A multi-parameter identification method for vehicle electric drive system

Abstract

The present invention discloses a multi-parameter identification method for an automotive electric drive system that takes into account SiC switching characteristics and motor magnetic saturation characteristics. For a SiC MOSFET voltage source inverter, multiple factors such as dead time, on-off delay, voltage drop, and output capacitance are fully considered, a nonlinear voltage error mathematical model is constructed, and an inverter model parameter offline identification method is designed to reduce the influence of inverter nonlinearity on the motor parameter identification accuracy. At the same time, the present invention considers the influence of magnetic saturation and cross-saturation phenomena of automotive permanent magnet synchronous motors, combines the d-axis current injection method with the step-by-step identification method to solve the lack-rank problem in multi-parameter online identification, and realizes the simultaneous online identification of the stator resistance, AC and DC axis apparent inductance and mutual inductance, and permanent magnet flux electromagnetic parameters, with an identification error not exceeding ±5%.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor control, and in particular relates to a multi-parameter identification method for a vehicle electric drive system taking into account SiC switching characteristics and motor magnetic saturation characteristics. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in new energy vehicles due to their high power density, high efficiency, and excellent dynamic performance. High-performance control of PMSMs relies on accurate motor parameters. However, the electromagnetic parameters of automotive motors are highly time-varying. Research into high-precision online parameter identification techniques is crucial for achieving high-performance control of PMSMs. Currently, online PMSM electromagnetic parameter identification schemes rely on mathematical models constructed from a set of voltage equations. Errors in the stator voltage and the lack of rank in the model are two major issues that affect parameter identification accuracy.

[0003] In the field of new energy vehicles, the switching frequency of voltage source inverters is generally 10-15kHz. Due to the limited sensor measurement bandwidth, the reference voltage output by the motor controller is often used as the identifier input. However, the nonlinear characteristics of the VSI (voltage source inverter) power devices can cause errors between the reference voltage and the actual voltage, resulting in parameter identification errors. VSI nonlinear voltage compensation approaches can be divided into three categories. The first category, such as the literature [Zhang Yong, Lu Ying, Shen Xiao, et al. Dead-zone compensation method for discontinuous pulse width modulation of heterogeneous switching three-phase three-level grid-connected converters [J / OL]. Journal of Power Supply], uses additional hardware circuits to detect the actual VSI output voltage online in real time and compare it with the reference voltage output by the controller to obtain the VSI nonlinear voltage. However, this method requires additional hardware. The second category, such as the literature [Jin Xuefeng, Tian Kai, Zhang Ce, et al. A dead-zone compensation method for voltage source inverters based on current prediction [J]. Electric Drive, 2015, 45(09): 25-29] uses observer observations to compensate for VSI nonlinear voltage, but this type of method requires online observation, which will increase the complexity of the system and limit its dynamic performance. The third type, such as the literature [Chen Youyan, Ji Wei, Yang Degang. Offline identification of inverter voltage compensation parameters and its application research [J]. Electronic Design Engineering], measures the VSI nonlinear voltage error under different current amplitudes offline, and then stores it as a table or fits it to the established VSI nonlinear voltage mathematical model, thereby compensating it to the controller.

[0004] There are two main solutions to the problem of under-ranking: the first is to increase the number of voltage equations, such as in the literature [Yang Gongde, Chen Yuxiang, Wang Peng. Comparative analysis of permanent magnet synchronous motor parameter identification based on signal injection method [J / OL]. Journal of Electronic Measurement and Instrumentation, 1-11], which increases the order of the motor equation by injecting DC signals, high-frequency signals, rotor position offset angles, etc. However, signal injection may have an adverse effect on the control system. The second is to reduce the number of parameters to be identified simultaneously, such as in the literature [Liu Huibo, Huang Qianzhu. Parameter identification of permanent magnet synchronous motor based on model predictive control [J]. Micromotors, 2021, 54(09): 70-77+100], which reduces the number of parameters to be identified simultaneously by fixing some parameters to nominal values and using two-step or three-step methods, but the inaccuracy of the fixed parameters will reduce the accuracy of identification.

[0005] In summary, a common shortcoming of existing methods is that they fail to consider cross-saturation effects, which affects the accuracy of identifying the DC-axis inductances. Furthermore, current research on VSI nonlinear voltage error modeling has mostly focused on inverters based on Si (silicon) IGBTs (insulated gate bipolar transistors), while limited research has been conducted on modeling VSI nonlinear voltage errors based on SiC (silicon carbide) MOSFETs (metal oxide semiconductor field-effect transistors). Summary of the Invention

[0006] In view of the above, the present invention provides a multi-parameter identification method for an automotive electric drive system that takes into account the SiC switching characteristics and the motor magnetic saturation characteristics. This method can solve the problems of nonlinear voltage compensation and multi-parameter online identification difficulties in electric drive systems using SiC MOSFET inverters and permanent magnet synchronous motors.

[0007] A multi-parameter identification method for a vehicle electric drive system considering SiC switching characteristics and motor magnetic saturation characteristics includes the following steps:

[0008] (1) Considering the switching physical characteristics of SiC MOSFET, a VSI nonlinear voltage error model under different currents is established and simplified;

[0009] (2) Calibrate the VSI nonlinear voltage error under different currents under offline conditions;

[0010] (3) Use the least squares method to fit the calibrated data to complete the parameter adjustment of the VSI nonlinear voltage error model;

[0011] (4) Using the VSI nonlinear voltage error model to perform online compensation for the nonlinear voltage, thereby establishing a PMSM mathematical model for parameter identification;

[0012] (5) Relevant data are collected during the online operation of the motor. The collected data are used to identify the parameters of the PMSM mathematical model through two least squares algorithms, and multiple parameters including stator resistance, permanent magnet flux, direct-axis apparent inductance, quadrature-axis apparent inductance, and cross-coupling apparent inductance are obtained.

[0013] Furthermore, the simplified VSI nonlinear voltage error model in step (1) is expressed as follows:

[0014]

[0015] Where: Δ u ( i x ) indicates the VSI x Phase nonlinear voltage error, I cc is the critical current, i x For VSI x Phase current, x =A,B,C, c 1~ c 4 is the parameter to be adjusted.

[0016] Furthermore, the critical current I cc The expression is as follows:

[0017]

[0018] in: U dc is the DC voltage of VSI, C is the parasitic capacitance of SiC MOSFET in VSI, U SiC is the on-state voltage drop of SiCMOSFET, U diode is the forward voltage drop of the SiC MOSFET anti-parallel diode, T d is the dead time.

[0019] Furthermore, in step (2), for any phase, multiple sets of calibration data are collected and measured through experiments under offline conditions, and each set of calibration data includes the corresponding Δ u ( i x )and i x ,The experiment is carried out under the condition that the motor rotor is blocked.

[0020] Furthermore, the expression of the PMSM mathematical model in step (4) is as follows:

[0021]

[0022] in: and are the d-axis voltage reference value and q-axis voltage reference value of PMSM respectively, Δ u d and Δ u q They are the d-axis voltage error and q-axis voltage error of the VSI (the three-phase nonlinear voltage error Δ is calculated by the VSI nonlinear voltage error model). u ( i A ),Δ u ( i B ),Δ u ( i C ) is obtained through coordinate transformation, R s is the stator resistance of the PMSM, ψ f is the permanent magnet flux linkage of PMSM, ω e is the electrical angular velocity of the PMSM, i d and i q are the d-axis current and q-axis current of PMSM respectively, L d_app 、 L q_app 、 L dq_app They are the direct-axis apparent inductance, quadrature-axis apparent inductance, and cross-coupling apparent inductance of PMSM respectively.

[0023] Furthermore, in step (5), the d-axis bias current is injected twice in succession during the online operation of the motor, and relevant data in this process including the dq-axis current, dq-axis voltage reference value and electrical angular velocity of the PMSM are collected; then, the collected data are used to perform parameter identification on the PMSM mathematical model through a step-by-step identification method, and during the first identification, the stator resistance and permanent magnet flux in the model are initialized using the nominal value or the result of offline measurement, and then the inductance parameters are identified using the recursive least squares method; finally, the identified inductance parameters are used as known quantities, and the dq-axis current, dq-axis voltage reference value and electrical angular velocity of the PMSM are used to perform a second identification using the recursive least squares method to obtain the stator resistance and permanent magnet flux.

[0024] Furthermore, when the dq axis current of the PMSM changes significantly, step (5) is repeated; if the dq axis current remains constant for a long time, the recursive least squares method is used to perform secondary identification every 10 seconds to update the values of the stator resistance and permanent magnet flux.

[0025] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the multi-parameter identification method for a vehicle electric drive system that considers SiC switching characteristics and motor magnetic saturation characteristics.

[0026] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the multi-parameter identification method for a vehicle electric drive system that takes into account SiC switching characteristics and motor magnetic saturation characteristics.

[0027] In the high-performance control of new energy vehicle motors, real-time and accurate online identification of motor parameters is very necessary. The present invention targets SiC MOSFET voltage source inverters, fully considers multiple factors such as dead time, on-off delay, voltage drop and output capacitance, constructs a nonlinear voltage error mathematical model, and designs an inverter model parameter offline identification method to reduce the impact of inverter nonlinearity on the accuracy of motor parameter identification. At the same time, the present invention considers the influence of magnetic saturation and cross-saturation phenomena of automotive permanent magnet synchronous motors, combines the d-axis current injection method with the step-by-step identification method to solve the rank deficiency problem in multi-parameter online identification, and realizes the simultaneous online identification of stator resistance, AC and DC axis apparent inductance and mutual inductance, and permanent magnet flux electromagnetic parameters. Compared with the prior art, the present invention has the following beneficial technical effects:

[0028] 1. Existing VSI nonlinear voltage mathematical models are often based on IGBT inverters. The present invention models the nonlinear voltage error of VSIs based on SiC MOSFETs, and fully considers the impact of the third quadrant operating mode of SiC MOSFETs.

[0029] 2. Existing PMSM parameter identification often fails to consider the impact of cross-saturation effects, and it is difficult to achieve simultaneous online identification of multiple parameters. However, this invention achieves simultaneous online identification of five electromagnetic parameters: stator resistance, AC and DC axis apparent inductance and mutual inductance, and permanent magnet flux linkage.

[0030] 3. Under the calibration method based on the input error criterion, the identification error of the VSI nonlinear error voltage of the present invention does not exceed ±5%. Under the calibration method based on the output error criterion, the identification error of the PMSM parameters under different operating conditions does not exceed ±5%. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1Schematic diagram of the structure of SiC MOSFET inverter.

[0032] Figure 2 This is the control system block diagram of the motor SiC MOSFET inverter.

[0033] Figure 3 Schematic diagram of the d-axis current reference value and d-axis current response under two current injections.

[0034] Figure 4 Schematic diagram of the flow of the parameter identification algorithm of the present invention.

[0035] Figure 5 Schematic diagram of the verification process of PMSM parameter identification results.

[0036] Figure 6 is the voltage error of phase A Δ u ( i A ) and phase A current i A And the B phase voltage error Δ u ( i B ) and B phase current i B Schematic diagram of the relationship.

[0037] Figure 7 is the voltage error Δ u ( i x ) and current i x Schematic diagram of the relationship and its fitting curve.

[0038] Figure 8 The diagram for verifying the inverter nonlinear voltage identification results is shown in the figure. t Indicates time, vertical axis u Indicates voltage.

[0039] Figure 9(a) to Figure 9(c) Respectively in i d =-20A, i q =50A, n =1000r / min three-phase current and speed, quadrature and direct axis current, experimental waveform diagram of identification results, horizontal axis t Indicates time, vertical axis u and i Indicates voltage and current.

[0040] Figure 10(a) to Figure 10(c) Respectively in i d =-50A,i q =200A, n =1000r / min three-phase current and speed, quadrature and direct axis current, experimental waveform diagram of identification results, horizontal axis t Indicates time, vertical axis u and i Indicates voltage and current. DETAILED DESCRIPTION

[0041] In order to describe the present invention more specifically, the technical solution of the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] The present invention provides a multi-parameter identification method for a vehicle electric drive system that considers SiC switching characteristics and motor magnetic saturation characteristics, comprising the following steps:

[0043] (1) The structure of SiC MOSFET inverter is as follows Figure 1 As shown, it is a three-phase six-bridge structure, including bridge arm switches S1~S6 and their corresponding anti-parallel diodes D1~D6, C H and C L are the total parasitic capacitances of the upper and lower bridge arms respectively.

[0044] Considering the influence of factors such as the conduction voltage drop of the switch tube and diode, the dead zone effect, the switch tube turn-on and turn-off delay, the parasitic capacitance charging and discharging when the switch tube is turned on, and the parasitic capacitance charging and discharging when the switch tube is turned off, the nonlinear voltage error of the inverter caused is summarized as shown in Table 1 (taking phase A as an example):

[0045] Table 1

[0046]

[0047] In the table: Δ u 1( i A )~Δ u 4( i A ) are the VSI nonlinear voltage errors caused by the conduction voltage drop of the switch tube and diode, the dead zone effect, the switch tube turn-on and turn-off delay, and the parasitic capacitance charge and discharge when the switch tube is turned on; Δ u 5( i A ) and Δ u 6( i A ) is the VSI nonlinear voltage error caused by the charging and discharging of the parasitic capacitance when the switch is turned off; C is the capacitance of the parasitic capacitor; I c1 It is the charge and discharge current of the parasitic capacitance when the switch tube is turned on; Icc is the critical current, and its expression is:

[0048]

[0049] When | i A |> I cc When the switch is turned off, the VSI nonlinear voltage error caused by the parasitic capacitance charging and discharging is only Δ u 5( i A ); when | i A |≤ I cc When the switch is turned off, the parasitic capacitance cannot be T d ' The charge and discharge are completed within the time, so the voltage error is caused by Δ u 5( i A ) and Δ u 6( i A ) consists of two parts; I c3 For in| i A |< I cc In the second stage when the switch tube is turned off, the charging and discharging current of the parasitic capacitance is

[0050] (2) Analyze and simplify the error voltage model obtained in step (1). The error voltage caused by various factors can be lumped together and written as:

[0051]

[0052] (3) Parameter self-tuning of the VSI nonlinear voltage model is performed. The specific process is as follows:

[0053] 3.1 In Figure 2 Under the control system structure shown, the experimental data is collected and the system uses the dq axis current reference value and As input, the dq axis voltage reference value is obtained through the current controller and .

[0054] In order to simplify the mathematical model, the experiment was carried out under the condition of rotor blocking. ω e =0, so the inductor voltage drop and back electromotive force parts in the voltage equation are both 0. By coordinate transformation, we can get:

[0055]

[0056] in: T abc,dq is the transformation matrix from the abc three-phase coordinate system to the dq coordinate system.

[0057] From the above formula we can get:

[0058]

[0059] There are three unknown quantities Δ u ( i A ),Δ u ( i B ) and Δ u ( i C ), it is difficult to identify them simultaneously, so let the given i d and i q Maintain special relationships to reduce the number of unknowns. The angle between the rotor electrical angle position and the axis of the A-phase winding when the motor is stalled satisfies θ = n (π / 3), where n =0,1,..5, θ ∈[0,360), combined with the mathematical properties obtained in step (2), the above formula can be simplified to:

[0060]

[0061] according to θ The value of is different. x is one of A, B, and C, and there is i x = i d , the corresponding relationship is shown in Table 2:

[0062] Table 2

[0063]

[0064] 3.2 Parameter identification method design.

[0065] exist θ When 0 or 5π / 3rad, the experimentally measured i A and Δ u ( i A ) Total data N Group, where the current is positive and negative data areN / 2 groups, recorded as ( i +A,k , Δ u ( i +A,k ))、( i -A,k , Δ u ( i -A,k )), k =1,2,…, N / 2; Least squares fitting can be used to obtain the formula in step (2) c 1. c 2. c 3. c 4. The process is as follows:

[0066] 1. i A , Δ u ( i A ) are horizontal and vertical coordinates respectively, draw a scatter plot, and determine Δ u ( i A )and i A The dividing line between linear and nonlinear relationship is estimated I cc .

[0067] 2. Set at 0< i A <2 I cc / 3 interval data total k Group, represented by ( i +A,1 , Δ u ( i +A,1 ))、( i +A,2 , Δ u ( i +A,2 )),…,( i +A,k , Δ u ( i +A,k )), using the least squares linear fitting method, we get c 3+ 、 c 4+ :

[0068]

[0069] 3. For -2 Icc / 3< i A <0 n Similarly, the least squares fitting is performed on the data set to obtain c 3- 、 c 4- .

[0070] 4. Located at i A >4 I cc / 3 interval data total m Group, represented by ( i +A,k+1 , Δ u ( i +A,k+1 ))、( i +A,k+2 , Δ u ( i +A,k+2 )),…,( i +A,k+m , Δ u ( i +A,k+m )),make X +A,l =1 / i +A,l , l = k , k +1,…, k + m ,use( X +A,k+1 , Δ u ( i A,k+1 ))、( X +A,k+2 , Δ u ( i +A,k+2 )),…,( X +A,k+m , Δ u ( i +A,k+m )) Perform a least squares fit of the function and get c 1+ 、 c 2+ :

[0071]

[0072] 5. For i A <-4 Icc / 3 m Similarly, the least squares fitting is performed on the data set to obtain c 1- 、 c 2- .

[0073] 6. Yes c 1+ 、 c 1- Taking the average, we get c 1. Similarly, we get c 2. c 3. c 4.

[0074] 7. Order c 3 i A - c 4=- c 1 / i A + c 2. Calculated i A for I cc .

[0075] (4) Verification of SiC MOSFET VSI nonlinear error voltage model:

[0076] The input error criterion is used to verify the model shown in step (2) and the parameter identification method in step 3.2. The experimental conditions are the same as those in step 3.1, that is, the quadrature and direct axis current reference values meet And the motor is stalled, the angle between the rotor electrical angle position and the axis of the A-phase winding satisfies θ = n (π / 3), where n =0,1,…,5, θ ∈[0, 360).

[0077] Under different current settings, according to the results obtained in step 3.2 c 1. c 2. c 3. c 4 and the mathematical model in step (2), calculate the Δ u ; Calculate using the voltage equation in step 3.1 , and the current controller output Compare them and get the relative error between the two. When the error is less than ±5%, the model is considered to have passed the verification.

[0078] (5) Methods for PMSM parameter identification

[0079] 5.1 Experimental Design

[0080] Under the condition of steady-state operation of the motor, the PMSM mathematical model used for parameter identification can be expressed as follows, where and In the following abbreviated as u d and u q .

[0081]

[0082] In the above formula, we have R s 、 L d_app 、 L q_app 、 L dq_app and ψ f There are five quantities to be identified. Among them, the inductance value may change significantly with the change of current due to the greater influence of the magnetic saturation effect on the inductance, while the resistance and permanent magnet flux are mainly affected by temperature. Therefore, the time-varying nature of the resistance and permanent magnet flux is weaker than that of the inductance. Based on this premise, a time-sharing method is used to design the identification experiment. During the identification process, the d-axis current reference value and the corresponding d-axis current response are as follows: Figure 3 As shown, it includes two stages. Stage 1 consists of M1 and M2 parts, which are used to identify the motor inductance parameters; Stage 2 consists of M3 and M4 parts, which are used to identify the motor resistance and permanent magnet flux amplitude; the amplitude of the two injected currents is Δ i d , the current before injection is i d0 , the current after injection is i d ’ .

[0083] In order to reduce the interference to the system operation, the duration of current injection should be as short as possible, but should be longer than the adjustment time of the d-axis current loop, and it needs to be kept for a sufficient time after the current stabilizes to sample enough data. Figure 3 As shown, assuming that the adjustment time of the current loop is T r , the sampling time is T s , the number of sampling times is x , then the current injection time Δ T = T r + xT s, combined with experiments, the specific Δ T .

[0084] Before the first bias current injection, the current response needs to reach a steady state, such as Figure 3 In the M1 segment, the motor current and speed are measured multiple times during this period and the reference voltage value is recorded. The data are averaged. i d And wait for the current to reach steady state, that is Figure 3 In the middle M2 section, the current, speed and voltage values are also measured and recorded and averaged. i d The value is small and the effect of incremental inductance is ignored.

[0085] 5.2 Identification Method

[0086] During the first identification, the model is initialized using nominal values or offline measurement results. R s and ψ f , and then identify the inductance parameters.

[0087] Combining the data obtained during the M1 and M2 periods and the voltage equation, we can obtain the least squares expression for the identification problem, namely:

[0088]

[0089]

[0090]

[0091] in: y and are the input and output matrices respectively, θ is the parameter matrix to be identified, and the superscripts M1 and M2 indicate that the data belongs to the M1 and M2 segments respectively. L d 、 L q 、 L dq are the direct-axis apparent inductance, quadrature-axis apparent inductance, and cross-coupling apparent inductance of the motor.

[0092] Using the recursive least squares method (RLS1), the identification expression is:

[0093]

[0094] in: is the estimated value of the parameter matrix, K and Pare correction coefficients, I is the identity matrix, λ For the forgetting factor, k is the number of recursions.

[0095] Furthermore, the identified inductance is used as a known quantity, and the recursive least squares method (RLS2) is used to identify the resistance and permanent magnet flux amplitude using the current, speed and reference voltage of the M3 segment. At this time, y 、 and θ They are:

[0096] , ,

[0097] Wherein: the superscript M3 indicates that the data belongs to the M3 segment.

[0098] The identification algorithm flow of the present invention is as follows Figure 4 As shown, whenever the current of the dq axis changes significantly, the parameter identification algorithm is activated to update R s 、 L d_app 、 L q_app 、 L dq_app and ψ f If the current of the dq axis remains constant for a long time, run the second step of identification every 10 seconds and update R s and ψ f The value of .

[0099] 5.3 Verification of PMSM parameter identification results

[0100] Using the output error criterion, under steady-state operation conditions, the identified R s 、 L d_app 、 L q_app 、 L dq_app and ψ f Substitute the current and speed obtained by real-time sampling into the mathematical model in step 5.1 to obtain u d and u q The corresponding calculation results and , the calculated result is compared with the voltage reference output by the current controller after the VSI nonlinear error voltage compensation. and Compare them and get the relative error between them, such as Figure 5 If the error is less than ±5%, the model is considered to have passed the verification.

[0101] In order to ensure the feasibility and effectiveness of this invention, we established an experimental system with a permanent magnet synchronous motor as the load, with a peak power of 150kW, a peak speed of 10000r / min, and a peak torque of 550Nm; the voltage source inverter is composed of a SiC module from Rohm Semiconductor, model BSM600D12P3G001, with a DC bus voltage of 700V and a control frequency of 10kHz.

[0102] According to the method of step (3), the experiments were carried out when the motor rotor locked angle was 0 and 2π / 3, and the results were Δ u ( i A )and i A , Δ u ( i B )and i B The relationship as Figure 6 As shown in the figure, the relationship between the current and error voltage of different phases is basically the same, so the experimental results of phase A can be extended to the three phases.

[0103] Using Δ u ( i A )and i A The data is numerically fitted to obtain c 1. c 2. c 3. c 4 are 6.1796, 4.0486, 0.526, and -0.0935, respectively, so we get I cc is 5.3044A, and we can get Δ u ( i A )and i A The expression of , and generalized to three phases:

[0104]

[0105] in: x For any phase among A, B, and C, the fitted curve is as follows Figure 7 shown.

[0106] Verify the obtained model and compare it at different currents and ,like Figure 8 As shown; it can be seen from the figure and The relative error is within ±5%, so the model is considered qualified.

[0107] The above results are used to compensate the reference voltage and perform online multi-parameter identification of PMSM, Δ i d The injection duration is set to 500ms, Δ i d The value is set to -3A, R s and ψ f The initial values of are set to 17.75mΩ and 0.1276Wb respectively according to the nominal values. According to the method of the present invention, online parameter identification is performed, the operating conditions of the motor are changed, and multiple groups of experiments are performed. Table 3 lists several groups of parameters obtained under different currents. L d_app 、 L q_app 、 L dq_app .

[0108] Table 3

[0109]

[0110] For the identified R s 、 L d_app 、 L q_app 、 L dq_app and ψ f For verification, the experimental waveform results are as follows Figure 9(a) to Figure 9(c) and Figure 10(a) to Figure 10(c) As shown in the figure, it can be seen that at different currents and speeds n Next, the dq axis reference voltage and The dq axis reference voltage calculated from the identified parameters and The relative error between them is within ±5%, so the identification result is considered qualified. In addition, it can be seen from the figure that injecting a small amplitude identification current into the d-axis current during the experiment does not have a significant impact on the current and speed.

[0111] The above description of the embodiments is intended to facilitate understanding and application of the present invention by those skilled in the art. It is apparent that those skilled in the art can readily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without requiring creative effort. Therefore, the present invention is not limited to the above embodiments. Any improvements or modifications made by those skilled in the art based on the disclosure of the present invention should fall within the scope of protection of the present invention.

Claims

1. A multi-parameter identification method for an automotive electric drive system considering SiC switching characteristics and motor magnetic saturation characteristics, comprising the following steps: (1) Considering the physical characteristics of SiC MOSFET switches, including the on-state voltage drop of the switch tube and body diode, dead zone effect, switch tube turn-on and turn-off delay, and the charging and discharging of parasitic capacitance when the switch tube is turned on and off, a VSI nonlinear voltage error model under different currents is established and simplified; The expression of the VSI nonlinear voltage error model is as follows: ; The simplified VSI nonlinear voltage error model is expressed as follows: ; ; in: Δ u ( i x ) indicates the VSI x Phase nonlinear voltage error, I cc is the critical current, i x For VSI x Phase current, x =A,B,C, c 1~ c 4 is the parameter to be adjusted, U dc is the DC voltage of VSI, C is the parasitic capacitance of SiC MOSFET in VSI, U SiC is the on-state voltage drop of SiC MOSFET, U diode is the forward voltage drop of the SiC MOSFET anti-parallel diode, T d is the dead time, Δ u 1( i x )~Δ u 4( i x ) are the VSI nonlinear voltage error caused by the conduction voltage drop of the switch tube and the diode, the dead zone effect, the switch tube turn-on and turn-off delay, and the parasitic capacitance charge and discharge when the switch tube is turned on. u 5( i x ) and Δ u 6( i x ) is the VSI nonlinear voltage error caused by the charging and discharging of the parasitic capacitance when the switch is turned off. I c1 is the charge and discharge current of the parasitic capacitance when the switch tube is turned on, I c3 For in| i x |< I cc The charge and discharge current of the parasitic capacitance during one of the stages when the switch tube is turned off; (2) Calibrate the VSI nonlinear voltage error under different currents under offline conditions; (3) Use the least squares method to fit the calibrated data to complete the parameter adjustment of the VSI nonlinear voltage error model; (4) The nonlinear voltage is compensated online using the VSI nonlinear voltage error model, thereby establishing a PMSM mathematical model for parameter identification. The expression is as follows: ; in: and are the d-axis voltage reference value and q-axis voltage reference value of PMSM respectively, Δ u d and Δ u q are the d-axis voltage error and q-axis voltage error of VSI respectively, R s is the stator resistance of the PMSM, ψ f is the permanent magnet flux linkage of PMSM, ω e is the electrical angular velocity of the PMSM, i d and i q are the d-axis current and q-axis current of PMSM respectively, L d_app 、 L q_app 、 L dq_app They are the direct-axis apparent inductance, quadrature-axis apparent inductance, and cross-coupling apparent inductance of the PMSM; (5) Relevant data is collected during the online operation of the motor, and the collected data is used to identify the parameters of the PMSM mathematical model through two least squares algorithms, and multiple parameters including stator resistance, permanent magnet flux, direct axis apparent inductance, quadrature axis apparent inductance, and cross-coupling apparent inductance are obtained. Specifically: during the online operation of the motor, the d-axis bias current is injected twice, and relevant data in this process are collected, including the dq axis current, dq axis voltage reference value, and electrical angular velocity of the PMSM; then, the collected data is used to identify the parameters of the PMSM mathematical model through a step-by-step identification method. During the first identification, the stator resistance and permanent magnet flux in the model are initialized using the nominal value or the result of offline measurement, and then the inductance parameters are identified using the recursive least squares method; finally, the identified inductance parameters are used as known quantities, and the dq axis current, dq axis voltage reference value, and electrical angular velocity of the PMSM are used to perform a second identification using the recursive least squares method to obtain the stator resistance and permanent magnet flux; When the dq axis current of the PMSM changes significantly, step (5) is repeated; if the dq axis current remains constant for a long time, the recursive least squares method is used for secondary identification every 10 seconds to update the values of the stator resistance and permanent magnet flux.

2. The multi-parameter identification method for a vehicle electric drive system considering SiC switching characteristics and motor magnetic saturation characteristics according to claim 1, characterized in that: In step (2), for any phase, multiple sets of calibration data are collected and measured through experiments under offline conditions, and each set of calibration data includes the corresponding Δ u ( i x )and i x ,The experiment is carried out under the condition that the motor rotor is blocked.

3. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: The processor is used to execute the computer program to implement a multi-parameter identification method for a vehicle electric drive system taking into account SiC switching characteristics and motor magnetic saturation characteristics as described in any one of claims 1 to 2.

4. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it implements a multi-parameter identification method for a vehicle electric drive system taking into account SiC switching characteristics and motor magnetic saturation characteristics as described in any one of claims 1 to 2.

Images