Composite plate hot rolling speed-deformation-pressure coupling control system and method

Abstract

The invention discloses a composite plate hot rolling speed-deformation-pressure coupling control system and method, and the method comprises the steps: building a time-space sequence prediction model through employing a long-short-term memory network based on the hot rolling data of a multi-layer composite plate, predicting the rolling speed, rolling force and deformation, and obtaining the deformation of the multi-layer composite plate according to the deviation between the prediction quality and the target quality. And carrying out feed-forward correction on the roll gap by adopting a multi-target balance optimization algorithm. On the basis of feed-forward correction, the residual error between the actual thickness value and the set outlet thickness is further considered, a roll gap feedback adjustment strategy is designed, and composite plate hot rolling speed-deformation-pressure coupling control is achieved. The composite plate hot rolling speed-deformation-pressure coupling control method is high in control precision, the defects of a traditional method are overcome, speed-deformation-rolling force dynamic coupling control is improved, and the method can be widely applied to multi-layer composite plate hot rolling.

Description

Technical Field

[0001] The present invention belongs to the technical field of hot rolling processing of metal materials, and specifically relates to a composite plate hot rolling speed-deformation-pressure coupling control system and method, which is suitable for improving the precision and stability of hot rolling forming of multi-layer composite plates. Background Art

[0002] Hot rolling of composite plates is a core technology that achieves metallurgical bonding between dissimilar metal layers of substrate and composite materials through hot rolling process. The thermal expansion coefficient, deformation resistance and other material properties of the substrate and composite materials are significantly different. The speed-deformation-pressure coupling control of hot rolling of composite plates is effectively achieved, which can ensure that the substrate and composite materials extend synchronously during rolling, avoiding quality problems such as warping, edge cracking or uneven thickness.

[0003] At present, initial progress has been made in the hot rolling process control of multi-layer composite plates. "CN113275381A" proposed a hot rolling composite method for preparing metal composite plates by using large plastic deformation; "CN113627469A" proposed a method for predicting the convexity of hot-rolled strip steel based on fuzzy reasoning algorithm to solve the technical problem of low temperature control accuracy of existing hot-rolled strip steel; "CN116371941B" proposed a method, device and electronic equipment for predicting the rolling force and thickness of each layer of metal composite plates, which determines the outlet thickness of soft metal slab and the outlet thickness of hard metal slab based on the reduction rate of soft metal slab, calculates the rolling force of soft metal plate and hard metal plate respectively, and takes the average rolling force of the two as the target rolling force. force; "CN115041529B" proposed a monitoring method, device, medium and electronic equipment for the hot rolling thickness control model. The self-learning coefficient generated by the self-learning module in the hot rolling thickness control model is used to monitor whether there is a problem with the hot rolling thickness control model, and the hot rolling thickness control model can be adjusted in time; "CN112845613A" proposed a thickness control method, device and terminal equipment for hot rolled strip steel. According to the characteristics of the raw materials and the rolling system during the rolling process, the rolling signal is optimized, and the eccentricity compensation is used to make the rolling signal avoid the influence of the roll eccentricity. The tail compensation is used to eliminate the thickness jump phenomenon due to loss of tension, which can effectively reduce the fluctuation of the thickness of the finished steel strip.

[0004] While the control methods for hot-rolling composite plates described in the aforementioned research have achieved some success, they still have drawbacks. Existing control methods often rely on independent control of a single parameter, lacking dynamic modeling of the coupled relationship between velocity, deformation, and rolling force. This can lead to defects such as interlayer misalignment, uneven thickness, and excessive residual stress during rolling. Furthermore, existing control methods exhibit a lag in responding to real-time operating conditions, making high-precision closed-loop control impossible. Summary of the Invention

[0005] The purpose of the present invention is to provide a composite plate hot rolling speed-deformation-pressure coupling control system and method, so as to respond to changes in working conditions in real time and achieve high-precision closed-loop control.

[0006] The object of the present invention is achieved by the following technical means: a composite plate hot rolling speed-deformation-pressure coupling control method, comprising the following steps: Step 1: Collect the hot-rolling production process data and product quality data of multi-layer composite plates to construct the original data set; Step 2: Preprocess the original data set; Step 3: Based on the preprocessed original data set, a convolutional-long short-term memory network spatiotemporal sequence prediction model is established to predict the outlet thickness; Step 4: Based on the prediction results of the prediction model, adopt the corresponding feedforward control strategy to correct the roll gap parameters; Step 5: Based on the actual thickness of the multi-layer composite plate during hot rolling, a feedback control strategy is adopted to adjust the roll gap; Step 6: Input the multi-layer composite plate hot rolling production process data and product quality data into the convolution-long short-term memory network spatiotemporal sequence prediction model to predict the outlet thickness. If the predicted result does not match the actual required outlet thickness, adjust the roll gap parameters and predict again until the predicted result matches the actual required outlet thickness. The convolution-long short-term memory network spatiotemporal sequence prediction model is used to predict the roller gap parameters used for production, and the feedback control strategy of step 5 is adopted during the production process to adjust the roller gap in real time.

[0007] The multi-layer composite plate hot rolling production process data and product quality data in step 1 specifically include: inlet thickness, set outlet thickness, deformation, reduction rate, inlet width, rolling temperature, rolling speed, rolling force, and outlet thickness.

[0008] The step 2 specifically includes the following steps: Step 2.1: Use the K-nearest neighbor interpolation algorithm to fill in the missing values ​​in the original data set. The formula is as follows:

[0009]

[0010]

[0011] Where, For the Samples to distance, It is The coordinates of the samples, For the The value of the sample, are the coordinates of the missing values ​​in the sample, is the missing value of the sample, For the The weight of the nearest neighbor points; Step 2.2: Use the Savitzky-Golay filtering method to filter the padded data to obtain the denoised data. The formula is as follows:

[0012]

[0013]

[0014] Where, is the unilateral degree of fitting, It's about of polynomial of order, , which is the integer index within the window, are the coefficients of the polynomial, , For index The value of the input data, is the sum of squares of the residuals between the fitted data points and the original data points, is the data after noise reduction; Step 2.3: Use maximum and minimum scaling to normalize the denoised data to obtain standardized data. The formula is as follows:

[0015] in represents the data after noise reduction, represents the standardized data, and Represents the maximum and minimum values ​​of the original data set after denoising; Step 2.4: Extract 80% of the standardized data as training data to build the prediction model, and the remaining data as test data to evaluate the model performance.

[0016] The step 3 specifically includes the following steps: Step 3.1: Build a convolutional-long short-term memory network spatiotemporal sequence pre-training model. The principle formula is as follows:

[0017] Where, Indicates the The weight corresponding to the convolution kernel position, Indicates the The bias corresponding to the convolution kernel of the position, represents the feature map of the previous layer, represents the feature map of the current layer, represents the activation function, represents a set of feature maps,

[0018]

[0019] Where, is the pooling result, is the pixel value of the input feature, and are the row and column index values ​​of the features, and are the row and column sizes of the pooling window,

[0020]

[0021]

[0022]

[0023] Where, is the weight matrix representing the input gate, is the bias of the input gate, is the sigmoid function, is the weight matrix of the forget gate, Indicates that the vector and vector Concatenate into a new vector, refers to the hidden state at time step t-1, refers to the input at time step t, is the bias of the forget gate, is the weight matrix of the output gate, is the bias of the output gate, is the temporary hidden state value, is the weight matrix when using the forget gate for feature screening, It is the bias when using forget gate for feature screening; Step 3.2: Use the inlet thickness, set outlet thickness, deformation, reduction ratio, inlet width, rolling temperature, rolling speed, and rolling force as model inputs, and the outlet thickness as model output; Step 3.3: Use grid search strategy and cross-validation method to determine the optimal hyperparameters of the spatiotemporal series prediction model to ensure the model has the best prediction performance; Step 3.4: Use the above model hyperparameters to build a quality prediction model for hot-rolled multi-layer composite plates.

[0024] The step 4 specifically includes the following steps: Step 4.1: Based on the deviation between the predicted outlet thickness and the target outlet thickness, a multi-objective balance optimization algorithm is used to perform feedforward correction on the roll gap; Step 4.2: Determine whether the adjustment amount of the last rolling parameter exceeds the limit; if it exceeds the limit, repeat step 4.1; if it does not exceed the limit, use the corrected roll gap for rolling.

[0025] The feedback control strategy in step 5 specifically includes the following steps: Step 5.1: Calculate the residual between the actual value of the composite plate thickness at the current moment and the set outlet thickness, and calculate the roller gap feedback adjustment amount at the next moment based on the residual value at the current moment; Step 5.2: Determine whether the roll gap feedback adjustment amount exceeds the limit; if it exceeds the limit, repeat step 5.1; if it does not exceed the limit, add the feedforward correction roll gap to the roll gap feedback adjustment amount and perform rolling.

[0026] A composite plate hot rolling speed-deformation-pressure coupling control system, comprising: The export thickness prediction system is used to predict the export thickness of the finished multi-layer composite board based on the production and process parameters. When the predicted export thickness does not meet the set export thickness requirements, the production and process parameters are modified until the predicted export thickness matches the set export thickness. Roll gap parameter feedback control system and feedback control strategy perform feedback control on the roll gap parameters of production equipment to ensure that the actual production results meet the requirements.

[0027] The beneficial effects of the present invention are as follows: for the constant roll gap control of hot rolling of multi-layer composite plates, a "feedforward plus feedback" dual-loop control is proposed. Based on the hot rolling data of multi-layer composite plates, a long short-term memory network is used to establish a spatiotemporal sequence prediction model to predict the rolling speed, rolling force and deformation, and according to the deviation between the predicted quality and the target quality, a multi-objective balance optimization algorithm is used to perform feedforward correction on the roll gap. On the basis of the feedforward correction, the residual between the actual thickness value and the set outlet thickness is further considered, and a roll gap feedback adjustment strategy is designed to realize the composite plate hot rolling speed-deformation-pressure coupling control. The composite plate hot rolling speed-deformation-pressure coupling control method proposed by the present invention has high control accuracy, makes up for the defects of the traditional method, improves the speed-deformation-rolling force dynamic coupling control, and can be widely used in the hot rolling of multi-layer composite plates. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1This is a flow chart of the composite plate hot rolling speed-deformation-pressure coupling control method of the present invention; Figure 2 This is a performance comparison chart of an embodiment of the present invention with a single prediction model; Figure 3 This is a comparison diagram before and after hot rolling force control in an embodiment of the present invention; The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0029] A composite plate hot rolling speed-deformation-pressure coupling control method comprises the following steps: Step 1: Collect the hot-rolling production process data and product quality data of multi-layer composite plates to construct the original data set; The multi-layer composite plate hot rolling production process data and product quality data described in step 1 specifically include: entry thickness, set exit thickness, deformation, reduction ratio, entry width, rolling temperature, rolling speed, rolling force, and exit thickness. The set exit thickness and exit thickness are the final product quality data; the remaining data are production process data.

[0030] Use historical production process data and product quality data for subsequent prediction model training.

[0031] Step 2: Preprocess the original data set; The step 2 specifically includes the following steps: Step 2.1: Use the K-nearest neighbor interpolation algorithm to fill in the missing values ​​in the original data set. The formula is as follows:

[0032]

[0033]

[0034] Where, For the Samples to distance, It is The coordinates of the samples, For the The value of the sample, are the coordinates of the missing values ​​in the sample, is the missing value of the sample, For the The weight of the nearest neighbor points; Step 2.2: Use the Savitzky-Golay filtering method to filter the padded data to obtain the denoised data. The formula is as follows:

[0035]

[0036]

[0037] Where, is the unilateral degree of fitting, It's about of polynomial of order, , which is the integer index within the window, are the coefficients of the polynomial, , For index The value of the input data (that is, the data after filling in step 2.1), is the sum of squares of the residuals between the fitted data points and the original data points, is the data after noise reduction; Step 2.3: Use maximum and minimum scaling to normalize the denoised data to obtain standardized data. The formula is as follows:

[0038] in represents the data after noise reduction, represents the standardized data, and Represents the maximum and minimum values ​​of the original data set after denoising; Step 2.4: Extract 80% of the standardized data as training data to build the prediction model, and the remaining data as test data to evaluate the model performance.

[0039] In this example, a hot-rolled strip steel dataset was constructed, as shown in Table 1. Missing values ​​were filled and normalized. 80% of the data were randomly extracted as training data, and the remaining data were used as test data to evaluate the model performance. Table 1

[0040] The parameters omitted in the columns of Table 1 are the set exit thickness, deformation, reduction rate, entry width, and rolling temperature. Since they are not used in the subsequent feedforward process, they are omitted in Table 1. The parameters omitted in the rows are data sets with different numbers, which are omitted due to the large number.

[0041] Step 3: Based on the preprocessed original data set, a convolutional-long short-term memory network spatiotemporal sequence prediction model is established to predict the outlet thickness; The step 3 specifically includes the following steps: Step 3.1: Build a CNN-LSTM pre-training model (i.e., a convolutional-long short-term memory network spatiotemporal sequence prediction model). The principle formula is as follows:

[0042] Where, Indicates the The weight corresponding to the convolution kernel position, Indicates the The bias corresponding to the convolution kernel of the position, represents the feature map of the previous layer, represents the feature map of the current layer, represents the activation function, represents a set of feature maps,

[0043]

[0044] Where, is the pooling result, is the pixel value of the input feature, and are the row and column index values ​​of the features, and are the row and column sizes of the pooling window,

[0045]

[0046]

[0047]

[0048] Where, is the weight matrix representing the input gate, is the bias of the input gate, is the sigmoid function, is the weight matrix of the forget gate, Indicates that the vector and vector Concatenate into a new vector, refers to the hidden state at time step t-1, refers to the input at time step t, is the bias of the forget gate, is the weight matrix of the output gate, is the bias of the output gate, is the temporary hidden state value, is the weight matrix when using the forget gate for feature screening, It is the bias when using forget gate for feature screening; Step 3.2: Use the inlet thickness, set outlet thickness, deformation, reduction ratio, inlet width, rolling temperature, rolling speed, and rolling force as model inputs, and the outlet thickness as model output; Step 3.3: Use the grid search strategy and cross-validation method to determine the optimal hyperparameters of the spatiotemporal series prediction model to ensure that the model has the best prediction performance; when the results of the grid search strategy and cross-validation method meet the design requirements, the corresponding hyperparameters are the optimal hyperparameters.

[0049] Step 3.4: Use the above model hyperparameters to build a quality prediction model for hot-rolled multi-layer composite plates.

[0050] The optimal hyperparameters are introduced into the CNN-LSTM pre-trained model to obtain the prediction model used for subsequent predictions.

[0051] In this embodiment, R 2 The fusion model, i.e., the convolution-long short-term memory network spatiotemporal sequence prediction model in step 3, is evaluated using the RMSE indicator. The formula is as follows:

[0052] In the formula is the coefficient of determination, is the measured value, is the predicted value, is the number of samples, is the average value of the measured variable. i is the symbol for the sample index in mathematical summation.

[0053]

[0054] In the formula is the root mean square error, is the measured value, is the predicted value, is the number of measured values. i is the symbol for the sample index in mathematical summation.

[0055] In this embodiment, for the data set (the original data set constructed by collecting the hot rolling production process data and product quality data of the multi-layer composite plate in step 1), the cross-validation method and grid search strategy are used to determine the main hyperparameters of all candidate models in the model candidate pool, as shown in Table 2.

[0056] Table 2

[0057] In this embodiment, the established CNN-LSTM spatiotemporal sequence model is used to test the real-time collected data set (the original data set in step 1), and is tested and compared with other spatiotemporal sequence models. The performance comparison of the best model corresponding to the data is shown in Table 3. Figure 2 As shown, the final prediction result R of CNN-LSTM is obtained 2 The value is 0.978, and the RMSE value is 2.933. Compared with the results of directly testing other spatiotemporal prediction strategy models using the original dataset, it is found that the CNN-LSTM model has better performance.

[0058] Table 3

[0059] Step 4: Based on the prediction results of the prediction model, adopt the corresponding feedforward control strategy to correct the roll gap parameters; the roll gap parameters specifically refer to the gap height between the two rolls of the rolling equipment, which is the control variable of the rolling process. By setting the roll gap, the thickness of the finished product, that is, the outlet thickness of the composite plate, is controlled.

[0060] The step 4 specifically includes the following steps: Step 4.1: Based on the deviation between the predicted outlet thickness and the target outlet thickness, a multi-objective balance optimization algorithm is used to perform feedforward correction on the roll gap;

[0061]

[0062]

[0063]

[0064] Where, To predict thickness deviation, is the predicted thickness of the convolution-long short-term memory network spatiotemporal sequence prediction model, is the target thickness, is the roll gap feedforward correction value of the kth iteration, is a multi-objective balance optimization function, is the roller gap adjustment constraint, is a set of process constraints, including rolling force, temperature, tension and other process constraints. is the feedforward correction roll gap after the kth iteration, is the initial roll gap, is the thickness prediction result after k iterations, It is a quality prediction model for hot-rolled multi-layer composite plates (i.e., convolution-long short-term memory network spatiotemporal sequence prediction model).

[0065] Step 4.2: Determine whether the adjustment amount of the last-pass rolling parameters (motor speed, roll gap opening, hydraulic pressure) exceeds the limit; if it exceeds the limit, repeat step 4.1; if it does not exceed the limit, use the corrected roll gap for rolling.

[0066]

[0067] Where, The thickness tolerance is the absolute value of the difference between the predicted thickness and the target thickness must be less than or equal to this threshold, that is, it does not exceed the limit.

[0068] Before actual production, the exit thickness of the multi-layer composite plate hot-rolled product is predicted based on the expected process parameters and other data using the convolution-long short-term memory network spatiotemporal sequence prediction model in step 3 to determine whether the predicted result meets the required exit thickness. The process parameters are the multi-layer composite plate hot-rolled production process data and product quality data in step 1, excluding the exit thickness, and specifically include: entry thickness, target exit thickness, deformation, reduction ratio, entry width, rolling temperature, rolling speed, and rolling force.

[0069] If it meets the requirements, actual production will begin. If it does not meet the requirements, the roll gap size will be modified through the multi-objective balance optimization algorithm, and then the prediction will be carried out again until the prediction result meets the requirements.

[0070] Step 5: Based on the actual thickness of the multi-layer composite plate during hot rolling, a feedback control strategy is adopted to perform feedback adjustment on the roll gap.

[0071] The feedback control strategy in step 5 specifically includes the following steps: Step 5.1: Calculate the residual between the actual thickness of the composite plate at the current moment and the set outlet thickness, and calculate the roller gap feedback adjustment amount at the next moment based on the residual value at the current moment;

[0072] Where, for The thickness residual at the moment, for The actual thickness of the moment.

[0073] The roll gap feedback adjustment amount is converted according to its own rigidity coefficient so that the converted value is equal to the residual value at the current moment.

[0074]

[0075] Where, For the next moment ( ) of the roller gap feedback adjustment amount, is the rigidity coefficient of the rolling mill.

[0076] The roll gap feedback adjustment amount is converted according to its own rigidity coefficient so that the converted value is equal to the residual value at the current moment.

[0077] For example, if the residual value is 0.01m overthick, after conversion based on the rigidity coefficient, the roller gap is reduced by 0.005m, which can reduce the thickness of the multi-layer composite plate by 0.01m.

[0078] Step 5.2: Determine whether the roll gap feedback adjustment amount exceeds the limit; if it exceeds the limit, repeat step 5.1; if it does not exceed the limit, add the feedforward correction roll gap to the roll gap feedback adjustment amount and perform rolling.

[0079]

[0080]

[0081] Where, is the absolute value of the maximum adjustment amount allowed by feedback control, for The final roll gap setting value at the moment, for Feedforward correction of roller gap value at the moment.

[0082] Taking the first sample of the original data set as an example, the predicted strip thickness is 3.43mm and the target thickness is 3.40mm. Based on the feedback of the deviation between the target thickness and the predicted thickness, the multi-objective balance optimization algorithm is used to perform feedforward control on the roll gap of each pass. The roll gap adjustment amount is calculated based on the deviation of the previous sample to obtain the final roll gap value of each pass, and then the corrected rolling force of each pass is calculated as follows: Figure 3 As shown, the adjustment limit is not exceeded. The output thickness after control is 3.41m, and the residual difference from the target thickness is only 0.01mm, achieving good strip thickness control effect, indicating that the present invention can accurately and effectively realize intelligent control of multi-layer composite plate hot-rolled products.

[0083] Step 6: Input the multi-layer composite plate hot rolling production process data and product quality data into the convolution-long short-term memory network spatiotemporal sequence prediction model to predict the outlet thickness. If the predicted result does not match the actual required outlet thickness, adjust the roll gap parameters and predict again until the predicted result matches the actual required outlet thickness. The convolution-long short-term memory network spatiotemporal sequence prediction model is used to predict the roller gap parameters used for production, and the feedback control strategy of step 5 is adopted during the production process to adjust the roller gap in real time.

[0084] After inputting the parameters in step 1 into the convolution-long short-term memory network spatiotemporal sequence prediction model for preliminary prediction to ensure that the final outlet thickness can theoretically meet the design requirements, the roller gap parameters are corrected through step 4 so that the final actual outlet thickness meets the design requirements. Then, the corrected roll gap parameters and other preset process parameters (i.e., the parameters in step 1, as well as the hydraulic pressure and motor speed) are used for production. During production, the feedback control strategy of step 5 is used to adjust the roll gap in real time to obtain the required multi-layer composite plate outlet thickness.

[0085] A composite plate hot rolling speed-deformation-pressure coupling control system, comprising: The export thickness prediction system is used to predict the export thickness of the finished multi-layer composite board based on the production and process parameters. When the predicted export thickness does not meet the set export thickness requirements, the production and process parameters are modified until the predicted export thickness matches the set export thickness. Production and process parameters, namely production process data and product quality data other than outlet thickness, specifically include: inlet thickness, set outlet thickness, deformation, reduction rate, inlet width, rolling temperature, rolling speed, and rolling force.

[0086] The roll gap parameter feedback control system and feedback control strategy provide feedback control of the roll gap parameters of the production equipment to ensure that the actual production results meet the requirements. The production results meet the requirements, which means that the actual composite plate outlet thickness meets the set outlet thickness requirements.

[0087] like Figure 1 As shown in the figure, the original set values, i.e. the pre-set production parameters and process parameters (i.e. inlet thickness, set outlet thickness, deformation, reduction rate, inlet width, rolling temperature, rolling speed and rolling force), are obtained through the prediction model to obtain the predicted results, i.e. the theoretically predicted outlet thickness; The predicted outlet thickness result is compared with the set outlet thickness to be achieved. According to the feedforward control strategy in step 4 of the method, the production and process parameters are feedforward corrected so that the predicted outlet thickness is consistent with the set outlet thickness. In other words, a composite plate with qualified thickness can be produced in theory.

[0088] Use the corrected parameters for production, detect the actual process parameters such as speed, deformation, rolling force (where speed corresponds to motor speed, deformation corresponds to roll gap opening, and rolling force corresponds to hydraulic pressure), and compare them with the target parameters. According to the feedback control strategy of step 5 of the method, feedback correction is performed on the roll gap opening, motor speed (motor speed is used to control the rolling speed), and hydraulic pressure.

[0089] Finally, a multi-layer composite board that meets the requirements is obtained.

Claims

1. A composite plate hot rolling speed-deformation-pressure coupling control method, characterized in that: The following steps are involved: Step 1: Collect the hot-rolling production process data and product quality data of multi-layer composite plates to construct the original data set; Step 2: Preprocess the original data set; Step 3: Based on the preprocessed original data set, a convolutional-long short-term memory network spatiotemporal sequence prediction model is established to predict the outlet thickness; Step 4: Based on the prediction results of the prediction model, adopt the corresponding feedforward control strategy to correct the roll gap parameters; Step 5: Based on the actual thickness of the multi-layer composite plate during hot rolling, a feedback control strategy is adopted to adjust the roll gap; Step 6: Input the multi-layer composite plate hot rolling production process data and product quality data into the convolution-long short-term memory network spatiotemporal sequence prediction model to predict the outlet thickness. If the predicted result does not match the actual required outlet thickness, adjust the roll gap parameters and predict again until the predicted result matches the actual required outlet thickness. The convolution-long short-term memory network spatiotemporal sequence prediction model is used to predict the roller gap parameters used for production, and the feedback control strategy of step 5 is adopted during the production process to adjust the roller gap in real time.

2. A composite plate hot rolling speed-deformation-pressure coupling control method according to claim 1, characterized in that: The multi-layer composite plate hot rolling production process data and product quality data in step 1 specifically include: inlet thickness, set outlet thickness, deformation, reduction rate, inlet width, rolling temperature, rolling speed, rolling force, and outlet thickness.

3. The method for controlling the speed, deformation and pressure of hot rolling of composite plates according to claim 1, wherein: The step 2 specifically includes the following steps: Step 2.1: Use the K-nearest neighbor interpolation algorithm to fill in the missing values ​​in the original data set. The formula is as follows: Where, For the Samples to distance, It is The coordinates of the samples, For the The value of the sample, are the coordinates of the missing values ​​in the sample, is the missing value of the sample, For the The weight of the nearest neighbor points; Step 2.2: Use the Savitzky-Golay filtering method to filter the padded data to obtain the denoised data. The formula is as follows: Where, is the unilateral degree of fitting, It's about of polynomial of order, , which is the integer index within the window, are the coefficients of the polynomial, , For index The value of the input data, is the sum of squares of the residuals between the fitted data points and the original data points, is the data after noise reduction; Step 2.3: Use maximum and minimum scaling to normalize the denoised data to obtain standardized data. The formula is as follows: in represents the data after noise reduction, represents the standardized data, and Represents the maximum and minimum values ​​of the original data set after denoising; Step 2.4: Extract 80% of the standardized data as training data to build the prediction model, and the remaining data as test data to evaluate the model performance.

4. A composite plate hot rolling speed-deformation-pressure coupling control method according to claim 3, characterized in that: The step 3 specifically includes the following steps: Step 3.1: Build a convolutional-long short-term memory network spatiotemporal sequence pre-training model. The principle formula is as follows: Where, Indicates the The weight corresponding to the convolution kernel position, Indicates the The bias corresponding to the convolution kernel of the position, represents the feature map of the previous layer, represents the feature map of the current layer, represents the activation function, represents a set of feature maps, Where, is the pooling result, is the pixel value of the input feature, and are the row and column index values ​​of the features respectively, and are the row and column sizes of the pooling window, Where, is the weight matrix representing the input gate, is the bias of the input gate, is the sigmoid function, is the weight matrix of the forget gate, Indicates that the vector and vector Concatenate into a new vector, refers to the hidden state at time step t-1, refers to the input at time step t, is the bias of the forget gate, is the weight matrix of the output gate, is the bias of the output gate, is the temporary hidden state value, is the weight matrix when using the forget gate for feature screening, It is the bias when using forget gate for feature screening; Step 3.2: Use the inlet thickness, set outlet thickness, deformation, reduction ratio, inlet width, rolling temperature, rolling speed, and rolling force as model inputs, and the outlet thickness as model output; Step 3.3: Use grid search strategy and cross-validation method to determine the optimal hyperparameters of the spatiotemporal series prediction model to ensure the model has the best prediction performance; Step 3.4: Use the above model hyperparameters to build a quality prediction model for hot-rolled multi-layer composite plates.

5. The method for controlling the speed, deformation and pressure of hot rolling of composite plates according to claim 1, characterized in that: The step 4 specifically includes the following steps: Step 4.1: Based on the deviation between the predicted outlet thickness and the target outlet thickness, a multi-objective balance optimization algorithm is used to perform feedforward correction on the roll gap; Step 4.2: Determine whether the adjustment amount of the last rolling parameter exceeds the limit; if it exceeds the limit, repeat step 4.1; if it does not exceed the limit, use the corrected roll gap for rolling.

6. The method for controlling the speed, deformation and pressure of hot rolling of composite plates according to claim 1, wherein: The feedback control strategy in step 5 specifically includes the following steps: Step 5.1: Calculate the residual between the actual value of the composite plate thickness at the current moment and the set outlet thickness, and calculate the roller gap feedback adjustment amount at the next moment based on the residual value at the current moment; Step 5.2: Determine whether the roll gap feedback adjustment amount exceeds the limit; if it exceeds the limit, repeat step 5.1; if it does not exceed the limit, add the feedforward correction roll gap to the roll gap feedback adjustment amount and perform rolling.

7. A composite plate hot rolling speed-deformation-pressure coupling control system according to any one of claims 1 to 6, characterized in that: include, The export thickness prediction system is used to predict the export thickness of the finished multi-layer composite board based on the production and process parameters. When the predicted export thickness does not meet the set export thickness requirements, the production and process parameters are modified until the predicted export thickness matches the set export thickness. Roll gap parameter feedback control system and feedback control strategy perform feedback control on the roll gap parameters of production equipment to ensure that the actual production results meet the requirements.

Images