Tire half-part composite extrusion control method, system, and storage medium

Abstract

The application provides a tire half-part composite extrusion control method, a system and a storage medium, and belongs to the field of tire industrial tread manufacturing equipment control. The method comprises the following steps: step 1, collecting composite extrusion production data and preprocessing to obtain a data set; step 2, obtaining a prediction model by using the data set, and establishing a linear regression model of the total width of the tread and the tread weight per meter; step 3, constructing a model predictive controller by using the prediction model and the linear regression model; step 4, a production debugging stage; and step 5, a stable production stage. The application uses a nonlinear model and a model predictive control to establish a targeted control model, solves the problem of the influence of multivariable parameters in the tread composite extrusion production process on the quality of the extruded semi-finished product, improves the stability, accuracy and intelligent degree of the tread composite extrusion system, and realizes an efficient, high-quality intelligent semi-finished tread production process.

Description

Technical Field

[0001] This invention belongs to the field of tire tread manufacturing equipment control, specifically relating to a composite extrusion control method, system, and storage medium for tire semi-components. Background Technology

[0002] With the rapid development of the automotive market, the variety of tire products is constantly increasing, and the requirements for tire wear resistance, oil resistance, surface aesthetics, and different physical and mechanical properties are also constantly improving. This has prompted the requirements for tire tread compound extrusion equipment to develop rapidly towards linkage, automation, and intelligence.

[0003] Since tire tread semi-finished products account for approximately one-third of the total tire weight, their quality stability and uniformity significantly affect the overall tire weight stability, as well as key quality evaluation indicators such as dynamic balance and uniformity. Composite extrusion processing, as a highly efficient method for producing tire tread semi-finished products, is increasingly emphasizing the intelligent development of its production process. However, rubber, as a high-molecular polymer, exhibits complex rheological properties during processing, and the dynamic co-extrusion process is also affected by equipment process parameters, raw material parameters, and processing parameters, leading to instability in viscoelastic fluid flow and thus reducing the quality stability of continuous extrusion production of tire tread semi-finished products.

[0004] Therefore, the current demand for product quality upgrades in tire manufacturers has raised the standard of control over the production process of semi-finished tire treads to a new level. The quality and precision of extruded tire treads directly affect the final quality and precision of the tire product. Due to fluctuations in rubber compound parameters, feeder feed fluctuations, and other factors, the weight of tread products within the same length range during extrusion can be unstable, directly affecting the dynamic balance and uniformity pass rate of the finished tire. In actual production, these process errors often change frequently and cause significant fluctuations in product quality. Operators need to promptly identify and adjust relevant production parameters for manual correction to ensure that the fluctuation range does not exceed the permissible range of the process standard. This control method, which relies on operator feedback, often suffers from poor control effectiveness and a high number of defective products. Furthermore, its heavy reliance on operator experience makes it difficult to achieve standardized and uniform production quality. Summary of the Invention

[0005] The purpose of this invention is to solve the problems existing in the prior art and provide a method, system and storage medium for controlling composite extrusion of tire semi-finished parts, thereby improving the stability, accuracy and intelligence of the composite extrusion system for tire treads and realizing an efficient and high-quality intelligent semi-finished tire tread production process.

[0006] This invention is achieved through the following technical solution:

[0007] A first aspect of the present invention provides a method for controlling the composite extrusion of tire semi-finished parts, the method comprising:

[0008] Step 1: Collect and preprocess compound extrusion production data to obtain a dataset;

[0009] Step 2: Use the dataset to obtain the prediction model and establish a linear regression model between the total tread width and the tread weight per meter.

[0010] Step 3: Construct a model predictive controller using the predictive model and the linear regression model;

[0011] Step 4, Production Debugging Phase;

[0012] Step 5, Stabilize Production Phase.

[0013] Preferably, the operation of collecting composite extrusion production data in step 1 includes: collecting the rotational speed and torque of all screws, as well as the tread weight per meter and total tread width measured at the front tread scale.

[0014] Preferably, the operation of obtaining the prediction model using the dataset in step 2 includes:

[0015] The prediction model adopts a multi-input multi-output nonlinear autoregressive model with measurable perturbation variables; the rotational speed of each screw is the input variable of the prediction model, the tread weight per meter and the total tread width measured at the front meter are the output variables of the prediction model, and the torque of each screw is the perturbation variable of the prediction model.

[0016] The prediction model is trained using the dataset to obtain the various parameters in the prediction model.

[0017] Preferably, the operation of establishing a linear regression model between the total tread width and the tread weight per meter in step 2 includes:

[0018] Based on historical data, a linear regression model was established for the total tread width and tread weight per meter for a specific production formula.

[0019] Preferably, step 3 includes: constructing a model predictive controller based on the cost function and constraints.

[0020]

[0021] U min ≤U(t)≤U max ,

[0022] △U min ≤U(t)≤ΔU max ,

[0023] Where J is the cost function, N C To control the time domain;

[0024] It controls the predicted tread weight per meter and total tread width obtained from the prediction model within the control time domain;

[0025] Y r (t)=[y r (t+1|t),y r (t+2|t),...,y r (t+N C |t)] T It controls the tread weight per meter and total tread width obtained from the linear regression model within the control time domain.

[0026] U(t)=[u(t+1|t),u(t+2|t),...,u(t+N C |t)] T It is the optimal control sequence in the control time domain to be solved;

[0027] ΔU(t)=[u(t|t)-u(t-1),u(t+1|t)-u(t|t),...,u(t+N C -1|t)-u(t+N C -2|t)] T Is it control time

[0028] Rate of change of the control sequence within the domain;

[0029] Matrix I, R, and S are the weight matrices for the output variable, input variable, and rate of change of the input variable, respectively.

[0030] Y min and Y max This is a multidimensional vector representing the minimum and maximum values ​​of the total tread width and tread weight per meter of the output.

[0031] U min and U max It is a multidimensional vector representing the minimum and maximum rotational speeds of each screw.

[0032] ΔU min and ΔU max It is a multidimensional vector representing the minimum and maximum values ​​of the rate of change of rotational speed of each screw.

[0033] Preferably, step 4 includes the following operations:

[0034] (41) Obtain the set tread weight per meter and total tread width based on the production formula and linear regression model;

[0035] (42) Set the values ​​of the weight matrices I, R, and S of the model prediction controller;

[0036] (43) The predicted tread weight per meter and total tread width are obtained by using the prediction model. Then, the control sequence is obtained by using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model prediction controller. That is, the control sequence is obtained by solving the minimum value optimization problem of the cost function under all constraints.

[0037] (44) Apply the first control vector in the control sequence to the compound extrusion device;

[0038] (45) Determine whether the error between the measured tread weight per meter and total tread width of the composite extrusion device and the set tread weight per meter and total tread width is stable within the tolerance range. If not, return to (42). If yes, end the production debugging stage.

[0039] Preferably, step 5 includes the following operations:

[0040] (51) The predicted tread weight per meter and total tread width are obtained by using the prediction model. Then, the control sequence is obtained by using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model prediction controller. That is, the control sequence is obtained by solving the minimum value optimization problem of the cost function under all constraints.

[0041] (52) Apply the first control vector in the control sequence to the compound extrusion device;

[0042] (53) Return to step (51).

[0043] A second aspect of the present invention provides a tread composite extrusion system, the system comprising: an industrial control computer, a model prediction control unit, a motor, and a composite extrusion device;

[0044] The compound extrusion unit transmits data to the industrial control computer, which then sends the processed data to the model predictive controller. The model predictive control unit uses the predictive model, the linear regression model, and the model predictive controller to generate a control sequence, and sends the first control vector in the control sequence as a control signal to the motor. The motor controls the rotational speed of each screw in the compound extrusion unit according to the first control vector.

[0045] Preferably, the model prediction control unit includes: a prediction model, a cost function setting module, a constraint setting module, and an online optimization module;

[0046] The prediction model is used to obtain the predicted tread weight per meter and total tread width;

[0047] The cost function setting module is used to input the set tread weight per meter and total tread width provided by the linear regression model, as well as the predicted tread weight per meter and total tread width, into the cost function;

[0048] The constraint setting module is used to input constraint conditions;

[0049] The online optimization module is used to solve the minimum cost function optimization problem to obtain the control sequence under the condition of satisfying the constraints.

[0050] A third aspect of the present invention provides a computer-readable storage medium storing at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the tire half-component composite extrusion control method.

[0051] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention conducts an in-depth analysis of various parameters and production process of the tread composite extrusion production line, and establishes a targeted control model by using nonlinear models and model predictive control. This solves the problem of the influence of multivariable parameters on the quality of extruded semi-finished products in the tread composite extrusion production process, improves the stability, accuracy and intelligence of the tread composite extrusion system, and realizes an efficient and high-quality intelligent semi-finished tread production process. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the overall process of the composite extrusion control method for tire semi-components of the present invention.

[0053] Figure 2 This is a schematic diagram of the training process of the prediction model in the method of the present invention.

[0054] Figure 3 This is a schematic diagram of the system of the present invention.

[0055] Figure 4 This is a schematic diagram of the workflow of the model prediction control unit in the system of the present invention.

[0056] Figure 5 The above represents the prediction results of the prediction model in this embodiment of the invention.

[0057] Figure 6 This is the fitting result of the tire tread weight per meter and total width based on the linear regression model in this embodiment of the invention. Detailed Implementation

[0058] The present invention will now be described in further detail with reference to the accompanying drawings:

[0059] In a specific embodiment of the present invention, a four-component tread extrusion production line is used to manufacture tire semi-components. The selected compound extrusion unit (also known as a compound extruder) is model XJF45 / 120 / 250 / 150, capable of achieving efficient four-component extrusion. Based on the required formulation, the extruder nozzle shape is rationally selected to improve material flow behavior and enhance molding quality. This embodiment, through precise formulation design, ensures that the performance and quality of the tire semi-components meet the expected range, providing a solid foundation for subsequent tire manufacturing processes.

[0060] like Figure 1 As shown, the present invention provides a method for controlling the composite extrusion of tire semi-components, specifically including the following steps:

[0061] Step 1: Collect and preprocess compound extrusion production data to obtain a dataset: In order to build a predictive model, it is necessary to extract the production process data of the compound extrusion system from the production database, mainly including key data such as screw speed, screw torque, current, die head temperature, die head pressure and semi-finished part mass.

[0062] The operation of collecting composite extrusion production data includes: analyzing the data and selecting parameters that have the greatest impact on the final quality of tire components and relevant parameters that reflect fluctuations in the production process from the database. The parameters selected in this invention include the rotational speed and torque of all screws, as well as the tread weight per meter and total tread width measured at the front meter scale (the front meter scale is a measuring device installed at the front of the entire production line to measure the weight and width of each meter of tread). For example, for a four-component extrusion unit, the selected parameters include: Φ45 screw speed, Φ120 screw speed, Φ250 screw speed, Φ150 screw speed, Φ45 screw torque, Φ120 screw torque, Φ250 screw torque, Φ150 screw torque, as well as tread weight per meter and total tread width.

[0063] The preprocessing steps include: removing missing values ​​from each parameter using existing data processing methods; identifying outliers using standard scores and replacing them with the mean; and finally standardizing all parameters (that is, normalizing industrial data of different dimensions and scales to a uniform range or distribution to eliminate the influence of dimensions and improve model adaptability, computational efficiency, and accuracy of data analysis; standardization can be performed using existing methods and will not be elaborated here). After this, continuous production process data for each parameter are obtained, and the continuous production process data for all parameters are combined into a dataset.

[0064] Step 2: Use the dataset to obtain the prediction model and establish a linear regression model between the total tread width and the tread weight per meter:

[0065] The specific steps in step 2 to obtain the prediction model using the dataset are as follows:

[0066] like Figure 2 As shown, a data-driven approach is used to establish the predictive model, that is, to build a data-driven model of the compound extrusion system using a dataset. The rotational speed of each screw in the compound extrusion process is the input variable of the predictive model, the tread weight per meter and the total tread width measured at the front weighing station are the output variables of the predictive model, and the screw torque, which reflects the fluctuation of the feed rate during production, is the disturbance variable of the predictive model. In addition, the setpoint of the die head temperature of each screw is determined by the production formula, and a PID controller is used to precisely control the temperature, while the die head temperature is regulated by cooling water.

[0067] Specifically, a multi-input multi-output nonlinear autoregressive model with measurable perturbation variables is designed according to the types of data in the dataset. Various existing multi-input multi-output nonlinear autoregressive models can be used. The model used in this embodiment is shown in formula (1):

[0068]

[0069] in, It is the output variable, where, Represents the n-dimensional real space, Let n represent the set of real numbers, and n represent the dimension of the output (using n-dimensional real number space to represent n output variables, such as 2 output variables, namely tire tread weight per meter and total tire width). It is an input variable. It represents an m-dimensional real space, where m represents the dimension of the input (using an m-dimensional real space to represent m input variables, for example, 4 input variables, i.e., the rotational speed of 4 screws); Let represent the noise sequence, l represent the dimension of the noise; f represent a nonlinear function or mapping that describes the dynamic relationship of the system, f(w(t)) represents the output y(t+1) generated by the current w(t) after passing through a nonlinear function f, which describes the relationship between w(t) and y(t+1); w(t) consists of the screw speed, torque, tread weight per meter, and total tread width measured at the current and historical times; n y and n u Let T represent the autoregressive order of the output variable and the autoregressive order of the input variable, respectively, and let T represent the transpose of the vector.

[0070] Equation (1) can be approximated by the following multi-input multi-output nonlinear autoregressive model:

[0071]

[0072] Formula (2) is the prediction model, which represents the relationship between the input and output variables. In formula (2), This represents the perturbation variable. Let v represent a v-dimensional real space, where v represents the dimension of the perturbation variable, i represents the index variable of the autoregression order, and k represents the index variable of the number of kernel functions. The predicted tread weight per meter and total tread width can be obtained using the prediction model.

[0073] Where, n y It is the autoregressive order of the output variable, n u It is the autoregressive order of the input variable, n. d Z is the autoregressive order of the disturbance variable (the autoregressive order represents the number of historical values ​​for each variable), all three are natural numbers greater than 1, and h is the number of radial basis functions; j,k It is the center of the radial basis functions; and These are weight coefficients of appropriate dimensions (appropriate dimensions for weight coefficients refer to the dimensions that the parameters need to match with the dimensions of the input, output, and perturbation variables to correctly describe the relationships between all variables in a multi-input multi-output system); ||·||2 represents the 2-norm of the vector. diag (diag represents a diagonal matrix) and dim (dim represents the dimension) are scaling factors; This represents the modeling error. j is an index variable used to distinguish different variable types, y represents the output variable (i.e., y(t)), u represents the input variable (i.e., u(t)), and d represents the disturbance variable (i.e., d(t)). The data-driven model includes feed fluctuations that affect the instability of the final tread quality, that is, incorporating the screw torque, which characterizes the change in feed rate, into the prediction model. In this embodiment, the radial basis function used is the Gaussian radial basis function. Besides the Gaussian radial basis function, a multinomial radial basis function or an inverse multinomial radial basis function can also be used, with the same input / output variables and parameters as the Gaussian radial basis function.

[0074] The dataset collected in step 1 is divided into a training set and a prediction set. The prediction model is trained using the training set and the prediction set to obtain the various parameters in the prediction model, such as... Figure 2 As shown, it specifically includes:

[0075] S1: Divide the parameters to be identified in the prediction model into a set of linear parameters and a set of nonlinear parameters:

[0076] The set of linear parameters is as follows:

[0077]

[0078] Formula (3) defines the set of linear parameters on the left side of the equation as the form on the right side, where n in Formula (3) jThe number of autoregressive orders is represented by k, which represents the number of radial basis functions. The number of radial basis functions is h, which indicates how many orders of historical data for each different variable (including output, input, and perturbation variables) are involved in the calculation in the current model. j is an index variable used to distinguish different variable types.

[0079] The collection of nonlinear parameters is as follows:

[0080]

[0081] Formula (4) means that the set of nonlinear parameters on the left side of the equation is defined in the form on the right side, where λ represents the width of the radial basis function and Z represents the center of the radial basis function.

[0082] Formula (2) in step 2 can be written in the following form:

[0083] y(t+1)=Φ(θ N ,Θ(t)) T θ L +ξ(t+1) (5)

[0084] Where, θ N Represents the set of nonlinear parameters, θ L Let Θ(t) represent the set of linear parameters, and Φ(θ) represent the set of state variables, indicating the system's data at time t (including the input, output, and disturbance variables at that time). N Θ(t) represents a nonlinear parameter set θ N The feature mapping function is related to the set of state variables Θ(t), where ξ(t+1) represents the error term and y(t) represents the tread weight per meter and the total tread width at time t.

[0085] S2: Determine the order of the nonlinear autoregressive model using the Akaike information criterion (AIC): Select the optimal order n of the nonlinear autoregressive model based on the AIC values ​​of models of different orders. y n u and n d :

[0086] AIC=2k-2ln(L) (6)

[0087]

[0088]

[0089] in, It is the actual value output at time t+1. It is the predicted value output at time t+1, e(t+1) represents the error between the actual value and the model prediction at time t+1, L is the maximum likelihood estimate of the model, and N is the number of parameters that need to be identified in the model, that is, the total number of linear and nonlinear parameters.

[0090] S3: Determine the initial values ​​of the nonlinear parameters: Determine the initial values ​​of the center points of the radial basis functions, i.e., the initial values ​​of the nonlinear parameters, using the existing K-means method.

[0091]

[0092] c i =arg min j ||x i -μ j || 2 (9)

[0093] Among them, J c It is the sum of the squared distances from all data points to their respective cluster centers, and the goal of K-means is to minimize this value; C j It is the set of all data points assigned to cluster j, |C j | is the number of data points in cluster j; c i It is data point x i The index of the cluster to which it belongs; repeat the assignment and update steps until the cluster centers no longer change (or the change is less than a predetermined threshold), or the maximum number of iterations is reached.

[0094] The initial value of the scaling factor is determined by the following formula:

[0095]

[0096] Where, ε k It represents a very small positive number and is usually used to avoid zero denominators or excessively large values ​​to ensure the stability of the value.

[0097] S4: Optimization of Linear and Nonlinear Parameters: After determining the initial values ​​of the nonlinear parameters, the existing least squares algorithm is used for linear parameter identification, and the existing Levenberg-Marquardt method is used for nonlinear parameter identification. That is, the two existing methods are used to obtain the linear and nonlinear parameters, which are briefly described below:

[0098] The parameter identification optimization problem can be represented as:

[0099]

[0100] in, As shown in equation (5); These are the actual values ​​of tire tread weight per meter and total width collected during the production process.

[0101] The optimal solution combining the linear and nonlinear parameters can be obtained as follows:

[0102]

[0103] S5: Determine if the training error is minimized or the maximum number of iterations has been reached. If yes, proceed to S6; otherwise, return to S4.

[0104] Specifically, the root mean square error (RMSE) between the predicted tread weight per meter and total tread width output by the prediction model and the actual measured tread weight per meter and total tread width output by the composite extrusion device (i.e., training error) is calculated. If the RMS error is less than the set threshold or the number of iterations reaches the maximum number of iterations, it is considered qualified, and the entire optimization process ends. At this point, the optimal set of linear and nonlinear parameters is the optimal parameter of the prediction model. Obtaining the optimal parameter means obtaining a qualified prediction model, and then proceeding to S6.

[0105] The root mean square error is calculated as follows:

[0106]

[0107] Where, N m This represents the total number of data points in the training set. This represents the actual output value at time i in the training set. This represents the model prediction value output by the model at time i.

[0108] The identification results of the prediction model are as follows Figure 5 As shown, the root mean square error (RMSE) of the predicted tread weight per meter is 0.0014158, and the RMS error of the total tread width is 0.0013885. Among these, Figure 5 This invention demonstrates the prediction model (i.e.) Figure 5 The comparison between the output of the autoregressive model with external input and the actual output shows that the predicted model fits the two outputs well. Figure 5 Output 1 represents the tread weight per meter, and output 2 represents the total tread width. The error is small for most of the time period. There is some deviation in the initial stage, which is due to rapid dynamic changes. This invention's nonlinear autoregressive model with external input can capture the dynamic trend of the system and has high prediction accuracy. This model provides accurate prediction and efficient modeling capabilities in industrial environments, and can efficiently handle external disturbances and noise interference, making it suitable for solving dynamic modeling problems in complex industrial systems.

[0109] S6: Save the parameters of the prediction model: Save the qualified prediction model as the prediction model in actual production control, i.e. Figure 1The phrase "using the identified model as the predictive model for model predictive control" is used in this context.

[0110] This invention uses a nonlinear autoregressive model with external input to perform data-driven modeling of the tire tread extrusion process. It establishes a predictive model that includes screw torque as a disturbance variable using historical production data, describing the impact of feed rate fluctuations on the extrusion process.

[0111] Meanwhile, this invention also establishes a nonlinear regression relationship between the tread weight per meter measured at the front weighing point of the composite extrusion system and the total tread width, providing a quantitative model for quality prediction and realizing dual-objective quality control of tread weight per meter and total width, overcoming the limitations of traditional methods that rely on a single quality index for control.

[0112] The steps in step 2 to establish a linear regression model between total tread width and tread weight per meter are as follows:

[0113] Based on historical data from stable production during the tread compounding process, a linear regression equation for a specific production formula is established, namely, the relationship between the total tread width and the tread weight per meter at the first meter of composite tread. The linear regression model for the total tread width and tread weight per meter established in this invention is shown in formula (14):

[0114] width=β0+β1weight+ε, (14)

[0115] Where β0 and β1 are linear regression coefficients; weight is the tread weight per meter; width is the total tread width; ε is the error term, and the results of the linear regression model are as follows: Figure 6 As shown. The purpose of establishing the linear regression model is to provide the set total tread width and tread weight per meter. Formula (14) is used for different production formulas, but the specific values ​​of β0, β1 and ε are different. The values ​​of these parameters are obtained from actual production data. Figure 6 This diagram illustrates a linear regression model between total tread width and tread weight per meter, a relationship derived offline from historical stable production data. The horizontal axis represents total tread width, and the vertical axis represents tread weight per meter. The scatter plots represent historical production data, and the fitted line represents the linear regression relationship between the two, i.e., the linear regression model. Figure 6 As can be seen, there is a certain positive correlation between the total tread width and the tread weight per meter. That is, as the tread weight per meter increases, the total tread width also tends to increase. This linear regression model provides data support for the subsequent optimization design of production control.

[0116] In actual use, before starting production, the tread weight per meter is determined according to the production formula using existing methods. This tread weight per meter is then used as the set tread weight per meter. The set tread weight per meter is then substituted into the linear regression model to calculate the total tread width. This total tread width is then used as the set total tread width. In this way, the set tread weight per meter and total tread width are obtained through the linear regression model, which are the target tread weight per meter and total tread width. These set values ​​are not modified throughout the entire production process of this production formula.

[0117] Step 3: Construct a model predictive controller using the predictive model and the linear regression model:

[0118] Using existing model predictive control methods, a model predictive controller based on cost functions and constraints is constructed for a tread compound extrusion system:

[0119]

[0120] U min ≤U(t)≤U max ,

[0121] ΔU min ≤U(t)≤ΔU max (15)

[0122] Where J is the cost function, i.e. the index to be optimized, and N... C To control the time domain, it is a natural number greater than 1, which can be set according to actual needs. Y represents the output variable, and U represents the input variable (i.e., the rotational speed of all screws), as follows:

[0123] It controls the predicted tread weight per meter and total tread width obtained from the prediction model within the control time domain;

[0124] Y r (t)=[y r (t+1|t),y r (t+2|t),...,y r (t+N C |t)] T It controls the set tread weight per meter and total tread width obtained from the linear regression model within the control time domain. After the set tread weight per meter and total tread width are determined according to the production formula and the linear regression model, the set values ​​do not change during the production process. This statement is to meet the requirement in the cost function J of formula (15) that "only vectors of the same dimension can be subtracted";

[0125] U(t)=[u(t+1|t),u(t+2|t),...,u(t+N C |t)] TIt is the optimal control sequence in the control time domain to be solved, i.e., the screw speed sequence;

[0126] ΔU(t)=[u(t|t)-u(t-1),u(t+1|t)-u(t|t),...,u(t+N C -1|t)-u(t+N C -2|t)] T It is the rate of change of the control sequence within the control time domain, i.e., the rate of change of the screw speed;

[0127] Matrix I, R, and S are the weight matrices for the output variable, input variable, and rate of change of the input variable, respectively.

[0128] Formula (15) is the core optimization objective of the model predictive controller, used to optimize the tread weight per meter and the total tread width Y based on the set values. r The optimal screw speed is calculated using (t) and weighting parameters I, R, and S. These weighting matrices achieve the control objective by indirectly influencing the controller's behavior; that is, by incorporating the error between the actual output and the output setpoint into the control sequence calculated using the cost function. In actual production, tread weight per meter and overall width are key indicators for quality control (i.e., the target value Y in Formula 15). r (t)). If either of these deviates from the target value, I, R, and S need to be adjusted in conjunction with production data analysis to optimize the controller's tracking performance, input smoothness, and range limitation, thereby achieving high-quality production control.

[0129] In practical application, based on the cost function in formula (15), and under the condition of satisfying three constraints, the optimal control solution is obtained by solving the minimum value optimization problem of the cost function, i.e., the control sequence. The control sequence is a multidimensional matrix (representing N). C The rotational speeds of all screws within a given moment (the dimension of which is n) u ×N C Controlled by time domain N C The control sequence consists of U(t), and the first control vector in the control sequence is the optimal screw speed, which includes the optimal speed values ​​of all screws.

[0130] In formula (15), the three expressions after “st” represent the constraint conditions, that is, the output variable constraint Y must be satisfied simultaneously. min and Y max (Y min and Y max This is a multi-dimensional vector representing the minimum and maximum values ​​of the total tread width and tread weight per meter of the output. Input variable constraints U min and U max (U min and U maxAs a multidimensional vector, representing the minimum and maximum rotational speeds of each screw, and the rate of change constraint ΔU of the input variables. min and ΔU max (ΔU min and ΔU max (This is a multi-dimensional vector representing the minimum and maximum rates of change of rotational speed for each screw).

[0131] For example, in this embodiment, the specific constraint value of the rotational speed of each screw is [4, 40] (corresponding to U). min and U max The constraint value for the output variable, tread weight per meter, is [0, 15] (corresponding to Y). min and Y max The constraint value for the output variable, total tread width, is [0, 800] (corresponding to Y). min and Y max The maximum and minimum values ​​of the above three constraints are all set manually based on actual production conditions. This invention also adds the rate of change of the input variable (i.e., the rate of change of the screw speed) as a constraint condition to the model predictive control. This avoids excessive fluctuations in the die head pressure caused by excessive changes in screw speed during production, which could lead to excessive fluctuations in product quality. For example, the constraint on the rate of change of the input variable is [-1, 1] (corresponding to ΔU). min and ΔU max ).

[0132] The model predictive controller uses the optimal control mode to ensure that the screw speeds of the composite extrusion system remain within the optimal operating range even when the feed rate fluctuates, and minimizes the error between the actual and target values ​​of the produced tread weight per meter and total tread width.

[0133] The core function of the weight parameters I, R, and S in the model predictive controller is to balance target tracking accuracy, control input range, and input smoothness. These parameters essentially reflect the trade-offs in the overall performance requirements of the extrusion system, specifically as follows: I mainly reflects the weight of the deviation from the target values ​​(tread weight per meter and total tread width), and is usually a fixed value directly related to the production target; R reflects the limitation on the amplitude of the control input, and is usually designed based on the physical range of screw speed and equipment operating limitations, and will not be frequently adjusted due to small changes in operating conditions; S reflects the limitation on input changes, and usually depends on the equipment response characteristics and process requirements, and is designed with versatility. Therefore, these parameters are designed for the global performance requirements of the system and have long-term effectiveness.

[0134] During the production commissioning phase, the weight parameters I, R, and S were adjusted multiple times through experiments to ensure the expected control effect was achieved. Once stable production was achieved, readjustment was not required for each production fluctuation. The optimal control solution, obtained by solving the cost function minimization problem in real time, automatically adapted to small-range production deviations. The weight matrix ensured strong versatility and robustness, requiring redesign or adjustment of I, R, and S only when production conditions underwent significant changes.

[0135] Step 4: Production debugging phase: Determine the weight matrices I, R, and S of the model predictive controller, that is, determine the appropriate controller parameters (i.e., weight matrices I, R, and S) through multiple experimental verifications and parameter optimizations.

[0136] During the production debugging phase (i.e., the phase between the start of production of the composite extrusion unit and stable production), I, R, and S in formula (15) are manually adjusted online according to the actual production process. Specifically, the weight of the output error weight matrix I can be increased according to the importance of the tread quality to ensure accurate control of the quality indicators; the weight of the input weight matrix R and the weight of the input change rate S should balance the smoothness of the screw speed and the system response speed. Frequent changes in screw speed will lead to mechanical wear, and the weight of S should be appropriately increased to ensure smooth input changes.

[0137] Step 4 includes the following operations:

[0138] (41) Obtain the set tread weight per meter and total tread width based on the production formula and linear regression model;

[0139] (42) Set the values ​​of the weight matrices I, R, and S of the model predictive controller; these matrices can be manually set based on multiple experiments and experience.

[0140] (43) The predicted tread weight per meter and total tread width are obtained by using the prediction model. Then, the control sequence is obtained by using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model prediction controller. That is, under the condition of satisfying the three constraints, the minimum value optimization problem of the cost function is solved to obtain the control sequence.

[0141] (44) Apply the first control vector in the control sequence to the compound extrusion unit, that is, send a control signal (which includes the first control vector) to the driver of the motor of the compound extrusion unit. The driver of the motor controls the rotation speed of each screw in the compound extrusion unit to be the same as the rotation speed of the corresponding screw in the first control vector.

[0142] (45) Determine whether the error between the measured tread weight per meter and total tread width of the composite extrusion device and the set tread weight per meter and total tread width is stable within the tolerance range (set two tolerance ranges for tread weight per meter and total tread width according to actual production needs. If either the error of tread weight per meter or total tread width is not within the tolerance range, it is determined as no. If both are within the tolerance range, it is determined as yes). If no, return to (42). If yes, end the production debugging stage. The weight matrix at this time is the optimal controller parameter. After entering the stable production stage, the optimal controller parameter can be used directly.

[0143] Step 5: Stabilized Production Stage: Real-time adjustment of the screw speeds of each screw in the compound extrusion unit using a model-based predictive controller.

[0144] The decision variable output by the model predictive controller is the control sequence. At each sampling time, the model predictive controller is run, and the first control vector in the solved control sequence (i.e., the optimal screw speed) is applied to the compound extrusion unit. Specifically, this includes:

[0145] (51) Obtain the predicted tread weight per meter and total tread width using the prediction model. Then, obtain the control sequence using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model predictive controller. That is, under the condition of satisfying three constraints, solve the minimum optimization problem of the cost function to obtain the control sequence.

[0146] (52) Apply the first control vector in the control sequence to the compound extrusion unit, that is, send a control signal (including the first control vector) to the motor driver of the compound extrusion unit, and the motor driver controls the rotation speed of each screw in the compound extrusion unit to be the same as the rotation speed of the corresponding screw in the first control vector.

[0147] (53) Return to step (51).

[0148] In this way, through steps (51) to (54), the rotational speed of each screw is optimized and controlled in real time during the stable production process. That is, the screw speed is dynamically adjusted by the actual measured values ​​of the tread weight per meter and the total tread width, thus ensuring the production quality.

[0149] Existing composite extrusion systems lack closed-loop control and rely on manual adjustment. The tread composite extrusion system of this invention, however... Figure 3As shown, the system includes an industrial control computer, a model predictive control unit, a motor, and a compound extrusion unit. The compound extrusion unit transmits data (including measured screw speed, screw torque, measured tread weight per meter, and total tread width) to the industrial control computer. The industrial control computer processes the data according to the requirements of the model predictive controller and then sends the processed data back to the model predictive controller. The model predictive control unit uses the predictive model, the linear regression model, and the model predictive controller to generate a control sequence and sends the first control vector (i.e., the optimal screw speed) in the control sequence as a control signal to the motor. The motor controls the speed of each screw in the compound extrusion unit according to the optimal screw speed in the first control vector to achieve adjustment of the speed of each screw.

[0150] Controlling the screw speed via motor can be achieved using existing control methods, which are briefly described below: The first control vector in the optimal control sequence matrix obtained from the optimization solution is sent to the driver of the corresponding screw motor. After receiving the control signal, the driver of each screw motor compares the received optimal screw speed for that screw with the actual motor speed. Based on the comparison result, the motor speed is adjusted (if the actual motor speed is less than the optimal screw speed, the motor speed is increased until it reaches the optimal screw speed). Since the motor's rotating shaft is connected to the screw via a coupling or directly, adjusting the motor speed adjusts the screw speed. In this way, the entire compound extrusion system continuously updates the current working state and adjusts the screw speed through closed-loop feedback to ensure that the error between the target tread weight per meter and the actual tread weight per meter and the actual tread width is stabilized within the tolerance range. Finally, the compound extrusion unit completes the tread semi-finished part extrusion process, outputting the tread semi-finished product that meets the process requirements.

[0151] The specific structure of the model prediction control unit is as follows: Figure 4 As shown, the system includes a prediction model, a cost function setting module, a constraint setting module, and an online optimization module. The prediction model is used to obtain the predicted tread weight per meter and total tread width. The cost function setting module is used to input the set tread weight per meter and total tread width provided by the linear regression model, as well as the predicted tread weight per meter and total tread width, into the cost function. The constraint setting module is used to input the constraint conditions. The online optimization module is used to solve the minimum value optimization problem of the cost function under the condition of satisfying the three constraints to obtain the control sequence. In this way, the model prediction control unit adjusts the equipment operating status in real time, effectively dealing with disturbances such as feed rate fluctuations. Ultimately, stable control of tread weight per meter and total tread width is achieved, improving the product quality of extruded tread semi-finished parts.

[0152] Existing control methods rely too heavily on operator experience in practical applications. The final quality of semi-finished tire tread products is influenced by the characteristics of the rubber material itself and its processing performance, making it challenging to establish a comprehensive control model based on mechanistic models. Furthermore, the compound extrusion production technology involves adjusting the speeds of multiple screws, leading to instability in tread quality-related parameters. This invention proposes a model predictive control based on a multi-input multi-output (MIMO) model of the compound extrusion process. It identifies parameters of a nonlinear autoregressive model with external inputs based on production data from the compound extrusion system, using the identified model as the predictive model for model predictive control. This approach eliminates the need for complex mechanistic models. By designing a model predictive controller, the product quality of semi-finished tire tread products from the compound extrusion system is improved even under conditions of feed fluctuations.

[0153] The method of this invention includes an offline process and an online process. The offline process is to establish a nonlinear autoregressive data-driven model with external input based on historical production process data. This model includes the screw torque, which reflects the fluctuation of the feed rate, as a disturbance variable, and establishes a linear regression relationship between the tread weight per meter and the total tread width at the front weighing station. The online part includes multi-input multi-output model predictive control under production condition constraints, which realizes the corresponding adjustment of the rotational speed of different screws according to the actual changes in tread weight per meter and total tread width.

[0154] The method of this invention solves the problem of unstable tread product quality caused by the manual monitoring and adjustment by on-site operators in existing tread extrusion methods, thus ensuring the stability of the weight per meter and total width of the final extruded semi-finished tread. It reduces reliance on operators in the tread composite extrusion production process and improves the product quality of the semi-finished tread composite extrusion, even considering fluctuations in the feed rate of each screw.

[0155] The above technical solution is only one embodiment of the present invention. For those skilled in the art, based on the principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the technical solutions described in the specific embodiments of the present invention. Therefore, the foregoing description is only a preferred option and is not restrictive.

Claims

1. A method for controlling the composite extrusion of tire semi-finished parts, characterized in that: The method includes: Step 1: Collect and preprocess compound extrusion production data to obtain a dataset; Step 2: Use the dataset to obtain the prediction model and establish a linear regression model between the total tread width and the tread weight per meter. Step 3: Construct a model predictive controller using the predictive model and the linear regression model; Step 4, Production Debugging Phase; Step 5, Stabilize Production Phase.

2. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: The operation of collecting compound extrusion production data in step 1 includes: collecting the rotational speed and torque of all screws, as well as the tread weight per meter and total tread width measured at the front tread scale.

3. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: Step 2, which involves obtaining the predictive model from the dataset, includes: The prediction model adopts a multi-input multi-output nonlinear autoregressive model with measurable perturbation variables; the rotational speed of each screw is the input variable of the prediction model, the tread weight per meter and the total tread width measured at the front meter are the output variables of the prediction model, and the torque of each screw is the perturbation variable of the prediction model. The prediction model is trained using the dataset to obtain the various parameters in the prediction model.

4. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: Step 2 involves establishing a linear regression model between total tread width and tread weight per meter, which includes: Based on historical data, a linear regression model was established for the total tread width and tread weight per meter for a specific production formula.

5. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: Step 3 involves constructing a model predictive controller based on the cost function and constraints. U min ≤U(t)≤U max , △U min ≤U(t)≤ΔU max , Where J is the cost function, N C To control the time domain; It controls the predicted tread weight per meter and total tread width obtained from the prediction model within the control time domain; Y r (t)=[y r (t+1|t),y r (t+2|t),...,y r (t+N C |t)] T It controls the tread weight per meter and total tread width obtained from the linear regression model within the control time domain. U(t)=[u(t+1|t),u(t+2|t),...,u(t+N C |t)] T It is the optimal control sequence in the control time domain to be solved; ΔU(t) = [u(t|t) - u(t - 1), u(t + 1|t) - u(t|t),..., u(t + N C - 1|t) - u(t + N C - 2|t)] T is for control Rate of change of the control sequence within the domain; Matrix I, R, and S are the weight matrices for the output variable, input variable, and rate of change of the input variable, respectively. Y min and Y max This is a multidimensional vector representing the minimum and maximum values ​​of the total tread width and tread weight per meter of the output. U min and U max It is a multidimensional vector representing the minimum and maximum rotational speeds of each screw. ΔU min and ΔU max It is a multidimensional vector representing the minimum and maximum values ​​of the rate of change of rotational speed of each screw.

6. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: Step 4 includes the following operations: (41) Obtain the set tread weight per meter and total tread width based on the production formula and linear regression model; (42) Set the values ​​of the weight matrices I, R, and S of the model predictive controller; (43) The predicted tread weight per meter and total tread width are obtained by using the prediction model. Then, the control sequence is obtained by using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model prediction controller. That is, the control sequence is obtained by solving the minimum value optimization problem of the cost function under all constraints. (44) Apply the first control vector in the control sequence to the compound extrusion device; (45) Determine whether the error between the measured tread weight per meter and total tread width of the composite extrusion device and the set tread weight per meter and total tread width is stable within the tolerance range. If not, return to (42). If yes, end the production debugging stage.

7. The method for controlling the composite extrusion of tire semi-components according to claim 1, characterized in that: Step 5 includes the following operations: (51) The predicted tread weight per meter and total tread width are obtained by using the prediction model. Then, the control sequence is obtained by using the set tread weight per meter and total tread width, the predicted tread weight per meter and total tread width, and the model prediction controller. That is, the control sequence is obtained by solving the minimum value optimization problem of the cost function under all constraints. (52) Apply the first control vector in the control sequence to the compound extrusion device; (53) Return to step (51).

8. A tread composite extrusion system, characterized in that: The system includes: an industrial control computer, a model prediction control unit, a motor, and a composite extrusion device; The compound extrusion unit transmits data to the industrial control computer, which then sends the processed data to the model predictive controller. The model predictive control unit uses the predictive model, the linear regression model, and the model predictive controller to generate a control sequence, and sends the first control vector in the control sequence as a control signal to the motor. The motor controls the rotational speed of each screw in the compound extrusion unit according to the first control vector.

9. The system according to claim 8, characterized in that: The model prediction control unit includes: a prediction model, a cost function setting module, a constraint setting module, and an online optimization module; The prediction model is used to obtain the predicted tread weight per meter and total tread width; The cost function setting module is used to input the set tread weight per meter and total tread width provided by the linear regression model, as well as the predicted tread weight per meter and total tread width, into the cost function; The constraint setting module is used to input constraint conditions; The online optimization module is used to solve the minimum cost function optimization problem to obtain the control sequence under the condition of satisfying the constraints.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the tire half-component composite extrusion control method as described in any one of claims 1-7.

Images