Pure electric vehicle motor multi-physics field collaborative optimization method and system based on particle swarm

Abstract

The invention relates to the technical field of motor optimization, and discloses a multi-physics field collaborative optimization method and system for a pure electric vehicle motor based on a particle swarm, and the method comprises the steps: building a multi-physics field coupling model of the pure electric vehicle motor; initializing parameters and a particle swarm of the improved multi-target particle swarm optimization algorithm, wherein each particle represents a group of motor design variables; calling a multi-physics field coupling model to simulate each particle, and calculating an electromagnetic performance index, a thermal performance index and a structural performance index; performing non-dominated sorting on each particle according to the electromagnetic performance index, the thermal performance index and the structural performance index, and updating a Pareto solution set; updating particle positions and speeds based on a Pareto solution set, and generating a new generation of particle swarms; and judging whether the updated Pareto solution set meets a convergence condition or not, and if yes, outputting the Pareto solution set as an optimization design scheme. According to the scheme, the comprehensive reliability of the motor can be improved, and an optimal design scheme considering multi-target performance is provided for the high-power-density electric vehicle motor.

Description

Technical Field

[0001] This invention belongs to the field of motor optimization technology, and particularly relates to a multi-physics field collaborative optimization method and system for pure electric vehicle motors based on particle swarm optimization. Background Technology

[0002] As new energy pure electric vehicles become increasingly mainstream, their drive motors need to achieve higher power output within a limited volume, while also ensuring low losses to extend range and meeting stringent requirements for low noise and wide-temperature stability. Multi-physics collaborative optimization has become the core path to overcome the performance bottleneck of motors and is crucial for improving the range, safety, and comfort of electric vehicles.

[0003] Traditional motor design methods typically have the following limitations:

[0004] Single-physics optimization is fragmented: Electromagnetic, thermal, and structural fields are often optimized independently, ignoring the coupling effects of multi-physics fields (such as increased temperature rise due to increased current density, and vibration induced by electromagnetic force), which leads to performance conflicts during actual operation.

[0005] Insufficient algorithm convergence and solution set quality: Standard multi-objective algorithms (such as NSGA-II) rely on fixed parameters and are prone to getting trapped in local optima; Pareto solution sets are prone to missing boundary solutions or being unevenly distributed;

[0006] Relying on experience and trial and error: Adjusting design variables requires manual intervention, the simulation-optimization loop is inefficient, and it is difficult to efficiently handle high-dimensional multi-objective problems.

[0007] Therefore, there is an urgent need to develop a multi-physics field collaborative optimization method and system for pure electric vehicle motors based on particle swarm optimization, which can improve the overall reliability of the motor and provide an optimized design scheme that takes into account multiple objectives for high power density electric vehicle motors. Summary of the Invention

[0008] To address the aforementioned technical problems, this invention provides a multi-physics field collaborative optimization method and system for pure electric vehicle motors based on particle swarm optimization. This method can improve the overall reliability of the motor and provide an optimized design scheme for high-power-density electric vehicle motors that takes into account multiple performance objectives.

[0009] This invention provides a multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization. The method includes the following steps:

[0010] S1. Establish a multi-physics coupling model for the electric motor of a pure electric vehicle. The multi-physics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model.

[0011] S2. Initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of motor design variables;

[0012] S3. Use the multiphysics coupling model to simulate each particle and calculate its electromagnetic performance, thermal performance and structural performance.

[0013] S4. Sort each particle in a non-dominated order based on its electromagnetic performance index, thermal performance index and structural performance index, and update the Pareto solution set.

[0014] S5. Update the particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarm, and repeat S3-S4.

[0015] S6. Determine whether the updated Pareto solution set satisfies the convergence condition. If it does, output the Pareto solution set as the optimal design scheme. If it does not, return to S3.

[0016] Furthermore, in S1, the electromagnetic field model includes:

[0017] ;

[0018] in, Let represent the Nabla operator, A represent the magnetic vector potential, μ represent the permeability, J represent the current density, σ represent the conductivity, and t represent time.

[0019] Furthermore, in S1, the temperature field model includes:

[0020] ;

[0021] in, Let ρ represent the Nabla operator, T represent the density, and C represent the temperature. p denoted by , b represents specific heat capacity, represents thermal conductivity, Q represents the heat source term, and t represents time.

[0022] Furthermore, in S1, the structural field model includes:

[0023] ;

[0024] in, Let ρ represent the Nabla operator, u represent the displacement vector, and σ represent the density. d Let F represent the stress tensor, F represent the volume force, and t represent time.

[0025] Furthermore, in S2, the improved multi-objective particle swarm optimization algorithm includes:

[0026] Using linearly decreasing dynamic inertia weights:

[0027] ;

[0028] Where k represents the current iteration number, K represents the maximum iteration number, w0 represents the initial inertia weight, and wend Indicates the termination inertia weight, w k The inertia weight represents the current iteration number;

[0029] Set learning factors, including individual cognitive coefficient c1 and social learning coefficient c2, and satisfy c1+c2≤4; where the individual cognitive coefficient is used to control the learning intensity of the particle towards its own historical best position, and the social learning coefficient is used to control the learning intensity of the particle towards the global best position of the group.

[0030] Furthermore, in S2, the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm are initialized, including:

[0031] S21. Initialize the parameters of the improved multi-objective particle swarm optimization algorithm, including population size, dynamic inertia weight, learning factor, maximum speed, and maximum number of iterations;

[0032] S22. Initialize the particle swarm attributes of the improved multi-objective particle swarm optimization algorithm, including the position, velocity, optimal position, and optimal fitness of each particle.

[0033] S23. Initialize the elite save file to an empty set.

[0034] Furthermore, in S3, electromagnetic performance indicators include efficiency and torque pulsation, thermal performance indicators include winding temperature rise, and structural performance indicators include vibration sound pressure level.

[0035] Furthermore, in S4, each particle is non-dominated and sorted according to its electromagnetic, thermal, and structural performance indices, and the Pareto solution set is updated, including:

[0036] S41. Sort the entire particle swarm according to the value of the first performance index. If the optimization objective of the performance index is to minimize it, sort in ascending order. If the optimization objective of the performance index is to maximize it, sort in descending order.

[0037] S42. Initialize the first non-dominated front. Add the first particle in the sorting to the first non-dominated front. Compare the remaining particles with each particle in the first non-dominated front. If the remaining particles are not dominated by any particle in the first non-dominated front, add the remaining particles to the first non-dominated front. Otherwise, continue to traverse the next remaining particle until the traversal is completed, forming the first non-dominated front.

[0038] S43. Remove the particles belonging to the first non-dominated front. Repeat S41-S43 for the remaining particles until all particles are assigned to a non-dominated front, resulting in several non-dominated fronts.

[0039] S44. Calculate the crowding degree of each particle within each non-dominated front.

[0040] S45. Compare the particles in the first non-dominated frontier with the original particles in the elite archive. If the original particles are dominated by the newly added particles, remove the original particles. If the newly added particles are not dominated by any original particles, add the newly added particles to the elite archive, which is the Pareto solution set.

[0041] S46. Construct a new population by adding particles from the first non-dominated front to the new population and determining whether the number of particles in the new population is greater than or equal to the population size. If not, continue adding particles from the next non-dominated front. If yes, stop adding new particles and obtain a new population.

[0042] S47. When the number of particles in the elite archive exceeds the preset threshold, or the number of particles in the new population exceeds the population size, remove particles in order of increasing crowding until the number of particles in the elite archive equals the preset threshold, or the number of particles in the new population equals the population size.

[0043] Furthermore, in S5, particle positions and velocities are updated based on the Pareto solution set to generate a new generation of particle swarms, including:

[0044] The velocity and position of each particle in the new population are updated to obtain a new generation of particle swarm;

[0045] The speed update formula is as follows:

[0046] ;

[0047] in, This represents the velocity of the i-th particle at iteration number k+1. Let r1 and r2 represent the velocity of the i-th particle at iteration number k, and r1 and r2 represent random numbers between [0,1]. Let represent the optimal position of the i-th particle itself. This represents the position of the i-th particle at iteration number k. This indicates the position of the guiding particle selected from the Pareto solution set for the i-th particle;

[0048] The position update formula is as follows:

[0049] ;

[0050] in, This represents the position of the i-th particle at iteration number k+1.

[0051] This invention also provides a particle swarm optimization system for multi-physics collaborative optimization of pure electric vehicle motors, used to execute the aforementioned particle swarm optimization method for multi-physics collaborative optimization of pure electric vehicle motors. The system includes the following modules:

[0052] The multiphysics construction module is used to build a multiphysics coupling model of a pure electric vehicle motor. The multiphysics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model.

[0053] The particle swarm optimization module is used to initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm. Each particle represents a set of motor design variables.

[0054] The simulation module is used to call the multiphysics coupling model to simulate each particle and calculate electromagnetic performance, thermal performance and structural performance indicators.

[0055] The sorting module is used to perform non-dominated sorting of each particle based on electromagnetic performance indicators, thermal performance indicators, and structural performance indicators, and update the Pareto solution set.

[0056] The update module is used to update particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarms, and repeat the simulation and sorting operations.

[0057] The output module is used to determine whether the updated Pareto solution set meets the convergence condition. If it does, the Pareto solution set is output as the optimization design scheme; otherwise, the updated particle swarm is returned.

[0058] The embodiments of the present invention have the following technical effects:

[0059] This invention establishes a coupled model of electromagnetic field, temperature field, and structural field, simultaneously optimizing performance indicators such as efficiency, torque ripple, winding temperature rise, and vibration sound pressure level. This overcomes the performance imbalance problem caused by traditional single-physics field optimization, ensuring synergistic improvement in electromagnetic performance, thermal stability, and mechanical reliability of the motor. It achieves efficient convergence by improving the multi-objective particle swarm optimization algorithm, balances the global exploration and local development capabilities of the algorithm through a linear decreasing strategy to avoid premature convergence, maintains the distribution and convergence of the Pareto solution set through non-dominated sorting and crowding degree screening to improve the quality of the solution set, and dynamically selects guiding particles from the Pareto solution set to drive the particle swarm towards the true Pareto front. While ensuring electromagnetic performance, it effectively suppresses temperature rise and vibration noise, improving the overall reliability of the motor and providing an optimized design scheme for high-power-density electric vehicle motors that considers multiple performance objectives. Attached Figure Description

[0060] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0061] Figure 1 This is a flowchart of the multi-physics field collaborative optimization method for pure electric vehicle motors based on particle swarm optimization provided in this embodiment of the invention;

[0062] Figure 2 This is a flowchart of a method for updating the Pareto solution set by non-dominated sorting of particle swarms, provided by an embodiment of the present invention.

[0063] Figure 3 This is a schematic diagram of the structure of the multi-physics field collaborative optimization system for pure electric vehicle motors based on particle swarm optimization provided in an embodiment of the present invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0065] This invention provides a multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization. Figure 1 This is a flowchart of the multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization provided in this embodiment of the invention. See also... Figure 1 The method includes the following steps:

[0066] S1. Establish a multi-physics coupling model for the electric motor of a pure electric vehicle. The multi-physics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model.

[0067] In this embodiment, a two-way coupling method is used for the multiphysics coupling model, considering the interactions between the various physical fields:

[0068] Electromagnetic-thermal coupling: Electromagnetic loss acts as a heat source in the temperature field, and temperature changes affect the electromagnetic properties of the material.

[0069] Electromagnetic-structural coupling: Electromagnetic force is used as an excitation source to input the structural field, and structural deformation affects the air gap distribution of the electromagnetic field;

[0070] Thermal-structural coupling: Temperature changes lead to thermal stress, which affects the deformation and stress distribution of the structure.

[0071] In some embodiments, electromagnetic field analysis is fundamental to motor performance evaluation, employing the finite element method to calculate the electromagnetic field distribution within the motor. The governing equations are Maxwell's equations, and considering the magnetic field effects within the motor, the electromagnetic field model includes:

[0072] ;

[0073] in, Let A represent the Nabla operator, μ represent the magnetic vector potential, J represent the current density, σ represent the conductivity, and t represent time. Through electromagnetic field analysis, the magnetic field distribution, electromagnetic force waveform, and energy losses (including iron and copper losses) of the motor can be obtained. These losses will be used as heat sources input into the temperature field model.

[0074] The temperature field model mainly focuses on the heat conduction and heat dissipation processes of the motor. The temperature field model includes:

[0075] ;

[0076] in, Let ρ represent the Nabla operator, T represent the density, and C represent the temperature. p Let represent specific heat capacity, b represent thermal conductivity, Q represent the heat source term, and t represent time. Temperature field analysis includes steady-state thermal analysis and transient thermal analysis, used to evaluate the temperature rise and temperature distribution of the motor after long-term operation, as well as temperature changes during startup, load changes, etc. Changes in the temperature field affect the electromagnetic properties (such as the resistivity of windings) and mechanical properties of materials, therefore it needs to be coupled with electromagnetic and structural fields.

[0077] The structural field model primarily analyzes the vibration characteristics and noise radiation of the motor. The structural field model includes:

[0078] ;

[0079] in, Let ρ represent the Nabla operator, u represent the displacement vector, and σ represent the density. d Let F represent the stress tensor, t represent the volume force, and t represent time. Structural field analysis includes modal analysis (determining the natural frequencies and mode shapes of the motor structure), static and dynamic mechanical analysis (evaluating the stress and deformation of the motor under static and dynamic loads), and fatigue analysis (predicting the fatigue life of the motor). The vibration response of the structural field can, in turn, affect the performance of the electromagnetic field; for example, changes in the air gap can lead to changes in the magnetic field distribution.

[0080] S2. Initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of motor design variables.

[0081] In some embodiments, the improved multi-objective particle swarm optimization algorithm includes:

[0082] A linearly decreasing dynamic inertia weight is used to balance global exploration and local exploitation capabilities:

[0083] ;

[0084] Where k represents the current iteration number, K represents the maximum iteration number, w0 represents the initial inertia weight, and w end Indicates the termination inertia weight, w k The inertia weight represents the current iteration number; when w k A larger value (e.g., 0.9) indicates a bias towards global exploration, meaning a broader search of new regions. k A smaller value (e.g., 0.4) tends to favor local development, i.e., a fine-grained search of the current region.

[0085] Set a learning factor, including an individual cognitive coefficient c1 and a social learning coefficient c2, and satisfy c1+c2≤4; wherein, the individual cognitive coefficient is used to control the learning intensity of the particle towards its own historical best position, and the social learning coefficient is used to control the learning intensity of the particle towards the global best position of the group; for example, the values ​​of c1 and c2 can be between 1.5 and 2.

[0086] Each particle represents a set of motor design variables, which may include, for example, geometric parameters such as stator inner / outer diameter, core length, air gap length, slot dimensions, permanent magnet dimensions, etc.; material parameters such as permanent magnet type, silicon steel sheet type, winding material, etc.; and control parameters such as current phase angle and current amplitude, etc.

[0087] In some embodiments, initializing the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm includes:

[0088] S21. Initialize the parameters of the improved multi-objective particle swarm optimization algorithm, including population size, dynamic inertia weight, learning factor, maximum speed, and maximum number of iterations.

[0089] Here, the swarm size represents the number of particles, affecting the search breadth and computational cost. Ideally, N should be between 50 and 200. A larger swarm provides greater exploration capability, but also higher computational overhead. The maximum velocity limits the absolute value of the particle velocity to prevent search oscillations or overflow of the design space. For example, it can be set to 10% to 20% of the variable's range.

[0090] S22. Initialize the particle swarm attributes of the improved multi-objective particle swarm optimization algorithm, including the position, velocity, optimal position, and optimal fitness of each particle.

[0091] Here, position represents a motor design scheme (such as air gap length, slot size, etc.), and velocity represents the particle search direction and step size, which are usually initialized to zero or a small random value. For example, during algorithm initialization, a set of particles are randomly generated within the design space, and their velocities and positions are initialized. The boundary of the design space is determined based on prior knowledge of the motor design, ensuring that all particles are within the feasible region. The individual optimal position is the particle's own historical optimal position, initialized to the current position; the individual optimal fitness is the objective function value (such as efficiency, temperature rise) corresponding to the particle's own historical optimal position, which can be initially calculated through simulation.

[0092] S23. Initialize the elite save file to an empty set.

[0093] The elite archive, also known as the Pareto solution set, is used to store the current non-dominated solutions (solutions that have not been surpassed by other solutions on all objectives). It is initially an empty set and is gradually filled with iterations.

[0094] S3. Use the multiphysics coupling model to simulate each particle and calculate its electromagnetic, thermal and structural performance indicators.

[0095] In some embodiments, electromagnetic performance indicators include efficiency and torque ripple, thermal performance indicators include winding temperature rise, and structural performance indicators include vibration sound pressure level. The optimization objective for efficiency is to maximize it, while the optimization objectives for torque ripple, winding temperature rise, and vibration sound pressure level are to minimize them.

[0096] Electromagnetic performance is a core performance indicator of a motor, mainly including: efficiency, which represents the ratio of the motor's output power to its input power, comprehensively reflecting the motor's energy conversion performance.

[0097] ;

[0098] in, P represents efficiency. out P represents the motor's output power. in T represents the input power of the motor. d Indicates the output torque. V represents the angular velocity, V represents the effective value of the voltage, and I represents the effective value of the current. Indicates the power factor;

[0099] Torque ripple indicates the degree of fluctuation in output torque, which affects the smoothness of motor operation.

[0100] ;

[0101] in, Indicates torque pulsation, T max T represents the maximum torque within one electrical cycle. min T represents the minimum torque within one electrical cycle.avg It represents the average torque over one electrical cycle.

[0102] Thermal performance indicators mainly focus on the temperature rise of the motor, which directly affects the motor's reliability and lifespan. Winding temperature rise indicates the temperature increase of the windings during operation.

[0103] ;

[0104] in, Indicates the temperature rise of the winding. This indicates the highest temperature of the winding. This indicates the ambient temperature. Excessive temperature rise will accelerate the aging of insulation materials, leading to motor failure.

[0105] Structural performance indicators mainly focus on the vibration and noise characteristics of the motor. Vibration sound pressure level represents a quantitative indicator of motor noise, reflecting the noise level during motor operation.

[0106] ;

[0107] Where SPL represents the vibration sound pressure level, P represents the sound pressure, and P0 represents the reference sound pressure. The vibration sound pressure level is closely related to the electromagnetic force waves and structural characteristics of the motor.

[0108] S4. Sort each particle in a non-dominated order based on its electromagnetic performance index, thermal performance index, and structural performance index, and update the Pareto solution set.

[0109] In some embodiments, S4 includes the following sub-steps:

[0110] S41. Sort the entire particle swarm according to the value of the first performance index. If the optimization objective of the performance index is to minimize it, sort in ascending order; if the optimization objective of the performance index is to maximize it, sort in descending order.

[0111] S42. Initialize the first non-dominated front. Add the first particle in the sorting to the first non-dominated front. Compare the remaining particles with each particle in the first non-dominated front. If the remaining particles are not dominated by any particle in the first non-dominated front, add the remaining particles to the first non-dominated front. Otherwise, continue to traverse the next remaining particle until the traversal is completed, forming the first non-dominated front.

[0112] In multi-objective optimization, for solution A to dominate solution B, two conditions must be satisfied simultaneously:

[0113] For all objectives, solution A is no worse than solution B;

[0114] For at least one objective, solution A is strictly superior to solution B.

[0115] The particle ranked first indicates that no other solution is better than this particle in terms of this performance index. Therefore, this particle cannot be dominated by other particles and is added to the first non-dominated front.

[0116] S43. Remove the particles belonging to the first non-dominated front. Repeat S41-S43 for the remaining particles until all particles are assigned to a non-dominated front, resulting in several non-dominated fronts.

[0117] Assuming efficiency is used as the primary performance indicator, and there are six particles A through F,

[0118] A: Efficiency 95%, winding temperature rise 65K, torque ripple 8.5%;

[0119] B: Efficiency 93.5%, winding temperature rise 60K, torque ripple 7.0%;

[0120] C: Efficiency 92%, winding temperature rise 55K, torque ripple 9.0%;

[0121] D: Efficiency 94%, winding temperature rise 70K, torque ripple 5.0%;

[0122] E: Efficiency 91%, winding temperature rise 50K, torque ripple 12.0%;

[0123] F: Efficiency 96%, winding temperature rise 75K, torque ripple 10.0%;

[0124] Based on this, it can be determined through step S42 that particle F is the first particle in the sorting, and particles A, D, and B are added to the first non-dominated front as non-dominated particles; particles C and E are both dominated by B, so they are retained.

[0125] Among the remaining particles, they are further sorted by efficiency. Particle C, ranked first, is added to the second non-dominated frontier. Particle E has a lower efficiency, torque pulsation, and better temperature rise than particle C. Therefore, particle E is not dominated by particle C and is also added to the second non-dominated frontier.

[0126] S44. Calculate the crowding degree of each particle within each non-dominated front.

[0127] In some embodiments, the congestion level is calculated as follows:

[0128] Sa, for each particle before non-dominated, sort them according to the value of each performance index and the optimization objective of that performance index;

[0129] Sb, Set the distance between the first and last particles in the sorting result to infinity or a maximum value (such as 999) in this performance index dimension.

[0130] Sc, calculate the distances to the middle particles respectively;

[0131] Taking the first non-dominated frontier = {F, A, D, B} as an example;

[0132] In terms of efficiency, the ranking is F, A, D, B;

[0133] In terms of temperature rise, the order is B, A, D, F;

[0134] In terms of torque ripple, the order is D, B, A, F;

[0135] Based on this, in the efficiency dimension, the distance between F and B is set to infinity. If particle A is ordered between F and D, then the distance of particle A = (efficiency value of particle D - efficiency value of particle F) / (maximum efficiency value among particles in the first non-dominated front - minimum efficiency value among particles in the first non-dominated front); if particle D is ordered between A and B, then the distance of particle D = (efficiency value of particle B - efficiency value of particle A) / (maximum efficiency value among particles in the first non-dominated front - minimum efficiency value among particles in the first non-dominated front).

[0136] The calculation methods for other performance metrics are similar.

[0137] Sd is the crowding degree of a particle, obtained by summing the distances of each particle across all performance metric dimensions.

[0138] S45. Compare the particles in the first non-dominated frontier with the original particles in the elite archive.

[0139] If an existing particle is dominated by a newly added particle, the existing particle is removed. If the newly added particle is not dominated by any existing particle, the newly added particle is added to the elite archive, which is the Pareto solution set.

[0140] The elite archive is used to store historical best solutions. In each iteration, the newly generated first non-dominated frontier particle is compared bidirectionally with its existing contents. In this way, the elite archive continuously accumulates high-quality solutions.

[0141] S46. Construct a new population by adding particles from the first non-dominated frontier to the new population and determining whether the number of particles in the new population is greater than or equal to the population size.

[0142] If not, continue adding particles from the next non-dominated frontier; if yes, stop adding new particles and obtain a new population. When constructing a new generation of population, prioritize selecting particles from the first non-dominated frontier. If the number is insufficient, supplement at each level until the population size requirement is met. Similarly, prune according to crowding to ensure that the new population has both superiority and diversity.

[0143] S47. When the number of particles in the elite archive exceeds a preset threshold, or the number of particles in the new population exceeds the population size, remove particles in ascending order of crowding until the number of particles in the elite archive equals the preset threshold, or the number of particles in the new population equals the population size. To control the archive size, when its number exceeds the preset threshold, remove individuals from dense areas based on their crowding from smallest to largest, retaining representative solutions with even distribution.

[0144] S5. Update the particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarm, and repeat S3-S4.

[0145] Particle swarm optimization (PSO) algorithms iteratively update the positions and velocities of particles, gradually bringing the particle swarm closer to the Pareto front. Each iteration requires determining whether the algorithm has converged to decide whether to terminate the optimization process.

[0146] In some embodiments, the velocity and position of each particle in the new population are updated to obtain a new generation of particle swarm;

[0147] The speed update formula is as follows:

[0148] ;

[0149] in, It represents the velocity of the i-th particle at iteration number k+1 (its essence is the displacement increment, i.e., the parameter adjustment step size, rather than the physical velocity). Let r1 and r2 represent the velocity of the i-th particle at iteration number k, and r1 and r2 represent random numbers between [0,1]. Let represent the optimal position of the i-th particle itself. This represents the position of the i-th particle at iteration number k (position represents a point in the solution space, such as a combination of motor design parameters, x = [stator inner diameter, air gap length, number of winding turns]). T ), This indicates the position of the guiding particle selected from the Pareto solution set for the i-th particle;

[0150] The position update formula is as follows:

[0151] ;

[0152] in, This represents the position of the i-th particle at iteration number k+1.

[0153] The actual meaning of the above formula is: new parameter = old parameter + parameter adjustment amount, so there is no problem of inconsistent dimensions.

[0154] After the velocity is updated, the particles adjust their coordinates in the design space according to the position update formula, i.e., the current combination of design variables. The displacement is determined by the latest calculated velocity, ensuring that each movement is comprehensively controlled by multiple factors. The entire update process is executed after each round of simulation and sorting, and the newly generated particle swarm will serve as the input for the next round of simulation, forming a closed-loop iteration. Through continuous velocity and position adjustments, the particle swarm gradually gathers towards a high-density, high-quality solution region, while maintaining solution diversity through elite archiving to avoid premature convergence. This mechanism effectively combines global exploration and local development capabilities, enabling the optimization process to steadily advance in a complex, multi-peaked solution space, ultimately obtaining a set of reasonably distributed and balanced candidate design schemes, providing strong support for the comprehensive performance improvement of pure electric vehicle motors.

[0155] S6. Determine whether the updated Pareto solution set satisfies the convergence condition. If it does, output the Pareto solution set as the optimal design scheme. If it does not, return to S3.

[0156] Convergence conditions include reaching the preset maximum number of iterations, the Pareto solution set not showing significant improvement for several consecutive generations, or the change in the objective function value for several consecutive generations being less than a threshold.

[0157] When the algorithm meets the convergence condition, it outputs a Pareto solution set as the optimal design scheme. The Pareto solution set provides a variety of design trade-offs, and engineers can choose the appropriate scheme from the solution set according to actual needs. For example, if efficiency is important, the scheme with the highest efficiency is selected; if temperature rise is a concern, the scheme with the lowest temperature rise is selected.

[0158] This invention establishes a coupled model of electromagnetic field, temperature field, and structural field, simultaneously optimizing performance indicators such as efficiency, torque ripple, winding temperature rise, and vibration sound pressure level. This overcomes the performance imbalance problem caused by traditional single-physics field optimization, ensuring synergistic improvement in electromagnetic performance, thermal stability, and mechanical reliability of the motor. It achieves efficient convergence by improving the multi-objective particle swarm optimization algorithm, balances the global exploration and local development capabilities of the algorithm through a linear decreasing strategy to avoid premature convergence, maintains the distribution and convergence of the Pareto solution set through non-dominated sorting and crowding degree screening to improve the quality of the solution set, and dynamically selects guiding particles from the Pareto solution set to drive the particle swarm towards the true Pareto front. While ensuring electromagnetic performance, it effectively suppresses temperature rise and vibration noise, improving the overall reliability of the motor and providing an optimized design scheme for high-power-density electric vehicle motors that considers multiple performance objectives.

[0159] This invention also provides a particle swarm optimization system for multi-physics collaborative optimization of pure electric vehicle motors, used to execute the aforementioned particle swarm optimization method for multi-physics collaborative optimization of pure electric vehicle motors. Figure 3 This is a schematic diagram of the multi-physics collaborative optimization system for pure electric vehicle motors based on particle swarm optimization provided in an embodiment of the present invention. See also... Figure 3 The system includes the following modules:

[0160] The multiphysics construction module is used to build a multiphysics coupling model of a pure electric vehicle motor. The multiphysics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model.

[0161] The particle swarm optimization module is used to initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm. Each particle represents a set of motor design variables.

[0162] The simulation module is used to call the multiphysics coupling model to simulate each particle and calculate electromagnetic performance, thermal performance and structural performance indicators.

[0163] The sorting module is used to perform non-dominated sorting of each particle based on electromagnetic performance indicators, thermal performance indicators, and structural performance indicators, and update the Pareto solution set.

[0164] The update module is used to update particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarms, and repeat the simulation and sorting operations.

[0165] The output module is used to determine whether the updated Pareto solution set meets the convergence condition. If it does, the Pareto solution set is output as the optimization design scheme; otherwise, the updated particle swarm is returned.

[0166] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization, characterized in that, The method includes the following steps: S1. Establish a multi-physics coupling model for the electric motor of a pure electric vehicle. The multi-physics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model. S2. Initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of motor design variables; S3. Use the multiphysics coupling model to simulate each particle and calculate the electromagnetic performance index, thermal performance index and structural performance index. S4. Sort each particle in a non-dominated order according to the electromagnetic performance index, thermal performance index and structural performance index, and update the Pareto solution set. S5. Update the particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarm, and repeat S3-S4. S6. Determine whether the updated Pareto solution set satisfies the convergence condition. If it does, output the Pareto solution set as the optimized design scheme. If it does not, return to S3.

2. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization as described in claim 1, characterized in that, In S1, the electromagnetic field model includes: ； in, Let represent the Nabla operator, A represent the magnetic vector potential, μ represent the permeability, J represent the current density, σ represent the conductivity, and t represent time.

3. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 1, characterized in that, In S1, the temperature field model includes: ； in, Let ρ represent the Nabla operator, T represent the density, and C represent the temperature. p denoted by , b represents specific heat capacity, represents thermal conductivity, Q represents the heat source term, and t represents time.

4. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization as described in claim 1, characterized in that, In S1, the structural field model includes: ； in, Let ρ represent the Nabla operator, u represent the displacement vector, and σ represent the density. d Let F represent the stress tensor, F represent the volume force, and t represent time.

5. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 1, characterized in that, In S2, the improved multi-objective particle swarm optimization algorithm includes: Using linearly decreasing dynamic inertia weights: ； Where k represents the current iteration number, K represents the maximum iteration number, w0 represents the initial inertia weight, and w end Indicates the termination inertia weight, w k The inertia weight represents the current iteration number; Set learning factors, including individual cognitive coefficient c1 and social learning coefficient c2, and satisfy c1+c2≤4; where the individual cognitive coefficient is used to control the learning intensity of the particle towards its own historical best position, and the social learning coefficient is used to control the learning intensity of the particle towards the global best position of the group.

6. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 5, characterized in that, In step S2, the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm are initialized, including: S21. Initialize the parameters of the improved multi-objective particle swarm optimization algorithm, including population size, dynamic inertia weight, learning factor, maximum speed, and maximum number of iterations; S22. Initialize the particle swarm attributes of the improved multi-objective particle swarm optimization algorithm, including the position, velocity, optimal position, and optimal fitness of each particle. S23. Initialize the elite save file to an empty set.

7. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 1, characterized in that, In S3, the electromagnetic performance indicators include efficiency and torque pulsation, the thermal performance indicators include winding temperature rise, and the structural performance indicators include vibration sound pressure level.

8. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 6, characterized in that, In step S4, each particle is non-dominated and sorted according to the electromagnetic performance index, thermal performance index, and structural performance index, and the Pareto solution set is updated, including: S41. Sort the entire particle swarm according to the value of the first performance index. If the optimization objective of the performance index is to minimize it, sort in ascending order. If the optimization objective of the performance index is to maximize it, sort in descending order. S42. Initialize the first non-dominated frontier, add the first particle in the sorting to the first non-dominated frontier, compare the remaining particles with each particle in the first non-dominated frontier, if the remaining particles are not dominated by any particle in the first non-dominated frontier, add the remaining particles to the first non-dominated frontier, otherwise continue to traverse the next remaining particle until the traversal is completed, forming the first non-dominated frontier. S43. Remove the particles belonging to the first non-dominated front. Repeat S41-S43 for the remaining particles until all particles are assigned to a non-dominated front, resulting in several non-dominated fronts. S44. Calculate the crowding degree of each particle within each non-dominated front. S45. Compare the particles in the first non-dominated front with the original particles in the elite archive; if the original particles are dominated by the newly added particles, then remove the original particles; if the newly added particles are not dominated by any original particles, then add the newly added particles to the elite archive, which is the Pareto solution set. S46. Construct a new population by adding particles from the first non-dominated front to the new population and determining whether the number of particles in the new population is greater than or equal to the population size. If not, continue adding particles from the next non-dominated front. If yes, stop adding new particles and obtain a new population. S47. When the number of particles in the elite archive is greater than a preset threshold, or the number of particles in the new population is greater than the population size, they are removed in order of increasing crowding until the number of particles in the elite archive is equal to the preset threshold, or the number of particles in the new population is equal to the population size.

9. The multi-physics collaborative optimization method for pure electric vehicle motors based on particle swarm optimization according to claim 8, characterized in that, In step S5, the particle positions and velocities are updated based on the Pareto solution set to generate a new generation of particle swarms, including: The velocity and position of each particle in the new population are updated to obtain a new generation of particle swarm; The speed update formula is as follows: ； in, This represents the velocity of the i-th particle at iteration number k+1. Let r1 and r2 represent the velocity of the i-th particle at iteration number k, and r1 and r2 represent random numbers between [0,1]. Let represent the optimal position of the i-th particle itself. This represents the position of the i-th particle at iteration number k. This indicates the position of the guiding particle selected from the Pareto solution set for the i-th particle; The position update formula is as follows: ； in, This represents the position of the i-th particle at iteration number k+1.

10. A particle swarm optimization-based multiphysics collaborative optimization system for pure electric vehicle motors, used to execute the particle swarm optimization-based multiphysics collaborative optimization method for pure electric vehicle motors according to any one of claims 1-9, characterized in that, The system includes the following modules: A multiphysics construction module is used to establish a multiphysics coupling model of a pure electric vehicle motor. The multiphysics coupling model includes an electromagnetic field model, a temperature field model, and a structural field model. The particle swarm optimization module is used to initialize the parameters and particle swarm of the improved multi-objective particle swarm optimization algorithm. Each particle represents a set of motor design variables. The simulation module is used to call the multiphysics coupling model to simulate each particle and calculate electromagnetic performance indicators, thermal performance indicators and structural performance indicators. The sorting module is used to perform non-dominated sorting of each particle according to the electromagnetic performance index, thermal performance index and structural performance index, and update the Pareto solution set. The update module is used to update the particle positions and velocities based on the Pareto solution set, generate a new generation of particle swarms, and repeatedly perform simulation and sorting operations; The output module is used to determine whether the updated Pareto solution set meets the convergence condition. If it does, the Pareto solution set is output as the optimization design scheme; otherwise, the updated particle swarm is returned.

Images