Layered optimization design method for dual three-phase permanent magnet synchronous motor

Abstract

The invention discloses a layered optimization design method for a dual three-phase permanent magnet synchronous motor, and the method comprises the steps: firstly building a motor agent model according to a motor optimization target, and carrying out the optimization of a motor body through a multi-target optimization algorithm, so as to obtain a Pareto leading edge of a motor parameter; then, Kp and Ki parameter setting is carried out on a motor model in the Pareto frontier, so that the dynamic performance of the motor is optimized; and finally, determining an optimal solution containing dynamic performance and a motor body optimization target by combining a table look-up method with a multi-target optimization algorithm. According to the method, through joint optimization of the motor body and the control system, the dynamic performance of the motor and the body optimization target can be considered in the design stage, the overall robustness and the performance consistency of the motor are improved, and it is ensured that the performance of a machining prototype is kept consistent with a simulation result.

Description

Technical Field

[0001] The invention belongs to the field of motor design and control optimization, and in particular relates to a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor. Technical Background

[0002] In new energy drive systems, the structural design of motors and the optimization of controller parameters need to meet multiple performance requirements (such as efficiency, dynamic response, and steady-state robustness, etc.), but existing methods usually face optimization difficulties caused by high computational costs and strong coupling. Taking electric vehicle drive systems as an example, the efficiency, dynamic response, weight, and cost of the system often restrict each other. How to control the computational cost while improving the overall performance of the system has become a technical difficulty.

[0003] In the existing technology, due to the strong coupling between the motor structure and the control parameters, it is difficult for traditional methods to take into account multiple performance indicators, and the cost of using finite element analysis and high-precision simulation solutions is also high. Summary of the invention

[0004] In view of the shortcomings of the prior art in motor design and control optimization, the present invention provides a hierarchical optimization design method for dual three-phase permanent magnet synchronous motors. The method optimizes the physical structure of the motor in the first layer to generate a set of Pareto front solutions; then, in the second layer, the controller gain is adjusted one by one for the motor parameters in the Pareto front to optimize the dynamic response. Finally, the global optimal solution is found among multiple performance indicators through the table lookup method, and the system-level performance is improved. The method optimizes the motor body size parameters and control parameters in layers, achieves system optimization in terms of dynamic response and steady-state performance, and significantly reduces the calculation cost, providing an efficient and easy-to-promote optimization method for new energy drive systems.

[0005] The technical solution of the present invention to solve the above technical problem is to design a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor, characterized in that the method comprises the following steps:

[0006] S1: Determine the motor optimization target based on motor design requirements

[0007] The total loss and mass of the motor are set as the first optimization target to achieve the comprehensive performance of high efficiency and light weight; the overshoot and adjustment time of the motor are set as the second optimization target to improve the stability and speed of vehicle operation; in addition, in order to ensure the design requirements of the magnetic field distribution and output power of the motor, the following constraints must be met: the peak value of the magnetic density of the stator teeth shall not exceed 1.8T to avoid magnetic saturation affecting efficiency and reliability; the peak value of the magnetic density of the stator yoke shall not exceed 1.6T to ensure the normal conduction of the magnetic flux path; the output power of the motor must be greater than 30kW to meet the power performance requirements of the vehicle;

[0008] S2: Determine the design variables and size set of the motor body

[0009] Set the basic specifications of the motor body, consider the response of the two optimization objectives when the size parameters change, and select the local sensitivity analysis method. The analysis equation can be expressed as:

[0010]

[0011] Where z i represents a certain size parameter, and f is the optimization objective function;

[0012] Select the size parameters with absolute sensitivity greater than 0.3 as design variables, and their number is n;

[0013] S3: Constructing the optimization model and constraints of the motor body

[0014] Considering the motor's loss, mass, dynamic performance, and basic specification limitations of the motor itself, determine the motor optimization objectives and constraints, where the constraints include:

[0015] 1) Considering that the motor current density is limited to 7A / mm 2 , the magnetic flux density of the stator teeth is limited to 1.8T, and the magnetic flux density of the stator yoke is limited to 1.6T;

[0016] 2) Considering the motor output power requirement, the motor output power is limited to be greater than 30kW to meet the performance requirements;

[0017] 3) Considering the stability of torque fluctuation during speed regulation, torque ripple needs to be limited to less than 5%, so it is no longer used as an optimization variable;

[0018] Combining the motor optimization target determined in step S1, the basic specifications of the motor body determined in step S2, and the above constraints, the motor optimization model can be obtained:

[0019]

[0020] Where B ps is the stator pole magnetic flux density, B cs is the magnetic flux density of the stator yoke, T ripple is the torque pulsation;

[0021] S4: Construct the motor ontology proxy model, which includes the following sub-steps:

[0022] S41, through Latin hypercube sampling, the design variables x1~x n 200 sample points are randomly selected in the determined n-dimensional design space, and the total loss, mass, stator pole flux density, stator yoke flux density, torque ripple, and power of all sample points are calculated using ANSYS Motor-cad;

[0023] S42, based on the sample points obtained in S41, an approximate model is constructed using an artificial neural network to obtain a motor body proxy model;

[0024] S5: Multi-objective optimization is performed based on the NSGAII algorithm. The Pareto frontier of motor optimization is obtained. The improved non-dominated sorting genetic algorithm is used for multi-objective optimization, which specifically includes the following steps:

[0025] S51: Use randomly generated individuals to form the initial population;

[0026] S52: Fast non-dominated sorting and calculation of crowding; Fast non-dominated sorting aims to divide individuals into levels according to non-dominated relationships, while calculation of crowding is to more finely identify the differences between individuals in the same dominance layer;

[0027] S53: Selection and genetic manipulation; The selection operation aims to screen out excellent individuals from the current population so that they can participate in subsequent genetic manipulation and thus breed better offspring. Genetic manipulation mainly includes two major links: crossover and mutation. The crossover operation simulates the mating process of species in nature, creating new offspring by randomly pairing two individuals and exchanging some genes;

[0028] S54: Reordering and elite strategy operation; first merge the parent population and the child population, and then reorder them through fast non-dominated sorting; then select the best individuals based on the hierarchy and crowding degree to form a new population, implement the elite strategy for this new population, retain the excellent individuals, so as to improve the convergence speed of the algorithm;

[0029] S55: Iteration termination and output of the optimal solution; if the current number of iterations exceeds the preset maximum value, the iteration ends and the Pareto frontier is output; otherwise, if the maximum number of iterations is not reached, the next round of iterations is continued;

[0030] S6: Use Simulink to build a dynamic control system model including a PI motor controller

[0031] When establishing the dynamic mathematical model of the motor, the following assumptions are made: (1) core loss, hysteresis loss and eddy current are ignored; (2) damping and inductance effects are ignored; (3) the magnetic field of the rotor permanent magnet is sinusoidally distributed in the air gap space; (4) the stator induced electromotive force is also a sine wave; the mathematical model of the dual three-phase permanent magnet synchronous motor is established in the vector space decoupled VSD coordinate system:

[0032] Stator voltage equation:

[0033]

[0034] Where Ud , U q , U x , U y are the d, q, x, and y axis components of the stator voltage respectively; i d 、i q 、i x 、i y are the d, q, x, and y axis components of the stator current, R is the stator resistance, and w e is the electrical angular velocity of the motor rotor; L d , L q are the d-axis and q-axis inductance components respectively; L z =L AA1 , is the leakage inductance; ψ f is the magnetic link;

[0035] Electromagnetic torque equation:

[0036] T e =3p n i q [i d (L d -L q )+ψ f ] (4)

[0037] Where T e is the electromagnetic torque of the motor, p n is the pole pair number;

[0038] Motor motion equation:

[0039]

[0040] Where w m is the mechanical angular velocity of the motor, J is the moment of inertia, B is the damping coefficient, T L is the load torque;

[0041] The control adopts id=0 current double closed loop control. The whole control system consists of current loop regulator ACR, speed loop regulator ASR, coordinate transformation, SVPWM module, inverter and motor.

[0042] S7: Get the functional relationship of Kp and Ki setting according to the control system transfer function

[0043] Since the xy plane of the dual three-phase permanent magnet synchronous motor in the VSD coordinate system does not participate in the electromechanical energy conversion, only the dq sub-plane is considered; since the d- and q-axis current inner loops are symmetrical and have similar system characteristics, only the parameter setting method of the q-axis current PI regulator is analyzed below. The parameter setting of the d-axis current PI regulator is similar to that of the q-axis. The parameter setting is mainly divided into the parameter setting of the current regulator and the speed regulator, as follows:

[0044] S71: Parameter setting of current regulator

[0045] Due to the existence of delay and small inertia link of SVPWM control, the transfer function of PI regulator is transformed into zero-pole form:

[0046]

[0047] The two small time constants Ts are combined and corrected by the PI regulator to obtain the open-loop transfer function of the current loop:

[0048]

[0049] The closed-loop transfer function can be obtained from the open-loop transfer function:

[0050]

[0051] When the switching frequency is high, it is considered that S 2 The coefficient of is zero and can be ignored. Therefore, the closed-loop transfer function of the q-axis current loop can be obtained as follows:

[0052]

[0053] According to the dynamic performance index parameter table of the typical type I system, it is the best second-order system when the damping ratio ξ=0.707, so:

[0054]

[0055] The PI parameters of the current loop can be derived from the above formula:

[0056]

[0057] S72: Speed ​​regulator parameter setting

[0058] The load torque T l As a disturbance, the adhesion friction coefficient can be ignored in engineering; since the speed outer loop contains the current inner loop, the current inner loop is equivalent to a small time constant and a small time constant T of the speed signal sampling s Merge and get T es =5T s , and ignoring the load torque first, the open-loop transfer function is obtained as:

[0059]

[0060] The intermediate frequency bandwidth of the speed outer loop is h s for:

[0061]

[0062] According to the parameter setting relationship of the typical type II system, we can get:

[0063]

[0064] Take the intermediate frequency bandwidth h s =5, and substituting it into formula (18) we can get:

[0065]

[0066] Then the PI parameters of the speed outer loop can be derived as:

[0067]

[0068] S8: For each group of motor design variables in the Pareto frontier solution set, the controller parameters Kp and Ki are adjusted one by one to obtain the dynamic performance of each group of solutions; mainly including the following steps:

[0069] S81: Bring back the motor body design variables in the 100 sets of Pareto frontiers obtained from S6 to the finite element analysis to obtain the motor's resistance R, d-axis inductance Ld, q-axis inductance Lq, z-axis inductance Lz, moment of inertia J, and pole pair number n p , Magnetic Link

[0070] S82: R, Ld, Lq, Lz, J, n obtained from S81 p , Bring it into the motor mathematical model of Simulink and adjust k one by one li , k lp , k vi , k vp ;

[0071] S9: Create a comprehensive scoring function lookup table system, which specifically includes the following steps:

[0072] S91: Define the comprehensive scoring function

[0073] Set multiple optimization goals and assign weights to each goal. The comprehensive scoring function is expressed as:

[0074] J=0.3f 损耗 +0.3f 质量 +0.2f 超调量 +0.2f 调节时间 (18)

[0075] S92: Collect data and generate table lookup system

[0076] For each set of motor design variables and their corresponding control parameters k li , k lp , kvi , k vp Conduct simulation or experiment to obtain dynamic performance indicators such as overshoot, adjustment time, and motor body performance indicators such as total loss and total mass; construct all results into a table lookup system, in which each row of the table records a set of motor body design variables and control parameters and the corresponding comprehensive score;

[0077] S10: Look up the table and select the optimal solution

[0078] In the table lookup system, the motor body design variables and controller parameters are input to calculate the corresponding comprehensive score; the scores are sorted and the motor design scheme with the highest score is selected as the optimal motor design scheme.

[0079] Compared with the prior art, the present invention has the following beneficial effects:

[0080] 1. The present invention can optimize the motor performance and the dynamic response of the control system by combining the motor body size parameter design with the controller parameter optimization. The influence of the controller adjustment on the dynamic performance of the motor is considered in the motor design stage, thereby improving the performance of the overall motor system.

[0081] 2. The proxy model-based optimization method adopted in the present invention realizes efficient parameter optimization in the motor body size parameter optimization stage, while avoiding the problem of high-dimensional complexity, and is suitable for the optimization of new motors in the early design stage.

[0082] 3. The present invention processes the motor body and controller optimization problems in layers through a hierarchical optimization method, and optimizes the optimization target of the motor body and the dynamic performance of the controller parameters respectively, thereby effectively reducing the complexity of the optimization problem and improving the optimization accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 The present invention is a flowchart of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0084] Figure 2 The results of sensitivity analysis of size parameters of a dual three-phase permanent magnet synchronous motor body according to an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention are shown.

[0085] Figure 3 The figure is a schematic diagram of the position of the size parameters of the dual three-phase permanent magnet synchronous motor body part of an embodiment of a layered optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention.

[0086] Figure 4 A motor body proxy model fitting diagram of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention.

[0087] Figure 5 The present invention is a flowchart of a multi-objective optimization method for a dual three-phase permanent magnet synchronous motor based on the NSGAII algorithm in one embodiment of the present invention.

[0088] Figure 6 A Pareto front diagram obtained by performing multi-objective optimization based on the NSGAII algorithm in one embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention.

[0089] Figure 7 This is a motor control block diagram with id=0 of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention.

[0090] Figure 8 The present invention is a block diagram of the q-axis current loop structure of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0091] Fig. 9 The present invention is a simplified q-axis current loop structure block diagram of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0092] Fig.10 The present invention is a speed loop approximate control block diagram of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0093] Fig.11 The present invention is a speed outer loop control block diagram of an embodiment of a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0094] Fig.12 It is an embodiment of the invention and shows a core loss diagram before and after optimization using a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor.

[0095] Fig.13 The present invention is an embodiment of an eddy current loss diagram before and after optimization using a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention.

[0096] Fig.14 The figure is a comparison diagram of the quality of various motor components before and after optimization using a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention in accordance with an embodiment of the present invention.

[0097] Fig.15 The present invention is an embodiment of a dynamic performance diagram before and after optimization using a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor of the present invention. DETAILED DESCRIPTION

[0098] The specific embodiments of the present invention are given below. The specific embodiments are only used to further illustrate the present invention in detail and do not limit the protection scope of the claims of this application.

[0099] The motor used in the embodiment of the present invention is a 48-slot / 8-pole dual three-phase permanent magnet motor. Compared with the three-phase permanent magnet motor, the dual three-phase motor adopts multiple phase groups in parallel, which can improve the current balance and system robustness when the motor is running. Due to the increase in the number of phases, the motor can achieve more precise current regulation, thereby improving the efficiency and output torque density of the motor. Compared with the traditional three-phase motor, the dual three-phase motor can reduce torque pulsation and improve the torque quality of the motor through reasonable phase arrangement and multi-phase drive advantages, especially under low speed and high load conditions. Compared with the traditional three-phase motor, it can provide higher power density and smaller size, and is suitable for applications with strict requirements on space and weight, such as electric vehicles and industrial automation equipment.

[0100] The present invention provides a hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor, the method comprising the following steps:

[0101] S1: Determine the motor optimization target based on motor design requirements

[0102] On the one hand, automotive motors need to take into account both energy efficiency and dynamic performance during operation. Therefore, the total loss (including copper loss and iron loss) and mass of the motor are set as the first optimization target to achieve the comprehensive performance of high efficiency and lightweight; on the other hand, in order to meet the dynamic response requirements of the control system, the overshoot and adjustment time of the motor are set as the second optimization target to improve the smoothness and rapidity of vehicle operation. In addition, in order to ensure the design requirements of the magnetic field distribution and output power of the motor, the following constraints must be met: the peak value of the stator tooth flux density shall not exceed 1.8T (T is the special gammon, abbreviated as a unit) to avoid magnetic saturation affecting efficiency and reliability. The peak value of the stator yoke flux density shall not exceed 1.6T to ensure the normal conduction of the magnetic flux path. The output power of the motor must be greater than 30kW to meet the power performance requirements of the vehicle.

[0103] S2: Determine the motor body design variables and size set

[0104] The basic specifications of the motor body are set. The sensitivity analysis of the variables can be used to analyze the sensitivity of the system output changes to the changes in the system size parameters. For multi-objective optimization, the sensitivity analysis can be used to obtain the sensitivity of different optimization objectives to the optimization parameters.

[0105] Sensitivity analysis mainly includes two methods, namely local sensitivity analysis and global sensitivity analysis. Local sensitivity analysis is based on the partial derivative analysis method, which mainly focuses on the impact of a single parameter on the result, while global sensitivity analysis considers the degree of change in the output of the entire system when the input parameters change. For the motor studied in this paper, the response of the two optimization objectives when the size parameters change is mainly considered, so the local sensitivity analysis method is selected, and its analysis equation can be expressed as:

[0106]

[0107] Where z i represents a certain size parameter, and f is the optimization objective function.

[0108] In this example, sensitivity analysis was performed on 14 size parameters, and the results are as follows: Figure 2 As shown in the figure, it can be seen that the optimization objective has different sensitivities to different size parameters.

[0109] In order to improve the quality of optimization, this paper selects dimensional parameters with absolute sensitivity greater than 0.3 as design variables. Finally, in the optimization process, six dimensional parameters including stator inner diameter, slot depth, permanent magnet length, air gap, V-pole angle and shaft length are selected as design variables of the motor body.

[0110] Figure 3 Schematic diagram of the positions of the five design variables of this embodiment, where a is the stator inner diameter R si , b is the groove depth H s2 , c is the length of the permanent magnet I N35UH , d is the air gap g, e is the V pole angle α pm The shaft length L is the axial length of the stator.

[0111] According to the initial design, the basic specifications of the motor body and the initial values ​​and variation ranges of the design variables are shown in Table 1 and Table 2.

[0112] Table 1 Basic specifications of the motor body

[0113] name Numeric unit Rated Power 30 kW Rated frequency 200 Hz Rated speed 3000 rpm Rated line voltage 330 V Rated torque 90 Nm Number of phases 6 \ Topology Built-in V-type \

[0114] Table 2 Design variables of the motor body

[0115] name Initial Value Range of variation unit Stator inner diameter 148 135-150 mm Groove depth 21 18-25 mm Permanent magnet length 18.5 16-19 mm Air Gap 1 0.5-2 mm V pole angle 135 120-140 mm Axis length 150 90-150 mm

[0116] S3: Constructing the optimization model and constraints of the motor body

[0117] Considering the motor's losses, mass, dynamic performance (overshoot and adjustment time), and the basic specification limitations of the motor itself, determine the motor optimization goals and constraints, where the constraints include:

[0118] 1) Considering that the motor current density in the embodiment is limited to 7A / mm 2 The magnetic flux density of the stator teeth is limited to 1.8T, and the magnetic flux density of the stator yoke is limited to 1.6T.

[0119] 2) Considering the motor output power requirement of the embodiment, the motor output power is limited to be greater than 30kW to meet the performance requirements.

[0120] 3) Considering the stability of torque fluctuation during speed regulation, the torque ripple needs to be limited to less than 5%, so it is no longer used as an optimization variable.

[0121] Combining the motor optimization target determined in step S1, the basic specifications of the motor body determined in step S2, and the above constraints, the motor optimization model can be obtained:

[0122]

[0123] Where B ps is the stator pole magnetic flux density, B cs is the magnetic flux density of the stator yoke, T ripple is the torque pulsation.

[0124] S4: Construct the motor ontology proxy model, which includes the following sub-steps:

[0125] S41. 200 sample points are randomly selected in the 6-dimensional design space determined by the design variables x1~x6 through Latin hypercube sampling, and the total loss, mass, stator pole flux density, stator yoke flux density, torque pulsation, and power of all sample points are calculated using ANSYS Motor-cad.

[0126] S42, based on the sample points obtained in S41, an approximate model is constructed using an artificial neural network to obtain a proxy model of the motor body. The proxy model is fitted as follows: Figure 4 As shown in Figure 3, the accuracy of the proxy model meets the subsequent optimization requirements.

[0127] S5: Multi-objective optimization based on NSGAII algorithm to obtain the Pareto frontier of motor optimization

[0128] In the process of solving multi-objective optimization problems, a commonly used method is to set some optimization objectives as constraints, thereby converting the multi-objective optimization problem into a single-objective optimization problem. However, this does not take into account the influence relationship between the optimization objectives and will lead to local optimal solutions, which is not suitable for the multi-objective optimization model of this paper.

[0129] In order to improve the optimization quality, this embodiment adopts an improved nondominated sorting genetic algorithm (NSGAⅡ) for multi-objective optimization. The NSGAⅡ algorithm has the characteristics of being able to search the entire population and having strong globality. The process of the algorithm is as follows: Figure 5 As shown, the specific steps include:

[0130] S51: Use randomly generated individuals to form an initial population.

[0131] S52: Fast non-dominated sorting and calculation of crowding. Fast non-dominated sorting aims to divide individuals in the population into levels according to non-dominated relationships, while calculation of crowding is to more finely distinguish the differences between individuals in the same dominance layer.

[0132] S53: Selection and genetic manipulation. The selection operation aims to screen out excellent individuals from the current population so that they can participate in subsequent genetic manipulation and thus produce better offspring. Genetic manipulation mainly includes two major links: crossover and mutation. The crossover operation simulates the mating process of species in nature, creating new offspring by randomly pairing two individuals and exchanging some genes.

[0133] S54: Reordering and elite strategy operation. First, merge the parent population and the child population, and then reorder them through fast non-dominated sorting. Then select the best individuals based on the level and crowding degree to form a new population. Implement the elite strategy for this new population and retain the excellent individuals to improve the convergence speed of the algorithm.

[0134] S55: Iteration termination and output of optimal solution. If the current number of iterations exceeds the preset maximum value, the iteration ends and the Pareto frontier is output; otherwise, if the maximum number of iterations is not reached, the next round of iterations will continue.

[0135] The above NSGA-II algorithm is used to optimize the proxy model. The optimized results are as follows: Figure 6 As shown in the figure, the Pareto frontier shows that the total motor loss and the total motor mass are mutually constrained, that is, when the loss decreases, the generator mass will increase, and when the motor mass decreases, the electromagnetic loss will increase. The two cannot achieve the optimal results at the same time.

[0136] A total of 100 groups of partial data of the Pareto frontier obtained in this embodiment are shown in Table 3.

[0137] Table 3 Pareto frontier obtained in this example

[0138]

[0139] S6: Use Simulink to build a dynamic control system model including a PI motor controller

[0140] When establishing the dynamic mathematical model of the motor, in order to simplify the complexity of the model analysis, the following assumptions are usually made: (1) Ignore core loss, hysteresis loss and eddy current; (2) Ignore damping and inductance; (3) The magnetic field of the rotor permanent magnet is sinusoidally distributed in the air gap space; (4) The stator induced electromotive force is also a sine wave. The mathematical model of the dual three-phase permanent magnet synchronous motor is established in the vector space decoupling (VSD) coordinate system:

[0141] Stator voltage equation:

[0142]

[0143] Where U d , U q , U x , U y are the d, q, x, and y axis components of the stator voltage respectively; i d 、i q 、i x 、i y are the d, q, x, and y axis components of the stator current, R is the stator resistance, and w e is the electrical angular velocity of the motor rotor; L d , L q are the d-axis and q-axis inductance components respectively; L z =L AA1 , is the leakage inductance; ψ f For magnetic link.

[0144] Electromagnetic torque equation:

[0145] T e =3p n i q [i d (L d -L q )+ψ f ] (4)

[0146] Where T e is the electromagnetic torque of the motor, p n is the pole pair number.

[0147] Motor motion equation:

[0148]

[0149] Where w m is the mechanical angular velocity of the motor, J is the moment of inertia, B is the damping coefficient, T Lis the load torque.

[0150] The control adopts id=0 current double closed loop control, the control diagram is as follows Figure 7 As shown, n* is the given desired speed, and the entire control system consists of a current loop regulator ACR, a speed loop regulator ASR, a coordinate transformation, a SVPWM module, an inverter and a motor.

[0151] S7: Get the functional relationship of Kp and Ki setting according to the control system transfer function

[0152] Since the xy plane of the dual three-phase permanent magnet synchronous motor in the VSD coordinate system does not participate in the electromechanical energy conversion, the following analysis only considers the dq sub-plane. Since the d- and q-axis current inner loops are symmetrical and have similar system characteristics, the following only analyzes the parameter setting method of the q-axis current PI regulator. The parameter setting of the d-axis current PI regulator is similar to that of the q-axis. The parameter setting is mainly divided into the parameter setting of the current regulator and the speed regulator, as follows;

[0153] S71: Parameter setting of current regulator

[0154] Due to the existence of delay and small inertia link of SVPWM control, the q-axis current loop structure diagram is as follows Figure 8 As shown in the figure, Ts represents the sampling period of the current loop. For the convenience of analysis, the influence of disturbance is not considered first, and the transfer function of the PI regulator is converted into a zero-pole form:

[0155]

[0156] Combining the two small time constants Ts, we get the following simplified block diagram of the q-axis current loop: Fig. 9 As shown in the figure, after correction by the PI regulator, the open-loop transfer function of the current loop is:

[0157]

[0158] The closed-loop transfer function can be obtained from the open-loop transfer function:

[0159]

[0160] When the switching frequency is high, T s If the value of is small enough, we can think that S 2 The coefficient is zero and can be ignored. Therefore, the closed-loop transfer function of the q-axis current loop can be obtained as:

[0161]

[0162] According to the dynamic performance index parameter table of the typical type I system, it is the best second-order system when the damping ratio ξ=0.707, so:

[0163]

[0164] The PI parameters of the current loop can be derived from the above formula:

[0165]

[0166] S72: Speed ​​regulator parameter setting

[0167] The load torque T l As a disturbance, since the adhesion friction coefficient can be ignored in engineering, the speed loop approximate control block diagram can be obtained, as shown in Fig.10 As shown, since the speed outer loop contains the current inner loop, the current inner loop is equivalent to the small time constant and the small time constant T of the speed signal sampling. s Merge and get T es =5T s , and ignoring the load torque first, the structural block diagram of the speed outer loop is obtained as follows: Fig.11 As shown by Fig.11 , the open-loop transfer function is obtained as:

[0168]

[0169] The intermediate frequency bandwidth of the speed outer loop is h s for:

[0170]

[0171] According to the parameter setting relationship of the typical type II system, we can get:

[0172]

[0173] In engineering applications, the intermediate frequency bandwidth h is generally taken s =5, and substituting it into formula (18) we can get:

[0174]

[0175] Then the PI parameters of the speed outer loop can be derived as:

[0176]

[0177] The above is the theoretical analysis and calculation of the PI regulator parameters for the current inner loop and the speed outer loop.

[0178] S8: For each group of motor design variables in the Pareto frontier solution set, the controller parameters Kp and Ki are adjusted one by one to obtain the dynamic performance of each group of solutions. It mainly includes the following steps:

[0179] S81: Bring back the motor body design variables in the 100 sets of Pareto frontiers obtained from S6 to the finite element analysis to obtain the motor's resistance R, d-axis inductance Ld, q-axis inductance Lq, z-axis inductance Lz, moment of inertia J, and pole pair number n p , Magnetic Link

[0180] S82: R, Ld, Lq, Lz, J, n obtained from S81 p , Bring it into the motor mathematical model of Simulink and adjust k one by one li , k lp , k vi , k vp Some of the data obtained are shown in Table 4 below.

[0181] Table 4k li , k lp , k vi , k vp

[0182]

[0183] S9: Create a comprehensive scoring function lookup table system

[0184] In order to more efficiently evaluate and select motor design and its control parameters, a comprehensive scoring function lookup table system is established. The core of this system is to use a comprehensive scoring function to convert the efficiency, loss, overshoot and adjustment time of multiple optimization target motor bodies into a unified scoring standard. By looking up the table, the optimal design can be quickly extracted from the pre-calculated data. The specific steps include:

[0185] S91: Define the comprehensive scoring function

[0186] Set multiple optimization goals and assign weights to each goal. The comprehensive scoring function can be expressed as:

[0187] J=w1f 损耗 +w2f 质量 +w3f 超调量 +w4f 调节时间 (17)

[0188] Among them, f 损耗 、f 质量 、f 超调量 、f 调节时间 They are the performance indicators of the motor body and the control system respectively, and w1, w2, w3, and w4 are their weights.

[0189] Total loss is a key factor in the efficiency and endurance of electric vehicles. Although its weight is not as high as when it is lightweight or high-efficiency, an appropriate weight of 0.3 ensures that energy efficiency is not ignored. Total mass affects power, acceleration and endurance. Lightweighting helps improve acceleration and endurance. To balance energy efficiency and performance, the weight of mass is set to 0.3, which is equal to total loss. Overshoot reflects the stability and response speed of the control system. A moderate overshoot can improve the response speed, but too much overshoot may cause instability. In view of the stability requirements, the overshoot weight is set to 0.2. Adjustment time affects the dynamic response of the system. A shorter adjustment time helps to respond quickly, but in comprehensive optimization, electric vehicles pay more attention to stability and moderate response, so the weight is 0.2.

[0190] After determining the weights, the comprehensive scoring function is adjusted to:

[0191] J=0.3f 损耗 +0.3f 质量 +0.2f 超调量 +0.2f 调节时间 (18)

[0192] S92: Collect data and generate table lookup system

[0193] For each set of motor design variables and their corresponding control parameters k li , k lp , k vi , k vp Simulate or experiment to obtain the dynamic performance indicators overshoot, adjustment time, and motor body performance indicators total loss and total mass. All results are constructed into a lookup table system, and each row of the table records a set of motor body design variables and control parameters and the corresponding comprehensive score.

[0194] S10: Look up the table and select the optimal solution

[0195] In the table lookup system, the motor design variables and controller parameters are input to calculate the corresponding comprehensive score. The motor design with the highest score is selected as the optimal motor design according to the score value, which can not only meet the performance requirements of the motor body, but also achieve good dynamic performance of the control system. The final motor design is shown in Table 5.

[0196] Table 5 Optimal motor design scheme

[0197] Motor body design parameters Optimization value Control parameters Setting value Stator inner diameter 137.6286mm <![CDATA[k li ]]> 22.179 Groove depth 21.9604mm <![CDATA[k lp ]]> 0.6935 Permanent magnet length 18.7316mm <![CDATA[k vi ]]> 235.909 Air Gap 0.6707mm <![CDATA[k vp ]]> 5.8929 V pole angle 120.0439° Axis length 137.7228mm

[0198] The loss, mass overshoot and adjustment time of the above-mentioned motor design scheme are evaluated through finite element simulation and Simlulink simulation.

[0199] The total motor loss mainly includes winding loss, core loss and permanent magnet eddy current loss. The permanent magnet synchronous motor winding loss refers to the energy loss caused by the resistance of the current passing through the wire in the permanent magnet synchronous motor winding. The main factors affecting this loss are the resistance of the wire and the current flowing through the winding. Ideally, the copper loss of the drive motor winding can be expressed as:

[0200] P Cu =mI 2 R (19)

[0201] In the formula, m is the number of winding phases; I is the winding phase current; and R is the winding resistance. The winding loss before optimization can be calculated as 632.4W, and the winding loss after optimization is 588.7W, a decrease of 6.9%.

[0202] Fig.12 This is a comparison of the motor core loss performance before and after optimization of the embodiment of the present invention. It can be seen that the motor core loss after optimization is reduced from 245.75W to 223.63W, a decrease of 9%.

[0203] Fig.13 The eddy current loss performance of the motor permanent magnet before and after optimization of the embodiment of the present invention is compared. It can be seen that the eddy current loss of the motor permanent magnet after optimization is reduced from 173.2028mW to 96.65mW, a decrease of 44.19%. The total loss is reduced by 7.49% after the comparison of the loss before and after optimization.

[0204] Fig.14 The weight of the motor components before and after optimization in the embodiment of the present invention is 34.478 kg, and the total weight after optimization is 30.496 kg, with a weight reduction of 11.55%. Moreover, the performance distribution of the motor after optimization meets the established design requirements.

[0205] Then the overshoot and adjustment time are verified by simulink simulation. Fig.15 It can be seen that its adjustment time is advanced from 0.369s to 0.308s, and the overshoot is reduced from 3.1737% to 0.4224%.

[0206] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A hierarchical optimization design method for a dual three-phase permanent magnet synchronous motor, characterized in that: The method comprises the following steps: S1: Determine the motor optimization target based on motor design requirements The total loss and mass of the motor are set as the first optimization target to achieve the comprehensive performance of high efficiency and light weight; the overshoot and adjustment time of the motor are set as the second optimization target to improve the stability and speed of vehicle operation; in addition, in order to ensure the design requirements of the magnetic field distribution and output power of the motor, the following constraints must be met: the peak value of the magnetic density of the stator teeth shall not exceed 1.8T to avoid magnetic saturation affecting efficiency and reliability; the peak value of the magnetic density of the stator yoke shall not exceed 1.6T to ensure the normal conduction of the magnetic flux path; the output power of the motor must be greater than 30kW to meet the power performance requirements of the vehicle; S2: Determine the design variables and size set of the motor body Set the basic specifications of the motor body, consider the response of the two optimization objectives when the size parameters change, and select the local sensitivity analysis method. The analysis equation can be expressed as: Where z i represents a certain size parameter, and f is the optimization objective function; Select the size parameters with absolute sensitivity greater than 0.3 as design variables, and their number is n; S3: Constructing the optimization model and constraints of the motor body Considering the motor's loss, mass, dynamic performance, and basic specification limitations of the motor itself, determine the motor optimization objectives and constraints, where the constraints include: 1) Considering that the motor current density is limited to 7A / mm 2 , the magnetic flux density of the stator teeth is limited to 1.8T, and the magnetic flux density of the stator yoke is limited to 1.6T; 2) Considering the motor output power requirement, the motor output power is limited to be greater than 30kW to meet the performance requirements; 3) Considering the stability of torque fluctuation during speed regulation, torque ripple needs to be limited to less than 5%, so it is no longer used as an optimization variable; Combining the motor optimization target determined in step S1, the basic specifications of the motor body determined in step S2, and the above constraints, the motor optimization model can be obtained: Where B ps is the stator pole magnetic flux density, B cs is the magnetic flux density of the stator yoke, T ripple is the torque pulsation; S4: Construct the motor ontology proxy model, which includes the following sub-steps: S41, through Latin hypercube sampling, the design variables x1~x n 200 sample points are randomly selected in the determined n-dimensional design space, and the total loss, mass, stator pole flux density, stator yoke flux density, torque ripple, and power of all sample points are calculated using ANSYS Motor-cad; S42, based on the sample points obtained in S41, an approximate model is constructed using an artificial neural network to obtain a motor body proxy model; S5: Multi-objective optimization is performed based on the NSGAII algorithm. The Pareto frontier of motor optimization is obtained. The improved non-dominated sorting genetic algorithm is used for multi-objective optimization, which specifically includes the following steps: S51: Use randomly generated individuals to form the initial population; S52: Fast non-dominated sorting and calculation of crowding; Fast non-dominated sorting aims to divide individuals into levels according to non-dominated relationships, while calculation of crowding is to more finely identify the differences between individuals in the same dominance layer; S53: Selection and genetic manipulation; The selection operation aims to screen out excellent individuals from the current population so that they can participate in subsequent genetic manipulation and thus breed better offspring. Genetic manipulation mainly includes two major links: crossover and mutation. The crossover operation simulates the mating process of species in nature, creating new offspring by randomly pairing two individuals and exchanging some genes; S54: Reordering and elite strategy operation; first merge the parent population and the child population, and then reorder them through fast non-dominated sorting; then select the best individuals based on the hierarchy and crowding degree to form a new population, implement the elite strategy for this new population, retain the excellent individuals, so as to improve the convergence speed of the algorithm; S55: Iteration termination and output of the optimal solution; if the current number of iterations exceeds the preset maximum value, the iteration ends and the Pareto frontier is output; otherwise, if the maximum number of iterations is not reached, the next round of iterations is continued; S6: Use Simulink to build a dynamic control system model including a PI motor controller When establishing the dynamic mathematical model of the motor, the following assumptions are made: (1) core loss, hysteresis loss and eddy current are ignored; (2) damping and inductance effects are ignored; (3) the magnetic field of the rotor permanent magnet is sinusoidally distributed in the air gap space; (4) the stator induced electromotive force is also a sine wave; the mathematical model of the dual three-phase permanent magnet synchronous motor is established in the vector space decoupled VSD coordinate system: Stator voltage equation: Where U d , U q , U x , U y are the d, q, x, and y axis components of the stator voltage respectively; i d 、i q 、i x 、i y are the d, q, x, and y axis components of the stator current, R is the stator resistance, and w e is the electrical angular velocity of the motor rotor; L d , L q are the d-axis and q-axis inductance components respectively; L z =L AA1 , is the leakage inductance; ψ f is the magnetic link; Electromagnetic torque equation: T e =3p n and q [i d (L d -L q )+ψ f ] (4) Where T e is the electromagnetic torque of the motor, p n is the pole pair number; Motor motion equation: Where w m is the mechanical angular velocity of the motor, J is the moment of inertia, B is the damping coefficient, T L is the load torque; The control adopts id=0 current double closed loop control. The whole control system consists of current loop regulator ACR, speed loop regulator ASR, coordinate transformation, SVPWM module, inverter and motor. S7: Get the functional relationship of Kp and Ki setting according to the control system transfer function Since the xy plane of the dual three-phase permanent magnet synchronous motor in the VSD coordinate system does not participate in the electromechanical energy conversion, only the dq sub-plane is considered; since the d- and q-axis current inner loops are symmetrical and have similar system characteristics, only the parameter setting method of the q-axis current PI regulator is analyzed below. The parameter setting of the d-axis current PI regulator is similar to that of the q-axis. The parameter setting is mainly divided into the parameter setting of the current regulator and the speed regulator, as follows: S71: Parameter setting of current regulator Due to the existence of delay and small inertia link of SVPWM control, the transfer function of PI regulator is transformed into zero-pole form: The two small time constants Ts are combined and corrected by the PI regulator to obtain the open-loop transfer function of the current loop: The closed-loop transfer function can be obtained from the open-loop transfer function: When the switching frequency is high, it is considered that S 2 The coefficient of is zero and can be ignored. Therefore, the closed-loop transfer function of the q-axis current loop can be obtained as follows: According to the dynamic performance index parameter table of the typical type I system, it is the best second-order system when the damping ratio ξ=0.707, so: The PI parameters of the current loop can be derived from the above formula: S72: Speed ​​regulator parameter setting The load torque T l As a disturbance, the adhesion friction coefficient can be ignored in engineering; since the speed outer loop contains the current inner loop, the current inner loop is equivalent to a small time constant and a small time constant T of the speed signal sampling s Merge and get T es =5T s , and ignoring the load torque first, the open-loop transfer function is obtained as: The intermediate frequency bandwidth of the speed outer loop h s for: According to the parameter setting relationship of the typical type II system, we can get: Take the intermediate frequency bandwidth h s =5, and substituting it into formula (18) we can get: Then the PI parameters of the speed outer loop can be derived as: S8: For each group of motor design variables in the Pareto frontier solution set, the controller parameters Kp and Ki are adjusted one by one to obtain the dynamic performance of each group of solutions; mainly including the following steps: S81: Bring back the motor body design variables in the 100 sets of Pareto frontiers obtained from S6 to the finite element analysis to obtain the motor's resistance R, d-axis inductance Ld, q-axis inductance Lq, z-axis inductance Lz, moment of inertia J, and pole pair number n p , Magnetic Link S82: R, Ld, Lq, Lz, J, n obtained from S81 p , Bring it into the motor mathematical model of Simulink and adjust k one by one li , k lp , k vi , k vp ; S9: Create a comprehensive scoring function lookup table system, which specifically includes the following steps: S91: Define the comprehensive scoring function Set multiple optimization goals and assign weights to each goal. The comprehensive scoring function is expressed as: <h2 style=";text-align:left;direction:ltr">J=0.3f<h2 style=";text-align:left;direction:ltr"> 损耗 <h2 style=";text-align:left;direction:ltr"> +0.3f<h2 style=";text-align:left;direction:ltr"> 质量 <h2 style=";text-align:left;direction:ltr"> +0.2f<h2 style=";text-align:left;direction:ltr"> 超调量 <h2 style=";text-align:left;direction:ltr"> +0.2f<h2 style=";text-align:left;direction:ltr"> 调节时间 <h2 style=";text-align:left;direction:ltr"> (18) S92: Collect data and generate table lookup system For each set of motor design variables and their corresponding control parameters k li , k lp , k vi , k vp Conduct simulation or experiment to obtain dynamic performance indicators such as overshoot, adjustment time, and motor body performance indicators such as total loss and total mass; construct all results into a table lookup system, in which each row of the table records a set of motor body design variables and control parameters and the corresponding comprehensive score; S10: Look up the table and select the optimal solution In the table lookup system, the motor body design variables and controller parameters are input to calculate the corresponding comprehensive score; the scores are sorted and the motor design scheme with the highest score is selected as the optimal motor design scheme.

Images