roll hardness, working roll surface roughness, working roll thermal wear, working roll thermal expansion; The rolled product parameters include: strip inlet thickness, strip outlet thickness, strip width, strip inlet temperature, strip outlet temperature, and strip hardness; The process parameters include: strip inlet speed, strip outlet speed, rolling force, working roll cooling water flow, and inter-stand cooling water flow.

3. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 1, characterized in that: The step 2 comprises: Step 2.1: Divide the collected hot rolling historical production data according to the rolling cycle, and convert the rolling kilometers into a time series; Step 2.2: Use the 3σ principle to detect outliers in the historical production data of each rolling cycle and remove outliers; Step 2.3: Use linear interpolation to fill in the missing data after removing missing values ​​and outliers; Step 2.4: Perform Z-score standardization on the data. The calculation formula is as follows: ； in, x and x * They are original historical production data and standardized historical production data, and ρ are the mean and standard deviation of the original data respectively; Step 2.5: Divide the standardized historical production data into a training set and a test set.

4. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 1, characterized in that: The step 3 comprises: Step 3.1: Assume that the input data is a two-dimensional matrix X , the expression is as follows: ； In the formula, x i Indicates i input samples, one sample consists of a set of historical hot rolling production data. n represents the number of samples, d represents the feature dimension of the input sample, R represents the set of real numbers; Step 3.2: Encode the input data and map it to a uniform dimension through CNN. The calculation formula is as follows: ； In the formula, X input Represents the encoded input information. , and They represent feature coding, location information coding and time information coding respectively; Step 3.3: Construct the encoder in Informer, which includes the ProbSparse self-attention mechanism, convolution, activation and pooling operations; The calculation formula of the ProbSparse self-attention mechanism is as follows: ； In the formula, Indicates input information X input The coefficient query matrix after sparsity measurement; K and V Respectively represent the input information X input The key matrix and value matrix extracted from ; is the weight matrix of the coefficient query matrix, W K is the weight matrix of the key matrix, W V is the weight matrix of the value matrix; Soft max() represents the normalization function; The calculation formulas for the convolution, activation and pooling operations are as follows: ； In the formula, Indicates j The feature matrix of the layer through the ProbSparse self-attention mechanism, Convld () represents a one-dimensional convolution operation, ELU () is the activation function, Maxpool () represents the maximum pooling operation; Step 3.4: Construct the decoder in Informer and input information X input The output of the encoder after the masked ProbSparse self-attention mechanism Perform the attention mechanism calculation together to get the output of the decoder ; Step 3.5: Construct the fully connected layer in Informer Input to the fully connected layer to get the output of the fully connected layer ; Step 3.6: Build a random forest and transform the encoder output And the output of the fully connected layer Connect them together as the input of random forest, build multiple decision trees according to the number of input samples and feature dimensions, perform feature selection and division on each decision tree, and obtain the prediction results of each decision tree; average or weighted average the prediction results of each decision tree to obtain the final output result The output results include the predicted values ​​of thermal wear of the working roll and the predicted values ​​of thermal expansion of the working roll. w i Indicates that according to i The predicted value of work roll thermal wear obtained by input samples, Δ E i Indicates that according to i The predicted value of thermal expansion of the work roll obtained by the input samples; Step 3.7: Construct the working roll thermal wear and thermal expansion prediction model Informer-RF which is a combination of Informer and random forest.

5. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 4, characterized in that: The step 4 comprises: Step 4.1: Set the hyperparameters in the Informer-RF prediction model, including the number of cycles, learning rate, optimizer, batch size, dropout, number of attention heads in Informer, output sequence length in Informer, start token length of decoder in Informer, prediction sequence length of decoder in Informer, number of trees in RF, and maximum depth allowed in RF; Step 4.2: Establish a loss function, the expression of the loss function is: ； In the formula, Indicates i The predicted value of samples, Indicates i The measured values ​​of samples; Step 4.3: Input the training set data into the Informer-RF prediction model in batches for training until the maximum number of cycles is reached.

6. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 1, characterized in that: The cyclic roller shifting strategy of step 6 is: When the working roll thermal wear prediction value w satisfy: w When ≤100um, the roller shifting stroke is -150mm~150mm, and the roller shifting step length is 15mm; When the working roll thermal wear prediction value w Meet: 100um< w When ≤200um, the roller shifting stroke is -120mm~120mm, and the roller shifting step length is 12mm; When the working roll thermal wear prediction value w Meet: 200um< w When ≤300um, the roller shifting stroke is -80mm~80mm, and the roller shifting step length is 9mm; When the working roll thermal wear prediction value w satisfy: w When the diameter is >300um, the roller shifting stroke is -50mm~50mm, and the roller shifting step length is 5mm.

7. The method for controlling thermal wear and thermal expansion of a work roll for improving the plate shape quality according to claim 1, characterized in that: The flow strategy of the working roll cooling water in step 7 is: When the predicted value of thermal expansion of the working roll Δ E Satisfaction: Δ E When the pressure is ≤10um, close the cooling water nozzles on both sides of the working roll, and set the flow rate of the cooling water nozzle in the middle of the working roll to 536.51m 3 / h; When the predicted value of thermal expansion of the working roll Δ E Satisfy: 100um<Δ E When the pressure is ≤20um, close the cooling water nozzles on both sides of the working roll, and set the flow rate of the cooling water nozzle in the middle of the working roll to 554.16m 3 / h; When the predicted value of thermal expansion of the working roll Δ E Satisfy: 100um<Δ E When the diameter is ≤20um, the flow rate of the cooling water nozzles on both sides of the working roll is set to 447.15m 3 / h; the flow rate of the working roll intermediate cooling water nozzle is set to 599.25 m 3 / h.

8. The method for controlling thermal wear and thermal expansion of a work roll for improving the quality of plate shape according to claim 1, characterized in that: The step 8 is specifically as follows: Step 8.1: Collect the measured convexity of the strip at the exit of the finishing mill and calculate the difference Δ between it and the target convexity C : ； in, C mea is the measured convexity of the strip at the exit of the finishing mill. C tar Target convexity; Step 8.2: Calculate the adjustment amount of the bending roll force according to the real-time production data; ； ； in, m is the number of stands of the finishing mill, Δ C j For the j The convexity adjustment of the rack is in μm; Δ F j For the j The adjustment amount of the frame bending roll force, in KN; W j For the j The strip steel width of the rack, in mm; For the j The efficiency factor of the strip width of the rack, in μm / mm; P j For the j The rolling force of the stand, in kN; For the j The efficiency coefficient of the rolling force of the stand, in μm / kN; S j For the j The roller shifting amount of the frame, in mm; For the j The efficiency coefficient of the roll shifting of the frame, in μm / mm; For the j The entrance convexity of the rack, in μm; For the j The efficiency coefficient of the rack inlet convexity; T j For the j The inlet temperature of the rack, in °C; For the j The efficiency factor of the rack inlet temperature; H j For the j The inlet thickness of the rack, in mm; For the j The efficiency coefficient of the rack inlet thickness, in μm / mm; w j For the j Thermal wear of the working roll of the stand, in μm; For the j Efficiency coefficient of thermal wear of working roll of the stand; Δ E j For the j Thermal expansion of the working roll of the frame, in μm; For the j Efficiency coefficient of thermal expansion of the working rolls of the frame.

Images