A control optimization method for steering of palletizing robot

Abstract

The invention discloses a control optimization method for steering of a palletizing robot, which belongs to the field of steering control of a palletizing robot. The method comprises the following specific steps: S1, constructing a steering control system of the palletizing robot; S2, improving a standard weighted average optimization algorithm; S3, optimizing control parameters of a PID controller of the steering control system of the palletizing robot online and in real time by using the improved weighted average optimization algorithm; S4, the PID controller outputting a control quantity according to a set of optimal control parameters obtained in S3, and controlling the robot to rotate to a specified position by using a servo driver module; the PID controller in the steering control system of the palletizing robot is optimized by using the improved weighted average optimization algorithm, thereby improving the control accuracy and robustness of the system.

Description

Technical Field

[0001] The present application relates to the field of steering control of a palletizing robot, and in particular to a control optimization method for steering a palletizing robot. Background Art

[0002] Steering control of a palletizing robot is a key step in achieving automated palletizing operations. Its technical background stems from the demand for high-efficiency, high-precision handling operations in modern logistics and manufacturing. With the development of multi-degree-of-freedom robotic arm structures and servo drive technology, palletizing robots must not only perform basic pick-and-place operations but also possess flexible and precise steering capabilities to adapt to the operational requirements of different workstations, different postures, and complex stacking paths. Therefore, achieving smooth steering and accurate positioning of the end effector within a limited space has become a key technical challenge in the design of palletizing robot control systems. A PID controller is typically used to control the robot's steering. By adjusting the proportional (Kp), integral (Ki), and differential (Kd) parameters, a balance between rapid response to the target angle and steady-state accuracy is achieved. However, due to the complex working conditions and frequent load changes in palletizing operations, traditional PID control methods based on empirical parameter tuning often struggle to balance system response speed and robustness, and are prone to problems such as large overshoot, slow response, and unstable control. Therefore, combining intelligent optimization algorithms to optimize PID parameters online or offline has become an important technical approach to improve steering control performance and adapt to the needs of multi-scenario operations.

[0003] The weighted average optimization algorithm (WAA) is a new type of swarm intelligence optimization method. Its core idea is to perform a weighted average of the positions of several individuals in the population according to their individual fitness in each iteration process, and construct a guiding reference position to replace the single optimal solution guidance mechanism in traditional algorithms. The number of candidate individuals in WAA gradually increases with the number of iterations, thereby achieving a dynamic transition from global exploration to local development. In the development stage, the algorithm designs three weight combination methods to update the current individual, guiding it to approach the weighted position, the global optimal position, and the individual historical optimal position; in the exploration stage, WAA combines the Levy flight mechanism and the random reinitialization strategy to enhance the ability to escape the local optimum. The overall algorithm achieves good convergence and search diversity while maintaining a simple structure. However, when optimizing engineering problems in some complex scenarios, there will be problems with low search performance stability and easy trapping in local optimal solutions. Summary of the Invention

[0004] In response to the problems existing in the above-mentioned background technology, the present invention proposes a control optimization method for the steering of a palletizing robot, which aims to combine the traditional PID controller with the improved weighted average optimization algorithm and apply it to the steering control system of the palletizing robot. The improved weighted average optimization algorithm can better adapt to the application scenarios of the steering control of the palletizing robot, has higher accuracy in the optimization process, and significantly improves the adaptability and control stability of the PID controller. Through the method proposed in the present invention, the control accuracy and response speed of the steering control system of the palletizing robot are improved as a whole, the dynamic error of the robot during the rotation process is reduced, and a stable working state can be maintained under different industrial control conditions.

[0005] The present invention proposes a control optimization method for steering of a palletizing robot, which comprises the following steps:

[0006] S1. Construct a palletizing robot steering control system, which includes: an angle signal receiving module, a PID controller, an improved weighted average optimization algorithm module, a servo driver module, a robot rotation module, and a position detection sensor module.

[0007] S2. Improve the standard weighted average optimization algorithm and adopt two improvement strategies, including:

[0008] S21. Introducing a fractional-order hybrid perturbation mechanism based on information entropy modulation to improve the first strategy of the exploration phase of the standard weighted average optimization algorithm, wherein the fractional-order hybrid perturbation mechanism based on information entropy modulation is used to adaptively update the individual position, wherein the construction of the perturbation step size simultaneously refers to the global optimal position, population statistical information, and individual historical information, and combines the fractional order calculated based on information entropy to complete the dynamic adjustment of the perturbation direction and amplitude;

[0009] S22. A nonlinear interactive dual-domain weighting development mechanism is used to improve the first mobile strategy in the development stage of the standard weighted average optimization algorithm. The nonlinear interactive dual-domain weighting development mechanism constructs a multivariate combination structure consisting of global guidance, mean perturbation and individual response, and combines it with a state-driven dynamic weighting method to achieve position adjustment, thereby replacing the traditional linear weighted position update method.

[0010] S3. The control parameters of the PID controller of the palletizing robot steering control system are optimized online and in real time through the improved weighted average optimization algorithm, and the optimal set of Kp, Ki, and Kd parameters are found under the set number of iterations.

[0011] S4, the PID controller outputs the control quantity through a set of optimal control parameters obtained in S3, and controls the robot to rotate to the specified position through the servo driver module.

[0012] Preferably, the operation process of the palletizing robot steering control system described in S1 is as follows: the position detection sensor module detects the position information between the object to be transported and the robot in real time, sends the detected position information to the angle signal receiving module, calculates the target angle that the robot needs to rotate, and performs error calculation with the current angle of the robot to obtain the angle error e(t) that needs to be adjusted, and inputs the angle error e(t) into the PID controller. The PID controller optimizes the parameters through the improved weighted average optimization algorithm module and outputs the control quantity U(t). The control quantity U(t) is input into the servo driver module. The servo driver module adjusts the control current of the motor of the robot rotation module through the control quantity U(t), and completes the angle adjustment of the robot through the rotation of the motor. The mathematical model of the robot rotation module is:

[0013] (1);

[0014] In formula (1), J represents the moment of inertia of the system, B represents the damping coefficient of the system, θ(t) represents the rotation angle of the manipulator, Kt represents the torque constant of the motor, U(t) represents the control quantity output by the PID controller, and the calculation formula is shown in formula (2). TL(t) represents the external load torque;

[0015] (2);

[0016] In formula (2), Kp represents the proportional gain, Ki represents the integral gain, Kd represents the differential gain, and e(t) represents the angle error.

[0017] Preferably, the first strategy of the exploration phase of the standard weighted average optimization algorithm is improved by introducing a fractional-order hybrid perturbation mechanism based on information entropy modulation as described in S21. First, the distribution of the current population in each dimension is statistically analyzed, and the probability distribution of each dimension is calculated according to the equidistant partitioning method, and the global information entropy is obtained. Based on the information entropy, a normalized fractional-order parameter is generated, which is used as an order adjustment factor to control the perturbation intensity and step size. Finally, the mean of the global optimal position, perturbation compensation and population position is integrated to form an update formula with multi-source perturbation and self-adjusting step size capabilities, as shown in formula (3);

[0018] (3);

[0019] In formula (3), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, λ1 represents the perturbation weight 1, which is a random number between [0,1], λ2 represents the perturbation weight 2, which obeys the normal distribution, S(iter) represents the perturbation compensation, and the calculation formula is shown in formula (4). represents the mean of the population position, γ represents the fractional order, and the calculation formula is shown in formula (5);

[0020] (4);

[0021] In formula (4), S(iter) represents disturbance compensation, η represents the disturbance amplitude coefficient, which is 0.5, and rand represents a random number between [0, 1]. represents the mean of the population position, represents the historical optimal position of the individual position under the current number of iterations, and γ represents the fractional order;

[0022] (5);

[0023] In formula (5), dim represents the problem dimension, which is 3, Hd represents the information entropy, and the calculation formula is shown in formula (6). K represents the interval divided by each dimension;

[0024] (6);

[0025] In formula (6), p represents the probability distribution value of each dimension in the population.

[0026] Preferably, after introducing the fractional-order hybrid perturbation mechanism based on information entropy modulation, the constructed optimization method has a richer structural hierarchy and better diversity of perturbation patterns in terms of population diversity regulation, adaptive perturbation capability and local jump exploration. This improved method is applied to the PID control parameter optimization task in the steering control system of the palletizing robot. It can achieve deep mining and effective exploration of the multi-parameter search space based on the high coupling and nonlinear characteristics of the control system, adapt to the response characteristic adjustment requirements in complex environments, and is conducive to improving the coverage and convergence stability of parameter optimization.

[0027] Preferably, the use of a nonlinear interactive dual-domain weighting development mechanism described in S22 improves the first mobile strategy in the development stage of the standard weighted average optimization algorithm. The mechanism includes three parts: first, a global guidance item is constructed using a hyperbolic tangent function to form a nonlinear offset between the current individual and the optimal individual; second, a disturbance guidance item is constructed by the mean of the population position and the random individuals in the population; the disturbance guidance item is superimposed with a dynamic adjustment substructure formed by periodic disturbances and normal micro-perturbations; third, a nonlinear interactive response item is generated by the distribution difference of the individual position relative to the group structure. In terms of the weights of the three parts, a dual-domain dynamic weighting structure driven by relative distance is adopted. A proportional normalization relationship is constructed based on the distribution difference between the individual position and the global optimal solution and the individual and individual position mean, so as to generate non-fixed combination weights in real time. The specific mathematical model is:

[0028] (7);

[0029] In formula (7), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, represents the mean of the population position, α(iter) represents the dynamic adjustment factor, α(iter)=(iter / max_iter), iter represents the current number of iterations, max_iter represents the maximum number of iterations, ω1 represents the weight factor 1, and the calculation formula is shown in formula (8), ω2 represents the weight factor 2, ω2=exp(iter / max_iter), ω3 represents the weight factor 3, ω3=1-ω1-ω2, ψ(iter) represents the two-way difference disturbance factor, and the calculation formula is shown in formula (9), σ(iter) represents the nonlinear interaction enhancement factor, and the calculation formula is shown in formula (10);

[0030] (8);

[0031] In formula (8), X(iter) represents the current individual position, represents the global optimal individual, represents the mean of the population position, and ε represents a very small constant to prevent the denominator from being 0;

[0032] (9);

[0033] In formula (9), rand represents a random number between [0, 1], X(iter) represents the current individual position, Xrand represents the position of a randomly selected individual in the population, μ represents the perturbation amplitude coefficient, and randn() represents the normal distribution function;

[0034] (10);

[0035] In formula (10), represents the global optimal individual, represents the mean of the population position, X(iter) represents the current individual position, β represents the response intensity factor, and its value range is [0.5, 2]. sign() represents the sign function, which outputs 1, -1, or 0 according to the sign of the passed value.

[0036] Preferably, the algorithm is improved by using a nonlinear interactive dual-domain weight allocation development mechanism, which improves the adaptability of the algorithm in the development stage, especially in the local development stage in the later stage of the algorithm. Nonlinear perturbations can effectively prevent the algorithm from falling into local optimal solutions, and by dynamically allocating weights, the flexibility of the algorithm in the optimization process is enhanced. In the application scenario of manipulator steering control requiring high control accuracy, it can quickly adapt to the scenario, flexibly adjust the position information of the optimization individual, improve the optimization accuracy of the algorithm, and the control accuracy of the PID controller.

[0037] Preferably, in said S3, the control parameters of the PID controller of the palletizing robot steering control system are optimized online and in real time by using an improved weighted average optimization algorithm, and the specific steps are as follows:

[0038] S31. Initialize the parameters of the improved weighted average optimization algorithm, including the population size N, the problem dimension dim, the maximum number of iterations max_iter, the upper limit ub and lower limit lb of the search space, and initialize the population position;

[0039] S32. Applying the improved weighted average optimization algorithm to the PID controller parameter tuning process, by establishing a mapping relationship between the individual position vectors of the optimization algorithm and the PID parameters, so that the three components of the individual position vector correspond to the proportional coefficient Kp, the integral coefficient Ki and the differential coefficient Kd respectively. During the iterative evolution of the algorithm, each individual continuously moves in the solution space, and its corresponding PID parameter combination is dynamically updated accordingly. Through the optimization of the algorithm, the optimal control effect is gradually approached, and finally a set of optimal parameters of the PID controller is obtained;

[0040] S33, setting a fitness value function of the improved weighted average optimization algorithm, calculating the fitness value corresponding to the individual position in the initial population through the fitness value function, and sorting the fitness values ​​of the individuals, selecting the global optimal position in the population, and the specific fitness value function is;

[0041] (11);

[0042] In formula (11), J represents the fitness value, e(t) represents the angle error value, and T represents the system operation time;

[0043] S34. The positions of individuals in the population are updated using the mathematical model of the improved weighted average optimization algorithm. A greedy strategy is used to determine whether the fitness value of the current individual after the update is better than the fitness value before the update. Individuals with better fitness values ​​are retained, and the ranking of the fitness values ​​of the population is updated. The mathematical model of the improved weighted average optimization algorithm is divided into an exploration phase and an development phase.

[0044] S35. Determine whether the current number of iterations has reached the maximum number of iterations. If not, return to execute S34 to continue optimizing. If so, exit the optimization algorithm and output the optimal solution.

[0045] Compared with the existing technology, the present invention effectively enhances the dynamic control capability of the weighted average optimization algorithm in the exploration and development stages by introducing a fractional-order hybrid perturbation mechanism based on information entropy modulation and a nonlinear interactive dual-domain weighted development mechanism, achieving more comprehensive coverage of complex search spaces and improving the ability to escape local optimality. Especially in the PID parameter optimization of the steering control system of the stacking robot, the present invention can dynamically adjust the search strategy and step size according to the nonlinear and multivariable coupling characteristics of the system, achieve efficient convergence and stability improvement of parameter optimization, thereby significantly improving the response speed and accuracy of the control system, meeting the application requirements in multiple scenarios and multiple working conditions, and has broad engineering application prospects and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of a control optimization method for steering of a palletizing robot.

[0047] Figure 2 This is the model diagram of the steering control system of the palletizing robot.

[0048] Figure 3 A comparison chart of the changes in fitness values ​​during the optimization process of the improved weighted average optimization algorithm and the standard weighted average optimization algorithm.

[0049] Figure 4 Comparison chart of the response of the PID controller in the steering control system of the palletizing robot optimized by the improved weighted average optimization algorithm and the standard weighted average optimization algorithm. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work shall fall within the scope of protection of the present invention.

[0051] The present invention provides a technical solution: a control optimization method for steering a palletizing robot, which specifically includes the following steps: Figure 1 shown.

[0052] S1, construct a palletizing robot steering control system, the control system includes: an angle signal receiving module, a PID controller, an improved weighted average optimization algorithm module, a servo drive module, a robot rotation module, and a position detection sensor module, such as Figure 2shown.

[0053] Furthermore, the operation process of the palletizing robot steering control system described in S1 is as follows: the position detection sensor module detects the position information between the object to be transported and the robot in real time, sends the detected position information to the angle signal receiving module, calculates the target angle that the robot needs to rotate, and calculates the error with the current angle of the robot to obtain the angle error e(t) that needs to be adjusted, and inputs the angle error e(t) into the PID controller. The PID controller optimizes the parameters through the improved weighted average optimization algorithm module and outputs the control quantity U(t). The control quantity U(t) is input into the servo driver module. The servo driver module adjusts the control current of the motor of the robot rotation module through the control quantity U(t), and completes the angle adjustment of the robot through the rotation of the motor. The mathematical model of the robot rotation module is:

[0054] (1);

[0055] In formula (1), J represents the moment of inertia of the system, B represents the damping coefficient of the system, θ(t) represents the rotation angle of the manipulator, Kt represents the torque constant of the motor, U(t) represents the control quantity output by the PID controller, and the calculation formula is shown in formula (2). TL(t) represents the external load torque;

[0056] (2);

[0057] In formula (2), Kp represents the proportional gain, Ki represents the integral gain, Kd represents the differential gain, and e(t) represents the angle error.

[0058] Furthermore, the simulation code of the palletizing robot steering control system model is implemented in MATLAB, specifically:

[0059] function [t, theta] = mecharm_pid(Kp, Ki, Kd, ​​time, ref)

[0060] % Input parameters:

[0061] % Kp, Ki, Kd - PID controller proportional, integral, and derivative gains

[0062] %time simulation time

[0063] %ref target value

[0064] % Output parameters:

[0065] % t - time vector

[0066] % theta - the robot angle response at the corresponding time point

[0067] % System parameter initialization

[0068] J = 0.01; % moment of inertia (kg·m^2)

[0069] B = 0.1; % Damping coefficient (N·m·s)

[0070] Kt = 0.05; % motor torque constant (N·m / A)

[0071] TL = 10; % load moment (N·m)

[0072] % Set simulation time and step size

[0073] tspan = [0 time];

[0074] dt = 0.001; % time step

[0075] time = tspan(1):dt:tspan(2);

[0076] % Pre-allocated variables

[0077] theta = zeros(size(time)); % angle

[0078] omega = zeros(size(time)); % angular velocity

[0079] integral_e = 0; % Error integral term initialization

[0080] % Target angle (step input)

[0081] theta_ref = ref; % target angle

[0082] for k = 2:length(time)

[0083] % Calculation Error

[0084] e = theta_ref - theta(k-1);

[0085] % Integral item accumulation

[0086] integral_e = integral_e + e dt;

[0087] % differential term

[0088] derivative_e = (e - (theta_ref - theta(max(k-2,1)))) / dt;

[0089] % PID controller output

[0090] U = Kp e + Ki integral_e + Kd derivative_e;

[0091] % Discretization of manipulator dynamics differential equations (Euler method)

[0092] alpha = (Kt U-B omega(k-1) - TL) / J; % angular acceleration

[0093] % Update angular velocity and angle

[0094] omega(k) = omega(k-1) + alpha dt;

[0095] theta(k) = theta(k-1) + omega(k) dt;

[0096] end

[0097] t = time;

[0098] end;

[0099] The system is optimized through an improved weighted average optimization algorithm, and the obtained Kp, Ki, and Kd parameter values ​​are passed into the mecharm_pid() function. The angle that the manipulator needs to adjust is obtained by calculation and returned to the system for the next optimization.

[0100] S2. Improve the standard weighted average optimization algorithm and adopt two improvement strategies, including:

[0101] S21. Introducing a fractional-order hybrid perturbation mechanism based on information entropy modulation to improve the first strategy of the exploration phase of the standard weighted average optimization algorithm, wherein the fractional-order hybrid perturbation mechanism based on information entropy modulation is used to adaptively update the individual position, wherein the construction of the perturbation step size simultaneously refers to the global optimal position, population statistical information, and individual historical information, and combines the fractional order calculated based on information entropy to complete the dynamic adjustment of the perturbation direction and amplitude;

[0102] S22. A nonlinear interactive dual-domain weighting development mechanism is used to improve the first mobile strategy in the development stage of the standard weighted average optimization algorithm. The nonlinear interactive dual-domain weighting development mechanism constructs a multivariate combination structure consisting of global guidance, mean perturbation and individual response, and combines it with a state-driven dynamic weighting method to achieve position adjustment, thereby replacing the traditional linear weighted position update method.

[0103] Furthermore, the first strategy of the exploration phase of the standard weighted average optimization algorithm is improved by introducing a fractional-order hybrid perturbation mechanism based on information entropy modulation as described in S21. First, the distribution of the current population in each dimension is statistically analyzed, and the probability distribution of each dimension is calculated according to the equidistant partitioning method, and the global information entropy is obtained. Based on the information entropy, a normalized fractional-order parameter is generated, which is used as an order adjustment factor to control the perturbation intensity and step size. Finally, the mean of the global optimal position, perturbation compensation and population position is integrated to form an update formula with multi-source perturbation and self-adjusting step size capabilities, as shown in formula (3);

[0104] (3);

[0105] In formula (3), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, λ1 represents the perturbation weight 1, which is a random number between [0,1], λ2 represents the perturbation weight 2, which obeys the normal distribution, S(iter) represents the perturbation compensation, and the calculation formula is shown in formula (4). represents the mean of the population position, γ represents the fractional order, and the calculation formula is shown in formula (5);

[0106] (4);

[0107] In formula (4), S(iter) represents disturbance compensation, η represents the disturbance amplitude coefficient, which is 0.5, and rand represents a random number between [0, 1]. represents the mean of the population position, represents the historical optimal position of the individual position under the current number of iterations, and γ represents the fractional order;

[0108] (5);

[0109] In formula (5), dim represents the problem dimension, which is 3, Hd represents the information entropy, and the calculation formula is shown in formula (6). K represents the interval divided by each dimension;

[0110] (6);

[0111] In formula (6), p represents the probability distribution value of each dimension in the population.

[0112] Furthermore, the first mobile strategy in the development phase of the standard weighted average optimization algorithm is improved by using a nonlinear interactive dual-domain weighting development mechanism as described in S22. The mechanism includes three parts: first, a global guidance term is constructed using a hyperbolic tangent function to form a nonlinear offset between the current individual and the optimal individual; second, a disturbance guidance term is constructed by the mean of the population position and the random individuals in the population; the disturbance guidance term is superimposed with a dynamic adjustment substructure formed by periodic disturbances and normal micro-perturbations; third, a nonlinear interactive response term is generated by the distribution difference of the individual position relative to the group structure. In terms of the weights of the three parts, a dual-domain dynamic weighting structure driven by relative distance is adopted. The proportional normalization relationship is constructed based on the distribution difference between the individual position and the global optimal solution and the individual and individual position mean, so as to generate non-fixed combination weights in real time. The specific mathematical model is:

[0113] (7);

[0114] In formula (7), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, represents the mean of the population position, α(iter) represents the dynamic adjustment factor, α(iter)=(iter / max_iter), iter represents the current number of iterations, max_iter represents the maximum number of iterations, ω1 represents the weight factor 1, and the calculation formula is shown in formula (8), ω2 represents the weight factor 2, ω2=exp(iter / max_iter), ω3 represents the weight factor 3, ω3=1-ω1-ω2, ψ(iter) represents the two-way difference disturbance factor, and the calculation formula is shown in formula (9), σ(iter) represents the nonlinear interaction enhancement factor, and the calculation formula is shown in formula (10);

[0115] (8);

[0116] In formula (8), X(iter) represents the current individual position, represents the global optimal individual, represents the mean of the population position, and ε represents a very small constant to prevent the denominator from being 0;

[0117] (9);

[0118] In formula (9), rand represents a random number between [0, 1], X(iter) represents the current individual position, Xrand represents the position of a randomly selected individual in the population, μ represents the perturbation amplitude coefficient, and randn() represents the normal distribution function;

[0119] (10);

[0120] In formula (10), represents the global optimal individual, represents the mean of the population position, X(iter) represents the current individual position, β represents the response intensity factor, and its value range is [0.5, 2]. sign() represents the sign function, which outputs 1, -1, or 0 according to the sign of the passed value.

[0121] S3. The control parameters of the PID controller of the palletizing robot steering control system are optimized online and in real time through the improved weighted average optimization algorithm, and the optimal set of Kp, Ki, and Kd parameters are found under the set number of iterations.

[0122] Furthermore, in said S3, the control parameters of the PID controller of the palletizing robot steering control system are optimized online and in real time by using an improved weighted average optimization algorithm, and the specific steps are as follows:

[0123] S31. Initialize the parameters of the improved weighted average optimization algorithm, including the population size N, the problem dimension dim, the maximum number of iterations max_iter, the upper limit ub and lower limit lb of the search space, and initialize the population position;

[0124] S32. Applying the improved weighted average optimization algorithm to the PID controller parameter tuning process, by establishing a mapping relationship between the individual position vectors of the optimization algorithm and the PID parameters, so that the three components of the individual position vector correspond to the proportional coefficient Kp, the integral coefficient Ki and the differential coefficient Kd respectively. During the iterative evolution of the algorithm, each individual continuously moves in the solution space, and its corresponding PID parameter combination is dynamically updated accordingly. Through the optimization of the algorithm, the optimal control effect is gradually approached, and finally a set of optimal parameters of the PID controller is obtained;

[0125] S33, setting a fitness value function of the improved weighted average optimization algorithm, calculating the fitness value corresponding to the individual position in the initial population through the fitness value function, and sorting the fitness values ​​of the individuals, selecting the global optimal position in the population, and the specific fitness value function is;

[0126] (11);

[0127] In formula (11), J represents the fitness value, e(t) represents the angle error value, and T represents the system operation time;

[0128] S34. The positions of individuals in the population are updated using the mathematical model of the improved weighted average optimization algorithm. A greedy strategy is used to determine whether the fitness value of the current individual after the update is better than the fitness value before the update. Individuals with better fitness values ​​are retained, and the ranking of the fitness values ​​of the population is updated. The mathematical model of the improved weighted average optimization algorithm is divided into an exploration phase and an development phase.

[0129] S35. Determine whether the current number of iterations has reached the maximum number of iterations. If not, return to execute S34 to continue optimizing. If so, exit the optimization algorithm and output the optimal solution.

[0130] S4, the PID controller outputs the control quantity through a set of optimal control parameters obtained in S3, and controls the robot to rotate to the specified position through the servo driver module.

[0131] Furthermore, in order to verify that the present invention has certain advantages, an experiment is carried out in MATLAB. First, the mathematical model of the standard weighted average optimization algorithm is improved, and the function is named IWAA(N, max_iter, lb, ub, dim, fobj). The standard weighted average optimization algorithm function is named WAA(N, max_iter, lb, ub, dim, fobj), where N represents the population size, max_iter represents the maximum number of iterations, the search space upper limit ub, the lower limit lb, dim represents the problem dimension, fobj represents the fitness value calculation function handle, which points to the fitness value function fitness(e(t)) constructed by formula (11), and the input parameter is the angle error. A simulation model of the steering control system of the palletizing robot is constructed in simulink, and the controlled field of the steering control system of the palletizing robot is designed as a second-order transfer function. The transfer function is obtained by Laplace transformation as follows: , s represents the variable in the complex frequency domain. In the main function, initialize the algorithm parameters, max_iter=20, N=50, dim=3, [lb, ub]=[0.001, 100], call the improved and standard weighted average optimization algorithm functions respectively, and pass the obtained parameters to the system simulation program mecharm_pid(Kp, Ki, Kd, ​​time, ref) function. At the same time, start the simulink simulation model and keep the optimization record for visualization, such as Figure 3 and Figure 4 shown.

[0132] Furthermore, Figure 3 is a comparison of the changes in fitness values ​​during the optimization process of the improved weighted average optimization algorithm and the standard weighted average optimization algorithm. It can be seen from the figure that after the number of iterations reaches 8, the standard weighted average optimization algorithm falls into a local optimum and does not continue to optimize until the end of the iteration. However, the improved weighted average optimization algorithm can quickly jump out and continue to optimize after falling into a local optimal solution, and the fitness value of the solution finally found is better. Figure 4 is a comparison of the responses of the PID controller in the steering control system of the palletizing robot optimized by the improved weighted average optimization algorithm and the standard weighted average optimization algorithm. The target value is set to a unit value. It can be seen from the figure that the response curve output by the PID controller of the steering system of the palletizing robot optimized by the improved weighted average optimization algorithm has a lower overshoot, better stability, and stronger robustness.

Claims

1. A control optimization method for steering of a palletizing robot, characterized in that: The specific steps are as follows: S1. Construct a palletizing robot steering control system, which includes: an angle signal receiving module, a PID controller, an improved weighted average optimization algorithm module, a servo driver module, a robot rotation module, and a position detection sensor module; S2. Improve the standard weighted average optimization algorithm and adopt two improvement strategies, including: S21. A fractional-order hybrid perturbation mechanism based on information entropy modulation is introduced to improve the first strategy of the exploration phase of the standard weighted average optimization algorithm. The fractional-order hybrid perturbation mechanism based on information entropy modulation is used to adaptively update the individual position, wherein the construction of the perturbation step size simultaneously refers to the global optimal position, population statistical information, and individual historical information, and combines the fractional order calculated based on information entropy to complete the dynamic adjustment of the perturbation direction and amplitude, specifically: First, the distribution of the current population in each dimension is statistically analyzed, and the probability distribution of each dimension is calculated based on the equidistant partitioning method. The global information entropy is obtained from this information entropy. Based on this information entropy, a normalized fractional order parameter is generated, which is used as an order adjustment factor to control the disturbance intensity and step size. Finally, the mean of the global optimal position, disturbance compensation, and population position is integrated to form an update formula with multi-source disturbance and self-adjusting step size capabilities, as shown in formula (3). (3)； In formula (3), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, λ1 represents the perturbation weight 1, which is a random number between [0,1], λ2 represents the perturbation weight 2, which obeys the normal distribution, S(iter) represents the perturbation compensation, and the calculation formula is shown in formula (4). represents the mean of the population position, γ represents the fractional order, and the calculation formula is shown in formula (5); (4)； In formula (4), S(iter) represents disturbance compensation, η represents the disturbance amplitude coefficient, which is 0.5, and rand represents a random number between [0,1]. represents the mean of the population position, represents the historical optimal position of the individual position under the current number of iterations, and γ represents the fractional order; (5)； In formula (5), dim represents the problem dimension, which is 3, Hd represents the information entropy, and the calculation formula is shown in formula (6). K represents the interval divided by each dimension; (6)； In formula (6), p represents the probability distribution value of each dimension in the population; S22. A nonlinear interactive dual-domain weighting development mechanism is used to improve the first mobile strategy in the development phase of the standard weighted average optimization algorithm. The nonlinear interactive dual-domain weighting development mechanism constructs a multivariate combination structure consisting of global guidance, mean perturbation and individual response, and combines it with a state-driven dynamic weighting method to achieve position adjustment, thereby replacing the traditional linear weighted position update method. The mechanism includes three parts: first, a global guidance item is constructed using a hyperbolic tangent function to form a nonlinear offset between the current individual and the optimal individual; second, a perturbation guidance item is constructed by the mean of the population position and the random individuals in the population. The perturbation guidance item is superimposed with a dynamic adjustment substructure formed by periodic perturbations and normal micro-perturbations; third, a nonlinear interactive response item is generated by the distribution difference of the individual position relative to the group structure. In terms of the weights of the three parts, a relative distance-driven dual-domain dynamic weighting structure is adopted. A proportional normalization relationship is constructed based on the distribution difference between the individual position and the global optimal solution and the individual and individual position mean, so as to generate non-fixed combination weights in real time. The specific mathematical model is: (7)； In formula (7), X(iter+1) represents the updated position of the individual, X(iter) represents the current position of the individual, represents the global optimal individual, represents the mean of the population position, α(iter) represents the dynamic adjustment factor, α(iter)=(iter / max_iter), iter represents the current number of iterations, max_iter represents the maximum number of iterations, ω1 represents the weight factor 1, and the calculation formula is shown in formula (8), ω2 represents the weight factor 2, ω2=exp(iter / max_iter), ω3 represents the weight factor 3, ω3=1-ω1-ω2, ψ(iter) represents the two-way difference disturbance factor, and the calculation formula is shown in formula (9), σ(iter) represents the nonlinear interaction enhancement factor, and the calculation formula is shown in formula (10); (8)； In formula (8), X(iter) represents the current individual position, represents the global optimal individual, represents the mean of the population position, and ε represents a very small constant to prevent the denominator from being 0; （9）； In formula (9), rand represents a random number between [0, 1], X(iter) represents the current individual position, Xrand represents the position of a randomly selected individual in the population, μ represents the perturbation amplitude coefficient, and randn() represents the normal distribution function; （10）; In formula (10), represents the global optimal individual, represents the mean of the population position, X(iter) represents the current individual position, β represents the response intensity factor, and its value range is [0.5, 2]. sign() represents the sign function, which outputs 1, -1, or 0 according to the sign of the input value. S3. Optimize the control parameters of the PID controller of the palletizing robot steering control system online and in real time using an improved weighted average optimization algorithm, and find the optimal set of Kp, Ki, and Kd parameters under a set number of iterations; S4, the PID controller outputs the control quantity through a set of optimal control parameters obtained in S3, and controls the robot to rotate to the specified position through the servo driver module.

2. A control optimization method for steering a palletizing robot according to claim 1, characterized in that: In S1, the operation process of the palletizing robot steering control system is as follows: the position detection sensor module detects the position information between the object to be transported and the robot in real time, sends the detected position information to the angle signal receiving module, calculates the target angle that the robot needs to rotate, and performs error calculation with the current angle of the robot to obtain the angle error e(t) that needs to be adjusted, and inputs the angle error e(t) into the PID controller. The PID controller optimizes the parameters through the improved weighted average optimization algorithm module and outputs the control quantity U(t). The control quantity U(t) is input into the servo driver module. The servo driver module adjusts the control current of the motor of the robot rotation module through the control quantity U(t), and completes the angle adjustment of the robot through the rotation of the motor.

3. The control optimization method for steering a palletizing robot according to claim 2, characterized in that: In S3, the control parameters of the PID controller of the palletizing robot steering control system are optimized online and in real time by using an improved weighted average optimization algorithm. The specific steps are as follows: S31. Initialize the parameters of the improved weighted average optimization algorithm, including the population size N, the problem dimension dim, the maximum number of iterations max_iter, the upper limit ub and lower limit lb of the search space, and initialize the population position; S32. Applying the improved weighted average optimization algorithm to the PID controller parameter tuning process, by establishing a mapping relationship between the individual position vectors of the optimization algorithm and the PID parameters, so that the three components of the individual position vector correspond to the proportional coefficient Kp, the integral coefficient Ki and the differential coefficient Kd respectively. During the iterative evolution of the algorithm, each individual continuously moves in the solution space, and its corresponding PID parameter combination is dynamically updated accordingly. Through the optimization of the algorithm, the optimal control effect is gradually approached, and finally a set of optimal parameters of the PID controller is obtained; S33, setting a fitness value function of the improved weighted average optimization algorithm, calculating the fitness values ​​corresponding to the individual positions in the initial population through the fitness value function, sorting the fitness values ​​of the individuals, and selecting the global optimal position in the population; S34. The positions of individuals in the population are updated using the mathematical model of the improved weighted average optimization algorithm. A greedy strategy is used to determine whether the fitness value of the current individual after the update is better than the fitness value before the update. Individuals with better fitness values ​​are retained, and the ranking of the fitness values ​​of the population is updated. The mathematical model of the improved weighted average optimization algorithm is divided into an exploration phase and an development phase. S35. Determine whether the current number of iterations has reached the maximum number of iterations. If not, return to S34 and continue to search for the optimal solution. If so, exit the search and output the optimal solution.

Images