Voltage Distribution and Insulation Weak Point Prediction Method Based on the Flat Wire Winding of Variable Frequency Motors

Abstract

The invention discloses a method for predicting voltage distribution and insulation weak points based on a flat wire winding of a variable-frequency motor, belonging to the technical field of motor design and insulation evaluation. This method comprehensively considers the geometric shape, insulation structure, and working voltage of the motor stator, establishes a complete finite element model of the flat wire winding motor, extracts distribution parameters such as capacitance, inductance, and resistance of the cable and motor winding, constructs a winding 'I'-type high-frequency equivalent circuit model, and performs field-circuit coupled electromagnetic analysis using finite element simulation software. Moreover, the above processes can all be automatically realized through the compiled program. Only the user needs to provide the original motor model and working parameters, which can efficiently and relatively accurately predict the voltage distribution of the flat wire motor winding under variable-frequency drive and generate a radar chart that can intuitively display the insulation weak points, providing new ideas for locating the insulation weak points of the motor, guiding the optimization of motor design, and improving the evaluation efficiency and accuracy of the motor insulation performance.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor design and insulation evaluation, and specifically relates to a method for predicting voltage distribution and insulation weak points based on a flat wire winding of a variable frequency motor. Background Art

[0002] With the development of new energy electric vehicles, the demand for the power density of the electric drive system is increasing continuously, and the drive motor is developing towards high speed, high frequency, and high voltage. To further improve the power density and motor efficiency, new energy vehicles tend to use flat wire winding motors. The flat wire winding usually inserts one turn of the coil axially into the slot and then performs transposition and welding at the end. Due to its advantages such as high slot fill factor, good thermal conductivity, and low vibration and noise, it is favored in the drive motors of electric vehicles. With the more mature manufacturing process, the existing forming and welding equipment can meet the requirements of its mass production.

[0003] The introduction of the third-generation wide bandgap semiconductor switching device represented by SiC Mosfet, the increase in the bus voltage level, and the short rise time make the motor excitation have the characteristic of high voltage change gradient (dv / dt), resulting in a larger overvoltage at the three-phase input end of the motor; in addition, the high-frequency dv / dt pulse voltage generally reaches several MHz to dozens of MHz. At this time, parameters such as the distributed capacitance and distributed inductance of the motor winding cannot be ignored. Due to impedance mismatch and the influence of the distributed parameters of the winding at high frequencies, the pulse voltage is unevenly distributed inside the motor winding.

[0004] The transmission line model is usually used to simulate the wave propagation effect in the inverter-fed motor winding and the long motor supply cable, which is represented by a lumped parameter equivalent circuit and is modeled as a series of equivalent Γ-type or π-type sections. The high-frequency components need to be processed by a distributed parameter circuit model, and these high-frequency components are considered up to the maximum frequency fmax, which is usually greater than the cut-off frequency of the FFT spectrum of the supply voltage signal (t r is the rise time of the pulsed stator voltage caused by reflection). Take The minimum wavelength of the reflected waveform along the transmission line is approximately: λ min = ν c / f max ≈ ν c ·t r , where v c is the wave propagation speed. For a cable, this speed is about half of the speed of light in a vacuum v c ≈ 150 m / μs. For the motor winding, due to the core lamination effect, the speed v c <75 m / μs. The rise time of the pulsed voltage at the inlet end of the motor winding is usually t r= 200...300 ns. At a propagation speed of 75 m / μs, the minimum wavelength of the reflected electromagnetic wave is denoted as λ min ≈ ν c ·t r = 75 m / μs · 200 ns = 15 m. When the minimum wavelength takes at least 5…10 Γ-type or π-type sections, the transmission line is properly modeled. Thus, the section length λ min / 10 ≈ 1.5 m. For standard induction motors (up to 100 kW), this value is larger than the length of one turn of the winding. For larger machines (>100 kW), it is more appropriate to take several Γ-type or π-type sections for each winding.

[0005] In addition to the distributed parameters affecting the winding voltage distribution, there are various winding connection structure schemes during the motor design stage, and different connection schemes will also result in different voltage distribution situations. The long-term exposure to overvoltage shocks and uneven voltage distribution greatly increases the probability of premature failure of the motor insulation system during service. The waveforms of the inter-turn / phase-to-phase / phase-to-ground voltages borne by the motor during operation are as Figure 4 shown. In particular, partial inter-turn / phase-to-phase / phase-to-ground insulation is prone to forming local high field strengths. When vacuum impregnation cannot completely eliminate air gaps, partial discharge (PD) will occur with a relatively high probability in the insulation area where the field strength is concentrated, accelerating insulation aging. Partial discharge is one of the main reasons for the insulation damage of variable-frequency motors. During the operation of electrical equipment, the generation of partial discharge should be avoided in most cases. Therefore, in the factory inspection and acceptance tests of electrical equipment, the partial discharge inception voltage (PDIV) and the partial discharge extinction voltage (PDEV) are two important parameters for evaluating insulation performance. Once the voltage borne between the motor windings exceeds its partial discharge inception voltage, it will cause rapid deterioration of the motor insulation and greatly shorten the service life of the motor. And the inter-turn short-circuit fault of the winding is the most common and serious type among various motor faults. According to statistics, the inter-turn short-circuit faults of motors account for about 30% - 40% of various motor faults. To evaluate the weak points of the motor insulation, the International Electrotechnical Commission (IEC) has formulated corresponding standards, stipulating that PDIV tests need to be carried out on different insulation parts of the stator before the motor leaves the factory and starts operation to ensure that partial discharge (PD) will not occur under working conditions.

[0006] In this case, it is particularly important to study the voltage distribution and insulation weak points of the flat wire winding motor. However, the traditional voltage distribution test methods mainly rely on manual testing or manual modeling and simulation. Manual testing is complex in operation. For the motor to be tested, additional wires need to be led out for testing. In the case of no lead-out wires, the insulation paint film needs to be worn to expose the copper conductor for voltage testing. Moreover, manual testing is for the already formed motor and cannot meet the insulation evaluation requirements in the motor design stage. For manual modeling and simulation, there are problems such as low automation level, time-consuming and laborious. Taking a 54-slot, 6-layer motor as an example, the complete high-frequency equivalent circuit of the three-phase winding is a complex network including 324 simulation units, hundreds of resistors, inductors and thousands of capacitors. In addition, the existing technologies have problems of insufficient efficiency and unintuitive results in predicting voltage distribution and locating insulation weak points.

[0007] To sum up, in order to realize the automatic prediction of the motor voltage distribution, improve the evaluation efficiency and accuracy, and meet the requirements in the motor design stage, it is of great significance to develop a set of automatic voltage distribution prediction system. At the same time, understanding the voltage distribution laws between the motor winding and the ground, between turns and between phases, clarifying the existing insulation weak points and intuitively locating them have important application values for the safe operation and optimized design of the motor. Summary of the Invention

[0008] The present invention aims to provide a method for predicting the voltage distribution and insulation weak points based on the flat wire winding of a variable-frequency motor to solve the technical problems in the background technology. Through this method, the voltage distribution of the motor winding can be predicted efficiently and relatively accurately. In addition, using the insulation weak point positioning radar chart of the present invention, the physical position of the insulation weak points of the motor can be intuitively displayed.

[0009] To solve the technical problems, the technical solution of the present invention is as follows:

[0010] A method for predicting the voltage distribution and insulation weak points based on the flat wire winding of a variable-frequency motor, the method comprising:

[0011] S1: Based on a preset ladder circuit parameter calculation program, determine the number of branches and the corresponding frequencies of the ladder circuit model according to the number of cable layers and length, obtain the resistance and inductance data of the cable corresponding to each branch or frequency, and calculate the cable-to-ground capacitance data according to the distance to the ground, that is, obtain the cable n-branch ladder circuit model;

[0012] S2: Perform size and shape recognition, classification and marking processing on the electrostatic field calculation model, and generate a material input interface with color markings. When the material information is input, the inter-turn capacitance and cable-to-ground capacitance values of the slot part are automatically calculated;

[0013] S3: Identify, classify, and label the size and shape of the eddy current field calculation model, and generate a parameter input interface. After inputting the parameter information, automatically calculate the end AC resistance and inductance values; the parameters include: wire, paint film material, and frequency;

[0014] S4: Number the slots and layers of the transient magnetic field model, and determine whether they are consistent with the number of slots and layers in the winding layout diagram. If not, generate an error prompt, that is, the stator model or the winding layout diagram is incorrect, please re-enter; if consistent, name each conductor in the transient magnetic field model according to the arrangement order of the winding layout diagram, and assign coils to the conductors, and couple the coils in the external circuit with the conductors in the magnetic field region; after the excitation is set, set the boundary, divide the mesh, and add the solution analysis to the transient magnetic field model to complete the construction of the transient magnetic field model of the flat wire winding motor;

[0015] S5: Perform secondary identification processing on the winding layout diagram, connect the inter-turn capacitance and the voltmeter on the basis of the high-frequency equivalent circuit of the winding to obtain a complete high-frequency equivalent circuit of the winding, integrate the cable n-branch ladder circuit obtained in step S1, and supply power through a pulse source to complete the construction of the high-frequency equivalent circuit model;

[0016] S6: Obtain the pulse voltage working parameters and the data calculated in steps S1, S2, and S3, and generate a parameter table;

[0017] S7: Import the parameter table into the constructed high-frequency equivalent circuit, and perform field-circuit coupling finite element analysis in combination with the transient magnetic field model of the flat wire winding motor to calculate the time-domain voltage distribution of the motor;

[0018] S8: Classify the calculated time-domain voltage distribution results according to different insulation structures and extract the voltage amplitudes to generate a voltage amplitude table for positioning the winding by slot number and layer number;

[0019] S9: Measure the PDIV and PDEV values of the motor to be measured, and import the obtained voltage amplitude table and the measured PDIV and PDEV values into the insulation weak point positioning program, and a radar chart that can visually display the insulation weak points of the motor can be generated, that is, the voltage distribution and insulation strength of each measurement point are visualized.

[0020] Further, before the step S1, the method further includes:

[0021] Obtain the cable model, length, number of layers, distance to the ground, the motor stator model, and the winding layout diagram, and calculate the preliminary cable n-branch ladder circuit model, electrostatic field calculation model, eddy current field calculation model, transient magnetic field model, and high-frequency equivalent circuit for subsequent parameter calculation and model optimization.

[0022] Further, the step S1 specifically includes:

[0023] Before the head end of the high-frequency equivalent circuit and after the power supply, a cable model is added, including resistance, inductance, and capacitance-to-ground parameters; the cable is represented as a combination of concentric tubular sub-conductors connected in parallel infinitely and simplified to a finite number of n-branch models; for each branch, the resistance-inductance (R-L) elements do not change with frequency and are connected step by step to generate an equivalent impedance, representing the change of the resistance and inductance of the cable with frequency; the modeling effect is good using the data set given at proportionally spaced frequencies, ensuring fewer branches and accurately representing the frequency dependence of R and L;

[0024] By using the CableConstants program of the EMTP platform or the finite element analysis method, the cable R-L parameters at different frequencies are obtained. The steps to determine the n-branch circuit parameters from the set of known cable R-L parameters at different frequencies are as follows:

[0025] Considering the special case of the two-branch model, the model parameters are determined by formulas (1)-(6):

[0026] ΔR = R C2 -R C1 (1)

[0027] ΔL = L C1 -L C2 (2)

[0028]

[0029] where R1, L1, R2, and L2 are the resistances and inductances of the first and second branches of the model; R C1 , L C1 , R C2 , L C2 are the input data, which are the resistances and inductances of the cable at known frequencies ω1 and ω2; through (1)-(6), if the R-L values of the cable corresponding to two frequencies are known, the parameters of the two-branch model can be analytically determined;

[0030] The calculation of the R-L circuit parameters with three or more branches is realized by using an additional program for sequential simplification of the sequential circuit:

[0031] Obtain two pairs of data related to the two highest frequencies ω n and ω n-1 from the data set of n pairs of resistances and inductances: (R Cn , L Cn ) and (R Cn-1 , L Cn-1 ), and then calculate the n-branch model corresponding to the external branch (R n , Ln ) parameters;

[0032] The n - branch model equivalent circuit consists of the external nth branch and an equivalent impedance (R′ eq(ωi) and L′ eq(ωi) ), where the latter represents the combination of the remaining n - 1 branches at all frequencies, and the total impedance of the circuit matches the known cable data at a given frequency; thus, the equivalent parameters of the internal branch combination are determined by formulas (7)(8):

[0033]

[0034] After determining the R - L for n - 1 pairs of frequencies ω i (i = 1, 2, … n - 1), the internal equivalent conductor formed by the n - 1 branches is fully characterized; following the calculation steps of the nth branch, the parameters of the n - 1 groups of equivalent conductors are used to determine the (n - 1)th branch, and so on, repeating the above steps until only two datasets corresponding to frequencies remain, at which time the inductance and resistance parameters (R1, L1) of the first branch are calculated using (5) and (6).

[0035] Furthermore, the step S2 specifically includes:

[0036] Model and mesh the single slot of the motor in the finite - element platform, perform mesh refinement on the thin insulation structure to ensure calculation accuracy, set the model boundary, and add electrostatic - field solution analysis;

[0037] In the slot region, the stator core is considered as the grounded conductor "0", and the capacitance to ground between each turn and the stator core is expressed as C i.0 = C slot i.0 (i = 1...n T ), n T is the total number of turns of the conductor in one slot, the capacitance between turns is expressed as C i,j = C slot i,j , (i≠j, j = 1...n T ); in the end region, the capacitance between turns C ov i,j is considered; after simplification, the capacitance to ground C g i.0 and the capacitance between turns C T i,j are calculated as follows:

[0038]

[0039] Use the electrical - energy method to calculate the capacitance of a conductor system located in a linear dielectric. For n conductors and a reference ground 0, the electrical energy is written as formula (11):

[0040]

[0041] where C i.0 and C i,j are the capacitances of the system, u i and u j are the electric potentials of the i-th and j-th conductors respectively; to determine the capacitance of a multi-conductor system with n conductors and a reference ground 0, independent combinations of the electric potentials u i and u j of 0 V and 1 V are needed to transform the energy expression (11) into a system of equations:

[0042]

[0043] where, W el represents the energy vector, C represents the capacitance vector, and k u represents the connection matrix between the two vectors W el and C, and the capacitance can be extracted from the system of equations:

[0044]

[0045] Furthermore, the specific steps of step S3 include:

[0046] The end part presents an irregular state due to the special twist angle of the flat wire winding and the welding process. 3D modeling of the end part is selected, and 3 adjacent conductors are taken to simplify the modeling. The calculation of the AC resistance and inductance of the end part is realized through the eddy current field, and the finite element method is used for analysis and calculation;

[0047] Inductance is generated by magnetic field energy:

[0048]

[0049] where, i j and i k (j, k = 1, …, n T ) are the current sources of conductors j and k, which are combinations of normalized sinusoidal current sources of 0 A and 1 A, and are introduced as initial conditions into the finite element partial differential equation;

[0050] Resistance is generated by dissipated power:

[0051]

[0052] At high frequencies (10 kHz…10 MHz), the finite element method divides the model region into a large number of three-dimensional meshes and uses the magnetic field energy W m and the copper loss P elThe calculation method continuously solves the inductance and resistance of each turn through different current combinations of 0A and 1A; finally, the inductance and AC resistance of the end part are regarded as constant values equal to each turn, and the average value of 3 conductors is taken for the calculation result.

[0053] Further, in step S4, the transient magnetic field model includes: three-phase windings and a complete stator core, and the B-H curve of the actual material is imported into the simulation software to fully consider the influence of the core saturation nonlinearity on the winding inductance parameters; based on finite element analysis and mesh encryption processing, the skin effect and proximity effect of the conductor at high frequencies are fully considered; the inductance and resistance parameters of the conductor in the transient magnetic field model will be calculated automatically in the subsequent field-circuit coupled finite element simulation.

[0054] Further, step S5 specifically includes:

[0055] Generate the primary circuit: taking each half-turn winding as a circuit unit, determine the total number of circuit units, where the total number of circuit units = number of slots × number of layers; connect each circuit unit in ascending order according to the winding number sequence. For example, for a motor with 54 slots and including three phases of UVW, the connection sequence is: U1i>U 1o>U 2i>U 2o……>U 54i>U 54o, V1i>V 1o>V 2i>V 2o……>V 54i>V 54o, W1i>W 1o>W 2i>W2o……>W 54i>W 54o;

[0056] Determine the capacitance parameters: identify the winding layout diagram, determine the position of the winding in the stator slot according to the winding layout diagram, and then determine the capacitance parameters corresponding to each winding unit, and assign the capacitance parameters to the corresponding circuit unit;

[0057] Connect the remaining circuits: complete the connection of the voltmeter on the basis of the primary circuit, integrate the cable n-branch ladder circuit obtained in step S1, and supply power through a pulse source to complete the construction of the high-frequency equivalent circuit model.

[0058] Further, when performing simulation calculations on the voltage distribution of the flat wire winding of a variable-frequency motor, step S6 specifically includes:

[0059] Use Ansys Maxwell software to perform simulation calculations on the voltage distribution. The basis of finite element analysis is Maxwell's equations, which consist of the following four laws: Ampere's circuital law, Faraday's law of electromagnetic induction, Gauss's law of electric flux, and Maxwell's equations, and derive the differential equations for solving electromagnetic problems using finite elements, corresponding to equations (16) to (19) respectively:

[0060]

[0061] The above equations include two curl equations (16), (17) and two divergence equations (18), (19); the relationships between the field quantities E, D, B, and H are determined by the characteristics of the medium; for a linear medium, the relationships are as follows:

[0062] D = εE (20)

[0063] B = μH (21)

[0064] J = σE (22)

[0065] where ε is the permittivity of the medium, F / m; μ is the permeability of the medium, H / m; σ is the conductivity of the medium, S / m; for an isotropic medium, ε, μ, and σ are scalars; for an anisotropic medium, they are tensors;

[0066] Regarding the problem that the finite element solver decouples the Maxwell equations, the circuit model and the finite element model are coupled by the field-circuit coupling method to solve the magnetic field and the electric field simultaneously.

[0067] Further, the step S8 includes:

[0068] Obtain the calculation result of the time-domain voltage distribution, export it in csv format to get the ungrouped original data, use the MATLAB program to read the groups in a loop, classify them according to different insulation structures, and for each classification, create a new data structure to store the slot numbers, layer numbers of each winding and the corresponding voltage amplitudes. Considering the margin of insulation evaluation, the voltage amplitude takes the maximum value within one period; traverse the original data set, integrate the obtained voltage amplitudes according to the actual positions of the windings in the stator slots, and generate a voltage amplitude table that can locate the specific windings through the slot numbers and layer numbers.

[0069] Further, the step S9 includes:

[0070] The PDIV / PDEV test method is measured using a preset PDIV / PDEV test platform by selecting appropriate pulse voltage parameters and wiring methods according to the IEC standard.

[0071] The radar chart generation program for insulation weak points converts the voltage distribution results from tabular form into a radar chart that is more similar to the actual structure of the motor, and integrates the voltage amplitudes corresponding to their positions in the motor onto one chart. Each axis of the radar chart represents a slot number, and the PDIV, PDEV values, and voltage distribution data of different layer numbers are distinguished by different colors or line styles. The voltage amplitude is represented by a radial line from the center outwards, and the farther away from the center, the higher the voltage amplitude. The extreme values of each axis represent the maximum and minimum voltages in that slot. Finally, by comparing the magnitudes of the voltage distributions with the PDIV test values, it can be analyzed whether discharge occurs under general conditions. Among them, the points where the voltage distribution is near or outside PDEV and PDIV can be regarded as the insulation weak points of the motor.

[0072] Compared with the prior art, the advantages of the present invention are as follows:

[0073] 1. The present invention is reasonably designed, has high integration, low cost, and convenient operation;

[0074] 2. High evaluation efficiency: Compared with traditional test and measurement methods, the present invention can automatically complete voltage distribution prediction and insulation evaluation, greatly shortening the evaluation cycle;

[0075] 3. High accuracy of the evaluation result of the insulation performance of the variable-frequency motor: Through the motor simulation modeling module, finite element software analysis module, external circuit generation module, etc., the present invention can more accurately predict the voltage distribution of the stator winding of the variable-frequency motor under different design parameters and different working conditions, thereby providing more reliable data support for motor design and insulation evaluation;

[0076] 4. Optimize motor design: Based on the prediction results, designers can improve the structure, materials, etc. of the motor winding, improve the overall insulation performance and durability of the motor, extend the service life of the motor, and reduce the risk of accidents caused by insulation failure.

[0077] 5. Intuitively locate insulation weak points: By visualizing the winding voltage distribution as a radar chart according to its corresponding position in the motor and comparing it with the corresponding PDIV and PDEV, the present invention can more intuitively identify potential insulation weak points in the motor winding. Description of the Drawings

[0078] Figure 1 System diagram for voltage distribution prediction applicable to the flat wire winding of a variable-frequency drive motor;

[0079] Figure 2 、 Four PDIV test platform of Sichuan University;

[0080] Figure 3 Effect example of the radar chart for locating insulation weak points applicable to the flat wire winding of a variable-frequency drive motor;

[0081] Figure 4 , the voltage waveforms borne by each insulation part of the motor;

[0082] Figure 5 , the minimum unit of the high-frequency equivalent circuit model. Specific embodiments

[0083] The specific embodiments of the present invention will be described below in conjunction with the embodiments:

[0084] It should be noted that the structures, ratios, sizes, etc. shown in this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the conditions under which the present invention can be implemented. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.

[0085] At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" cited in this specification are only for the convenience of clear narration, and are not used to limit the scope under which the present invention can be implemented. The change or adjustment of their relative relationship, without substantial change in the technical content, should also be regarded as the scope under which the present invention can be implemented.

[0086] Embodiment 1:

[0087] The embodiment further explains the content of the present invention, specifically as Figure 1 shown. Based on the voltage distribution and insulation weak point prediction method of the flat wire winding of the variable frequency motor, the method includes:

[0088] S1: Obtain the cable model, length, number of layers and distance to the ground, the motor stator model and the winding layout diagram. Based on ① the transmission line theory, ② the three-dimensional geometric processing software, ③ the three-dimensional reconstruction algorithm based on deep learning (using the machine to automatically extract and calculate the internal features of the three-dimensional model), ④ the table structure recognition algorithm GFTE (a novel Graph-Neural-Network-based algorithm) based on the graph neural network (GCN), and ⑤ the high-frequency equivalent circuit generation program, obtain the preliminary cable n-branch ladder circuit model, electrostatic field calculation model, eddy current field calculation model, transient magnetic field model and high-frequency equivalent circuit for subsequent parameter calculation and model optimization;

[0089] S2: Based on the preset ladder circuit parameter calculation program, determine the number of branches and the corresponding frequencies of the ladder circuit model according to the number of layers and the length of the cable, obtain the resistance and inductance data of the cable corresponding to each branch or frequency, and calculate the cable-to-ground capacitance data according to the distance to the ground, that is, obtain the cable n-branch ladder circuit model;

[0090] S3: A geometric feature extraction algorithm based on the contour shape (implemented by calculating and comparing the contour features of the 3D model to obtain the geometric similarity of the 3D model). Based on the straight part of the slot located in the stator core, perform size and shape recognition, classification, and marking processing on the electrostatic field sub-model, and generate a material input interface with color markings. After inputting the material information, calculate the inter-turn capacitance and ground capacitance values of the slot based on the finite element theory;

[0091] S4: A geometric feature extraction algorithm based on the contour shape. Based on the bent part of the end that extends beyond the stator core, perform size and shape recognition, classification, and marking processing on the eddy current field calculation model, and generate a parameter input interface. After inputting the parameter information, automatically calculate the AC resistance and inductance values of the end; The parameters include: wire, paint film material, and frequency;

[0092] S5: Number the slots and layers of the transient magnetic field model, and determine whether the number of slots and layers is consistent with that of the winding layout diagram. If not, generate an error message "The stator model / winding layout diagram is incorrect, please re-enter"; If consistent, name each conductor in the transient magnetic field model according to the arrangement order of the winding layout diagram, assign coils to the conductors, and couple the coils in the external circuit with the conductors in the magnetic field region; After the excitation settings are completed, set the boundaries, divide the mesh, and add a solution analysis to the transient magnetic field model to complete the construction of the transient magnetic field model of the flat wire winding motor;

[0093] S6: Perform secondary recognition processing on the winding layout diagram, connect the inter-turn capacitance and voltmeter on the basis of the initially generated high-frequency equivalent circuit of the winding to obtain a complete high-frequency equivalent circuit of the winding, integrate the cable n-branch ladder circuit obtained in S2, and supply power through a pulse source to complete the construction of the circuit part model;

[0094] S7: Obtain the pulse voltage operating parameters and the data obtained in steps S2, S3, and S4, and generate a parameter table;

[0095] S8: Import the parameter table into the established high-frequency equivalent circuit, and perform field-circuit coupled finite element analysis in combination with the transient magnetic field model of the flat wire winding motor to calculate the time-domain voltage distribution of the motor;

[0096] S9: Classify the calculated time-domain voltage distribution results according to different insulation structures (main insulation, phase insulation, inter-turn insulation) and extract the voltage amplitudes to generate a voltage amplitude table that locates the winding by slot number and layer number;

[0097] S10: Measure the PDIV and PDEV values of the motor to be tested, import the obtained voltage amplitude table and the measured PDIV and PDEV values into the insulation weak point location program, and a radar chart that can visually display the insulation weak points of the motor can be generated, that is, the voltage distribution and insulation strength of each measurement point are visualized.

[0098] In one implementation, the step S2 specifically includes:

[0099] In practice, there is a section of cable between the frequency converter and the motor. To simulate the actual working conditions, a cable model including resistance, inductance, and capacitance to ground parameters should be set before the power supply after the head of the high-frequency equivalent circuit. To include the frequency dependence of the parameters, the cable can be represented as a combination of concentric tubular sub-conductors connected in parallel infinitely, and each sub-conductor has specific inductance and resistance values. Theoretically, it can be represented as an equivalent circuit with an infinite number of R-L branches. Since an infinite number of branches is impractical in reality, it is simplified to a finite number of "n-branch" models. For each branch, the R-L elements do not change with frequency and are connected step by step in a ladder manner, thus generating an equivalent impedance representing the change of the resistance and inductance of the cable with frequency. It should be noted that the number of cable branches is not always the more the better. The best configuration scheme is to have the fewest branches and accurately represent the frequency dependence of R and L throughout the cycle. If too many branches are used, some branch parameters may be negative, which is contrary to the fact that the equivalent resistance and inductance of the model are always positive. Research shows that a dataset given at proportionally spaced frequencies (ω n / ω n-1 =ω n-1 / ω n-2 =…=ω2 / ω1) has a good modeling effect, and this scheme is also adopted in the present invention.

[0100] By using the "Cable Constants" program of the EMTP platform or the finite element analysis method, the cable R-L parameters at different frequencies are obtained. The steps to determine the n-branch circuit parameters from the set of known cable R-L parameters at different frequencies are as follows:

[0101] Considering the special case of the double-branch model, the model parameters can be determined by formulas (1)-(6):

[0102] ΔR=R C2 -R C1 (1)

[0103] ΔL=L C1 -L C2 (2)

[0104]

[0105] wherein, R1, L1, R2, and L2 are the resistances and inductances of the first and second branches of the model, respectively; R C1 , L C1 , R C2 , L C2 is the input data, which is the resistance and inductance of the cable at known frequencies ω1 and ω2. Through (1)-(6), if the R-L values of the cable corresponding to two frequencies are known, the parameters of the double-branch model can be analytically determined;

[0106] If more sets of cable parameters are available, a more accurate representation can be obtained. However, the R-L circuit parameters with three or more branches cannot be simply obtained using the analytical method in (1)-(6), but rather require an additional procedure of sequential circuit simplification:

[0107] 1. Obtain two pairs of data related to the two highest frequencies ω n and ω n-1 from the dataset of n pairs of resistance and inductance data: (R Cn , L Cn ) and (R Cn-1 , L Cn-1 ). Then, calculate the parameters of the n-branch model corresponding to the external branch (R n , L n ) through (1)-(4);

[0108] 2. The equivalent circuit of the n-branch model consists of the external nth branch and an equivalent impedance (R′ eq(ωi) and L′ eq(ωi) ), where the latter represents the combination of the remaining n - 1 branches at all frequencies, and the total impedance of the circuit matches the known cable data at the given frequency. Therefore, determine the equivalent parameters of the internal branch combination through formulas (7)(8):

[0109]

[0110] 3. After determining the R-L at n - 1 pairs of frequencies ω i (i = 1, 2, … n - 1), the internal equivalent conductor formed by n - 1 branches is fully characterized. According to the calculation steps of the nth branch, use the parameters of n - 1 sets of equivalent conductors to determine the (n - 1)th branch, and so on. Repeat the above steps until only the dataset corresponding to two frequencies remains. At this time, use (5) and (6) to calculate the inductance and resistance parameters (R1, L1) of the first branch.

[0111] In one embodiment, the step S3 specifically includes:

[0112] Model a slot of the motor and perform mesh division on the finite element platform. Mesh encryption is performed on the insulation structure with a small thickness to ensure calculation accuracy. Set the model boundaries and add electrostatic field solution analysis.

[0113] Principle of distributed capacitance calculation: In the slot region, the stator core is considered as a grounded conductor "0", and the capacitance to the ground between each turn and the stator core is expressed as C i.0 = C slot i.0 (i = 1...n T ),n T is the total number of turns of conductors in a slot, and the capacitance between turns is expressed as C i,j = C slot i,j ,(i≠j,j = 1...n T ); In the end region, the capacitance between turns C ov i,j is considered; After simplification, the capacitance to the ground C g i.0 and the capacitance between turns C T i,j are calculated as follows:

[0114]

[0115] Use the electrical energy method to calculate the capacitance of a conductor system located in a linear dielectric. For n conductors and a reference ground 0, the electrical energy is written as formula (11):

[0116]

[0117] where C i.0 and C i,j are the capacitances of the system, u i and u j are the electric potentials of the i-th and j-th conductors respectively; To determine the capacitance of a multi-conductor system with n conductors and a reference ground 0, independent combinations of 0V and 1V conductor electric potentials u i and u j are required to convert the energy expression (11) into a system of equations:

[0118]

[0119] where, W el represents the energy vector, C represents the capacitance vector, k u represents the connection matrix between the two vectors W el and C, and the capacitance can be extracted from the system of equations:

[0120]

[0121] In one embodiment, step S4 specifically includes:

[0122] Due to the special twist angle of the flat wire winding and the welding process, the end part presents an irregular state. 3D modeling is selected for the end part. To simplify the modeling, 3 adjacent conductors are taken, and the calculation of the AC resistance and inductance of the end part is realized through the eddy current field. The actual inductance and resistance are not constant but depend on the frequency. Therefore, the finite element method is used for analysis and calculation.

[0123] Inductance is generated by magnetic field energy:

[0124]

[0125] where i j and i k , j = 1, …, n T and k = 1, …, n T , are the current sources of conductors j and k, which are a combination of normalized sine current sources of 0 A and 1 A, and are introduced as the initial conditions into the finite element partial differential equation;

[0126] Resistance is generated by dissipated power:

[0127]

[0128] At high frequencies (10 kHz … 10 MHz), the finite element method divides the model region into a large number of three-dimensional meshes, and uses the calculation methods of magnetic field energy W m and copper loss P el to continuously solve the inductance and resistance of each turn through different current combinations of 0 A and 1 A; finally, the inductance and AC resistance of the end part are regarded as constant values equal to each turn, and the calculation results take the average value of 3 conductors.

[0129] In one embodiment, the winding layout diagram specifically includes:

[0130] The winding layout diagram provides the position information, connection scheme, and transposition method of each turn of the motor winding, and distinguishes different phases by different colors. Number interpretation: For example, for a motor with 54 slots and including three phases of U, V, and W, the element "blue 30i" in the 4th column and 6th row of its winding layout diagram represents that the 30th winding unit of phase U is located in the 6th layer of slot 4 of the motor, and the current direction is positive; the element "red 35o" in the 4th column and 5th row represents that the 35th winding unit of phase W is located in the 5th layer of slot 4 of the motor, and the current direction is negative;

[0131] Functions of the winding layout diagram: Provide the numbering basis for the transient magnetic field model described in S5; provide the wiring method of the winding part of the high-frequency equivalent circuit.

[0132] In one embodiment, step S5 specifically includes:

[0133] The transient magnetic field model: includes all three-phase windings and the complete stator core, and imports the B-H curve of the actual material into the simulation software, fully considering the influence of core saturation nonlinearity on the winding inductance parameters; based on finite element analysis and mesh encryption processing, fully considering the skin effect and proximity effect of conductors at high frequencies; the inductance and resistance parameters of the conductors in the transient magnetic field model will be calculated automatically in the subsequent field-circuit coupled finite element simulation and do not need to be added to the parameter table.

[0134] In one embodiment, the high-frequency equivalent circuit specifically includes:

[0135] ① Construct a circuit unit model based on the transmission line theory: The transmission line model is usually used to simulate the wave propagation effect in the inverter-fed motor winding and the long motor supply cable, represented by a distributed parameter equivalent circuit, and is modeled as a series of equivalent Γ-type or π-type sections. Taking each half-turn winding as a circuit unit, the equivalent circuit model of a single-turn winding is as Figure 5 shown, where u i (i = 1, 2, …) is the node voltage of the i-th unit, and i i is the current flowing through the i-th unit; the distributed parameters are T i including the self-inductance L i and the AC resistance R i , M 1,2 is the mutual inductance between the straight parts; the lumped parameters include the distributed capacitance, the AC resistance and inductance of the end part, C i0 is the capacitance to ground of each unit, grounded through the stator core; C i,m is the inter-turn capacitance between adjacent windings; R ei and L ei are the AC resistance and inductance of the end part respectively. From the above analysis, it can be seen that the distributed parameters required to establish the high-frequency equivalent circuit include: inter-turn capacitance and capacitance to ground, AC resistance and inductance parameters of the end part, and AC resistance and inductance parameters of the straight part. Assuming that the motor has n flat wire windings, that is, it contains 2n units, according to Kirchhoff's current law:

[0136]

[0137] According to Kirchhoff's voltage law:

[0138]

[0139] From formula (18), the differential equation RI + Z(dI / dt) = U can be obtained, where,

[0140] R = [R1 + R e1 , …, R m+R em ,…,R 2n +R e2n (19)

[0141] I = [i1,…,i m ,…,i 2n (20)

[0142] Z is the winding inductance matrix, with self-inductance (L i +L ei ) on the main diagonal and mutual inductance M i,m At the (i, m) position, i and m are determined according to the winding connection structure.

[0143] ② Calculate the total number of circuit units and generate a primary circuit containing only the winding module: The total number of circuit units = the number of slots × the number of layers. Connect each circuit unit in ascending order of the winding number to obtain three independent phase winding modules of U, V, and W. Since in each SVPWM non-zero vector state, the three-phase windings are always connected in series with two phases and then in parallel with the other phase, this connection method is also adopted to connect the three phase winding modules of U, V, and W, denoted as U-V / / W;

[0144] ③ Supplement the primary circuit: Determine the position of the winding in the stator slot according to the winding layout diagram, and then determine the capacitance parameters corresponding to each winding unit. Traverse all the windings adjacent to the same slot to obtain the inter-turn capacitance connection method, connect all the inter-turn capacitances to the existing primary circuit, and then add the voltmeter turn by turn to the circuit for measuring the voltage distribution. Finally, the complete high-frequency equivalent circuit consists of four parts: winding, voltmeter, cable, and power supply.

[0145] In one embodiment, the step S8 specifically includes:

[0146] The software used in the voltage distribution simulation calculation is Ansys Maxwell. This software platform is a comprehensive electromagnetic simulation software, suitable for the design and analysis of 3D / 2D structures of electrical and electromechanical devices such as motors, drivers, transformers, and others. Based on the finite element method (FEM), Maxwell can solve static, frequency-domain, and time-varying electromagnetic and electric fields. The basis of the field-circuit coupled finite element calculation is Maxwell's equations, which actually consist of 4 laws, namely: Ampere's circuital law, Faraday's law of electromagnetic induction, Gauss's law of electric flux, and Gauss's law of magnetic flux (also known as the law of magnetic flux continuity). Regardless of the medium and the distribution of the magnetic field strength H, the line integral of the magnetic field strength along any closed path in the magnetic field is equal to the total current passing through the surface Ω determined by this integral path, or rather, this line integral is equal to the total current enclosed by the integral path. The current here includes conduction current (generated by free charges) and displacement current (generated by the change of the electric field), and its mathematical expression is as shown in equation (21), and this expression is called Ampere's circuital law.

[0147]

[0148] Faraday's law of electromagnetic induction states that the induced electromotive force in a closed loop is proportional to the rate of change of the magnetic flux through this loop, and is expressed in integral form as (22):

[0149]

[0150] Regardless of the distribution of the electrolyte and the electric flux density vector in the electric field, the electric flux passing through any closed surface is equal to the electric charge enclosed by this closed surface. Here, it is pointed out that the electric flux is also the integral of the electric flux density vector over this closed surface, and this is called Gauss's law of electric flux. The integral form of this law can be expressed as follows:

[0151]

[0152] Regardless of the distribution of the medium and the magnetic flux density vector in the magnetic field, the magnetic flux passing through any closed surface is always equal to zero. Here, it is pointed out that the magnetic flux is the directed integral of the magnetic flux vector over this closed surface. The integral form of Gauss's law of magnetic flux is:

[0153]

[0154] Equations (21) to (24) constitute Maxwell's equations describing the electromagnetic field, which are expressed in integral form. In addition, they also have their own differential forms, and through them, differential equations that can be processed by the finite element for electromagnetic problems can be derived, corresponding to equations (25) to (28) respectively:

[0155]

[0156] The above equations include two curl equations (25), (26) and two divergence equations (27), (28). The relationships between the field quantities E, D, B, and H are determined by the properties of the medium. For a linear medium, the relationships are as follows:

[0157] D = εE (29)

[0158] B = μH (30)

[0159] J = σE (31)

[0160] where ε is the permittivity of the medium, F / m; μ is the permeability of the medium, H / m; σ is the conductivity of the medium, S / m; for an isotropic medium, ε, μ, and σ are scalars; for an anisotropic medium, they are tensors.

[0161] Regarding the problem that the finite element solver decouples the Maxwell equations, the circuit model and the finite element model can be coupled by means of field-circuit coupling to solve the magnetic field and the electric field simultaneously.

[0162] In one implementation, step S9 specifically includes:

[0163] Obtain the calculation result of the time-domain voltage distribution, export it in csv format to get the ungrouped original data, use the MATLAB program to read the groups in a loop, classify them according to different insulation structures (main insulation, phase-to-phase insulation, turn-to-turn insulation), and for each classification, create a new data structure (such as DataFrame) to store the slot number, layer number of each winding and the corresponding voltage amplitude. Considering the margin of insulation evaluation, the voltage amplitude takes the maximum value within one period. Traverse the original data set, integrate the obtained voltage amplitudes according to the actual position of the winding in the stator slot, and generate a voltage amplitude table that can locate the specific winding through the slot number and layer number.

[0164] In one implementation, step S10 includes:

[0165] The PDIV / PDEV test method: According to the IEC standard, select appropriate pulse voltage parameters and wiring methods, and use Figure 2 The PDIV / PDEV test platform of Sichuan University shown in the figure is measured; the radar chart generation program for insulation weak points converts the voltage distribution result from tabular form to a radar chart that is more similar to the actual structure of the motor, and integrates the voltage amplitudes corresponding to their positions in the motor onto one chart, which is more intuitive. As Figure 3As shown, each axis of the radar chart represents a slot number. The PDIV, PDEV values, and voltage distribution data of different layer numbers are distinguished by different colors or line styles. The voltage amplitude is represented by radial lines from the center outwards. The farther away from the center, the higher the voltage amplitude. The extreme values of each axis represent the maximum and minimum voltages in that slot. Finally, by comparing the magnitudes of the voltage distributions with the PDIV test values, it is possible to analyze whether discharge occurs under normal conditions. Among them, points where the voltage distribution is near or outside PDEV and PDIV can be regarded as weak insulation points of the motor.

[0166] Embodiment 2:

[0167] This Embodiment 2 is applied to Embodiment 1 and provides a detailed description during the user's usage. It is a specific implementation method applicable to the voltage distribution prediction system of the flat wire winding of a variable-frequency drive motor. First, the user imports the cable model, length, number of layers, distance to the ground, the motor stator model, and the winding layout diagram into the input module. The system initially generates an "n"-branch ladder circuit model of the cable based on the cable model, extracts the electrostatic field calculation model and eddy current field calculation model for calculating the winding distribution parameters from the motor stator model, establishes a transient magnetic field model of the flat wire winding motor in combination with the winding layout diagram, and establishes a high-frequency equivalent circuit model of the three-phase winding based on the winding layout diagram. After generating the required models, the system prompts the user to input information such as materials and frequencies required for calculating the distribution parameters. After the user finishes inputting, the calculation starts automatically, and the calculated cable and winding distribution parameter values, as well as the input working voltage parameters, are summarized in a parameter table. Finally, the parameter table is imported into the established high-frequency equivalent circuit, and a field-circuit coupling finite element analysis is performed in combination with the motor transient magnetic field model to calculate the motor voltage distribution.

[0168] Such as Figure 3 , a specific implementation method of a radar chart for locating weak insulation points of the flat wire winding of a variable-frequency drive motor. According to the IEC standard, appropriate pulse voltage parameters and wiring methods are selected, and the PDIV and PDEV values of the motor under test are measured using the Figure 2 Sichuan University PDIV test platform shown. Finally, the calculated voltage distribution results and the measured PDIV and PDEV values are imported into the weak insulation point location program, and a radar chart that can visually display the weak insulation points of the motor can be generated, visualizing the voltage distribution and insulation strength of each measurement point.

[0169] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0170] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one or more of the processes and / or blocks Figure 1 or multiple blocks.

[0171] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one or more of the processes and / or blocks Figure 1 or multiple blocks.

[0172] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one or more of the processes and / or blocks Figure 1 or multiple blocks. The preferred embodiments of the present invention have been described in detail above, but the present invention is not limited to the above embodiments. Within the knowledge of those of ordinary skill in the art, various changes can be made without departing from the gist of the present invention.

[0173] Many other changes and modifications can be made without departing from the concept and scope of the present invention. It should be understood that the present invention is not limited to specific embodiments, and the scope of the present invention is defined by the appended claims.

Claims

1. A method for predicting voltage distribution and insulation weak points of flat-wire windings of variable-frequency motors, characterized in that: The method comprises: S1: Based on the preset ladder circuit parameter calculation program, the number of branches and the corresponding frequency of the ladder circuit model are determined according to the number of cable layers and length, the resistance and inductance data of the cable corresponding to each branch or frequency are obtained, and the cable-to-ground capacitance data is calculated according to the distance to the ground, that is, the n-branch ladder circuit model of the cable is obtained; S2: Perform size and shape recognition, classification and identification processing on the electrostatic field calculation model, and generate a material input interface with color identification. After the material information is input, the inter-turn capacitance and ground capacitance of the slot are automatically calculated; S3: Perform size and shape recognition, classification and identification processing on the eddy current field calculation model, and generate a parameter input interface. After the parameter information is input, the end AC resistance and inductance values ​​are automatically calculated; the parameters include: conductor, paint film material and frequency; S4: Number the slots and layers of the transient magnetic field model to determine whether they are consistent with the number of slots and layers in the winding arrangement diagram. If they are inconsistent, an error message is generated, that is, the stator model or the winding arrangement diagram is incorrect, please re-enter; if they are consistent, name the conductors in the transient magnetic field model one by one according to the arrangement order of the winding arrangement diagram, and assign coils to the conductors to couple the coils in the external circuit with the conductors in the magnetic field area; after the excitation setting is completed, set the boundary of the transient magnetic field model, divide the grid and add solution analysis to complete the construction of the transient magnetic field model of the flat wire winding motor; S5: Perform secondary identification processing on the winding arrangement diagram, complete the connection of the inter-turn capacitor and the voltmeter on the basis of the high-frequency equivalent circuit of the winding, obtain a complete high-frequency equivalent circuit of the winding, integrate the cable n-branch ladder circuit obtained in step S1, and power it through a pulse source to complete the construction of the high-frequency equivalent circuit model; S6: Obtain pulse voltage working parameters and data calculated in steps S1, S2 and S3, and generate a parameter table; S7: importing the parameter table into the constructed high-frequency equivalent circuit, combining the transient magnetic field model of the flat wire winding motor to perform field-circuit coupling finite element analysis, and calculating the motor time domain voltage distribution; S8: classifying the calculated time-domain voltage distribution results according to different insulation structures and extracting voltage amplitudes, and generating a voltage amplitude table for locating windings by slot number and layer number; S9: Determine the PDIV and PDEV values ​​of the motor to be tested, and import the obtained voltage amplitude table and the measured PDIV and PDEV values ​​into the insulation weak point positioning program to generate a radar map that can intuitively display the insulation weak points of the motor, that is, realize the visualization of the voltage distribution and insulation strength of each measurement point.

2. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1 is characterized in that: Before step S1, the method further includes: The cable model, length, number of layers and distance to the ground, motor stator model and winding layout are obtained, and a preliminary cable n-branch ladder circuit model, electrostatic field calculation model, eddy current field calculation model, transient magnetic field model and high-frequency equivalent circuit are calculated for subsequent parameter calculation and model optimization.

3. The method for predicting voltage distribution and insulation weak points based on flat wire windings of variable frequency motors according to claim 1, characterized in that: The step S1 specifically includes: Before the head end of the high-frequency equivalent circuit and after the power supply, a cable model is added, including resistance, inductance and capacitance to ground parameters; the cable is represented as a combination of infinitely parallel connected concentric tubular sub-conductors, simplified to a finite number of n-branch models; for each branch, the resistance-inductance RL element does not change with frequency, and is connected step by step to produce an equivalent impedance, representing that the resistance and inductance of the cable change with frequency; the data set given by proportionally spaced frequencies has a good modeling effect, ensuring that the number of branches is small and the frequency dependence of R and L can be accurately represented; The steps to obtain the cable RL parameters at different frequencies by using the CableConstants program of the EMTP platform or the finite element analysis method are as follows: Considering the special case of the two-branch model, the model parameters are determined by formulas (1)-(6): ΔR=R C2 -R C1 (1) ΔL=L C1 -L C2 (2) Where R1, L1, R2, L2 are the resistance and inductance of the first and second branches of the model respectively; R C1 ,L C1 ,R C2 ,L C2 is the input data, which is the resistance and inductance of the cable at known frequencies ω1 and ω2. Through (1)-(6), if the RL values ​​of the cable corresponding to the two frequencies are known, the parameters of the dual-branch model can be determined analytically. The calculation of the parameters of RL circuits with three or more branches is achieved using additional procedures for sequential circuit order simplification: Get the two highest frequencies ω from a data set of n pairs of resistors and inductors n and ω n-1 Two pairs of related data: (R Cn , L Cn ) and (R Cn-1 , L Cn-1 ), and then calculate the n-branch model corresponding to the external branch (R n , L n ) parameters; The equivalent circuit of the n-branch model consists of an external n-th branch and an equivalent impedance (R′ eq(ωi) and L′ eq(ωi) ), the latter represents the combination of the remaining n-1 branches at all frequencies, and the total impedance of the circuit matches the known cable data at a given frequency; therefore, the equivalent parameters of the internal branch combination are determined by formula (7) (8): Determine n-1 pairs of frequencies ω i After RL under (i=1,2,...n-1), the internal equivalent conductor formed by the n-1 branches is fully characterized; according to the calculation steps of the n-th branch, the parameters of the n-1 groups of equivalent conductors are used to determine the n-1th branch, and so on. The above steps are repeated until only two data sets corresponding to the frequencies are left. At this time, (5) and (6) are used to calculate the inductance and resistance parameters (R1, L1) of the first branch.

4. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1, characterized in that: The step S2 specifically includes: Model and mesh the individual slots of the motor in the finite element platform, encrypt the mesh of the thin insulation structure to ensure the calculation accuracy, set the model boundary, and add electrostatic field solution analysis; In the slot area, the stator core is considered as a grounded conductor "0", and the capacitance to ground between each turn and the stator core is expressed as C i.0 =C slot i.0 (i=1…n T ), n T is the total number of conductor turns in a slot, and the inter-turn capacitance is expressed as C i，j =C slot i，j , (i≠j, j=1…nT); consider the inter-turn capacitance C at the end ov i，j ; After simplification, the capacitance to ground of each turn is C g i.0 and the inter-turn capacitance C T i，j The calculation is as follows: The electrical energy method is used to calculate the capacitance of a conductor system located in a linear dielectric. For n conductors and reference ground 0, the electrical energy is written as formula (11): Among them C i.0 and C i，j is the capacitance of the system, u i and u j are the potentials of the i-th and j-th conductors, respectively; to determine the capacitance of a multiconductor system with n conductors and reference ground 0, we need Independent conductor potentials u of 0V and 1V i and u j The energy expression (11) is transformed into a system of equations: Among them, W el represents the energy vector, C represents the capacitance vector, k u Represents two vectors W el The connection matrix between and C, the capacitance can be extracted from the equations:

5. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1, characterized in that: The step S3 specifically includes: The end is in an irregular state due to the special torsion angle of the flat wire winding and the welding process. Therefore, 3D modeling is performed on the end. Three adjacent conductors are taken to simplify the modeling. The calculation of the AC resistance and inductance of the end is realized through the eddy current field, and the finite element method is used for analysis and calculation. Inductance is generated by the energy of the magnetic field: Among them, i j and i k (j, k = 1, ..., n T ) is the current source of conductors j and k, which is a normalized combination of 0A and 1A sinusoidal current sources, and is introduced into the finite element partial differential equation as an initial condition; The resistance is caused by the power dissipated: At high frequencies, the finite element method divides the model region into a large number of three-dimensional grids and uses the magnetic field energy W m and copper loss P el The calculation method is used to continuously solve the inductance and resistance of each turn through different current combinations of 0A and 1A; finally, the inductance and AC resistance at the end are regarded as constant values ​​of equal value per turn, and the calculation result is the average of the three conductors.

6. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1, characterized in that: In step S4, the transient magnetic field model includes: three-phase windings and a complete stator core, and the BH curve of the actual material is imported into the simulation software, fully considering the influence of the core saturation nonlinearity on the winding inductance parameters; based on finite element analysis and mesh encryption processing, the skin effect and proximity effect of the conductor at high frequency are fully considered; the inductance and resistance parameters of the conductor in the transient magnetic field model will be automatically calculated in the subsequent field-circuit coupled finite element simulation.

7. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1, characterized in that: The step S5 specifically includes: Generate primary circuit: take each half-turn winding as a circuit unit, determine the total number of circuit units, total number of circuit units = number of slots × number of layers; connect each circuit unit in the order of winding numbers from small to large, for example, for a 54-slot motor containing UVW three-phases, the connection order is: U1i>U 1o>U 2i>U 2o……>U 54i>U 54o, V1i>V 1o>V 2i>V2o……>V 54i>V 54o, W1i>W 1o>W 2i>W2o……>W 54i>W 54o; Determine capacitance parameters: Identify the winding arrangement diagram, determine the position of the winding in the stator slot according to the winding arrangement diagram, and then determine the capacitance parameters corresponding to each winding unit, and assign the capacitance parameters to the corresponding circuit unit; Remaining circuit connections: Complete the connection of the voltmeter based on the primary circuit, integrate the cable n-branch ladder circuit obtained in step S1, and power it with a pulse source to complete the construction of the high-frequency equivalent circuit model.

8. The method for predicting voltage distribution and insulation weak points based on flat wire windings of variable frequency motors according to claim 1, characterized in that: When simulating and calculating the voltage distribution based on the flat wire winding of the variable frequency motor, the step S6 specifically includes: The voltage distribution is simulated and calculated using Ansys Maxwell software. The basis of finite element analysis is the Maxwell equations, which consist of the following four laws: Ampere's circuit law, Faraday's law of electromagnetic induction, Gauss's flux law, and Maxwell's equations. The differential equations for electromagnetic problems using finite elements are derived, corresponding to equations (16) to (19) respectively: The above equations include two curl equations (16), (17) and two divergence equations (18), (19). The relationship between the field quantities E, D, B, and H is determined by the properties of the medium. For linear media, the relationship is: D = εE (20) B = μH (21) J = σE (22) Where ε is the dielectric constant of the medium, F / m; μ is the magnetic permeability of the medium, H / m; σ is the electrical conductivity of the medium, S / m; for isotropic media, ε, μ, σ are scalars; for anisotropic media, they are tensors; To address the problem that the finite element solver will decouple the Maxwell equations, the circuit model is coupled with the finite element model through field-circuit coupling to simultaneously solve the magnetic field and electric field.

9. The method for predicting voltage distribution and insulation weak points based on flat wire windings of variable frequency motors according to claim 1, characterized in that: The step S8 comprises: The calculation results of the time-domain voltage distribution are obtained and exported in csv format to obtain the ungrouped raw data. The MATLAB program is used to read the groups in a loop and classify them according to different insulation structures. For each classification, a new data structure is created to store the slot number, layer number and corresponding voltage amplitude of each winding. Considering the margin of insulation evaluation, the voltage amplitude takes the maximum value within a cycle; the original data set is traversed, and the obtained voltage amplitude is integrated according to the actual position of the winding in the stator slot to generate a voltage amplitude table that can locate the specific winding by slot number and layer number.

10. The voltage distribution and insulation weak point prediction method based on the flat wire winding of the variable frequency motor according to claim 1, characterized in that: The step S9 comprises: The PDIV / PDEV test method is measured by selecting appropriate pulse voltage parameters and wiring methods according to the IEC standard using a preset PDIV / PDEV test platform; The insulation weak point radar map generation program converts the voltage distribution results from a tabular form into a radar map that is more similar to the actual structure of the motor, and integrates each voltage amplitude into a map according to its position in the motor. Each axis of the radar map represents a slot number. The PDIV, PDEV values, and voltage distribution data of different layer numbers are distinguished by different colors or line styles. The voltage amplitude is represented by radial lines from the center to the outside. The farther away from the center, the higher the voltage amplitude. The extreme value of each axis represents the maximum and minimum voltage in the slot. Finally, by comparing the size of each voltage distribution with the PDIV test value, it is possible to analyze whether discharge occurs under general conditions. Among them, points where the voltage is distributed near or outside PDEV and PDIV can be regarded as insulation weak points of the motor.

Images