A model predictive current control method for three-level inverter permanent magnet synchronous motor

Abstract

This invention relates to a model predictive current control method for a three-level inverter permanent magnet synchronous motor, comprising: establishing a new voltage space vector table; obtaining the electrical angle θ(k) and electrical angular velocity ω(k) of the permanent magnet synchronous motor at time k, and obtaining the stator current dq-axis component i. d (k) and i q (k); obtain the q-axis reference current i q * Given the d-axis reference current i d * =0; This yields the voltage space vector values ​​u along the α and β axes. α (k) and u β (k); Determine the sector where the voltage space vector is located, select the optimal voltage space vector, and then output the switching sequence through the switching table. This invention effectively suppresses common-mode voltage by discarding voltage space vectors with large common-mode voltages; by constructing a virtual voltage space vector, it effectively suppresses midpoint potential fluctuations, predicts the reference voltage space vector in the next cycle, reduces the selection range of candidate voltage space vectors, and greatly reduces the amount of computation.

Description

Technical Field

[0001] This invention relates to the field of motor drive and control technology, and in particular to a model predictive current control method for a three-level inverter permanent magnet synchronous motor. Background Technology

[0002] Inverters play a crucial role in PMSM speed control systems, directly affecting the control performance of the motor system. Traditional two-level inverters are widely used in various applications due to their simple topology and mature control methods. However, with the development of high-voltage, high-power devices, two-level inverters are increasingly unable to meet the requirements of control systems. Compared to two-level inverters, three-level inverters have lower switching losses, higher waveform output quality, and more flexible and versatile control, making them suitable for applications involving high-power motors.

[0003] However, NPC-type three-level inverters suffer from neutral point imbalance, which directly affects the output performance of the inverter system. Neutral point potential imbalance not only increases the harmonic content of the output voltage waveform but also increases the voltage across the switching transistors in the inverter, reducing the device's lifespan and potentially damaging switching devices and DC capacitors. In the inverter circuit, a common-mode voltage is generated on the output side. While this common-mode voltage does not affect the inverter's output characteristics, its high-frequency oscillations cause heat loss in the motor, consuming some input power; it also induces a high shaft voltage on the motor shaft, forming bearing current and affecting the bearing's mechanical lifespan; and the high-frequency common-mode voltage and its rate of change generate common-mode coupling current, affecting the motor's insulation performance and accelerating insulation aging.

[0004] Model predictive current control (MPCC) has attracted widespread attention from researchers due to its fast response speed, multi-objective optimization, and simple principle. Traditional MPCC requires rolling optimization of all voltage vectors, which includes many "redundant vectors," making the calculation process complex. From the perspective of voltage amplitude, single-vector MPCC selects the optimal voltage vector from 19 basic voltage vectors and applies it throughout the control cycle. However, it only applies one voltage vector within the sampling period, resulting in poor steady-state performance. Dual-vector MPCC uses two vectors to apply throughout the control cycle, and although its output current waveform is better than that of single-vector MPCC, it requires online calculation of the duty cycle, adding a burden to the processor. Summary of the Invention

[0005] To address the issues of poor steady-state performance of traditional single-vector MPCC, long computation time of dual-vector MPCC, and the need to add common-mode voltage and midpoint voltage value functions to suppress common-mode voltage and midpoint voltage fluctuations in NPC inverters, the present invention aims to provide a model predictive current control method for three-level inverter permanent magnet synchronous motors that can effectively suppress midpoint voltage fluctuations and common-mode voltage, improve the output waveform, and greatly reduce computational load.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a model predictive current control method for a three-level inverter permanent magnet synchronous motor, the method comprising the following sequential steps:

[0007] (1) Establish a new voltage space vector table: First, establish the original voltage space vector map and divide it into sectors. In the original voltage space vector, discard the voltage space vector with a large common-mode voltage and construct a virtual voltage space vector with zero midpoint current to replace the original voltage space vector and expand the voltage space vector.

[0008] (2) Obtain the electrical angle θ(k) and electrical angular velocity ω(k) of the permanent magnet synchronous motor at time k, and convert the three-phase stator current , , The dq-axis components of the stator current are obtained through Clark and Park transforms. and ;

[0009] (3) Set the reference rotational speed ω * The difference between the current speed ω of the permanent magnet synchronous motor and the current speed ω is input to the outer loop speed PI controller to obtain the q-axis reference current. * Given the d-axis reference current * =0;

[0010] (4) Calculate the voltage space vector values ​​of the d and q axes using the deadbeat current predictive control method. and By using the inverse Park transformation, the voltage space vector values ​​along the α and β axes are obtained. and ;

[0011] (5) Determine the sector where the voltage space vector is located. The voltage space vector in the sector interval is selected by value function rolling optimization to select the optimal voltage space vector, and then the switching sequence is output through the switching table.

[0012] Step (1) specifically includes the following steps:

[0013] (1a) Establish the original voltage space vector diagram:

[0014] The voltage space vector synthesized from the three-phase voltages in the stator is:

[0015] U=u a +u b +u c (1)

[0016] In the formula, j is the imaginary part, and U is the voltage space vector. These are the three-phase phase voltages;

[0017] Given the DC side voltage of an NPC-type three-level inverter Then the output voltage of the A, B, and C phase arms of the NPC-type three-level inverter is / 2 or 0 or - / 2, the switching states of the three-phase bridge arms are denoted as P, O, and N respectively; using This indicates the switching state of each phase arm, and the voltage of each phase is expressed as:

[0018] (2)

[0019] In the formula, ={-1, 0, 1}, x=a, b, c;

[0020] The calculation shows that there are a total of 3×3×3=27 switching states for the three-phase output. Substituting equation (2) into equation (1) yields:

[0021] U= [(2S a -S b -S c )+ (3)

[0022] Substituting the 27 switching states into equation (3), the direction and amplitude of the corresponding voltage space vector in the stationary coordinate system are calculated to obtain the voltage space vector diagram. The voltage space vector of the voltage space vector diagram is denoted as Vi, i=0,1,...,26. According to the different amplitudes, the voltage space vectors are divided into four categories: zero vector, whose amplitude is zero; small vector, whose amplitude is... / 3; the medium vector, whose magnitude is / 3; a large vector with an amplitude of 2. / 3;

[0023] (1b) Sector division: The sector is determined by the angle θ between the voltage space vector U and the α axis, and is denoted as follows: First sector, 0≤θ<60°; Second sector, 60°≤θ<120°; Third sector, 120°≤θ<180°; Fourth sector, 180°≤θ<240°; Fifth sector, 240°≤θ<300°; Sixth sector, 300°≤θ<360°.

[0024] (1c) Calculate the common-mode voltage The calculation formula is as follows:

[0025] (4)

[0026] The original voltage space vector is divided into | based on the magnitude of the common-mode voltage. |= / 2,| |= / 3,| |= / 6,| |=0 (four categories), discarding the absolute value of the common-mode voltage. / 2 and / 3 of the voltage space vector, the remaining 19 basic voltage space vectors, for the first major sector, the remaining voltage space vectors are V0 [OOO], V1 [PN N], V2 [POO], V3 [PON], V4 [OON], V5 [PPN];

[0027] (1d) Synthesizing the virtual voltage space vector:

[0028] NPC three-level inverter midpoint current i np The instantaneous value is expressed as:

[0029] i np = (5)

[0030] when When =1 and -1, the midpoint current i np =0; when When = 0, the midpoint current i np Not equal to 0;

[0031] Based on the analysis of the flow direction of the midpoint current when each voltage space vector is applied, it is found that the zero vector and the large vector affect the midpoint potential of the inverter, while the small vector and the medium vector generate the midpoint current. Analyzing the voltage vector of the first large sector, the corresponding midpoint currents are V0, V1, V5=0, V2=-ia, V3=ib, V4=-ic.

[0032] A virtual voltage space vector with zero midpoint current is constructed to replace the original voltage space vector with non-zero midpoint current. For the first major sector, V[PNO] and V[OON] are combined to replace V[POO] as V2, V[PNN] and V[PPN] are combined to replace V[PON] as V3, and V[POO] and V[ONP] are combined to replace V[OON] as V4. It is necessary to ensure that the amplitude and angle of the voltage space vector before the combination are the same as those of the new voltage space vector after the combination. The relationships are V[POO] = 1 / 2 * (V[PNO] + V[OON]), V[PON] = 1 / 2 * (V[PNN] + V[PPN]), and V[OON] = 1 / 2 * (V[POO] and V[ONP]).

[0033] (1e) Expand the voltage space vector and synthesize a new voltage space vector using a constant duty cycle method. The relationship between them satisfies the following formula:

[0034] Vx = a * Vy + b * Vz

[0035] Where Vx is the synthesized voltage space vector, Vy and Vz are the original voltage space vectors, and a+b=1;

[0036] For the first major sector, the composite voltage space vector is V19 = 1 / 4 * (V5) + 3 / 4 * (V2), V20 = 1 / 4 * (V4) + 3 / 4 * (V1), V21 = 1 / 3 * (V[OPN] + V[PON] + V[PNO]), V22 = 1 / 4 * (V1) + 3 / 4 * (V4), V23 = 1 / 4 * (V2) + 3 / 4 * (V5);

[0037] (1f) Establish a new voltage space vector table. For the first sector, the voltage space vectors it contains are {V0, V1, V2, V3, V4, V5, V19, V20, V21, V22, V23}.

[0038] Step (4) specifically includes the following steps:

[0039] (4a) Using the deadbeat current predictive control method, the voltage space vector values ​​of the d and q axes are calculated according to the following formula. and :

[0040] (6)

[0041] (7)

[0042] In the formula, Sampling time, For stator resistance, , They are respectively Reference current values ​​of the stator current along the d and q axes at any given time. , They are respectively The d-axis and q-axis reference current values ​​at time t. , for The d and q components of the stator current at any given time. , They are respectively The d-axis and q-axis components of the stator voltage obtained through prediction calculation at each moment. , These are the d-axis and q-axis inductances of the motor stator windings, respectively. yes The electric angular velocity of the motor at any given time, For permanent magnet flux linkage;

[0043] (4b) According to the inverse Park transformation formula, the voltage space vector values ​​of the d and q axes and Voltage space vector values ​​converted to α and β axes and .

[0044] Step (5) specifically includes the following steps:

[0045] (5a) Determine the sector where the voltage space vector U is located by comparing its relationship with the sector boundary:

[0046] |U|= (8)

[0047] In the formula, U is the voltage space vector. , For voltage space vector in , The components of the axis;

[0048] when >0 and - At that time, U was in the first major sector.

[0049] when >0 and - At that time, U was in the second largest sector.

[0050] when >0 and + At that time, U was in the third largest sector.

[0051] when <0 and - At that time, U was in the fourth sector.

[0052] when <0 and + At that time, U was in the fifth sector.

[0053] when <0 and + At that time, U was in the sixth sector.

[0054] (5b) Perform rolling optimization of the voltage space vector in the interval according to the following value function:

[0055] g j =( - j ) 2 +( - j ) 2 (9)

[0056] In the formula, , These are the voltage space vector values ​​along the α and β axes. j , j Let g be the α and β axis components of the j-th voltage space vector within the sector. For the first large sector, g min ={g0, g1, g2, g3, g4, g5, g 19 g 20 g 21 g 22 g 23}, g min It is the minimum value in the set;

[0057] (5c) According to g min The optimal voltage space vector for this interval is obtained, and the corresponding switching sequence is output by looking up the switching table. 0 represents the switch being off, and 1 represents the switch being on. The specific actions of the four switches in the A, B, and C arms of the inverter, corresponding to the switching states P, O, and N, are as follows:

[0058] P: [1 1 0 0]

[0059] O: [0 1 1 0]

[0060] N: [0 0 1 1].

[0061] As can be seen from the above technical solution, the beneficial effects of the present invention are as follows: First, the present invention effectively suppresses common-mode voltage by discarding voltage space vectors with large common-mode voltages based on the original voltage space vector; Second, the present invention effectively suppresses midpoint potential fluctuations by constructing a virtual voltage space vector and replacing voltage space vectors with non-zero midpoint currents. It predicts the reference voltage space vector in the next cycle through a deadbeat current prediction control method and reduces the selection range of candidate voltage space vectors by sector judgment, greatly reducing the amount of computation; Third, the present invention uses a fixed duty cycle method to synthesize a new voltage space vector to expand the voltage space vector and improve the coverage of candidate vectors. Compared with dual-vector MPCC, the duty cycle is directly given, and online calculation is not required, which greatly reduces the amount of complex calculation. The vector coverage of the extended vector MPCC is larger, and it can achieve more accurate control effects than single-vector MPCC. Attached Figure Description

[0062] Figure 1 This is a block diagram illustrating the current control principle of the present invention;

[0063] Figure 2 This is a vector diagram of the original voltage space of the NPC-type three-level inverter of the present invention;

[0064] Figure 3 , 4 All of these are voltage space vector diagrams of the first major sector newly established in this invention;

[0065] Figure 5 , 6 Figures 7 and 8 are simulation diagrams of predicted current control for the three-level inverter permanent magnet synchronous motor model provided by this invention. Detailed Implementation

[0066] A model-predictive current control method for a three-level inverter permanent magnet synchronous motor, comprising the following sequential steps:

[0067] (1) Establish a new voltage space vector table: First, establish the original voltage space vector diagram and divide it into sectors. In the original voltage space vector, discard the voltage space vector with a large common-mode voltage and construct a virtual voltage space vector with zero midpoint current to replace the original voltage space vector, and expand the voltage space vector.

[0068] (2) Obtain the electrical angle θ(k) and electrical angular velocity ω(k) of the permanent magnet synchronous motor at time k, and convert the three-phase stator current , , The dq-axis components of the stator current are obtained through Clark and Park transforms. and ;

[0069] (3) Set the reference rotational speed ω * The difference between the current speed ω of the permanent magnet synchronous motor and the current speed ω is input to the outer loop speed PI controller to obtain the q-axis reference current. * Given the d-axis reference current * =0;

[0070] (4) Calculate the voltage space vector values ​​of the d and q axes using the deadbeat current predictive control method. and By using the inverse Park transformation, the voltage space vector values ​​along the α and β axes are obtained. and ;

[0071] (5) Determine the sector where the voltage space vector is located. The voltage space vector in the sector interval is selected by value function rolling optimization to select the optimal voltage space vector, and then the switching sequence is output through the switching table.

[0072] Step (1) specifically includes the following steps:

[0073] (1a) Establish the original voltage space vector diagram:

[0074] The voltage space vector synthesized from the three-phase voltages in the stator is:

[0075] U=u a +u b +u c (1)

[0076] In the formula, j is the imaginary part, and U is the voltage space vector. These are the three-phase phase voltages;

[0077] Given the DC side voltage of an NPC-type three-level inverter Then the output voltage of the A, B, and C phase arms of the NPC-type three-level inverter is / 2 or 0 or - / 2, the switching states of the three-phase bridge arms are denoted as P, O, and N respectively; using This indicates the switching state of each phase arm, and the voltage of each phase is expressed as:

[0078] (2)

[0079] In the formula, ={-1, 0, 1}, x=a, b, c;

[0080] The calculation shows that there are a total of 3×3×3=27 switching states for the three-phase output. Substituting equation (2) into equation (1) yields:

[0081] U= [(2S a -S b -S c )+ (3)

[0082] Substituting the 27 switching states into equation (3), the direction and magnitude of the corresponding voltage space vector in the stationary coordinate system are calculated, resulting in the voltage space vector diagram, as follows: Figure 2 As shown, the voltage space vector of the voltage space vector diagram is denoted as Vi, i=0,1,...,26. Based on different amplitudes, the voltage space vectors are divided into four categories: zero vectors, whose amplitude is zero; small vectors, whose amplitude is... / 3; the medium vector, whose magnitude is / 3; a large vector with an amplitude of 2. / 3;

[0083] (1b) Sector division: The sector is determined by the angle θ between the voltage space vector U and the α axis, and is denoted as follows: First sector, 0≤θ<60°; Second sector, 60°≤θ<120°; Third sector, 120°≤θ<180°; Fourth sector, 180°≤θ<240°; Fifth sector, 240°≤θ<300°; Sixth sector, 300°≤θ<360°.

[0084] (1c) Calculate the common-mode voltage The calculation formula is as follows:

[0085] (4)

[0086] The original voltage space vector is divided into | based on the magnitude of the common-mode voltage. |= / 2,| |= / 3,| |= / 6,| The four categories with |=0 are shown in Table 1; the absolute value of the common-mode voltage is discarded. / 2 and / 3 of the voltage space vector, the remaining 19 basic voltage space vectors, for the first major sector, the remaining voltage space vectors are V0 [O OO], V1 [PNN], V2 [POO], V3 [PON], V4 [OON], V5 [PPN], and so on for other major sectors;

[0087] Table 1

[0088]

[0089] (1d) Synthesizing the virtual voltage space vector:

[0090] NPC three-level inverter midpoint current i np The instantaneous value is expressed as:

[0091] i np = (5)

[0092] when When =1 and -1, the midpoint current i np =0; when When = 0, the midpoint current i np Not equal to 0;

[0093] Based on the analysis of the flow direction of the midpoint current when each voltage space vector is applied, it is found that the zero vector and the large vector affect the midpoint potential of the inverter, while the small vector and the medium vector generate the midpoint current. Analyzing the voltage vector of the first large sector, the corresponding midpoint currents are V0, V1, V5=0, V2=-ia, V3=ib, V4=-ic; and so on for other large sectors.

[0094] The midpoint current corresponding to the voltage vector is shown in Table 2 below:

[0095] Table 2

[0096]

[0097] A virtual voltage space vector with zero midpoint current is constructed to replace the original voltage space vector with non-zero midpoint current, such as... Figure 3 As shown, for the first major sector, V[PNO] and V[OON] are combined to replace V[POO] to become V2, V[PN N] and V[PPN] are combined to replace V[PON] to become V3, and V[POO] and V[ONP] are combined to replace V[OON] to become V4. It is necessary to ensure that the magnitude and angle of the voltage space vector before the combination are the same as those of the new voltage space vector after the combination. The relationships are V[POO] = 1 / 2 * (V[PNO] + V[OON]), V[PON] = 1 / 2 * (V[PNN] + V[PPN]), and V[OON] = 1 / 2 * (V[POO] and V[ONP]). The same logic applies to other major sectors.

[0098] (1e) Expand the voltage space vector and synthesize a new voltage space vector using a constant duty cycle method. The relationship between them satisfies the following formula:

[0099] Vx = a * Vy + b * Vz

[0100] Where Vx is the synthesized voltage space vector, Vy and Vz are the original voltage space vectors, and a+b=1;

[0101] For the first major sector, such as Figure 4 As shown, the synthesized voltage space vectors are V19 = 1 / 4 * (V5) + 3 / 4 * (V2), V20 = 1 / 4 * (V4) + 3 / 4 * (V1), V21 = 1 / 3 * (V[OPN] + V[PON] + V[PNO]), V22 = 1 / 4 * (V1) + 3 / 4 * (V4), and V23 = 1 / 4 * (V2) + 3 / 4 * (V5).

[0102] (1f) Establish a new voltage space vector table, as shown in Table 3. For the first sector, the voltage space vectors it contains are {V0, V1, V2, V3, V4, V5, V19, V20, V21, V22, V23}:

[0103] Table 3

[0104]

[0105] Step (4) specifically includes the following steps:

[0106] (4a) Using the deadbeat current predictive control method, the voltage space vector values ​​of the d and q axes are calculated according to the following formula. and :

[0107] (6)

[0108] (7)

[0109] In the formula, Sampling time, For stator resistance, , They are respectively Reference current values ​​of the stator current along the d and q axes at any given time. , They are respectively The d-axis and q-axis reference current values ​​at time t. , for The d and q components of the stator current at any given time. , They are respectively The d-axis and q-axis components of the stator voltage obtained through prediction calculation at each moment. , These are the d-axis and q-axis inductances of the motor stator windings, respectively. yes The electric angular velocity of the motor at any given time, For permanent magnet flux linkage;

[0110] (4b) According to the inverse Park transformation formula, the voltage space vector values ​​of the d and q axes and Voltage space vector values ​​converted to α and β axes and .

[0111] Step (5) specifically includes the following steps:

[0112] (5a) Determine the sector where the voltage space vector U is located by comparing its relationship with the sector boundary:

[0113] |U|= (8)

[0114] In the formula, U is the voltage space vector. , For voltage space vector in , The components of the axis;

[0115] when >0 and - At that time, U was in the first major sector.

[0116] when >0 and - At that time, U was in the second largest sector.

[0117] when >0 and + At that time, U was in the third largest sector.

[0118] when <0 and - At that time, U was in the fourth sector.

[0119] when <0 and + At that time, U was in the fifth sector.

[0120] when <0 and + At that time, U was in the sixth sector.

[0121] (5b) Perform rolling optimization of the voltage space vector in the interval according to the following value function:

[0122] g j =( - j ) 2 +( - j ) 2 (9)

[0123] In the formula, , These are the voltage space vector values ​​along the α and β axes. j , j Let g be the α and β axis components of the j-th voltage space vector within the sector. For the first large sector, g min ={g0, g1, g2, g3, g4, g5, g 19 g 20 g 21 g 22 g 23}, g min The minimum value in the set; the same applies to other large sectors;

[0124] (5c) According to g min The optimal voltage space vector for this interval is obtained, and the corresponding switching sequence is output by looking up the switch table (Table 4). 0 represents the switch being off, and 1 represents the switch being on. The specific actions of the four switches in the A, B, and C arms of the inverter, corresponding to switch states P, O, and N respectively, are as follows:

[0125] P: [1 1 0 0]

[0126] O: [0 1 1 0]

[0127] N: [0 0 1 1].

[0128] Table 4

[0129]

[0130] like Figure 1 As shown, given the motor reference speed ω * The difference between the current speed ω and the q-axis reference current is obtained through the outer loop speed PI controller. * Given the d-axis reference current * =0, the three-phase stator current of the motor , , The dq-axis components are obtained through Clark and Park transformations. and The voltage space vector values ​​of the d and q axes are calculated using the deadbeat current prediction control module. and Then, the voltage space vector values ​​along the α and β axes are obtained through inverse Park transformation. and Based on the calculated voltage space vector value and The value is used to determine the sector in which the voltage space vector is located. Candidate vectors for that interval are selected from the newly established voltage space vectors, and value function rolling optimization is performed on them to select the optimal voltage space vector. Finally, the switching sequence output by the switching table is applied to the NPC type three-level inverter to drive the permanent magnet synchronous motor.

[0131] The reference speed is set to 1000 r / min, and the DC side voltage is... The voltage is 600V, and the given load is 2N*m. Figure 5 It can be seen that the present invention can effectively suppress midpoint potential fluctuations, from Figure 6 It can be seen that the maximum common-mode voltage stabilizes at ± / 6 (±100V) and below, from Figure 7 and Figure 8 It can be seen that the present invention has good tracking performance for the speed and torque of permanent magnet synchronous motor.

[0132] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A model predictive current control method for a three-level inverter permanent magnet synchronous motor, characterized in that: The method includes the following steps in sequence: (1) Establish a new voltage space vector table: First, establish the original voltage space vector map and divide it into sectors. In the original voltage space vector, discard the voltage space vector with a large common-mode voltage and construct a virtual voltage space vector with zero midpoint current to replace the original voltage space vector and expand the voltage space vector. (2) Obtain the electrical angle θ(k) and electrical angular velocity ω(k) of the permanent magnet synchronous motor at time k, and convert the three-phase stator current , , The dq-axis components of the stator current are obtained through Clark and Park transforms. and ; (3) Set the reference rotational speed ω * The difference between the current speed ω of the permanent magnet synchronous motor and the current speed ω is input to the outer loop speed PI controller to obtain the q-axis reference current. * Given the d-axis reference current * =0; (4) Calculate the voltage space vector values ​​of the d and q axes using the deadbeat current predictive control method. and By using the inverse Park transformation, the voltage space vector values ​​along the α and β axes are obtained. and ; (5) Determine the sector where the voltage space vector is located. The voltage space vector in the sector interval is selected by rolling optimization of the value function to select the optimal voltage space vector, and then the switching sequence is output through the switching table. Step (1) specifically includes the following steps: (1a) Establish the original voltage space vector diagram: (1b) Sector division: The sector is determined by the angle θ between the voltage space vector U and the α axis, and is denoted as follows: First sector, 0≤θ<60°; Second sector, 60°≤θ<120°; Third sector, 120°≤θ<180°; Fourth sector, 180°≤θ<240°; Fifth sector, 240°≤θ<300°; Sixth sector, 300°≤θ<360°. (1c) Calculate the common-mode voltage The calculation formula is as follows: （4） The original voltage space vector is divided into | based on the magnitude of the common-mode voltage. |= / 2,| |= / 3,| |= / 6,| |=0 (four categories), discarding the absolute value of the common-mode voltage. / 2 and / 3 of the voltage space vectors, the remaining 19 basic voltage space vectors, for the first major sector, the remaining voltage space vectors are V0 [OOO], V1 [PNN], V2 [POO], V3 [PON], V4 [OON], V5 [PPN]; (1d) Synthesizing the virtual voltage space vector: NPC three-level inverter midpoint current i np The instantaneous value is expressed as: i np = （5） when When =1 and -1, the midpoint current i np =0; when When = 0, the midpoint current i np Not equal to 0; Based on the analysis of the flow direction of the midpoint current when each voltage space vector is applied, it is found that the zero vector and the large vector affect the midpoint potential of the inverter, while the small vector and the medium vector generate the midpoint current. Analyzing the voltage vector of the first large sector, the corresponding midpoint currents are V0, V1, V5=0, V2=-ia, V3=ib, V4=-ic. A virtual voltage space vector with zero midpoint current is constructed to replace the original voltage space vector with non-zero midpoint current. For the first major sector, V[PNO] and V[OON] are combined to replace V[POO] as V2, V[PNN] and V[PPN] are combined to replace V[PON] as V3, and V[POO] and V[ONP] are combined to replace V[OON] as V4. It is necessary to ensure that the amplitude and angle of the voltage space vector before combination and the new voltage space vector after combination are the same. The relationships are V[POO] = 1 / 2 * (V[PNO] + V[OON]), V[PON] = 1 / 2 * (V[PNN] + V[PPN]), and V[OON] = 1 / 2 * (V[POO] and V[ONP]). (1e) Expand the voltage space vector and synthesize a new voltage space vector using a constant duty cycle method. The relationship between them satisfies the following formula: Vx = a * Vy + b * Vz Where Vx is the synthesized voltage space vector, Vy and Vz are the original voltage space vectors, and a+b=1; For the first major sector, the composite voltage space vector is V19 = 1 / 4 * (V5) + 3 / 4 * (V2), V20 = 1 / 4 * (V4) + 3 / 4 * (V1), V21 = 1 / 3 * (V[OPN] + V[PON] + V[PNO]), V22 = 1 / 4 * (V1) + 3 / 4 * (V4), V23 = 1 / 4 * (V2) + 3 / 4 * (V5); (1f) Establish a new voltage space vector table. For the first sector, the voltage space vectors it contains are {V0, V1, V2, V3, V4, V5, V19, V20, V21, V22, V23}. Step (5) specifically includes the following steps: (5a) Determine the sector where the voltage space vector U is located by comparing its relationship with the sector boundary: |U|= （8） In the formula, U is the voltage space vector. , For voltage space vector in , The components of the axis; when >0 and - At that time, U was in the first major sector. when >0 and - At that time, U was in the second largest sector. when >0 and + At that time, U was in the third largest sector. when <0 and - At that time, U was in the fourth sector. when <0 and + At that time, U was in the fifth sector. when <0 and + At that time, U was in the sixth sector. (5b) Perform rolling optimization of the voltage space vector in the interval according to the following value function: g j =( - j ) 2 +( - j ) 2 （9） In the formula, , These are the voltage space vector values ​​along the α and β axes. j , j Let g be the α and β axis components of the j-th voltage space vector within the sector. For the first large sector, g min ={g0, g1, g2, g3, g4, g5, g 19 g 20 g 21 g 22 g 23 }, g min It is the minimum value in the set.

2. The model predictive current control method for a three-level inverter permanent magnet synchronous motor according to claim 1, characterized in that: The step (1a) specifically refers to: The voltage space vector synthesized from the three-phase voltages in the stator is: U=u a +in b +in c (1) In the formula, j is the imaginary part, and U is the voltage space vector. These are the three-phase phase voltages; Given the DC side voltage of an NPC-type three-level inverter Then the output voltage of the A, B, and C phase arms of the NPC-type three-level inverter is / 2 or 0 or - / 2, the switching states of the three-phase bridge arms are denoted as P, O, and N respectively; using This indicates the switching state of each phase arm, and the voltage of each phase is expressed as: （2） In the formula, ={-1, 0, 1}, x=a, b, c; The calculation shows that there are a total of 3×3×3=27 switching states for the three-phase output. Substituting equation (2) into equation (1) yields: U= [(2S a -S b -S c )+ ] (3) Substitute the 27 switching states into equation (3) to calculate the direction and amplitude of the corresponding voltage space vector in the stationary coordinate system, and obtain the voltage space vector diagram. The voltage space vector of the voltage space vector diagram is denoted as Vi, i=0,1,...,26. According to the different amplitudes, the voltage space vector is divided into four categories: zero vector, whose amplitude is zero; Small vector, its magnitude is / 3; the medium vector, whose magnitude is / 3; Large vector, with an amplitude of 2 / 3.

3. The model predictive current control method for a three-level inverter permanent magnet synchronous motor according to claim 1, characterized in that: Step (4) specifically includes the following steps: (4a) Using the deadbeat current predictive control method, the voltage space vector values ​​of the d and q axes are calculated according to the following formula. and : (6) (7) In the formula, Sampling time, For stator resistance, , They are respectively Reference current values ​​of the stator current along the d and q axes at any given time. , They are respectively The d-axis and q-axis reference current values ​​at time t. , for The d and q components of the stator current at any given time. , They are respectively The d-axis and q-axis components of the stator voltage obtained through prediction calculation at each moment. , These are the d-axis and q-axis inductances of the motor stator windings, respectively. yes The electric angular velocity of the motor at any given time, For permanent magnet flux linkage; (4b) Voltage space vector values ​​on the d and q axes according to the inverse Park transformation formula and Voltage space vector values ​​converted to α and β axes and .

4. The model predictive current control method for a three-level inverter permanent magnet synchronous motor according to claim 1, characterized in that: According to g min The optimal voltage space vector for this interval is obtained, and the corresponding switching sequence is output by looking up the switching table. 0 represents the switch being off, and 1 represents the switch being on. The specific actions of the four switches in the A, B, and C arms of the inverter, corresponding to the switching states P, O, and N, are as follows: P：[1 1 0 0] O：[0 1 1 0] N：[0 0 1 1]。

Images