Linear rotation permanent magnet motor parameter identification method based on cuckoo optimization algorithm

Abstract

The invention discloses a Cuckoo optimization algorithm-based linear rotation permanent magnet motor parameter identification method, and relates to the technical field of motor parameter identification. Specifically, a parameter identification model of the linear rotating permanent magnet motor is established under a synchronous rotating coordinate system; parameters and populations of a cuckoo optimization algorithm are initialized through chaotic mapping, and the positions of nests are updated through Levy flight; obtaining a dynamic discovery probability P of the cuckoo nest; optimizing the dimension-by-dimension reverse learning strategy by adopting a nonlinear weight coefficient; establishing a fitness function according to an identification value of the identification model and an actual measurement value, so as to correct an error between the identification model and actual output and obtain an optimal identification parameter; and correcting an identification model and an actual output error according to the fitness function until the error is minimum, if a precision requirement is met, ending, otherwise, continuing iteration until the set maximum number of iterations is reached, stopping iteration to obtain an optimal identification parameter, and improving the accuracy of parameter identification of the linear rotation permanent magnet motor.

Description

Technical Field

[0001] The invention relates to the technical field of motor parameter identification, and in particular to a linear rotary permanent magnet motor parameter identification method based on a cuckoo optimization algorithm. Background Art

[0002] The linear and rotary permanent magnet motor (LRPM) is a motor that can achieve both linear and rotary motion. It combines the characteristics of permanent magnet rotary motors and permanent magnet linear motors and can achieve two degrees of freedom, i.e. linear motion and rotary motion. It is widely used in the fields of intelligent manufacturing equipment, transportation equipment, new energy power generation, national defense equipment, etc.

[0003] Parameter identification of linear rotary permanent magnet motors is an important step to ensure the accuracy of motor control systems. It includes parameters such as resistance, inductance, and flux linkage. Accurate parameter identification can improve the performance of motor control systems, achieve more precise positioning and more efficient energy utilization, and plays an important role in improving the dynamic performance of control systems.

[0004] At present, in the parameter identification of linear rotary permanent magnet motors, such as the high-speed rotary aerodynamic parameter identification method based on extended Kalman filtering proposed in Chinese patent announcement number CN109376364B, the extended Kalman filter algorithm is used to identify the parameters, but the results are greatly affected by the initial values. This will reduce the robustness of parameter identification, because the error of the initial value will be amplified through the iterative process of EKF, thereby affecting the final identification accuracy. In addition, if the system model is nonlinear, EKF needs to linearize the model in each iteration step, which also introduces additional errors.

[0005] Therefore, the present application aims to improve the accuracy of parameter identification of a linear rotating permanent magnet motor. Summary of the invention

[0006] The main purpose of the present invention is to provide a method for improving the accuracy of parameter identification of a linear rotating permanent magnet motor.

[0007] In order to achieve the above object, the present invention proposes a linear rotary permanent magnet motor parameter identification method based on a cuckoo optimization algorithm, comprising the following steps:

[0008] Step S1: establishing a parameter identification model of a linear rotating permanent magnet motor in a synchronous rotating coordinate system;

[0009] Step S2: Use chaos mapping to initialize the parameters and population of the cuckoo optimization algorithm, and use Levy flight to update the location of the nest, which is expressed as follows:

[0010]

[0011] x t+1,i Indicates the location of the nest after the update; x t,i represents the position of the nest before updating; β is the control factor of Levy flight; α′ is the improved variable step length control factor; the expression of α′ is:

[0012]

[0013] Among them, α represents the step size factor of the previous iteration; γ represents the coefficient controlling the exponential decay rate; t represents the current iteration number; t max represents the maximum number of iterations; k represents the nonlinear adjustment coefficient;

[0014] Step S3: Obtain the dynamic discovery probability P of the cuckoo's nest;

[0015] Step S4: Optimizing the dimension-by-dimension reverse learning strategy using a nonlinear weight coefficient to improve the convergence efficiency and global optimization capability of the algorithm;

[0016] Step S5: Establish a fitness function with the identification value of the parameter identification model and the actual measurement value to correct the error between the identification model and the actual output until the accuracy requirement is met or t=t max Stop the iteration when , and obtain the optimal identification parameters.

[0017] Furthermore, the parameter identification model is:

[0018]

[0019] Among them, u dr and u qr are the d-axis and q-axis voltages of the rotary motion unit respectively; u dl and u ql are the d-axis and q-axis voltages of the linear motion unit respectively; w e and v e are the electrical angular velocity and linear velocity of the motor respectively; R sr and R sl are the rotation and linear motion phase winding resistances respectively; is the magnetic flux of the permanent magnet of the rotating motion unit, is the magnetic flux of the permanent magnet of the linear motion unit; L dr and L qr are the d-axis and q-axis inductances of the stator winding of the rotary motion unit respectively; L dl and L ql are the d-axis and q-axis inductances of the stator winding of the linear motion unit respectively; i dr and i qr are the d-axis and q-axis currents of the stator of the rotary motion unit respectively; i dl and iql They are the d-axis and q-axis currents of the stator of the linear motion unit respectively; the top mark “^” represents the identification value of the identification model, and the top mark without mark represents the measured value of the actual model.

[0020] Furthermore, in step S2, the parameters and population of the cuckoo optimization algorithm are initialized using the Circle chaos map, and the expression of the Circle chaos map is:

[0021]

[0022] Among them, n i represents the i-th chaotic sequence number, mod(a, b) represents the modulo operation of a over b.

[0023] Furthermore, in step S2, the Levy random flight path expression is:

[0024]

[0025] Among them, u and v both represent standard normal random variables, Γ is the Gamma function, and Φ is the proxy parameter in the dynamic Levy variable step size strategy.

[0026] Further, in step S3, the dynamic discovery probability P adopts a sinusoidally decreasing dynamic discovery probability, and the expression of the dynamic discovery probability P is:

[0027]

[0028] Among them, P min is the minimum discovery probability; P max is the maximum discovery probability.

[0029] Furthermore, in step S4, the population size is N, the dimension is D, and the position when the iterative process selects the updated position is Temp(i); the upper and lower limits of each dimension are Ub(i) and Lb(i), respectively, and the solution of each dimension based on weighted optimization reverse learning is:

[0030] Temp'(i)=Ub(i)+Lb(i)-Temp(i)·w;

[0031] Temp′(i) is the position after reverse learning based on weight optimization; w is the nonlinear weight coefficient.

[0032] Furthermore, the expression of the nonlinear weight coefficient w is:

[0033]

[0034] Furthermore, in step S5, the square of the difference between the identification value of the identification model and the actual measurement value is used to establish a fitness function, and the expression of the function is:

[0035]

[0036] Among them, w1, w2, w3 and w4 are weighting factors of the fitness function.

[0037] The above technical solution has the following advantages:

[0038] The present invention establishes a mathematical model and a motor parameter identification model of a linear rotary permanent magnet motor; adopts a chaotic mapping to initialize a population and introduces an improved exponentially decreasing step size control factor α′ to improve the global optimization capability of the algorithm; then adopts an improved dynamic discovery probability P and an optimized dimension-by-dimension reverse learning strategy to improve the convergence speed and global optimization capability of the algorithm; finally, establishes a fitness function for parameter identification, and corrects the error between the identification model and the actual output according to the fitness function until it is minimized. If the accuracy requirement is met, the error ends; otherwise, the iteration continues until the maximum number of iterations is reached and the optimal identification parameters are obtained.

[0039] The present invention adopts circle chaotic mapping to initialize the population to generate a more uniform and orderly initial population. In the early stage of iteration, a larger w helps the algorithm to conduct global exploration and avoid falling into the local optimal solution. In the later stage of iteration, the gradually decreasing w helps the algorithm to conduct a more refined local search near the currently found potential optimal solution, thereby improving the accuracy of the solution.

[0040] Through simulation analysis, the method proposed in the present invention can not only accurately and effectively identify parameters, but also has the characteristics of high convergence accuracy, small number of iterations, and strong stability of identification results. It is suitable for linear rotary permanent magnet motor parameter identification system, has good practicality, and is worthy of promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention is described in detail below with reference to specific embodiments and accompanying drawings, wherein:

[0042] Figure 1 is a flow chart of the present invention;

[0043] Figure 2 It is a principle block diagram of the linear rotary permanent magnet motor parameter identification of the present invention;

[0044] Figure 3 It is a simulation block diagram of linear rotary permanent magnet motor parameter identification based on ICS algorithm of the present invention;

[0045] Figure 4 It is the particle distribution diagram of the Circle chaos mapping of the present invention;

[0046] Figure 5 This is a test function optimization effect diagram of the present invention;

[0047] Figure 6 A diagram of the evolutionary process for identification of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention is described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the following specific embodiments are only used to explain the present invention and do not constitute a limitation to the present invention.

[0049] Figure 1 This is a flow chart of a parameter identification method for a linear permanent magnet motor based on an improved cuckoo optimization algorithm. The parameter identification process is as follows: first, a vector control system model and a parameter identification model of a linear permanent magnet motor are established; secondly, the algorithm parameters are set, the Circle chaos map is used to initialize the population, and an improved exponentially decreasing step size control factor α′ is introduced to improve the global optimization ability of the algorithm; then, an improved dynamic discovery probability P and an optimized dimension-by-dimension reverse learning strategy are used to improve the convergence speed and global optimization ability of the algorithm; finally, a fitness function is established, and the error between the identification model and the actual output is corrected according to the fitness function until it is minimized, otherwise it continues to iterate until the maximum number of iterations is reached and the iteration is stopped to obtain the optimal identification parameters.

[0050] like Figure 1-Figure 3 As shown, the linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm includes the following specific steps:

[0051] Step S1: establishing a parameter identification model of a linear rotating permanent magnet motor in a synchronous rotating coordinate system;

[0052] In this application, the control system of the linear rotating permanent magnet motor is modeled, and its voltage equation in the synchronous rotating coordinate system (dq coordinate system) is:

[0053]

[0054] The flux equation of the linear rotary permanent magnet motor is:

[0055]

[0056] Among them, u dr and u qr are the d-axis and q-axis voltages of the rotary motion unit respectively; u dl and u ql are the d-axis and q-axis voltages of the linear motion unit respectively; w e and v e are the electrical angular velocity and linear velocity of the motor respectively; R sr and R sl are the rotation and linear motion phase winding resistances respectively; is the magnetic flux of the rotating permanent magnet, is the magnetic flux of the linear permanent magnet; L dr and L qr are the d-axis and q-axis inductances of the stator winding of the rotary motion unit respectively; L dl and L ql are the d-axis and q-axis inductances of the stator winding of the linear motion unit respectively; i dr and i qr are the d-axis and q-axis currents of the stator of the rotary motion unit respectively; i dl and i ql They are the d-axis and q-axis currents of the linear motion unit stator respectively;

[0057] The motor equation is discretized by the voltage equation and the flux linkage method, and the parameter identification model of the motor is obtained. The parameter identification model is:

[0058]

[0059] The top mark “^” indicates the identification value of the identification model, and the top mark without a mark indicates the measured value of the actual model.

[0060] Step S2: Use chaotic mapping to initialize the parameters and population of the cuckoo optimization algorithm, where the chaotic mapping can be Logistic mapping, Sine mapping, and Tent mapping. Circle mapping is usually defined as a function that maps the interval [0,1] to the unit circle, which has periodicity and rotation characteristics and can exhibit multiple periods and chaotic behaviors, while Tent mapping is a simpler linear piecewise function with simple chaotic behavior and a relatively simple bifurcation structure. Compared with Tent mapping, the periodicity and rotation characteristics of Circle mapping enable the search space to be covered more effectively, and due to its diverse dynamic behaviors, Circle mapping can guide the cuckoo algorithm to better explore the complex search space and avoid falling into the local optimum; secondly, the characteristics of Circle mapping allow flexible switching of exploration strategies at different stages, which can effectively balance global search and local search, improve convergence speed and accuracy, and enable the optimization algorithm to maintain a high robustness, increasing the chance of finding the global optimal solution.

[0061] like Figure 4 As shown, in order to ensure that the cuckoo optimization algorithm can produce a more uniform and orderly initialization population, it is preferred to use Circle mapping, and use Circle chaos mapping to initialize the parameters and population of the cuckoo optimization algorithm. The expression of Circle chaos mapping is:

[0062]

[0063] Among them, n i represents the i-th chaotic sequence number, mod(a, b) represents the modulo operation of a over b.

[0064] The cuckoo optimization algorithm updates the position of the nest using Levy flight, and its expression is as follows:

[0065]

[0066] The expression of the Levy random flight path is:

[0067]

[0068] Among them, both u and v represent standard normal random variables, Γ is the Gamma function, and Φ is a proxy parameter in the dynamic Levy variable step size strategy.

[0069] x t+1,i represents the position of the updated nest; x t,i represents the position of the nest before update; β is the control factor of Levy flight; α′ is an improved variable step size control factor; the expression of α′ is:

[0070]

[0071] Among them, α represents the step size factor of the previous iteration; γ represents the coefficient controlling the exponential decay rate; t represents the current iteration number; t max represents the maximum number of iterations; k represents the non-linear adjustment coefficient. When k > 1, the step size will decrease faster, and when 0 < k < 1, the step size will decrease slowly.

[0072] Step S3: Obtain the dynamic discovery probability P of the cuckoo nest; this method replaces the fixed discovery probability of the original algorithm. The exponential decay function and logarithmic decreasing function can be used. To make the algorithm balance between global search and local search, it is preferred that the dynamic discovery probability P adopts the sine decreasing dynamic discovery probability because it has the following advantages: First, P has stronger dynamics and can change the discovery probability more frequently, enabling more active exploration of the search space in the early stage. The existing Pa is relatively conservative, resulting in insufficient randomness in the initial exploration process, thereby affecting the diversity of solutions. Second, the non-linear change of P can accelerate the convergence speed. When facing complex motor parameter identification problems, it can quickly find a better solution and reduce the search time in the invalid area. Since Pa introduces less randomness, more iteration times are required to converge to an acceptable solution, especially when the solution space is complex, more time and computing resources are consumed. The expression of the decreasing dynamic discovery probability P is:

[0073]

[0074] Among them, P min is the minimum discovery probability; P maxis the maximum discovery probability. The dynamic discovery probability P is adjusted in real time, and the random number R∈(0,1) obeys the uniform distribution. The random number R is compared with the discovery probability P of the cuckoo nest. If R<P, the current nest position is retained; otherwise, R≥P, the undesirable nest position is discarded, and the Levy flight formula is used to update the nest position; the fitness value of the objective function is calculated, and the nest with better fitness is retained to the next generation.

[0075] Step S4: Use nonlinear weight coefficients to optimize the dimension-by-dimension reverse learning strategy to improve the convergence efficiency and global optimization ability of the algorithm; specifically, the population size is N, the dimension is D, and the position when the iterative process selects the update position is Temp(i); the upper and lower limits of each dimension are Ub(i) and Lb(i), respectively, and the solution of each dimension based on weight-optimized reverse learning is:

[0076] Temp'(i)=Ub(i)+Lb(i)-Temp(i)·w;

[0077] Temp′(i) is the position after reverse learning based on weight optimization; w is the nonlinear weight coefficient, and its expression is:

[0078]

[0079] During the algorithm iteration process, reverse learning is performed on each dimension of the updated solution, and the updated values ​​of each dimension are mapped to its reverse learning number for spatial search, which increases the spatial search range of the solution. The influence between the original solution and the solution after the reverse operation is balanced by using a nonlinear weight coefficient, so that the initial stage of the algorithm focuses more on global exploration, and in the later stage it focuses more on local search, thereby improving search efficiency and optimization ability, avoiding the algorithm from converging to the local optimal solution too early, and improving the global optimization ability.

[0080] Step S5: Establish a fitness function with the identification value of the parameter identification model and the actual measurement value to correct the error between the identification model and the actual output until the accuracy requirement is met or t=t max Stop the iteration when , and obtain the optimal identification parameters.

[0081] Specifically, the square of the difference between the identification value of the identification model and the actual measurement value is used to establish the fitness function, and the expression of the function is:

[0082]

[0083] Among them, w1, w2, w3 and w4 are weighted factors of the fitness function. According to the fitness function, the error between the identification model and the actual output is corrected until it is minimized. If the accuracy requirement is met, it ends. Otherwise, it continues to iterate until the maximum number of iterations is reached, and the optimal identification parameters are obtained.

[0084] In the fitness function calculation process, the smaller the fitness function value, the higher the PMSM parameter identification accuracy, thereby obtaining the optimal solution identification parameters that meet the accuracy requirements.

[0085] like Figure 5 As shown in Figure 1, the improved cuckoo optimization algorithm (ICS) is compared with the cuckoo optimization algorithm (CS). The number of iterations of the above algorithm is set to 2000 times, the population size is 50, and after the international optimization test function test in Table 1, the optimization effect of different algorithms on different test functions is shown in Figure 1. Figure 5 As shown in Table 1.

[0086] Table 1 Examples of internationally accepted continuous function extreme value optimization

[0087]

[0088] exist Figure 5 middle, Figure 5 (a) is the optimization effect diagram measured by the international test function Shere; Figure 5 (b) is the optimization effect diagram measured by the international test function Schwefel2.22; Figure 5 (c) is the optimization effect diagram measured by the international test function Rastrigin; Figure 5 (d) is the optimization effect diagram measured by the international test function Griewank;

[0089] Depend on Figure 5 It can be seen that the ICS algorithm is not only better than the CS algorithm in fitness convergence, but also has fewer iterations than the CS algorithm and converges faster. Figure 5 (a) and Figure 5 (b) It can be seen that the ICS algorithm converges significantly faster than the CS algorithm and the ICS algorithm converges to the optimal solution. Figure 5 (c) and Figure 5 As can be seen from (d), the ICS algorithm converges to the optimal solution, while the CS algorithm converges poorly. In addition, the evolution graph also reflects that the CS algorithm lacks activity during the evolution process and has a slow convergence iteration speed. In summary, the convergence speed and optimization accuracy of the ICS algorithm proposed in this paper are significantly improved compared with the CS algorithm.

[0090] When the speed is 500r / min and the load torque is 10Nm, the algorithm iteration number is 200 times and the population size is 50. The iterative identification process of the algorithm is as follows: Figure 6 shown. Figure 6 (a) and (b) are the evolutionary diagrams of the quadrature-axis and direct-axis inductance identification of the rotary motion unit, respectively; Figure 6 (c) Diagram of the evolution process of stator resistance identification of the rotary motion unit; Figure 6 (d) is the evolution process diagram of magnetic flux identification of the rotational motion unit; Figure 6 (e) and (f) are the evolutionary diagrams of the quadrature-axis and direct-axis inductance identification of the linear motion unit, respectively; Figure 6 (g) is the evolution process diagram of the stator resistance identification of the linear motion unit; Figure 6 (h) is the evolution process diagram of the magnetic flux identification of the linear motion unit;

[0091] in, Figure 6 (a), (b), (c) and (d) are iterative identification diagrams of the rotational motion unit. Figure 6 (e), (f), (g) and (h) are iterative identification diagrams of the linear motion unit. Figure 6 (a), (b), (c) and (d) show that the identification effect of the I CS algorithm proposed in this paper is closer to the actual value and can converge to the true value faster than the CS algorithm.

[0092] from Figure 6 It can be seen from (e), (f), (g) and (h) that the ICS algorithm proposed in this paper can not only identify the rotary motion unit faster and better, but also converge faster and better to the true value of the linear motion unit. In summary, the ICS algorithm proposed in this paper is effective and feasible for the parameter identification of the linear rotary permanent magnet synchronous motor system, and has the advantages of fast convergence speed and high accuracy compared with the CS algorithm. The error comparison of the linear rotary permanent magnet synchronous motor parameter identification algorithm is shown in Table 2:

[0093] Table 2 Comparison of linear-rotary permanent magnet synchronous motor parameter identification algorithms

[0094]

[0095] It can be seen from Table 2 that compared with the CS algorithm, the proposed I CS algorithm has better multi-parameter identification effect, and the results can converge to the true value with small errors. It can be seen that the I CS algorithm proposed in this paper has the characteristics of high convergence accuracy and small number of iterations in parameter identification.

[0096] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A linear rotary permanent magnet motor parameter identification method based on cuckoo optimization algorithm, characterized in that: The steps include: Step S1: establishing a parameter identification model of a linear rotating permanent magnet motor in a synchronous rotating coordinate system; Step S2: Use chaos mapping to initialize the parameters and population of the cuckoo optimization algorithm, and use Levy flight to update the location of the nest, which is expressed as follows: x t+1,i Indicates the location of the nest after update; x t,i represents the position of the nest before updating; β is the control factor of Levy flight; α′ is the improved variable step length control factor; the expression of α′ is: Among them, α represents the step size factor of the previous iteration; γ represents the coefficient controlling the exponential decay rate; t represents the current iteration number; t max represents the maximum number of iterations; k represents the nonlinear adjustment coefficient; Step S3: Obtain the dynamic discovery probability P of the cuckoo's nest; Step S4: Optimizing the dimension-by-dimension reverse learning strategy using a nonlinear weight coefficient to improve the convergence efficiency and global optimization capability of the algorithm; Step S5: Establish a fitness function with the identification value of the parameter identification model and the actual measurement value to correct the error between the identification model and the actual output until the accuracy requirement is met or t=t max The iteration is stopped when , and the optimal identification parameters are obtained.

2. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 1, characterized in that: The parameter identification model is: Among them, u dr and u qr are the d-axis and q-axis voltages of the rotary motion unit respectively; u dl and u ql are the d-axis and q-axis voltages of the linear motion unit respectively; w e and v e are the electrical angular velocity and linear velocity of the motor respectively; R sr and R sl are the rotation and linear motion phase winding resistances respectively; is the magnetic flux of the permanent magnet of the rotating motion unit, is the magnetic flux of the permanent magnet of the linear motion unit; L dr and L qr are the d-axis and q-axis inductances of the stator winding of the rotary motion unit respectively; L dl and L ql are the d-axis and q-axis inductances of the stator winding of the linear motion unit respectively; i dr and i qr are the d-axis and q-axis currents of the stator of the rotary motion unit respectively; i dl and i ql They are the d-axis and q-axis currents of the stator of the linear motion unit respectively; the top mark "^" represents the identification value of the identification model, and the top mark without mark represents the measured value of the actual model.

3. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 1 is characterized in that: In step S2, the parameters and population of the cuckoo optimization algorithm are initialized using the Circle chaos map, and the expression of the Circle chaos map is: Among them, n i represents the i-th chaotic sequence number, mod(a, b) represents the modulo operation of a over b.

4. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 1, 2 or 3, characterized in that: In step S2, the Levy random flight path expression is: Among them, u and v both represent standard normal random variables, Γ is the Gamma function, and φ is the proxy parameter in the dynamic Levy variable step size strategy.

5. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 1, characterized in that: In step S3, the dynamic discovery probability P adopts a sinusoidally decreasing dynamic discovery probability, and the expression of the dynamic discovery probability P is: Among them, P min is the minimum discovery probability; P max is the maximum discovery probability.

6. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 1, characterized in that: In step S4, the population size is N, the dimension is D, and the position when the iterative process selects the updated position is Temp(i); the upper and lower limits of each dimension are Ub(i) and Lb(i), respectively. The solution of each dimension based on weighted optimization reverse learning is: Temp'(i)=Ub(i)+Lb(i)-Temp(i)·w; Temp′(i) is the position after reverse learning based on weight optimization; w is the nonlinear weight coefficient.

7. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 6 is characterized in that: The expression of the nonlinear weight coefficient w is:

8. The linear rotary permanent magnet motor parameter identification method based on the cuckoo optimization algorithm according to claim 2, characterized in that: In step S5, the square of the difference between the identification value of the identification model and the actual measurement value is used to establish a fitness function, and the expression of the function is: Among them, w1, w2, w3 and w4 are weighting factors of the fitness function.

Images