Stator bar structure optimization method based on GA-PSO algorithm

Abstract

The application discloses a stator bar structure optimization method based on a GA-PSO algorithm, and is specifically implemented according to the following steps: a stator bar is simulated by using comsol software, and a stator bar model is established; motor operation conditions are added in the stator bar model to perform model electric field simulation, and an electric field distribution diagram is obtained; a mathematical model is established by improving a particle swarm algorithm; by the mathematical model, an insulation thickness d parameter and a conductor fillet radius r parameter are input, and according to the electric field distribution, a best value of an angle part electric field uneven coefficient f m And a minimum value of an angle part maximum electric field intensity E m Are calculated and output, and optimal structure parameters of the stator bar are obtained; the GA-PSO algorithm is used, the relationship between stator structure parameters, electric fields and potentials involved in the stator optimization process can be coupled, the calculation convergence speed is improved, the particle diversity is enhanced, and efficient and reliable global search is realized.

Description

Technical Field

[0001] This invention belongs to the technical field of motor structure design methods, specifically relating to a stator bar structure optimization method based on the GA-PSO algorithm. Background Technology

[0002] Stator bars are components used to fix and conduct current, and are widely used in various types of motors. In recent years, the design and optimization of stator bars has become a research hotspot in the field of motors, with new technologies and processes constantly being introduced to improve motor performance and efficiency.

[0003] Currently, most stator bar structure optimizations involve improving the structure and shape of the stator bars and employing the finite element method (FEM) for simulation experiments. However, for large-size motor stator bars, the large number of meshes required increases the simulation time. The computation and experiments are time-consuming and costly, making them unsuitable for effective prediction and optimization. Furthermore, FEM calculations cannot meet the requirements for rapid monitoring of the stator bar's internal structure during digital equipment manufacturing processes.

[0004] With the development of Particle Swarm Optimization (PSO), solving high-dimensional parameters and reducing simulation time has become possible. By establishing alternative models to replace finite element method (FEM) calculations with machine learning, and reducing the size of the FEM model, rapid and accurate prediction of parameters such as model temperature can be achieved, improving computational efficiency and accelerating the analysis of model performance. Since the conductivity of the stator rod is a nonlinear function that varies with the electric field, and PSO, through information sharing and cooperation among individuals in the swarm, can search for optimal solutions in complex solution spaces, it exhibits strong superiority, especially in solving optimization problems of multi-peak, high-dimensional functions. However, PSO often suffers from premature convergence due to its tendency to get trapped in local optima. Adding a Genetic Algorithm (GA) to PSO can improve computational convergence speed, enhance particle diversity, and achieve efficient and reliable global search.

[0005] However, there is no existing technology that applies the improved particle swarm optimization (GA-PSO) algorithm to the optimization of stator bar structures. Summary of the Invention

[0006] The purpose of this invention is to provide a stator bar structure optimization method based on the GA-PSO algorithm, which has the characteristics of improving computational efficiency by establishing an alternative model through the GA-PSO algorithm to replace the finite element calculation results.

[0007] The technical solution of this invention is a stator bar structure optimization method based on the GA-PSO algorithm, which is implemented according to the following steps:

[0008] Step 1: Use COMSOL software to simulate the stator bars and establish a stator bar model;

[0009] Step 2: Add motor operating conditions to the stator bar model to simulate the electric field and obtain the electric field distribution diagram;

[0010] Step 3: Based on the electric field value at each point in the electric field distribution diagram, use the improved particle swarm optimization algorithm to establish a mathematical model for the optimal solution;

[0011] Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars were obtained.

[0012] The present invention is further characterized in that: step 2 is specifically implemented according to the following steps:

[0013] Step 2.1: Use the current module and domain ordinary differential module in the simulation software to simulate the electric field of the model;

[0014] Step 2.2: Add motor operating conditions and set boundary conditions; the outer surface of the motor stator is selected as the boundary surface for solving the boundary conditions. The first and second types of boundary conditions are used when analyzing the internal electric field of the motor.

[0015] Step 2.3: Mesh the stator bar model;

[0016] Step 2.4: Set up the solution, view the results of the electric field and potential, and obtain the electric field distribution diagram.

[0017] In step 2.1, the current module is used to generate a potential in the copper core, and the relevant differential formula is:

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] Where H is the magnetic field strength, J is the current density, D is the electric displacement vector, E is the model electric field strength, B is the magnetic induction intensity, and ρ is the charge density;

[0024] The parametric equations representing the relationships between different field quantities are as follows:

[0025] D=εE(6)

[0026] J=σE(7)

[0027] B=μH(8)

[0028] Where ε is the relative permittivity, σ is the conductivity, and μ is the permeability.

[0029] Step 3 is implemented in the following steps:

[0030] Step 3.1, Basic parameter settings; Population size n, maximum number of iterations iter max The velocity boundary value V of the population particles min ~V max Learning factors c1 and c2;

[0031] Step 3.2: Parameter initialization. Initial selection number iter = 0. Randomly initialize the position and velocity of the particles and calculate the fitness value of each particle.

[0032] Step 3.3: Take the current position of each particle as the individual optimal P. id The particle with the best fitness among all particles is designated as the global optimal P. gd ;

[0033] Step 3.4: Calculate the inertia weight w and update the particle's velocity and position;

[0034] Step 3.5: Perform mutation operations on the population particles with a certain probability;

[0035] Step 3.6: Calculate the fitness value of each mutated particle to obtain the individual's optimal P. id and the global optimal P gd ;

[0036] Step 3.7: Let iter = iter + 1. If iter > iter max If the algorithm outputs the optimal solution, the algorithm ends; otherwise, proceed to step 3.1 for alternative calculation.

[0037] In step 3.1, the parameters are set to population size n = 20 and maximum number of iterations iter max =1000, particle velocity boundary V min ~V max Let it be [-1,+1], and the learning factors c1=c2=2.

[0038] In step 3.4, the inertia weight w is set as a linear function that decreases with the number of iterations of the improved particle swarm optimization algorithm, as shown in the formula:

[0039]

[0040] Step 4 is implemented in the following steps:

[0041] Step 4.1: Using the insulation thickness d and conductor corner radius r as the structural parameters to be optimized, the corner electric field non-uniformity coefficient f... m and the maximum electric field intensity E at the corner m To optimize the objective, where f m E m The parametric equation is

[0042]

[0043]

[0044] Step 4.2, Corner electric field non-uniformity coefficient f m Based on the iterative calculations of the improved particle swarm optimization algorithm, the corner electric field non-uniformity coefficient f is obtained as the number of iterations increases. m The optimal value, the maximum electric field strength E at the corner m Then, with the corner electric field non-uniformity coefficient f m Through iterative optimization, the maximum electric field intensity E at the corner is obtained. m The minimum value;

[0045] Step 4.3: At this point, based on the corner electric field non-uniformity coefficient f... m Optimal values ​​and maximum electric field strength E at the corner m By finding the minimum value of the insulation thickness d parameter and the conductor fillet radius r parameter, the optimal structural parameters of the stator bar can be obtained.

[0046] In step 4.1, the optimization range for insulation thickness d is 3mm to 6mm, and the optimization range for conductor fillet radius r is 2mm to 5mm.

[0047] The beneficial effects of this invention are:

[0048] 1. This invention establishes an alternative model to replace the finite element calculation results with machine learning, and reduces the size of the finite element model to achieve rapid and accurate prediction of parameters such as model temperature, thereby improving computational efficiency and accelerating the analysis of model performance.

[0049] 2. The simulation model established by this invention based on COMSOL software can accurately obtain the distribution of electric field and electric potential of stator bars;

[0050] 3. This invention utilizes the GA-PSO algorithm, which can couple the relationship between stator structural parameters, electric field and electric potential involved in the stator optimization process, improve the calculation convergence speed, enhance particle diversity, and achieve efficient and reliable global search. Attached Figure Description

[0051] Figure 1 This is a stator bar model diagram created in the COMSOL simulation software of this invention;

[0052] Figure 2 This invention relates to the electric field intensity distribution on the surface of the nonlinear anti-corona structure bar.

[0053] Figure 3 The maximum electric field E at the corner of the stator bar in this invention. m A graph showing the relationship between the chamfer radius r and the chamfer radius r.

[0054] Figure 4 This is a flowchart of the GA-PSO calculation process in this invention;

[0055] Figure 5 The corner non-uniformity coefficient f in this invention m Iterative graph;

[0056] Figure 6 The maximum electric field E at the corner of the stator bar in this invention. m With the corner non-uniformity coefficient f m Line graph;

[0057] In the diagram, 1. Low resistance band grounding, 2. Medium resistance band, 3. Medium-high resistance band, 4. High resistance band, 5. Copper core, 6. Main insulation. Detailed Implementation

[0058] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0059] This invention provides a stator bar structure optimization method based on the GA-PSO algorithm, which is implemented according to the following steps:

[0060] Step 1: Simulate the stator bars using COMSOL software and establish a stator bar model; for example... Figure 1 As shown; the stator bar model is not chamfered, i.e., the chamfer radius r is 0; the insulation layer thickness d is 8mm;

[0061] Step 2: Add motor operating conditions to the stator bar model to simulate the electric field of the model and obtain the electric field distribution diagram.

[0062] Electric field distribution cloud map, such as Figure 2 As shown, after adopting the anti-corona structure using nonlinear materials, two peaks appear on the surface electric field of the rod, occurring at the slot outlet and the interface between the low-resistivity layer and the medium-resistivity layer, respectively. The main reason for the strong concentration phenomenon at the interface of each layer is the abrupt change in resistivity at the interface, resulting in a large potential gradient at that location. In the electric field simulation, the insulation thickness d and the conductor fillet radius r of the rod have a crucial influence on the surface electric field distribution of the rod. These influences often involve the maximum value of the electric field and its distribution.

[0063] Step 2.1: Use the current module and domain ordinary differential module in COMSOL simulation software to simulate the electric field of the model;

[0064] The current module is used to generate a potential in the copper core, and the differential formula involved is:

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] Where H is the magnetic field strength, J is the current density, D is the electric displacement vector, E is the model electric field strength, B is the magnetic induction intensity, and ρ is the charge density;

[0071] The parametric equations representing the relationships between different field quantities are as follows:

[0072] D=εE (6)

[0073] J=σE (7)

[0074] B=μH (8)

[0075] Where ε is the relative permittivity, σ is the conductivity, and μ is the permeability;

[0076] Step 2.2: Add motor operating conditions and set boundary conditions; the outer surface of the motor stator is selected as the boundary surface for solving the boundary conditions. The first and second types of boundary conditions are used when analyzing the internal electric field of the motor.

[0077] Step 2.3: Mesh the stator bar model;

[0078] Step 2.4: Set up the solution, view the solved electric field and potential results, obtain the electric field distribution, and perform post-processing operations;

[0079] Step 3: Based on the electric field value at each point in the electric field distribution diagram, establish a mathematical model for the optimal solution using an improved particle swarm optimization algorithm, such as... Figure 4 As shown, by establishing the GA-PSO model, the current position of each particle is taken as the individual optimum, and the best fitness among the current particles is taken as the global optimum. The particles are continuously iterated and updated to continuously obtain the individual optimum and the global optimum.

[0080] Step 3.1, Basic parameter settings; Population size n, maximum number of iterations iter maxThe velocity boundary value V of the population particles min ~V max Learning factors c1 and c2; parameters are set to population size n = 20, and maximum number of iterations iter max =1000, particle velocity boundary V min ~V max Let the range be [-1, +1], and the learning factors be c1 = c2 = 2.

[0081] Step 3.2: Parameter initialization. Initial selection number iter = 0. Randomly initialize the position and velocity of the particles and calculate the fitness value of each particle.

[0082] Step 3.3: Take the current position of each particle as the individual optimal P. id The particle with the best fitness among all particles is designated as the global optimal P. gd ;

[0083] Step 3.4: Calculate the inertia weight w and update the particle's velocity and position;

[0084] The inertia weight w is set as a linear function that decreases with the number of iterations of the improved particle swarm optimization algorithm, as shown in the formula:

[0085]

[0086] Step 3.5: Perform mutation operations on the population particles with a certain probability;

[0087] Step 3.6: Calculate the fitness value of each mutated particle to obtain the individual's optimal P. id and the global optimal P gd ;

[0088] Step 3.7: Let iter = iter + 1. If iter > iter max If the algorithm outputs the optimal solution, the algorithm ends; otherwise, proceed to step 3.1 for alternative calculation.

[0089] like Figure 5 As shown, GA-PSO has a faster convergence speed and better optimization results. This is because the addition of random mutations to GA-PSO enhances the algorithm's global search capability, and the resulting optimization results are more in line with the optimization requirements.

[0090] Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars were obtained.

[0091] Step 4.1: The insulation thickness d and the conductor fillet radius r are the structural parameters to be optimized;

[0092] like Figure 3 As shown, increasing the fillet radius can effectively reduce the electric field non-uniformity coefficient and improve the electric field distribution. Theoretically, the larger the fillet radius, the more obvious the effect. However, excessively large fillet radii of strands and conductors after forming can affect the slot fill factor. A high slot fill factor indicates tight filling within the slot; a low slot fill factor indicates loose filling within the slot. From the perspective of fully utilizing motor materials and operational performance, a higher slot fill factor is naturally better. However, an excessively high slot fill factor can cause difficulties in winding and easily damage the insulation; a low slot fill factor causes the wires to loosen within the slots during motor operation, easily damaging the insulation. Furthermore, more gaps within the slots, due to the poor thermal conductivity of air, affect the heat dissipation of the coils, increasing the motor temperature rise. The slot fill factor is generally taken as 60%–80%. Based on this, the optimized range for insulation thickness d is 3mm–6mm, and the optimized range for conductor fillet radius r is 2mm–5mm.

[0093] Corner electric field non-uniformity coefficient f m and the maximum electric field intensity E at the corner m To optimize the objective, where f m E m The parametric equation is

[0094]

[0095]

[0096] Step 4.2, Corner electric field non-uniformity coefficient f m Based on the iterative calculations of the improved particle swarm optimization algorithm, the corner electric field non-uniformity coefficient f is obtained as the number of iterations increases. m The optimal value, such as Figure 6 As shown, the maximum electric field intensity E at the corner m Then, with the corner electric field non-uniformity coefficient f m Through iterative optimization, the maximum electric field intensity E at the corner is obtained. m The minimum value;

[0097] Step 4.3: At this point, based on the corner electric field non-uniformity coefficient f... m Optimal values ​​and maximum electric field strength E at the corner m By finding the minimum value of the insulation thickness d parameter and the conductor fillet radius r parameter, the optimal structural parameters of the stator bar can be obtained.

[0098] like Figure 4 As shown, the corner electric field non-uniformity coefficient is calculated iteratively according to GA-PSO. With increasing iteration count, the corner electric field non-uniformity coefficient f... m It eventually stabilized at 1.94. The maximum electric field E at the corner... m With the corner electric field non-uniformity coefficient f mThrough optimization, its value was continuously reduced to 0.61 kV / mm. The optimal fillet radius was 2.3 mm, and the main insulation thickness was 4.1 mm.

[0099] Example 1

[0100] The stator bar structure optimization method based on the GA-PSO algorithm is implemented according to the following steps:

[0101] Step 1: Simulate the stator bars using COMSOL software and establish a stator bar model; the stator bar model is not chamfered, i.e., the chamfer radius r is 0; the insulation layer thickness d is 8mm;

[0102] Step 2: Add motor operating conditions to the stator bar model to simulate the electric field and obtain the electric field distribution;

[0103] Step 2.1: Use the current module and domain ordinary differential module in the simulation software to simulate the electric field of the model;

[0104] In section 2.1, the current module is used to generate a potential in the copper core, and the differential formula involved is:

[0105]

[0106]

[0107]

[0108]

[0109]

[0110] Where H is the magnetic field strength, J is the current density, D is the electric displacement vector, E is the model electric field strength, B is the magnetic induction intensity, and ρ is the charge density;

[0111] The parametric equations representing the relationships between different field quantities are as follows:

[0112] D=εE (6)

[0113] J=σE (7)

[0114] B=μH (8)

[0115] Where ε is the relative permittivity, σ is the conductivity, and μ is the permeability;

[0116] In step 2.1, the current module is used to generate a potential in the copper core, and the relevant differential formula is:

[0117]

[0118]

[0119]

[0120]

[0121]

[0122] Where H is the magnetic field strength, J is the current density, D is the electric displacement vector, E is the model electric field strength, B is the magnetic induction intensity, and ρ is the charge density;

[0123] The parametric equations representing the relationships between different field quantities are as follows:

[0124] D=εE (6)

[0125] J=σE (7)

[0126] B=μH (8)

[0127] Where ε is the relative permittivity, σ is the conductivity, and μ is the permeability;

[0128] Step 2.2: Add motor operating conditions and set boundary conditions; the outer surface of the motor stator is selected as the boundary surface for solving the boundary conditions. The first and second types of boundary conditions are used when analyzing the internal electric field of the motor.

[0129] Step 2.3: Mesh the stator bar model;

[0130] Step 2.4: Set up the solution, view the solved electric field and potential results, obtain the electric field distribution, and perform post-processing operations;

[0131] Step 3: Based on the electric field value at each point in the electric field distribution, use the improved particle swarm optimization algorithm to establish a mathematical model for the optimal solution;

[0132] Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars were obtained.

[0133] Example 2

[0134] The stator bar structure optimization method based on the GA-PSO algorithm is implemented according to the following steps:

[0135] Step 1: Simulate the stator bars using COMSOL software and establish a stator bar model; the stator bar model is not chamfered, i.e., the chamfer radius r is 0; the insulation layer thickness d is 8mm;

[0136] Step 2: Add motor operating conditions to the stator bar model to simulate the electric field and obtain the electric field distribution diagram;

[0137] Step 3: Based on the electric field value at each point in the electric field distribution diagram, use the improved particle swarm optimization algorithm to establish a mathematical model for the optimal solution;

[0138] Step 3.1, Basic parameter settings; Population size n, maximum number of iterations iter max The velocity boundary value V of the population particles min ~V max Learning factors c1 and c2; parameters are set to population size n = 20, and maximum number of iterations iter max =1000, particle velocity boundary V min ~V max Let the range be [-1, +1], and the learning factors be c1 = c2 = 2.

[0139] Step 3.2: Parameter initialization. Initial selection number iter = 0. Randomly initialize the position and velocity of the particles and calculate the fitness value of each particle.

[0140] Step 3.3: Take the current position of each particle as the individual optimal P. id The particle with the best fitness among all particles is designated as the global optimal P. gd ;

[0141] Step 3.4: Calculate the inertia weight w and update the particle's velocity and position; the inertia weight w is set as a linear function that decreases with the number of iterations of the improved particle swarm optimization algorithm, as shown in the formula:

[0142]

[0143] Step 3.5: Perform mutation operations on the population particles with a certain probability;

[0144] Step 3.6: Calculate the fitness value of each mutated particle to obtain the individual's optimal P. id and the global optimal P gd ;

[0145] Step 3.7: Let iter = iter + 1. If iter > iter max If the algorithm outputs the optimal solution, the algorithm ends; otherwise, proceed to step 3.1 for alternative calculation.

[0146] Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars were obtained.

[0147] Example 3

[0148] The stator bar structure optimization method based on the GA-PSO algorithm is implemented according to the following steps:

[0149] Step 1: Simulate the stator bars using COMSOL software and establish a stator bar model; the stator bar model is not chamfered, i.e., the chamfer radius r is 0; the insulation layer thickness d is 8mm;

[0150] Step 2: Add motor operating conditions to the stator bar model to simulate the electric field and obtain the electric field distribution diagram;

[0151] Step 3: Based on the electric field value at each point in the electric field distribution diagram, use the improved particle swarm optimization algorithm to establish a mathematical model for the optimal solution;

[0152] Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars are obtained.

[0153] Step 4.1: Using the insulation thickness d and conductor corner radius r as the structural parameters to be optimized, the corner electric field non-uniformity coefficient f... m and the maximum electric field intensity E at the corner m To optimize the objective, where f m E m The parametric equation is

[0154]

[0155]

[0156] The optimized range for insulation thickness d is 3mm to 6mm, and the optimized range for conductor fillet radius r is 2mm to 5mm.

[0157] Step 4.2, Corner electric field non-uniformity coefficient f m Based on the iterative calculations of the improved particle swarm optimization algorithm, the corner electric field non-uniformity coefficient f is obtained as the number of iterations increases. m The optimal value, the maximum electric field strength E at the corner m Then, with the corner electric field non-uniformity coefficient f m Through iterative optimization, the maximum electric field intensity E at the corner is obtained. m The minimum value;

[0158] Step 4.3: At this point, based on the corner electric field non-uniformity coefficient f... m Optimal values ​​and maximum electric field strength E at the corner mBy finding the minimum value of the insulation thickness d parameter and the conductor fillet radius r parameter, the optimal structural parameters of the stator bar can be obtained.

[0159] This invention provides a stator bar structure optimization method based on the GA-PSO algorithm. The advantages are: by using the GA-PSO algorithm, the relationship between stator structure parameters, electric field and electric potential involved in the stator optimization process can be coupled, thereby improving the calculation convergence speed, enhancing particle diversity, and achieving efficient and reliable global search.

Claims

1. A method for optimizing stator bar structures based on the GA-PSO algorithm, characterized in that, The specific steps are as follows: Step 1: Use COMSOL software to simulate the stator bars and establish a stator bar model; Step 2: Add motor operating conditions to the stator bar model to simulate the electric field of the model and obtain the electric field distribution diagram; Step 3: Based on the electric field value at each point in the electric field distribution diagram, use the improved particle swarm optimization algorithm to establish a mathematical model for the optimal solution; Step 4: Using the mathematical model of the optimal solution, input the insulation thickness d and the conductor fillet radius r, calculate and output the corner electric field non-uniformity coefficient f. m and the maximum electric field intensity E at the corner m The optimal structural parameters of the stator bars are obtained. Step 3 is implemented in the following steps: Step 3.1, Basic parameter settings; Population size n, maximum number of iterations iter max The velocity boundary value V of the population particles min ~V max Learning factors c1 and c2; Step 3.2: Parameter initialization. Initial iteration count iter=0. Randomly initialize the position and velocity of the particles and calculate the fitness value of each particle. Step 3.3: Take the current position of each particle as the individual optimal P. id The particle with the best fitness among all particles is designated as the global optimal P. gd ; Step 3.4: Calculate the inertia weight w and update the particle's velocity and position; Step 3.5: Perform mutation operations on the population particles with a certain probability; Step 3.6: Calculate the fitness value of each mutated particle to obtain the individual's optimal P. id and the global optimal P gd ; Step 3.7: Let iter = iter + 1. If iter > iter max If the algorithm outputs the optimal solution, the algorithm ends; otherwise, it continues to step 3.1 iterative calculation. Step 4 is implemented in the following steps: Step 4.1: Using the insulation thickness d and conductor corner radius r as the structural parameters to be optimized, the corner electric field non-uniformity coefficient f... m and the maximum electric field intensity E at the corner m To optimize the objective, where f m E m The parametric equation is （10） （11）； Step 4.2, Corner electric field non-uniformity coefficient f m Based on the iterative calculations of the improved particle swarm optimization algorithm, the corner electric field non-uniformity coefficient f is obtained as the number of iterations increases. m The optimal value, the maximum electric field strength E at the corner m Then, with the corner electric field non-uniformity coefficient f m Through iterative optimization, the maximum electric field intensity E at the corner is obtained. m The minimum value; Step 4.3: At this point, according to the corner electric field non-uniformity coefficient f... m Optimal values ​​and maximum electric field strength E at the corner m By finding the minimum value of the insulation thickness d parameter and the conductor fillet radius r parameter, the optimal structural parameters of the stator bar can be obtained.

2. The stator bar structure optimization method based on the GA-PSO algorithm according to claim 1, characterized in that, Step 2 is implemented in the following steps: Step 2.1: Use the current module and domain ordinary differential module in the simulation software to simulate the electric field of the model; Step 2.2: Add motor operating conditions and set boundary conditions; the boundary conditions are selected from the outer surface of the motor stator as the boundary surface for solving, and the first and second types of boundary conditions are used when analyzing the internal electric field of the motor. Step 2.3: Mesh the stator bar model; Step 2.4: Set up the solution, view the results of the electric field and potential, and obtain the electric field distribution diagram.

3. The stator bar structure optimization method based on the GA-PSO algorithm according to claim 2, characterized in that, In step 2.1, the current module is used to generate a potential in the copper core, and the relevant differential formula is: （1） （2） （3） （4） （5） Where H is the magnetic field strength, J is the current density, D is the electric displacement vector, E is the model electric field strength, and B is the magnetic flux density. It is charge density; The parametric equations representing the relationships between different field quantities are as follows: （6） （7） （8） Where ε is the relative permittivity, It is electrical conductivity. It is the permeability.

4. The stator bar structure optimization method based on the GA-PSO algorithm according to claim 1, characterized in that, In step 3.1, the parameters are set to population size n=20 and maximum number of iterations iter max =1000, particle velocity boundary V min ~V max Let it be [-1,+1], and the learning factors c1=c2=2.

5. The stator bar structure optimization method based on the GA-PSO algorithm according to claim 1, characterized in that, In step 3.4, the inertia weight w is set as a linear function that decreases with the number of iterations of the improved particle swarm optimization algorithm, as shown in the formula: （9）。 6. The stator bar structure optimization method based on the GA-PSO algorithm according to claim 1, characterized in that, In step 4.1, the optimized range of insulation thickness d is 3mm to 6mm, and the optimized range of conductor fillet radius r is 2mm to 5mm.

Images