Cascade multi-level permanent magnet synchronous motor driving system loss optimization control method

Abstract

The invention discloses a loss optimization control method for a cascaded multilevel permanent magnet synchronous motor driving system, and the method comprises the steps: building an equivalent resistance circuit model considering motor loss, and constructing a motor broadband domain loss equivalent circuit model through linear superposition; and obtaining a dq-axis lower converter loss model according to a cascade multi-level carrier phase-shifting modulation strategy. According to the input and output power of the motor driving system before and after the LC filter is added under the working conditions of different rotating speeds and different loads and the dual Fourier series, calculating and outputting the harmonic voltage amplitude of each frequency, and identifying the equivalent resistance under the motor loss model by using a least square method; and a hysteresis component and an eddy current component are separated according to the fundamental frequency equivalent loss resistance, a motor system integrated loss model is constructed in conclusion, and optimal control is performed by applying a golden section algorithm. According to the method, the cascaded multi-level permanent magnet synchronous motor system loss model is established, efficiency optimization is realized by combining the golden section algorithm and applying the motor system loss model, and system-level loss optimization is realized.

Description

Technical Field

[0001] The present invention belongs to the field of application of motor and circuit theory, and in particular relates to a loss optimization control method for a cascaded multi-level permanent magnet synchronous motor drive system. Background Art

[0002] Compared to induction motors, permanent magnet synchronous motors (PMSMs) offer advantages such as high power density, high efficiency, and excellent control performance. They have been widely used in electric transmission, especially high-performance electric drives. The mathematical model of the motor is the foundation for motor control, so establishing an accurate mathematical model of the PMSM is key to achieving high-performance control. However, in the field of motor control, existing PMSM modeling often fails to account for the effects of eddy current reactions, that is, the motor's iron losses. This leads to discrepancies between the theoretical and actual models, especially in high-speed motor applications where iron losses account for a relatively high proportion. Compared to traditional two-level topologies, cascaded multilevel topologies offer low voltage distortion and low common-mode voltage in high-voltage inverter scenarios. Therefore, research on loss modeling and optimal control methods for PMSM systems based on cascaded multilevel power supply is of great value.

[0003] Existing permanent magnet motor modeling methods that account for motor losses typically directly introduce an equivalent iron loss resistor branch into the motor equivalent circuit model to simulate the motor's eddy current losses. However, these modeling methods typically only consider motor losses when the excitation source is at the fundamental frequency. They do not account for the impact of harmonic voltages on the motor when powered by a cascaded multilevel converter, nor do they consider the overall system losses associated with the coupling between the cascaded multilevel converter and the permanent magnet synchronous motor.

[0004] High-performance motor control requires an accurate mathematical model of the permanent magnet motor. However, existing modeling methods for permanent magnet motors powered by cascaded multilevel converters that consider motor losses only account for the motor's fundamental frequency losses. They fail to consider the impact of the converter's high-frequency power supply on motor losses, nor the coupling between the losses of the permanent magnet synchronous motor and the cascaded multilevel converter. Consequently, the modeling is inaccurate. Furthermore, conventional fundamental frequency iron loss models fail to account for the difference between hysteresis loss components and eddy current losses. Summary of the Invention

[0005] The purpose of the present invention is to provide a loss optimization control method for a cascaded multi-level permanent magnet synchronous motor drive system, establish an equivalent circuit for the motor's fundamental frequency iron loss and an equivalent circuit for the motor's high-frequency iron loss, which is more accurate than the traditional permanent magnet motor model that takes into account the fundamental frequency loss of eddy current reaction; based on the cascaded multi-level carrier phase-shift modulation strategy, a dq-axis converter loss model is established to achieve overall loss modeling of the motor drive system; for complex optimization problems, a golden section algorithm is used to compress the working range and converge to the optimal working point.

[0006] In order to achieve the above object, the solution of the present invention is:

[0007] A method for optimizing the control of losses in a cascaded multi-level permanent magnet synchronous motor drive system comprises the following steps:

[0008] Step 1: Establish a circuit model of equivalent resistance that takes into account motor losses. Using linear superposition, consider the impact of harmonic voltage on the high-frequency losses of the motor under converter power supply and construct a wide-frequency domain loss equivalent circuit for the motor.

[0009] Step 2: Based on the cascaded multi-level carrier phase-shift modulation strategy, the double Fourier series expansion is used to calculate the harmonic voltage amplitude of each frequency and establish the dq axis converter loss model;

[0010] Step 3: Based on the motor wide-frequency domain loss motor model of step 1 and the calculation formula of each frequency harmonic voltage amplitude of step 2, the fundamental frequency loss expression and the high-frequency loss expression are obtained;

[0011] Step 4: Measure the motor's input power and output power before and after adding LC filtering under multi-speed and multi-load conditions, and calculate offline data of the motor's fundamental frequency loss and high-frequency loss;

[0012] Step 5: Based on the offline loss data and the amplitudes of the harmonic voltages in step 2, the resistance value of the equivalent loss resistance of the motor in the wide frequency domain is identified using the least squares method;

[0013] Step 6: Separate the hysteresis component and the eddy current component from the fundamental frequency iron loss equivalent resistance in the equivalent loss resistance, and separate the resistance values ​​of each frequency from the high frequency iron loss resistance in the equivalent loss resistance, thereby obtaining a motor model that takes into account wide-frequency domain losses;

[0014] Step 7: construct an integrated loss model of the motor system, and use the golden section algorithm to optimize and control the common variable d-axis current in the dq-axis converter loss model and the motor model considering wide-band loss.

[0015] In step 1 above, the circuit model of the equivalent resistance considering the motor loss is established as follows:

[0016]

[0017] Among them, u d and u q are the d-axis and q-axis stator voltages, i d and i q are the d-axis and q-axis stator currents, L d and L q are the d-axis and q-axis inductances, R s is the stator resistance, ψ f is the permanent magnet flux linkage, ω is the electrical angular frequency, Ri is the equivalent iron loss resistance

[0018] In step 1 above, the motor wide-frequency domain loss model is constructed as follows:

[0019]

[0020] Among them, u df and u qf ,u dh and u qh are the dq axis fundamental frequency voltage and dq axis high frequency voltage respectively, i df and i qf ,i dh and i qh are the dq axis fundamental frequency current and dq axis high frequency current respectively, and p is the derivative symbol;

[0021] In the above step 2, the double Fourier series expansion is used to calculate the amplitude of each frequency harmonic voltage under carrier phase shift modulation.

[0022]

[0023] Among them, u0 is the output voltage of the converter; x is the phase of the triangular carrier wave, y is the phase of the modulation wave, A mn ,B mn Represents the coefficients of a double Fourier series.

[0024] For the cascaded multi-level carrier phase-shift modulation strategy, the converter loss model is established as follows:

[0025]

[0026] Where V on ,R on ,e on ,e off ,e rec All of them are standard experimental values ​​in the device manual. In the above formula, the current parameter I m is the three-phase current amplitude, f c is the switching frequency, V dc is the DC side voltage of the H-bridge unit. According to the Park amplitude invariant matrix, the current parameters can be converted into current components under the dq axis.

[0027] In step 3 above, the expressions for fundamental frequency loss and high frequency loss are as follows:

[0028]

[0029] Among them, P loss_f is the fundamental frequency loss, P loss_h is the high frequency loss, u o is the motor terminal voltage, udqh is the dq axis harmonic voltage, R ch is the loss equivalent resistance at the corresponding frequency

[0030] In step 4 above, the expression for separating the fundamental frequency loss and the high frequency loss in the experiment is:

[0031] P loss_f =P in_LC -P Cu -P out

[0032] P loss_h =P in -P in_LC

[0033] Among them, P loss_f is the fundamental frequency loss, P loss_h is the high frequency loss; P in is the converter input power without LC filtering; P in_LC is the converter input power after LC filtering, excluding high-frequency components; P out is the motor output power; P Cu Copper loss.

[0034] In the above step 5, the least square method is used to identify the parameters of the iron loss resistor using offline data.

[0035] J=(y-Xθ) T (y-Xθ)=y T y-θ T X T yy T Xθ+θ T X T Xθ

[0036] Where θ=(θ1,θ2,...,θ n ) is the identification parameter, and y is the output variable of the identification system. n ) is the identification system input variable;

[0037] The calculation method of the equivalent resistance considering the motor fundamental frequency loss is obtained through the motor fundamental frequency loss model.

[0038]

[0039] Among them, P loss_f is the motor fundamental frequency loss, R c is the equivalent resistance value of the traditional fundamental frequency iron loss equivalent circuit, u o is the motor terminal voltage

[0040] The calculation method of the equivalent resistance considering the high-frequency loss of the motor is obtained through the equivalent circuit of the high-frequency loss of the motor.

[0041]

[0042] in, It is the square of the harmonic voltage amplitude at 1 times the equivalent switching frequency. It is the square of the harmonic voltage amplitude at twice the equivalent switching frequency; R f1 is the equivalent resistance at 1 times the equivalent switching frequency, R f2 It is the equivalent resistance at twice the equivalent switching frequency.

[0043] In step 6 above, based on the separation expressions of hysteresis loss resistance and eddy current loss resistance:

[0044]

[0045] Among them, R c is the traditional fundamental frequency iron loss equivalent resistance, R e is the eddy current loss equivalent resistance, R h is the hysteresis loss equivalent resistance, ω m is the motor mechanical angular frequency;

[0046] In step 7 above, the integrated loss model of the motor system is constructed as follows:

[0047] P L =minP Cu +P Fe,f (i od )+P Fe,h (i od )+P inv (i od ,f sw )

[0048] Among them, P L is the system loss; P Cu is the copper loss, P Fe,f is the motor fundamental frequency loss, P Fe,h is the high-frequency loss of the motor, P inv is the converter loss, i od is the d-axis magnetizing current, f sw is the switching frequency.

[0049] In step 7 above, the golden section algorithm based on interval compression is applied, and the golden section ratio is used to search for the global optimal solution.

[0050] After adopting the above scheme, compared with the prior art, the present invention has the following advantages:

[0051] (1) The present invention establishes a wide-frequency motor loss model, combining the fundamental frequency iron loss equivalent circuit with the high-frequency iron loss equivalent circuit using linear superposition to obtain a wide-frequency motor equivalent circuit model. Compared to motor loss models that only consider the fundamental frequency excitation source, the modeling method proposed in the present invention is more accurate.

[0052] (2) Based on the wide-frequency domain loss model of the motor, the modeling method proposed in the present invention further divides the fundamental frequency iron loss equivalent resistance into hysteresis component and eddy current component. Compared with the traditional modeling method based on equivalent iron loss resistance, it takes into account the influence of speed change on the fundamental frequency iron loss model, and the model is more accurate.

[0053] (3) The present invention proposes a loss model for a permanent magnet synchronous motor system powered by a cascaded multilevel converter, and uses the decoupled current under the dq axis as a variable, which facilitates the construction of an integrated loss model of the permanent magnet synchronous motor and the cascaded multilevel converter based on unified variables, making it easier to subsequently control and optimize.

[0054] (4) The loss model of the permanent magnet synchronous motor system powered by a cascaded multi-level converter has complex nonlinear characteristics. The traditional fundamental frequency minimum loss control algorithm (Loss Minimization Control, LMC) direct analytical calculation method through gradient descent is no longer applicable. Therefore, the golden section algorithm is used for optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is the equivalent circuit diagram of an ideal permanent magnet motor without taking into account the effects of losses;

[0056] Where, (a) is the d-axis, (b) is the q-axis;

[0057] Among them, 1.1 is the stator resistance, 1.2 is the coupling potential of the q-axis to the d-axis, 1.3 is the d-axis stator voltage, 1.4 is the d-axis inductance, 1.5 is the coupling potential of the d-axis to the q-axis, 1.6 is the q-axis inductance, and 1.7 is the q-axis stator voltage;

[0058] Figure 2 This is the equivalent circuit diagram of a traditional permanent magnet motor taking into account the fundamental frequency iron loss;

[0059] Where, (a) is the d-axis, (b) is the q-axis;

[0060] Among them, 2.1 is the stator resistance, 2.2 is the coupling potential of the q-axis to the d-axis, 2.3 is the d-axis stator voltage, 2.4 is the equivalent iron loss resistance, 2.5 is the d-axis inductance, 2.6 is the coupling potential of the d-axis to the q-axis, 2.7 is the q-axis inductance, and 2.8 is the q-axis stator voltage;

[0061] Figure 3 This is a flow chart of a loss optimization control method for a cascaded multi-level permanent magnet synchronous motor drive system;

[0062] Among them, 3.1 is the step of establishing a permanent magnet synchronous motor, 3.2 is the step of establishing the motor wide-frequency domain loss equivalent circuit, 3.3 is the step of identifying the loss equivalent resistance of each frequency using the least squares method, 3.4 is the step of analyzing the cascaded multi-level modulation technology, 3.5 is the step of establishing the converter loss model in the dq coordinate system, 3.6 is the step of obtaining the motor drive system loss model, and 3.7 is the step of implementing optimization control based on the golden section algorithm;

[0063] Figure 4 This is the equivalent circuit diagram of a permanent magnet synchronous motor that takes into account the fundamental frequency iron loss and separates the eddy current component and the hysteresis component;

[0064] Where, (a) is the d-axis, (b) is the q-axis;

[0065] Among them, 4.1 is the stator resistance, 4.2 is the coupling potential of the q-axis to the d-axis, 4.3 is the d-axis stator voltage, 4.4 is the eddy current loss equivalent resistance, 4.5 is the hysteresis loss equivalent resistance, 4.6 is the d-axis inductance, 4.7 is the coupling potential of the d-axis to the q-axis, 4.8 is the q-axis inductance, and 4.9 is the q-axis stator voltage;

[0066] Figure 5 This is the equivalent circuit diagram of the permanent magnet synchronous motor with wide-frequency loss of the motor;

[0067] Where, (a) is the d-axis, (b) is the q-axis;

[0068] Wherein, 5.1 is the stator winding resistance, 5.2 is the coupling potential of the q-axis to the d-axis, 5.3 is the voltage applied to the stator d-axis when the inverter is powered, 5.4 is the equivalent resistance of the motor's wide-band loss, 5.5 is the d-axis inductance, 5.6 is the coupling potential of the d-axis to the q-axis, 5.7 is the voltage applied to the stator q-axis when the inverter is powered, and 5.8 is the q-axis inductance;

[0069] Figure 6 This is the topology diagram of the experimental setup for modeling the permanent magnet synchronous motor with wide-frequency losses.

[0070] Among them, 6.1 is a cascaded multilevel converter, 6.2 is an LC filter, 6.3 is a single-pole double-throw switch, 6.4 is a power analyzer, and 6.5 is a permanent magnet synchronous motor;

[0071] Figure 7 It is a schematic diagram of cascaded multi-level carrier phase-shift modulation with two units cascaded;

[0072] Among them, (a) is a schematic diagram of the output voltage waveforms of the two units when the modulation wave is positive in one fundamental wave cycle, and (b) is a schematic diagram of the output voltage waveforms of the two units when the modulation wave is negative in one fundamental wave cycle;

[0073] Figure 8This is a verification of the accuracy of voltage harmonic analysis and calculation, which includes the voltage waveform and Fourier transform spectrum distribution diagram obtained by dual Fourier series analysis and calculation and simulation operation;

[0074] Among them, (a) is the waveform diagram of the formula calculation, (b) is the corresponding Fourier decomposition spectrum diagram, (c) is the simulation voltage waveform diagram, and (d) is the corresponding Fourier decomposition spectrum diagram;

[0075] Figure 9 This is a least squares identification flow chart for fundamental frequency iron loss and high frequency iron loss, which includes one-dimensional identification for fundamental frequency loss model and two-dimensional identification for high frequency loss model;

[0076] Among them, (a) is the least squares identification of the motor's fundamental frequency iron loss, (b) is the two-dimensional least squares identification of the motor's high-frequency loss, and (c) is the error diagram under two-dimensional identification;

[0077] Figure 10 The accuracy of the motor fundamental frequency iron loss model is verified by cross-experiments under different speed and torque conditions, including absolute error diagrams and relative error diagrams;

[0078] Among them, (a) is the comparison between the fundamental frequency loss model prediction and experimental data at 120 rpm, (b) is the prediction error of the fundamental frequency loss model at 120 rpm, (c) is the comparison between the fundamental frequency loss model prediction and experimental data at 300 rpm, and (d) is the prediction error of the fundamental frequency loss model at 300 rpm;

[0079] Figure 11 The accuracy of the motor high-frequency iron loss model is verified by cross-test under different speed and torque conditions, including absolute error diagram and relative error diagram;

[0080] Among them, (a) is the comparison between the high-frequency loss model prediction and experimental data at 120 rpm, (b) is the high-frequency loss prediction error diagram at 120 rpm, (c) is the comparison between the high-frequency loss model prediction and experimental data at 300 rpm, and (d) is the high-frequency loss prediction error diagram at 300 rpm;

[0081] Figure 12 The accuracy of the converter loss model is verified by cross-tests at different switching frequencies and speeds, including absolute error graphs and relative error graphs.

[0082] Among them, (a) is the comparison between the converter loss model prediction and experimental data at different speeds and switching frequencies, and (b) is the converter loss model prediction error at different speeds and switching frequencies;

[0083] Figure 13 The accuracy of the converter loss model is verified by cross-tests at different switching frequencies and loads, including absolute error graphs and relative error graphs.

[0084] Among them, (a) is the comparison between the converter loss model prediction and experimental data under different load torques and switching frequencies, and (b) is the converter loss model prediction error under different load torques and switching frequencies;

[0085] Figure 14 This is a control block diagram based on the golden section algorithm applied to the loss modeling of the cascaded multi-level power supply permanent magnet synchronous motor system;

[0086] Among them, 14.1 is the golden section algorithm, 14.2 is the speed closed-loop PI regulator, 14.3 is the cascaded multilevel converter, 14.4 is the three-phase permanent magnet synchronous motor, 14.5 is the coordinate transformation module, 42.6 is the motor system loss model, 14.7 is the position encoder, and 14.8 is the speed calculation module;

[0087] Figure 15 This is the flow chart of the golden section algorithm;

[0088] Figure 16 Based on the loss modeling of the cascaded multi-level power supply permanent magnet synchronous motor system, the current trajectory comparison and corresponding loss line graph of system loss optimization and fundamental frequency loss optimization using the golden section algorithm under different load conditions at 300rpm are presented.

[0089] Among them, (a) is the current trajectory of fundamental frequency loss optimization and system loss optimization, and (b) is the motor loss of fundamental frequency loss optimization and system loss optimization. DETAILED DESCRIPTION

[0090] The technical solutions and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0091] The present invention provides a loss optimization control method for a cascaded multi-level permanent magnet synchronous motor drive system, comprising the following steps:

[0092] (1) Establish a circuit model of equivalent resistance that takes into account motor losses. Using linear superposition, consider the impact of harmonic voltage on the high-frequency loss of the motor under converter power supply, and construct a wide-frequency domain loss motor model.

[0093] (2) Based on the cascaded multi-level carrier phase-shift modulation strategy, the double Fourier series expansion is used to calculate the harmonic voltage amplitude of each frequency and construct the dq axis converter loss model;

[0094] (3) According to the circuit model in step (1) and the voltage harmonic calculation formula in step (2), the fundamental frequency loss expression and the high frequency loss expression are obtained;

[0095] (4) Using the LC filter experiment, the input power and output power of the motor before and after LC filtering under multi-speed and multi-load conditions are obtained, and the offline data of the motor fundamental frequency loss and high-frequency loss are calculated;

[0096] (5) Based on the offline loss data and the loss model in step (3), the resistance value of the equivalent loss resistance of the motor in the wide frequency domain is identified using the least squares method.

[0097] (6) Separating the hysteresis component and the eddy current component based on the fundamental frequency iron loss equivalent resistance in step (5), and separating the resistance values ​​of each frequency from the high frequency iron loss resistance, thereby obtaining a motor model that considers wide frequency domain losses;

[0098] (7) Combine steps (2) and (6) to construct an integrated loss model of the motor system, and use the golden section algorithm to optimize the control of the common variable d-axis current in the two models.

[0099] The equivalent circuit diagram of the motor model without considering the iron loss of the motor according to the present invention is as follows: Figure 1 As shown, the stator resistance 1.1, the q-axis coupled potential 1.2, the d-axis stator voltage 1.3, the d-axis inductance 1.4, the d-axis coupled potential 1.5, the q-axis inductance 1.6, and the q-axis stator voltage 1.7 are included. The d-axis and q-axis equivalent circuits are composed of resistance, inductance, and potential in series. Based on the equivalent circuit structure, the following motor model can be obtained:

[0100]

[0101] Among them, u d and u q are the d-axis and q-axis stator voltages, i d and i q are the d-axis and q-axis stator currents, L d and L q are the d-axis and q-axis inductances, R s is the stator resistance, ψ f is the permanent magnet flux linkage, and ω is the electrical angular frequency.

[0102] The equivalent circuit diagram of the conventional permanent magnet motor taking into account the fundamental frequency iron loss of the present invention is as follows: Figure 2 As shown, it includes stator resistance 2.1, q-axis to d-axis coupling potential 2.2, d-axis stator voltage 2.3, equivalent iron loss resistance 2.4, d-axis inductance 2.5, d-axis to q-axis coupling potential 2.6, q-axis inductance 2.7, and q-axis stator voltage 2.8. This type of model directly connects the equivalent iron loss resistance 2.4 in parallel with the inductance and potential branch (i.e., the induced potential branch) to account for eddy current reaction, but the parallel equivalent iron loss resistance usually only considers the iron loss generated when the excitation source is the fundamental frequency. According to the equivalent circuit, the motor model shown in formula (2) can be obtained:

[0103]

[0104] Among them, R i This type of permanent magnet motor modeling method based on equivalent iron loss resistance is usually suitable for steady-state analysis.

[0105] The modeling assumptions based on the motor wide-band loss described in the present invention include the following:

[0106] (1) Ignore spatial harmonics and assume that the winding magnetomotive force is distributed sinusoidally along the air gap;

[0107] (2) Ignore the magnetic circuit saturation effect, that is, the self-inductance of each winding and the mutual inductance between windings are constant;

[0108] (3) Ignore the impact of temperature rise and frequency on the motor parameters;

[0109] (4) Assume that the magnetic permeability inside the permanent magnet is the same as that of air, and assume that the permanent magnet flux is constant.

[0110] The flow chart of the permanent magnet synchronous motor modeling based on the wide frequency domain loss of the motor according to the present invention is as follows: Figure 3 As shown, it includes step 3.1 of establishing a permanent magnet synchronous motor model, step 3.2 of establishing an equivalent circuit of the motor's fundamental frequency iron loss, step 3.3 of establishing an equivalent circuit of the motor's high-frequency iron loss, step 3.4 of experimentally separating the fundamental frequency loss, step 3.5 of experimentally separating the high-frequency loss, step 3.6 of identifying the equivalent resistance of each frequency loss, and step 3.7 of establishing a wide-frequency domain loss model based on the circuit parameters.

[0111] The equivalent circuit diagram of the fundamental frequency iron loss of the improved permanent magnet motor of the present invention is as follows: Figure 4 As shown, it includes stator resistance 4.1, q-axis to d-axis coupling potential 4.2, d-axis stator voltage 4.3, eddy current loss equivalent resistance 4.4, hysteresis loss equivalent resistance 4.5, d-axis inductance 4.6, d-axis to q-axis coupling potential 4.7, q-axis inductance 4.8, and q-axis stator voltage. From this, the parallel equivalent method can be obtained. The separate expressions for hysteresis loss resistance and eddy current loss resistance are:

[0112]

[0113] Among them, R c is the traditional fundamental frequency iron loss equivalent resistance, R e is the eddy current loss equivalent resistance, R h is the hysteresis loss equivalent resistance, ω m is the motor's mechanical angular frequency. Considering the high-frequency voltage and its resulting high-frequency response, and applying the linear superposition principle, the motor voltage equation for the motor's wide-band loss is shown in Equation (4):

[0114]

[0115] Among them, u df and u qf ,u dh and u qh They are dq axis fundamental frequency voltage and dq axis high frequency voltage respectively, and the corresponding i df and i qf ,i dh and i qh are the dq axis fundamental frequency current and the dq axis high frequency current, respectively, and p is the derivative symbol. When the motor is running under high frequency excitation, two simple assumptions are made here: 1. According to the two-level voltage analysis, the high frequency is mainly the switching frequency and its surrounding frequency band, which is much higher than the motor fundamental frequency. Therefore, R can be ignored when calculating the motor winding impedance. s 2. The actual operation of the motor is at the base frequency, so the back electromotive force and cross coupling in the voltage equation are ignored when constructing the high-frequency loss equivalent circuit. Furthermore, the high-frequency iron loss equivalent voltage model can be calculated as follows:

[0116]

[0117] Among them, R ch It is the equivalent resistance value of high-frequency iron loss under high-frequency voltage. Therefore, the formula for calculating high-frequency iron loss is as follows:

[0118]

[0119] Among them, U2dqh is the effective value of the voltage at the iron loss equivalent resistance terminal.

[0120] Furthermore, in step 3.3 of the present invention for calculating the equivalent resistance of each frequency loss, an identification algorithm is shown in the following formula:

[0121] y(i)=θ1x1(i)+θ2x2(i)+…+θ n x n (i) i=1,2,…,m (7)

[0122] Where θ=(θ1,θ2,...,θ n ) is the identification parameter, and y is the output variable of the identification system. n ) is the input variable of the identification system. The least squares rule is to introduce the error variable ε=(ε1,ε2…,ε n ) T , we will find the identification value that satisfies the equation Subtracting the measured value, we can get the sum of squares of the identification parameter errors, as shown in the following formula:

[0123]

[0124] Then, matrix derivation is used to find the partial derivative of the identification parameters to obtain the value of θ that minimizes the sum of squared errors. According to the calculation principle of the least squares method, the least squares matrix for the motor fundamental frequency iron loss equivalent resistance identification is:

[0125]

[0126] Among them, R c is the equivalent resistance value of the traditional fundamental frequency iron loss equivalent circuit. According to the improved fundamental frequency iron loss equivalent circuit, the identification system is regarded as an input variable ω m , the output variable is Therefore, the matrix of the least squares of the improved fundamental frequency iron loss equivalent circuit can be expressed as:

[0127]

[0128] The fundamental frequency iron loss power is equal to the square of the terminal voltage divided by the iron loss equivalent resistance value, as shown in formula (11):

[0129]

[0130] Among them, P in is the inverter input power after LC filtering, excluding high-frequency components, and is measured by a power analyzer using the two-meter method in the experiment. out The motor output power is obtained by multiplying the motor speed by the torque. The speed is obtained by the photoelectric encoder, and the torque is obtained by the torque analyzer. Cu is the copper loss, which is calculated by the expression of copper loss, P mac is the mechanical loss, which is assumed to be constant when the speed remains unchanged. Therefore, the fundamental frequency loss value can be calculated. The power analyzer measures the inverter output power and the motor output power. The calculation formula for copper loss is shown in formula (12):

[0131]

[0132] Among them U AC is the AC two-phase line voltage, U BC is the BC two-phase line voltage, I A is the phase current of phase A, I B is the phase current of phase B, N is the motor speed, T o is the motor load torque, i d is the stator d-axis current, i q is the stator q-axis current.

[0133] Because there are two iron loss equivalent resistance parameters to be represented in the high-frequency iron loss model, a two-dimensional least squares method is used for identification. Therefore, the least squares model based on the high-frequency iron loss model is as follows:

[0134]

[0135] in, It is the square of the harmonic voltage amplitude at 1 times the equivalent switching frequency. It is the square of the harmonic voltage amplitude at 2 times the equivalent switching frequency. Experimental separation of high-frequency loss is shown in the following formula:

[0136] P loss_f =P in -P LC_in (14)

[0137] Among them, at the same speed and torque, P in is the inverter output power without LC filter filtering, P LC_in is the inverter output power after LC filter filtering, P loss_f It is the difference between the two under the same working conditions, which is the high-frequency loss.

[0138] The cascade multi-level modulation technology of the present invention is analyzed in step 3.4. Taking a two-unit cascade H-bridge as an example, the phase configuration of the carrier phase shift modulation strategy is as follows: Figure 7 As shown. Each unit output voltage u o1 、u o2 The superposition forms the phase voltage u x There is a 180° phase difference between the carrier waves of the units. For an n-module cascade system, the carrier phase is distributed according to the mathematical relationship of 360° / n. When the modulation wave m is at any position, it will be compared with the carrier wave of each unit. Therefore, each H-bridge unit will output a PWM wave at the same time, and the number of switching tube actions will also increase accordingly, resulting in greater switching losses. However, the equivalent switching frequency of the inverter is n times the carrier frequency. Therefore, under carrier phase-shift modulation, the output voltage harmonics are mainly concentrated near 1 times the equivalent switching frequency and 2 times the equivalent switching frequency. Assume that there are two continuous time variables and y = ω0t, where ω c and are the frequency and phase of the triangular carrier respectively, and ω0 is the frequency of the modulating wave. The cascaded multi-level output voltage is analyzed using a dual Fourier series expansion with two variables, as shown in Equation (15):

[0139]

[0140] Among them, m and n represent the frequency multiples of the modulation wave and the carrier wave respectively, so the harmonic frequency can be expressed as mω c +nω0. A mn and B mn is the coefficient of the double Fourier series, and the size of the coefficient can be calculated by the following formula.

[0141]

[0142] The intersection point of the modulated wave and the carrier wave is the upper and lower limits of the double Fourier series calculation, even when the switch tube is actuated under the natural sampling method. Therefore, the harmonic spectrum of the output voltage under carrier phase shift modulation is as follows (17):

[0143]

[0144] In step 3.5 of analyzing the dq-axis converter loss model of the present invention, based on cascaded multi-level carrier phase-shift modulation and according to the converter's conduction loss and switching loss generation mechanisms, the converter loss model proposed by the present invention is shown as follows:

[0145]

[0146] Where V on ,R on ,e on ,e off ,e rec All of them are standard experimental values ​​in the device manual. In the above formula, the current parameter I m is the three-phase current amplitude, f c is the switching frequency, V dc is the DC side voltage of the H-bridge unit. According to the Park amplitude invariant matrix, the current parameters can be converted into current components under the dq axis.

[0147] A calculation example of step 3.6 of establishing a wide-band loss model based on identifying resistance parameters according to the present invention is as follows. The general form of the wide-band loss model is shown below:

[0148]

[0149] Among them, P motorloss Represents the motor's broadband loss R s Represents the stator resistance, R c Represents the fundamental frequency iron loss equivalent resistance, i cd ,i cq Represents the fundamental frequency iron loss resistance branch current under dq axis, R f1， R f2 Represents the equivalent resistance of iron loss at 1 times the equivalent switching frequency and 2 times the equivalent switching frequency, u h1 ,u h2 Represents the harmonic voltage amplitude of 1 times the equivalent switching frequency and 2 times the equivalent switching frequency. Combined with the above converter loss, the motor system integrated loss is obtained.

[0150] A calculation example of applying the golden section algorithm step 3.7 based on the motor system loss model of the present invention is as follows. The wide-band loss optimization model is shown as follows:

[0151]

[0152] The optimization process of applying the golden section algorithm based on the wide-frequency domain loss of the motor system is as follows:

[0153] 1) Initialize optimization parameters: define the search interval boundary [a, b] and the convergence threshold ε, and set the golden section coefficients F1 = 0.382 and F2 = 0.618;

[0154] 2) First iteration (k=1), set the interval boundaries: i dmin (1) = a, i dmax (1) = b, calculate the interval length L1 = i dmax (1)-i dmin (1), determine the right trial point x r =i dmin (1)+F2 L1, change x r Substitute the loss model into formula (5.7) to obtain the total controllable loss P1(x r );

[0155] 3) Second iteration (k=2), inheriting the previous interval boundary: i dmin (2)=i dmin (1), idmax(2) = idmax(1), calculate the updated interval length L2 = i dmax (2)-i dmin (2) Determine the left trial point x1 = i dmin (2)+F1L2, substitute x1 into the loss model to obtain P2(x1);

[0156] 4) Loss performance comparison, if P2(x1) <P1(x r ), then discard the right subinterval and execute step 5); if P2(x1)>P1(x r ), then discard the left subinterval and execute step 6);

[0157] 5) During the right interval contraction process, the iteration counter increases: k←k+1, and the interval boundary is updated: i dmin (k)=i dmin (k-1), i dmax (k) = x r (k-1), calculate the new interval length L k , relocate the trial point: x r (k) = x1(k-1), x1(k) = i dmin (k)+F1Lk , if |x r (k)-x1(k)|<ε, go ​​to step 7), otherwise calculate P k (x1), return to step 4);

[0158] 6) Left interval contraction process: Iteration counter increment: k←k+1 Update interval boundary: i dmax (k)=i dmax (k-1), i dmin (k) = x1(k-1) to calculate the new interval length L k Then reposition the trial point: x1(k)=x r (k-1), x r (k)=i dmin (k)+F2L k If |x r (k)-x1(k)|<ε, go ​​to step 7), otherwise calculate P k (x r ), return to step 4);

[0159] 7) Take the final control current i d *=(x1(k)+x r (k)) / 2 is used as the current loop reference value and the optimization process is terminated.

[0160] The motor equivalent circuit diagram of the permanent magnet synchronous motor based on the motor wide frequency domain loss of the present invention is as follows: Figure 5 As shown. Including stator winding resistance 5.1, q-axis to d-axis coupling potential 5.2, d-axis voltage 5.3 under inverter power supply, motor wide-band iron loss equivalent resistance 5.4, d-axis inductance 5.5, d-axis to q-axis coupling potential 5.6, q-axis voltage 5.7 under inverter power supply, q-axis inductance 5.8, Figure 1 and Figure 2 The physical meaning of the iron loss resistance in the equivalent circuit shown is significantly different, and its representation of the wide-band iron loss makes the modeling more accurate.

[0161] The topology diagram of the permanent magnet synchronous motor modeling experimental device for the wide frequency domain loss of the motor described in the present invention is as follows: Figure 6 As shown, it includes a cascaded multilevel converter 6.1, an LC filter 6.2, a single-pole double-throw switch 6.3, a power analyzer 6.4, and a permanent magnet synchronous motor 6.5. Figure 6The experiment shows how the motor's fundamental frequency iron loss and high-frequency loss are separated. The motor input power is obtained through a power analyzer, while the LC filter removes the high-frequency component of the supply voltage. A single-pole double-throw switch is used for a control experiment. One experiment includes the motor's high-frequency loss test, while the other does not. The motor's speed is obtained from the motor encoder, and the torque is obtained from the torque analyzer. The motor's output power is calculated from this, and the motor's fundamental frequency loss and high-frequency loss are derived from this.

[0162] The double Fourier series analytical calculation verification diagram of voltage harmonics based on cascaded multi-level carrier phase shift modulation of the present invention is shown in the figure below: Figure 8 As shown, including the formula to calculate the voltage waveform and its Fourier decomposition spectrum and the simulated voltage waveform and the corresponding Fourier decomposition diagram. Figure 8 It can be seen that the spectra of the two are consistent, and the amplitudes at the corresponding frequencies are also the same, which verifies the accuracy of the double Fourier analysis calculation.

[0163] The present invention describes the use of MATLAB based on offline experimental data to identify the small resistance least squares method of the iron loss of the motor wide frequency domain loss, such as Figure 9 As shown in the figure above, the least squares method is used to identify the motor's fundamental frequency iron loss. The intercept represents the mechanical loss under fixed operating conditions, and the slope, based on the least squares matrix, represents the inverse of the equivalent resistance. The figure below shows the least squares method for identifying the motor's high-frequency iron loss. Because there are two parameters to be identified in the high-frequency iron loss model, a surface model was fitted. The error between the fitted surface and the measured data is within 10%.

[0164] The present invention uses the result of cross experiment verification as shown in Figure 10 and Figure 11 As shown, Figure 10 The accuracy of the motor fundamental frequency iron loss model is verified by cross-experiments under different speed and torque conditions, including absolute error diagrams and relative error diagrams; Figure 11 The accuracy of the high-frequency iron loss model of the motor was verified by cross-experiments under different speed and torque conditions, including absolute and relative error plots. In the absolute error plot of fundamental frequency loss, the experimental loss sampling points closely match the loss curve fitted with the equivalent resistance. The relative error plot also shows that the prediction error is kept below 5%. In the absolute error plot of high-frequency loss, the experimental loss sampling points are also largely on the fitted curve, and the actual prediction error is within 10%, which is very accurate compared to other methods.

[0165] The loss test results of the cascaded multi-level converter described in the present invention are as follows: Figure 12 and Figure 13 As shown, Figure 12 This is a comparison chart of the converter loss model prediction value and experimental value at different switching frequencies and speeds, including absolute error chart and relative error chart; Figure 13 This is a comparison chart of the converter loss model prediction value and experimental value under different switching frequencies and load torques, including absolute error chart and relative error chart.

[0166] The block diagram of the loss optimization control of the permanent magnet synchronous motor system based on cascade multi-level power supply of the present invention is as follows: Figure 14 As shown, 14.1 is the golden section algorithm, 14.2 is the speed closed-loop PI regulator, 14.3 is the three-phase two-level inverter, 14.4 is the three-phase permanent magnet synchronous motor, 14.5 is the coordinate transformation module, 14.6 is the wide-frequency domain loss model of the cascaded multi-level power supply permanent magnet synchronous motor system, 14.7 is the position encoder, and 14.8 is the speed calculation module.

[0167] The flowchart of the golden section algorithm based on the loss model of the cascaded multi-level power supply permanent magnet synchronous motor system of the present invention is as follows: Figure 15 shown.

[0168] The current trajectory and the motor system loss line graph of the motor system loss optimization using the golden section algorithm and the traditional fundamental frequency loss optimization based on the cascaded multi-level power supply permanent magnet synchronous motor system loss model of the present invention are as follows: Figure 16 shown.

[0169] The parameters of the permanent magnet motor in the experiment are as follows: the number of pole pairs is 10, the q-axis inductance is 12.773mH, the d-axis inductance is 8.268mH, the permanent magnet flux is 0.10397Wb, the stator resistance is 0.56Ω, the sampling frequency is 5kHz, and the switching frequency is 500Hz. Figure 8 and Figure 9 The experimental results show that the permanent magnet synchronous motor modeling method based on the wide-frequency domain loss of the cascaded multi-level motor proposed in this patent has higher accuracy than the conventional modeling method. Figure 13 It can be seen that the motor system loss is optimized at 300rpm. Under different load conditions, the motor system loss is smaller than that under the fundamental frequency loss optimization control, verifying the effectiveness of the golden section algorithm optimization control.

[0170] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0171] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0172] These computer program instructions may 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 produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0173] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0174] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0175] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for optimizing the loss control of a cascaded multi-level permanent magnet synchronous motor drive system, characterized in that The steps include: Step 1: Establish a circuit model of equivalent resistance that takes into account motor losses. Using linear superposition, consider the impact of harmonic voltage on the high-frequency losses of the motor under converter power supply and construct a wide-frequency domain loss equivalent circuit for the motor. Step 2: Based on the cascaded multi-level carrier phase-shift modulation strategy, the double Fourier series expansion is used to calculate the harmonic voltage amplitude of each frequency and construct the dq axis converter loss model according to the modulation strategy; Step 3: Based on the motor wide-band loss motor equivalent circuit of step 1 and the calculation formula of each frequency harmonic voltage amplitude of step 2, the fundamental frequency loss expression and the high-frequency loss expression are obtained; Step 4: Calculate the offline data of the motor's fundamental frequency loss and high-frequency loss based on the motor's input power and output power before and after adding LC filtering under multi-speed and multi-load conditions; Step 5: Based on the offline loss data and the harmonic voltage amplitude in step 2, the resistance value of the equivalent loss resistance of the motor in the wide frequency domain is identified using the least squares method; Step 6: Separate the hysteresis component and the eddy current component from the fundamental frequency iron loss equivalent resistance in the equivalent loss resistance, and separate the resistance values ​​of each frequency from the high frequency iron loss resistance in the equivalent loss resistance, thereby obtaining a motor loss model that considers a wide frequency domain; Step 7: Build an integrated loss model for the motor system and apply the golden section algorithm to optimize the d-axis current, a common variable of the dq-axis converter loss model in step 2 and the motor model considering wide-band loss in step 1, for control.

2. The method according to claim 1, wherein: In step 1, the voltage model of the equivalent resistance considering the fundamental frequency loss of the motor is established as follows: Among them, u d and u q are the d-axis and q-axis stator voltages, i d and i q are the d-axis and q-axis stator currents, L d and L q are the d-axis and q-axis inductances, R s is the stator resistance, ψ f is the permanent magnet flux linkage, ω is the electrical angular frequency, R i is the equivalent iron loss resistance.

3. The method according to claim 1, wherein: In step 1, considering the high-frequency voltage and its generated high-frequency response, the linear superposition principle is applied to construct the motor wide-frequency domain loss voltage model as follows: Among them, u df and u qf ,u dh and u qh are the dq axis fundamental frequency voltage and dq axis high frequency voltage respectively, i df and i qf ,i dh and i qh are the dq axis fundamental frequency current and dq axis high frequency current respectively, and p is the derivative symbol; The high-frequency iron loss equivalent voltage model is as follows: Among them, R ch It is the equivalent resistance value of high-frequency iron loss under the action of high-frequency voltage.

4. The method according to claim 1, wherein: In step 2, the amplitude of each frequency harmonic voltage under carrier phase shift modulation is calculated using double Fourier series expansion. Among them, u0 is the output voltage; x is the phase of the triangular carrier wave, y is the phase of the modulation wave, A mn ,B mn Represents the coefficients of a double Fourier series.

5. The method according to claim 1, wherein: In step 2, the converter loss model is as follows: Among them, V on ,R on ,e on ,e off ,e rec All are standard experimental values ​​in the device manual; current parameter I m is the three-phase current amplitude, f c is the switching frequency, V dc is the DC side voltage of the H-bridge unit.

6. The method according to claim 1, wherein: In step 4, the fundamental frequency loss expression and the high frequency loss expression are: P loss_f =P in_LC -P Cu -P out P loss_h =P in -P in_LC Among them, P loss_f is the fundamental frequency loss, P loss_h is the high frequency loss; P in is the converter input power without LC filtering; P in_LC is the converter input power after LC filtering, excluding high-frequency components; P out is the motor output power; P Cu Copper loss.

7. The method according to claim 1, wherein: In step 5, the least squares method is used to identify the parameters of the iron loss resistance using offline data. J=(y-Xθ) T (y-Xθ)=y T y-θ T X T yyy T Xθ+θ T X T Xth Where θ=(θ1,θ2,...,θ n ) is the identification parameter, y is the output variable of the identification system, X=(x1,x2,...,x n ) is the identification system input variable; The calculation method of the equivalent resistance considering the motor fundamental frequency loss is obtained through the motor fundamental frequency loss model. Among them, P loss_f is the motor fundamental frequency loss, R c is the equivalent resistance value of the traditional fundamental frequency iron loss equivalent circuit, u o is the motor terminal voltage; The calculation method of the equivalent resistance considering the high-frequency loss of the motor is obtained through the harmonic voltage of the double Fourier series expansion and the high-frequency loss model of the motor. in, It is the square of the harmonic voltage amplitude at 1 times the equivalent switching frequency. It is the square of the harmonic voltage amplitude at twice the equivalent switching frequency; R f1 is the equivalent resistance at 1 times the equivalent switching frequency, R f2 It is the equivalent resistance at twice the equivalent switching frequency.

8. The method according to claim 1, wherein: The specific process of step 6 is based on the separation expression of hysteresis loss resistance and eddy current loss resistance: Among them, R c is the traditional fundamental frequency iron loss equivalent resistance, R e is the eddy current loss equivalent resistance, R h is the hysteresis loss equivalent resistance, ω m is the motor mechanical angular frequency.

9. The method according to claim 1, wherein: In step 7, the motor system integrated loss model is constructed as follows: P L =minP Cu +P Fe,f (i od )+P Fe,h (i od )+P inv (i od ,f sw ) Among them, P L is the system loss; P Cu is the copper loss, P Fe,f is the motor fundamental frequency loss, P Fe,h is the high-frequency loss of the motor, P inv is the converter loss, i od is the d-axis magnetizing current, f sw is the switching frequency.

10. The method according to claim 1, wherein: In step 7, a golden section algorithm based on interval compression is applied, and the golden section ratio is used to search for a global optimal solution.

Images