Temperature parameter self-tuning model prediction control method in rubber mixing process

Abstract

The invention relates to a model prediction control method for temperature parameter self-tuning in a rubber mixing process. The method comprises the following steps: S1, constructing a mechanism model and a rotor motor model in a mixing heat transfer process; s2, constructing a prediction model under an incremental control method based on the mechanism model and the rotor motor model; s3, constructing an MPC controller based on the prediction model, wherein the controller adopts a genetic algorithm to set parameters; s4, the MPC controller obtains the control target, the controlled quantity, the state quantity and the disturbance quantity after discrete sampling, the control target, the controlled quantity, the state quantity and the disturbance quantity are input into a prediction model, the prediction model outputs a prediction sequence, the prediction sequence is input into a rolling optimization module of the MPC controller to obtain an optimal control sequence, and the rolling optimization module adopts the parameters set in the S3; and S5, inputting the optimal control sequence into the internal mixer for control. Compared with the prior art, the method has the advantages that the problems that high-precision control over the internal mixing temperature is difficult, control parameters are difficult to set, and real-time observation is difficult are solved.

Description

Technical Field

[0001] The invention relates to the technical field of model predictive control, in particular to a model predictive control method for parameter self-tuning of temperature in a rubber mixing process. Background Art

[0002] The internal mixing process is a key step in the rubber production process. The quality of the rubber mix produced is directly related to the performance of the rubber product. Therefore, lean control of the internal mixing process must be achieved. During the internal mixing process, the internal mixer rotor stirs and mixes the rubber in a closed internal mixing chamber, causing the rubber raw materials to undergo a series of complex physical and chemical changes, and ultimately producing the desired rubber mix product. This process generates a large amount of heat, causing the temperature of the internal mixing chamber to rise. The appropriate temperature is conducive to the full progress of the reaction, while too high or too low a temperature will seriously affect the quality of the rubber mix. Therefore, it is of great significance to achieve precise control of the temperature during the internal mixing process. The current methods for controlling the temperature of the internal mixing process have the following main shortcomings:

[0003] (1) In the current actual mixing production, the control method of mixing temperature is open-loop control, which has poor control accuracy and poor anti-disturbance ability of the control system, and cannot effectively guarantee the quality of the mixed rubber products.

[0004] (2) Due to the high complexity of the mixing process, it is difficult to achieve effective control using existing conventional control methods, and the tuning of controller parameters often relies on experience, which further affects the control effect.

[0005] (3) The mixing process is quite closed, and it is difficult to observe the internal operation of the production equipment during the production process. Summary of the Invention

[0006] The purpose of the present invention is to solve the problems of high-precision control of mixing temperature, difficulty in setting control parameters, and difficulty in real-time observation, and to provide a model predictive control method for parameter self-tuning of the temperature of the rubber mixing process.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A model predictive control method for parameter self-tuning of temperature in a rubber mixing process, the method comprising the following steps:

[0009] S1. Construct the mechanism model of the mixing heat transfer process and the rotor motor model;

[0010] S2. Construct a prediction model under the incremental control method based on the mechanism model and the rotor motor model;

[0011] S3. Constructing an MPC controller based on the prediction model. The controller uses a genetic algorithm to tune parameters. The tuned parameters include a control domain N, a prediction domain P, an error weight matrix Q, and a control variable weight matrix R.

[0012] S4, MPC controller obtains the control target y after discrete sampling ref (k), controlled quantity u(k), state quantity x(k) and disturbance quantity d(k), where the control target is the current refining target temperature e ref , the controlled quantity is the armature voltage E of the rotor motor a , the state quantity includes the armature current i of the rotor motor a , motor speed n, mixing temperature e3, mixing chamber outer wall temperature e4 and high temperature part cooling water temperature e5, the disturbance includes the load torque T of the rotor motor L and cooling water flow q, the controlled variable u(k), state variable x(k) and disturbance variable d(k) are input into the prediction model, and the prediction model outputs the mixing temperature prediction sequence Y(k). The prediction sequence is input into the rolling optimization module of the MPC controller to obtain the optimal control increment sequence ΔU(k) ​​of the rotor motor armature voltage. The rolling optimization module adopts the parameters set in S3;

[0013] S5. The optimal control sequence is input into the internal mixing machine for control.

[0014] Furthermore, the prediction model under the incremental control method is:

[0015]

[0016] Among them, Δx(k+1) represents the state quantity increment at the next moment, A is the state transfer matrix, Δx(k) represents the state quantity increment at the current moment, Bu is the controlled quantity control matrix, Δu(k) represents the control quantity increment at the current moment, B d is the disturbance control matrix, y(k) represents the control quantity at the current moment, C is the observation matrix, and y(k-1) represents the control quantity at the previous moment;

[0017] The calculation process of the mixing temperature prediction sequence Y(k) is:

[0018] The prediction model is calculated N times on the prediction domain and outputs the control quantity y(k) at N discrete sampling moments after the current moment k. The control quantities y(k) at these N discrete sampling moments form a vector of length N, which is the prediction sequence Y(k).

[0019] Furthermore, the specific steps of inputting the prediction sequence into the rolling optimization module of the MPC controller to obtain the optimal control sequence are as follows:

[0020] The predicted sequence is substituted into the quadratic programming problem in the rolling optimization module, and the active set method is used to solve the quadratic programming problem to obtain the optimal control sequence.

[0021] Furthermore, the quadratic programming problem is:

[0022]

[0023] Among them, y(k+i|k) represents the control amount at time i after the current time k in the prediction sequence Y(k), y ref (k+i|k) represents the control target at time i after the current time k, and Δu(k+j-1|k) represents the control increment at time j-1 after the current time k;

[0024] The optimal control increment sequence ΔU(k) ​​of the armature voltage of the rotor motor is a vector composed of the optimal control increments at P discrete sampling moments obtained by solving the quadratic programming problem.

[0025] Furthermore, the specific steps of using genetic algorithm to tune the controller parameters are as follows:

[0026] Analyze the control domain N, prediction domain P, error weight matrix Q, and control variable weight matrix R. The error weight matrix Q = diag{q,q,…,q,f}, and the control variable weight matrix R = diag{r,r,…,r}, where q is the tracking error weight at all times in the prediction domain except the final state, f is the tracking error weight at the final state of the prediction domain, and r is the weight of the incremental modulus of the controlled variable.

[0027] The five parameters N, P, q, f, and r are used as parameter combinations. The genotype of the parameter combination is represented by a genetic code consisting of n-bit binary numbers. After randomly initializing m population genotypes, each parameter is binary decoded.

[0028] Calculate the fitness, and use the random selection method to select the excellent genotype of the population based on the fitness. For the excellent genotype, select the random position on the gene code with the crossover probability to perform genotype crossover, and then use the mutation probability P m Random positions on the gene code are selected for genotype variation, and the optimal parameter combination corresponding to the best fitness of the population is obtained by iterative calculation. The optimal parameter combination is the tuned parameter.

[0029] Furthermore, the parameters for binary decoding are:

[0030]

[0031] Among them, x∈[a,b] is the value range of parameter x, d i is the binary value of the i-th bit, and a and b are the upper and lower limits of the parameter.

[0032] Furthermore, the fitness is:

[0033]

[0034] Among them, f i represents the fitness of the i-th genotype, k j is the weight coefficient corresponding to the jth indicator under each set of parameters, N i,j is the jth index of the i-th genotype.

[0035] Furthermore, the indicators consist of overshoot, rise time, adjustment time and steady-state error.

[0036] Furthermore, when using the random selection method to select the best genotypes in the population, the probability of selection is:

[0037]

[0038] Where P(i) represents the probability of selecting the i-th genotype.

[0039] Furthermore, the mechanism model of the internal mixing heat transfer process is:

[0040]

[0041] Where n is the rotor speed, q is the cooling water flow rate, K1 and K2 are the heat transfer coefficients, Sf is the heat flow source generated by the input power of the main motor, C3, C4, and C5 are the equivalent heat capacities of the rubber compound, the mixing chamber wall, and the cooling water in the high-temperature part, respectively; R6, R7, and R8 are the thermal resistances between the rubber compound and the mixing chamber wall, the mixing chamber wall and the cooling water, and the mixing chamber wall and the air, respectively; e3, e4, and e5 are the mixing temperature, the mixing chamber outer wall temperature, and the high-temperature part cooling water temperature, respectively;

[0042] The rotor motor model is:

[0043]

[0044] Among them, E a ,i a ,R a ,L a are the armature voltage, current, resistance and inductance of the motor respectively; Δu is the brush voltage drop; K e ,K T are the motor back electromotive force coefficient and torque coefficient respectively; T L is the load torque, J is the moment of inertia of the motor and load, B is the viscous friction coefficient, and n is the motor speed.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) Based on the analysis of the internal mixer heat transfer process, this paper constructs a simulation model of the internal mixer temperature model predictive control (MPC) system. Compared with the model-free control method used in actual production, the constructed model can provide a theoretical design basis for more complex, model-dependent advanced control systems. Compared with the existing model, the addition of the internal mixer rotor motor model further improves the model.

[0047] (2) To address the difficulty of tuning model predictive controller parameters, a genetic algorithm is used to optimize and tune controller parameters offline. The effectiveness of this method is verified through simulation experiments. Compared with traditional controllers, model predictive controller parameters are more complex and have strong coupling. Traditional parameter tuning methods that rely on manual experience consume huge labor costs and are ineffective. The method proposed in this paper can effectively improve these problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention provides a predictive control system structure for the internal mixing temperature model;

[0049] Figure 2 Simulink model for predictive control system of mixing process temperature model;

[0050] Figure 3 is the MPC controller in the simulation model;

[0051] Figure 4 It is the mixing temperature control process in the simulation model;

[0052] Figure 5 Optimize the control parameter process for genetic algorithm;

[0053] Figure 6 Control effect curve for parameter optimization experiment;

[0054] Figure 7 Run the pipeline for the twin visualization model. ;

[0055] Figure 8 Unity 3D model of internal mixer;

[0056] Figure 9 For twin visualization experiments. DETAILED DESCRIPTION

[0057] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0058] The present invention proposes a model predictive control method for parameter self-tuning of the rubber mixing process temperature, and realizes twin visualization of the control process, which is used to achieve lean control of the quality of the mixing product. The solution includes the following contents: a simulation model of the mixing temperature model predictive control (MPC) system is constructed, a genetic algorithm-based mixing temperature MPC controller parameter optimization tuning method is invented, and a twin 3D visualization model of the mixing machine is constructed. The present invention realizes effective control of the mixing process temperature. The selection of controller parameters does not need to rely on manual experience, and the production process can be observed in real time. Compared with traditional control methods, the control accuracy is improved, and real-time observation of the production process also improves safety. Taking a certain type of mixing machine as an example, the effectiveness of the method is verified.

[0059] The present invention proposes a model predictive control method for parameter self-tuning of the temperature of the rubber mixing process in tire production, and realizes twin visualization of the control process, which can solve the problems of high-precision control of mixing temperature, difficulty in adjusting control parameters, and difficulty in real-time observation.

[0060] In order to achieve the above object, the design scheme of the present invention includes the following steps:

[0061] (1) Based on the analysis of the internal mixing heat transfer process, a simulation model of the internal mixing process temperature model predictive control (MPC) system is constructed.

[0062] (2) To address the difficulty in tuning the parameters of the model predictive controller, a genetic algorithm is used to achieve offline optimization and tuning of the controller parameters; the effectiveness of the method is verified through simulation experiments.

[0063] (3) Build a twin 3D model of the mixing process based on the Unity 3D platform and realize the twin visualization of the control process.

[0064] This invention is divided into three parts, which describe in detail an offline adaptive model predictive control (MPC) method for the temperature of a rubber mixing process. (1) The mixing process temperature control system is modeled, the MPC controller of the mixing process temperature control system is designed, and a simulation model is constructed. (2) A genetic algorithm is used to automatically optimize and set the parameters of the MPC controller of the mixing process temperature, and the method is verified through simulation experiments. (3) A Unity 3D model of the mixing machine is constructed, and a twin visualization of the control process is constructed through a networked approach.

[0065] Part 1:

[0066] During the mixing process, under the action of the rotating action of the mixer rotor, the heat generated by the friction between the rubber material, the rotor and the mixing chamber wall will cause the mixing chamber to heat up; at the same time, the air and cooling water flowing through the mixing chamber wall and the rotor will take away the excess heat, causing the mixing chamber to cool down. Based on the analysis of the mixing heat transfer process, the mechanism model of the mixing heat transfer process can be obtained based on the bond graph theory as shown in formula (1):

[0067]

[0068] Where n is the rotor speed, q is the cooling water flow rate, K1 and K2 are heat transfer coefficients, Sf is the heat flow source generated by the input power of the main motor, Se is the equivalent heat potential source of the cooling water in the water inlet part; C3, C4, and C5 are the equivalent heat capacities of the rubber compound, the mixing chamber wall, and the cooling water in the high-temperature part respectively; R6, R7, and R8 are the thermal resistances between the rubber compound and the mixing chamber wall, the mixing chamber wall and the cooling water, and the mixing chamber wall and the air respectively; e3, e4, and e5 are the mixing temperature, the mixing chamber outer wall temperature, and the high-temperature part cooling water temperature respectively.

[0069] The temperature control of the mixing process is mainly achieved by adjusting the rotor speed. The rotor motor type commonly used in the current production process is a brushed DC motor, and its model is shown in formula (2):

[0070]

[0071] Where, E a ,i a ,R a ,L a are the armature voltage, current, resistance and inductance of the motor respectively; Δu is the brush voltage drop; K e ,K T are the motor back electromotive force coefficient and torque coefficient respectively; T L is the load torque, J is the moment of inertia of the motor and load, B is the viscous friction coefficient, and n is the motor speed.

[0072] By connecting the rotor motor model and the mixing heat transfer process model in series, the mathematical model of the mixing temperature control system can be obtained, and on this basis, the model predictive control system can be further designed.

[0073] Figure 1 The structure of the control system for the internal mixing temperature model prediction.

[0074] The prediction model is a discrete state space model. For the rotor motor, the state quantity x1=(i a n) T , input u1=(E a T L ) T , the model can be transformed into a second-order state space model; for the heat transfer process, the state quantity x2=(e3 e4 e5) T , input u2=(nq) T , since the state quantity and control quantity at each moment on the desired temperature curve of the mixing process satisfy the relationship: Therefore, by performing a first-order Taylor expansion on f at any point on the desired curve, the heat transfer model can be linearized and converted into a third-order state space model. The above two models are connected in series, discretized using the one-step Euler method, and converted into a prediction model under the incremental control method as shown in formula (3):

[0075]

[0076] The control input is the armature voltage E of the rotor motor a , the disturbance input is the motor load torque T L and cooling water flow q, the output is the temperature Q of the mixing chamber (i.e., the state quantity e3 in the heat transfer model). The prediction model outputs the mixing temperature prediction sequence Y(k). When the prediction model corresponding to Formula 3 performs the prediction step, it needs to iterate N times on the prediction domain. The mixing temperature prediction sequence output by the prediction model represents the predicted mixing temperature at the next N discrete sampling moments. During the iterative calculation process, each time △x(k+1) is calculated, △x(k+1) is used as the △x(k) for the next calculation. Finally, N y(k) can be obtained through Formula 3. y(k) forms a vector of length N, that is, the prediction sequence Y(k), where N is the length of the prediction domain.

[0077] The rolling optimization problem is a quadratic programming problem in the form of formula (4):

[0078]

[0079] Where N and P are the prediction domain and control domain respectively, Q and R are weight coefficient matrices. This optimization problem is solved by the active set method.

[0080] Figure 2 This is a simulation model of the mixing process temperature model predictive control system built in the Simulink toolbox under MATLAB R2024a version. The model consists of two parts: the model predictive controller and the mixing temperature control process.

[0081] Figure 3 It is a model predictive controller module. This module first calculates the matrix parameters of the prediction model and the matrix parameters of the optimization function, and at the same time receives the control target, controlled quantity, state quantity and disturbance quantity information after discrete sampling. Then, it calculates the predicted output sequence through the prediction model. Finally, it calculates the optimal control sequence through the rolling optimization module and outputs the optimal rotor motor voltage control quantity according to the requirements of the production control cycle.

[0082] Figure 4This module simulates the actual production process, using the rotor motor voltage as the control input, the motor load torque and cooling water flow rate as the disturbance inputs, the mixer temperature as the controlled output, and other predictive model state variables as state outputs. Using differential equation modeling, simulation models for the DC rotor motor and the mixer heat transfer process are constructed separately, and the two are connected in series to create the controlled object.

[0083] Part II:

[0084] In order to solve the parameter tuning problem of the MPC controller, a genetic algorithm is used to optimize the model parameters. The process is as follows: Figure 5 The specific process is as follows:

[0085] In an MPC controller, there are four sets of parameters that need to be tuned: the control domain N, the prediction domain P, the error weight matrix Q = diag{q,q,…,q,f}, and the control variable weight matrix R = diag{r,r,…,r}. Therefore, the parameters that need to be optimized are N, P, q, f, and r. Assume that the genotypes of the five parameters are all represented by n-bit binary numbers. After randomly initializing m population genotypes, first perform binary decoding on each parameter, as shown in formula (5):

[0086]

[0087] Where x∈[a,b] is the range of the parameter x, d i is the binary value of the i-th bit.

[0088] Assign parameters and call the Simulink simulation model to obtain the step response curve of temperature control and determine whether the response converges. If the response converges, calculate the following control indicators: overshoot, rise time, adjustment time, steady-state error. The above indicators are recorded as N j (j=1,2,3,4), then the fitness function of the genetic algorithm can be defined as shown in formula (6):

[0089]

[0090] Where k j It is the weight coefficient corresponding to each indicator under each set of parameters, which represents the degree of attention paid to each indicator in the response curve. Adjusting this value can adjust the incentive or inhibition of the corresponding control indicator.

[0091] The excellent genotypes in the population are selected using the random selection method. The probability of each genotype being selected is calculated according to formula (7):

[0092]

[0093] For the selected population genotype, with crossover probability Pc Select a random position on the gene code for genotype crossover with a mutation probability P m Random positions on the gene code were selected for genotyping.

[0094] The above process is repeated cyclically, and the optimal fitness of the population is calculated until the optimal genotype is found. The corresponding control parameters are a set of optimal control parameters.

[0095] Taking a certain type of internal mixer as an example, the internal mixer model parameters are shown in Table 1. The discretization period of the prediction model is 1ms, the control period of the MPC controller is 1s, and the target control temperature is 140℃.

[0096] Table 1 Refining model parameters

[0097]

[0098]

[0099] The optimization running environment of the genetic algorithm is Python 3.11.5 (conda), and the key parameters are shown in Table 2.

[0100] Table 2 Genetic algorithm parameters

[0101]

[0102] The control parameters obtained by genetic algorithm optimization are N=5, P=4, q=95842.10, f=3754678668.22, and r=95.82. The control effect curve is as follows: Figure 6 As shown in the figure, the actual temperature of the mixing chamber reaches the target temperature value in about 50s, with no steady-state error and no overshoot in the transition process, which is consistent with the actual requirements of the mixing process. It proves that the MPC parameters obtained by optimization can achieve good temperature control effect.

[0103] Part III:

[0104] The twin visualization model of the mixing process can help engineers visually observe the operation of the system. Based on the Unity 3D platform version 2022.3.8f1c1, a 3D model of the mixer is built, and a twin is constructed through the networked module. The operation process of the twin visualization model is as follows: Figure 7 shown.

[0105] The Simulink simulation model and Unity 3D model of the mixing process use their own communication modules to achieve UDP communication between different devices on the same network. Simulink packages and sends real-time mixer operating data to Unity. After unpacking and processing, code controls the movement of the 3D model, achieving twin visualization. Users are responsible for maintaining the IP addresses and ports of the communication modules on both ends.

[0106] The 3D model of the internal mixer was modeled by SolidWorks software and converted into a standard 3D format (.fbx) file and imported into Unity3D. Figure 8 The description of each submodule of the 3D model is shown in Table 3.

[0107] Table 3 Description of submodules of the internal mixer Unity 3D model

[0108]

[0109] This 3D model of the internal mixer can demonstrate the three main movements of the internal mixing process: the opening and closing of the blanking plate, the raising and lowering of the upper bolt, and the rotation of the rotor. All of these movements are driven by C# code scripts.

[0110] In terms of twin construction, the Simulink side uses the "Byte Pack" and "UDP Send" modules to package and send the blanking plate switch signal, upper bolt displacement and rotor speed data to the specified IP address and port number; the Unity side uses the C# script communication coroutine to receive and unpack the data, and controls the movement of the 3D model accordingly, realizing real-time visualization of the working status of the internal mixer. Figure 9 shown.

[0111] The present invention constructs a model predictive control (MPC) system simulation model of the mixing process temperature, and combines it with a genetic algorithm to effectively solve the problems of precise temperature control and parameter adjustment in the mixing process, and can effectively realize the mixing temperature control; a twin visualization model is constructed to effectively solve the problem of real-time observation of the closed production process, making the production and control process more intuitive, and has guiding significance for the control of the mixing production process.

[0112] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A model predictive control method for parameter self-tuning of temperature in a rubber mixing process, characterized in that: The method comprises the following steps: S1. Construct the mechanism model of the mixing heat transfer process and the rotor motor model; S2. Construct a prediction model under the incremental control method based on the mechanism model and the rotor motor model; S3. Constructing an MPC controller based on the prediction model. The controller uses a genetic algorithm to tune parameters. The tuned parameters include a control domain N, a prediction domain P, an error weight matrix Q, and a control variable weight matrix R. S4, MPC controller obtains the control target y after discrete sampling ref (k), controlled quantity u(k), state quantity x(k) and disturbance quantity d(k), where the control target is the current refining target temperature e ref , the controlled quantity is the armature voltage E of the rotor motor a , the state quantity includes the armature current i of the rotor motor a , motor speed n, mixing temperature e3, mixing chamber outer wall temperature e4 and high temperature part cooling water temperature e5, the disturbance includes the load torque T of the rotor motor L and cooling water flow q, the controlled variable u(k), state variable x(k) and disturbance variable d(k) are input into the prediction model, and the prediction model outputs the mixing temperature prediction sequence Y(k). The prediction sequence is input into the rolling optimization module of the MPC controller to obtain the optimal control increment sequence ΔU(k) ​​of the rotor motor armature voltage. The rolling optimization module adopts the parameters set in S3; S5. The optimal control sequence is input into the internal mixing machine for control.

2. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 1, characterized in that: The prediction model under the incremental control method is: Among them, Δx(k+1) represents the state increment at the next moment, A is the state transfer matrix, Δx(k) represents the state increment at the current moment, B u is the controlled variable control matrix, Δu(k) represents the control variable increment at the current moment, B d is the disturbance control matrix, y(k) represents the control quantity at the current moment, C is the observation matrix, and y(k-1) represents the control quantity at the previous moment; The calculation process of the mixing temperature prediction sequence Y(k) is: The prediction model is calculated N times on the prediction domain and outputs the control quantity y(k) at N discrete sampling moments after the current moment k. The control quantities y(k) at these N discrete sampling moments form a vector of length N, which is the prediction sequence Y(k).

3. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 2, characterized in that: The specific steps for inputting the prediction sequence into the rolling optimization module of the MPC controller to obtain the optimal control sequence are as follows: The predicted sequence is substituted into the quadratic programming problem in the rolling optimization module, and the active set method is used to solve the quadratic programming problem to obtain the optimal control sequence.

4. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 3, characterized in that: The quadratic programming problem is: Among them, y(k+i|k) represents the control amount at time i after the current time k in the prediction sequence Y(k), y ref (k+i|k) represents the control target at time i after the current time k, and Δu(k+j-1|k) represents the control increment at time j-1 after the current time k; The optimal control increment sequence ΔU(k) ​​of the armature voltage of the rotor motor is a vector composed of the optimal control increments at P discrete sampling moments obtained by solving the quadratic programming problem.

5. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 1, characterized in that: The specific steps of using genetic algorithm to tune the controller parameters are as follows: Analyze the control domain N, prediction domain P, error weight matrix Q, and control variable weight matrix R. The error weight matrix Q = diag{q,q,…,q,f}, and the control variable weight matrix R = diag{r,r,…,r}, where q is the tracking error weight at all times in the prediction domain except the final state, f is the tracking error weight at the final state of the prediction domain, and r is the weight of the incremental modulus of the controlled variable. The five parameters N, P, q, f, and r are used as parameter combinations. The genotype of the parameter combination is represented by a genetic code consisting of n-bit binary numbers. After randomly initializing m population genotypes, each parameter is binary decoded. Calculate the fitness, and use the random selection method to select the excellent genotype of the population based on the fitness. For the excellent genotype, select the random position on the gene code with the crossover probability to perform genotype crossover, and then use the mutation probability P m Random positions on the gene code are selected for genotype variation, and the optimal parameter combination corresponding to the best fitness of the population is obtained by iterative calculation. The optimal parameter combination is the tuned parameter.

6. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 5, characterized in that: The parameters for binary decoding are: Among them, x∈[a,b] is the value range of parameter x, d i is the binary value of the i-th bit, and a and b are the upper and lower limits of the parameter.

7. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 6, characterized in that: The fitness is: Among them, f i represents the fitness of the i-th genotype, k j is the weight coefficient corresponding to the jth indicator under each set of parameters, N i,j is the jth index of the i-th genotype.

8. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 7, characterized in that: The indicators are composed of overshoot, rise time, settling time and steady-state error.

9. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 8, characterized in that: When using random selection to select the best genotypes in a population, the probability of selection is: Where P(i) represents the probability of selecting the i-th genotype.

10. The model predictive control method for parameter self-tuning of temperature in a rubber mixing process according to claim 8, characterized in that: The mechanism model of the internal mixing heat transfer process is: Where n is the rotor speed, q is the cooling water flow rate, K1 and K2 are the heat transfer coefficients, Sf is the heat flow source generated by the input power of the main motor, C3, C4, and C5 are the equivalent heat capacities of the rubber compound, the mixing chamber wall, and the cooling water in the high-temperature part, respectively; R6, R7, and R8 are the thermal resistances between the rubber compound and the mixing chamber wall, the mixing chamber wall and the cooling water, and the mixing chamber wall and the air, respectively; e3, e4, and e5 are the mixing temperature, the mixing chamber outer wall temperature, and the high-temperature part cooling water temperature, respectively; The rotor motor model is: Among them, E a ,i a ,R a ,L a are the armature voltage, current, resistance and inductance of the motor respectively; Δu is the brush voltage drop; K e ,K T are the motor back electromotive force coefficient and torque coefficient respectively; T L is the load torque, J is the moment of inertia of the motor and load, B is the viscous friction coefficient, and n is the motor speed.

Images