Method for balancing the neutral point potential of a three-level t-type inverter based on mpcc-dsvm

Abstract

This invention addresses the problems of inconsistent switching frequency and unbalanced midpoint potential in traditional model predictive control (MMC) and three-level T-type inverters. It proposes a midpoint potential balancing method for three-level T-type inverters based on MPCC-DSVM. By generating virtual vectors through linear combination of real vectors, the output current ripple is effectively reduced, achieving a fixed switching frequency. The introduced specific virtual vectors eliminate low-frequency oscillations in the midpoint potential, avoiding the need for weight factor selection. The feasibility and effectiveness of this control strategy are verified using Matlab / Simulink. Simulation results show that this invention achieves a fixed switching frequency, effectively improves output current quality, enables midpoint potential balancing, and ensures the system's dynamic response performance.

Description

Technical Field

[0001] This invention belongs to the field of power electronics, and in particular relates to a method for balancing the midpoint potential of a three-level T-type inverter based on MPCC-DSVM. Background Technology

[0002] In the context of the rapid development of new energy power generation and electric vehicles, three-level inverters play a crucial role. Compared with three-phase two-level inverters, three-level inverters have advantages such as lower output voltage and current harmonics, lower voltage stress on switching power devices, and higher power density. They have been widely used in industrial applications, such as renewable power generation systems and motor drives. Their control performance directly affects the safe and reliable operation of the system. Control methods for three-level T-type inverters are currently a research hotspot. Commonly used controllers include proportional-integral (PI) control, proportional-resonant (PR) control, and deadbeat control. However, PI controllers struggle to balance steady-state and transient performance, and PR controllers exhibit poor control performance when the frequency of the output current changes. Deadbeat control is significantly affected by system parameters; its control performance deteriorates when system parameters change. Finite control set-model predictive control (FCS-MPC) has been widely applied in power electronics in recent years due to its fast dynamic response, simple control, good robustness, and ability to handle multiple control objectives and constraints. However, FCS-MPC also has many drawbacks. As the number of voltage vectors increases, it leads to a very high computational burden. Furthermore, the variable switching frequency of FCS-MPC increases the difficulty of filter design.

[0003] The variable switching frequency has always been a major challenge in model predictive control (MPC). In recent years, multi-vector model predictive current control (MPC) has been extensively studied in three-level converters. A method for minimizing tracking error is used to calculate the optimal duty cycle for each sector, and then a symmetrical seven-segment voltage vector sequence is applied to achieve a fixed switching frequency. However, selecting a vector from 24 sectors involves a large computational load, and the performance of the tracking reference cannot be guaranteed. A four-stage fixed-frequency MPC method is proposed, which determines the required sector through sector and sub-sector selection, relatively reducing the computational load; however, the four-step calculation process is complex. Existing technologies have proposed deadbeat predictive current control for T-type inverters, effectively suppressing grid-connected current distortion caused by control delay and achieving fixed-frequency control of FCS-MPC. However, this method requires high system sampling accuracy, suffers from significant overshoot during startup, and affects the reliable operation of the entire system. To address the shortcomings of traditional MPC algorithms, a novel model predictive control algorithm based on discrete space vector modulation is proposed. By introducing virtual vectors, multiple voltage vectors are output within a single switching cycle, achieving a fixed switching frequency and improving grid-connected current ripple.

[0004] Another fundamental problem with three-level inverters is neutral point voltage balance (NPVB). Generally, NPVB is achieved by adding an extra term to the cost function, and the performance of each control objective is coupled with others. Therefore, adjusting the weighting factor is a difficult problem in multi-objective function design, and there is currently no clear method to guide this. Existing technology proposes a current model predictive control algorithm for the neutral point potential fluctuation problem in T-type three-level grid-connected inverters, effectively solving the DC fluctuation phenomenon caused by load imbalance or unequal capacitance values. However, the optimization process is complex, leading to cumbersome controller calculations. Furthermore, only candidate vectors adjacent to the reference vector are selected, effectively reducing the number of candidate vectors. Secondly, suitable redundant small vectors are selected to balance their NPV, eliminating the weighting factor.

[0005] For three-level T-type inverters, an MPC method using Discrete Space Vector (DSV) has been proposed, which can reduce computation time and output current ripple. However, the DSV design does not consider its impact on NPV, and a separate objective function is required to balance the DC-side capacitor voltage. An improved FCS-MPC based on DSVM has been proposed, achieving lower output current THD and a balanced midpoint potential, but it only considers six virtual vectors, limiting overall control performance compared to traditional MPC. Redundant voltage vectors can also be used to regulate NPV oscillation without adding a DC-link capacitor voltage balancing term to the cost function, but this method is not applicable if the candidate vector selected by MPC is the midpoint vector, thus limiting the operating region of MPC to certain modulation indices. Applying virtual vectors that do not affect NPV to three-level voltage source inverters achieves DC-side capacitor voltage balance while avoiding weight factor adjustment, but using multiple virtual vectors increases switching losses. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention designs a method for balancing the midpoint potential of a three-level T-type inverter based on MPCC-DSVM.

[0007] The method for balancing the neutral point potential of a three-level T-type inverter based on MPCC-DSVM specifically includes the following steps:

[0008] Step 1: Model the three-level T-type inverter topology system according to Kirchhoff's voltage law, then perform coordinate system transformation and discretization on the modeling results, and finally write the objective function equation of FCS-MPC according to the principle of FCS-MPC.

[0009] Step 1.1: The differential form of the output voltage of the three-level T-type inverter in the abc coordinate system is as follows:

[0010]

[0011] Where V xn (x = a, b, c) represents the voltage of phase x relative to the neutral point, V xO (x = a, b, c) represents the voltage at the x-phase output point relative to the DC side midpoint O; V On It is the voltage at point O on the DC side relative to the neutral point n; i ox (x = a, b, c) represents the load current of phase x; L is the load inductance;

[0012] Step 1.2: Transform Equation 1 from the abc coordinate system to the αβ coordinate system to obtain:

[0013]

[0014] Among them, V α and V β i represents the output voltage of the inverter in the α and β coordinate systems. oα and i oβ These are the inverter's output currents in the α and β coordinate systems, respectively; R is the load resistance.

[0015] Step 1.3: Each arm of the inverter generates three different output voltages, which are VDC voltages relative to the DC-side midpoint potential of the inverter topology. dc / 2, 0 and -V dc / 2, the inverter generates 27 switching states, where "1", "0" and "-1" correspond to the three voltage levels output by the inverter, i.e., V. dc / 2, 0, -V dc / 2; Set the switching function T i Defined as:

[0016]

[0017] Where T i1 and T i4 T represents the upper and lower switching transistors of a T-type inverter. i2 and T i3 Represents the two middle switching transistors;

[0018] The output voltage of the inverter is then expressed as:

[0019]

[0020] Where T a T b T c These represent the switching states of phases a, b, and c of the T-type inverter, respectively; V dc It is the DC side voltage.

[0021] Step 1.4: Predict the current value i at time (k+1) by applying the forward Euler formula to Equation 2. oα (k+1) and i oβ (k+1) are respectively:

[0022]

[0023] Where i oα (k) and i oβ (k) represents the output current of the inverter at time k in the αβ coordinate system, T S The sampling period;

[0024] The current flowing through the DC-side capacitor of a three-level T-type inverter topology is expressed as:

[0025]

[0026] Where i c1 and i c2 These represent the current flowing through the upper capacitor and the lower capacitor on the DC side, respectively; V up It is the voltage across the capacitor, V low C is the voltage across the capacitor; C is the DC-side voltage regulator capacitor value.

[0027] Step 1.5: Applying the forward Euler formula to Equation 6, the predicted values ​​of the DC-side upper and lower capacitor voltages are:

[0028]

[0029] The current flowing through the upper and lower capacitors at time k can be expressed as:

[0030]

[0031] Where i dc (k) represents the DC current at time k, i c,a (k), i c,b (k), i c,c (k) represents the branch current flowing through the three-phase bridge arms a, b, and c at time k;

[0032] If T i ="1",H 1x =1, otherwise H 1x =0

[0033] If T i ="-1",H 2x =1, otherwise H 2x =0.

[0034] x = {a, b, c}.

[0035] To track the reference current while balancing the voltage across the capacitor, the objective function of the traditional FCS-MPC is defined as:

[0036]

[0037] i oα (k+1) and i oβ (k+1) represent the predicted current value at time k+1 in the αβ coordinate system; ΔV dc (k+1) is the difference in voltage between the upper and lower capacitors on the DC side at time k+1, expressed as V. up (k+1)-V low (k+1), λ np It is a weighting factor; and Let be the reference current values ​​of the inverter at time k+1 in the αβ coordinate system, which are calculated using the third-order Lagrange extrapolation equation as follows:

[0038]

[0039] in and These are the reference current values ​​of the inverter at time k in the αβ coordinate system; and These are the reference current values ​​of the inverter at time k-1; and These are the reference current values ​​of the inverter at time k-2; and These are the reference current values ​​of the inverter at time k-3.

[0040] Step 2: Based on the discretized model of the predicted current and the objective function of FCS-MPC obtained in Step 1, vector synthesis and MPCC-DSVM strategy are carried out according to the construction principle of virtual vector. Finally, the optimal switching state at the next moment is determined by the objective function without weight factors.

[0041] Step 2.1: Using MPC combined with DSVM, assuming the sampling period is divided into N equal parts, then the virtual vector V vir From real vector Synthesis, represented as follows:

[0042]

[0043] t1+t2+...+t N =T S (12)

[0044]

[0045] The new virtual vector needs to satisfy the condition that the sum of the currents flowing into and out of the midpoint of the DC side in the inverter topology is zero within one sampling period;

[0046] All of the causes i O Non-zero original vectors are replaced by equivalent virtual vectors with different magnitudes and angles, but the resulting i O Zero;

[0047] Step 2.2: In the new spatial voltage vector diagram, analyze the synthesis principle of each constructed virtual vector and its impact on the voltage balance of the DC-side capacitor;

[0048] Considering the principle of achieving a fixed switching frequency and minimizing the number of switching operations, only (0,0,0) is considered when selecting the zero vector. Secondly, from the perspective of reducing common-mode voltage, (0,0,0) does not generate common-mode voltage.

[0049] New virtual small vectors V8~V 13 Each is acted upon by its own positive and negative small vectors, respectively. S Synthesizing at equal time intervals (e.g., V8) follows the following principle:

[0050]

[0051] Where V S1 [i a [] is a small positive vector with a switching state of (1,0,0), producing a midpoint current of i. a , and V S1 [-i a ] is a negative small vector with a switching state of (0, -1, -1), and produces a midpoint current of -i. a Each vector is applied 0.5T. S , then i O The value will be zero, and will not affect the midpoint potential;

[0052] The newly synthesized virtual vectors V2 to V7 have vector magnitudes corresponding to their respective virtual subvectors V8 to V7. 13 Half of the, producing i O It is also zero; taking V2 as an example:

[0053]

[0054] New vector V 14 ~V 19 Each vector is synthesized from a pair of positive and negative small vectors and the fundamental medium vector in each sector, with each vector acting for an equal time, i.e., each vector acts for 1 / 3T. S The sector is a regular hexagon in the space vector diagram of a three-level T-type inverter, divided into 6 equal sectors, each 60° apart, denoted as sectors I to VI; V 14 The synthesis is as follows:

[0055]

[0056] Step 2.3: A novel medium vector V 20 ~V 25 The midpoint currents generated by the equal-time synthesis of the fundamental midpoint vector of the current sector and the midpoint vectors of the two adjacent sectors are i. a i b and i cTherefore, the average midpoint current generated by the synthesized new virtual vector is 0; with sector I V 20 For example, vector synthesis is as follows:

[0057]

[0058] Where V M1 V M2 and V M6 These are the basic mid-vectors of sectors I, II, and VI, respectively, corresponding to the switching states of the three phases of the inverter as (1, 0, -1), (0, 1, -1), and (1, -1, 0).

[0059] V 26 ~V 31 Located at the position where the large and small vectors are equally divided in the vector diagram, with sector I's V... 26 For example, the synthesis is as follows:

[0060]

[0061] Step 2.4: A new type of vector V, located within the inscribed circle with the magnitude of the median vector as its radius, and symmetrically distributed around the line containing the median vector as its axis. 32 ~V 43 V of sector I 32 V 33 For example, vector synthesis is shown in Equations 19 and 20;

[0062]

[0063]

[0064] Virtual large vector V 44 ~V 49 As a fundamental large vector, the resulting midpoint current is zero and does not affect the midpoint potential;

[0065] Step 2.5: In Equation 9, since the objective function of traditional MPC contains a midpoint potential balance term, it is difficult to adjust the weighting factor. To avoid this process, an objective function without weighting factors is used as shown in Equation 21:

[0066]

[0067] Beneficial technical effects of the present invention:

[0068] This invention proposes a novel MPCC-DSVM method. The constructed virtual vector completely eliminates low-frequency oscillations at the midpoint potential and avoids the need for weight factor selection. Simultaneously, it achieves a fixed switching frequency, effectively reducing output current ripple and improving output current quality. The proposed method addresses the problem of inverter output level degradation to two levels when the tracking reference current amplitude is large in Model Predictive Voltage Control (MPVC) methods. Attached Figure Description

[0069] Figure 1 A topology diagram of a three-level T-type inverter according to an embodiment of the present invention;

[0070] Figure 2 A new spatial vector diagram of a three-level T-type inverter according to an embodiment of the present invention;

[0071] Figure 3 Embodiment V of the present invention 33 Vector synthesis principle diagram;

[0072] Figure 4 Figure a of the embodiment of the present invention generates V 16 A schematic diagram of the vector gate signal, Figure b generates V. 28 A schematic diagram of a vector gate signal;

[0073] Figure 5 This invention includes a block diagram of a three-level T-type inverter MPCC-DSVM control system according to an embodiment of the present invention.

[0074] Figure 6 The voltage diagram of the upper and lower capacitors in a conventional MPC according to an embodiment of the present invention;

[0075] Figure 7 DC-side capacitor voltage diagram of the model predictive voltage control (MPVC) method in this embodiment of the invention;

[0076] Figure 8 Point potential waveform diagram in an embodiment of the present invention;

[0077] Figure 9 Enlarged view of DC-side capacitor voltage in an embodiment of the present invention;

[0078] Figure 10 Output line voltage diagram of an embodiment of the present invention;

[0079] Figure 11 Line voltage diagram of the model predictive voltage control (MPVC) method in this embodiment of the invention;

[0080] Figure 12 Output current diagram of a conventional MPC according to an embodiment of the present invention;

[0081] Figure 13The current THD diagram of a conventional MPC according to an embodiment of the present invention;

[0082] Figure 14 Output current diagram of the model predictive voltage control (MPVC) method in this embodiment of the invention;

[0083] Figure 15 Output current diagram of an embodiment of the present invention;

[0084] Figure 16 Output current THD diagram of the model predictive voltage-controlled MPVC method in this embodiment of the invention;

[0085] Figure 17 Output current THD diagram of the method proposed in this embodiment of the invention;

[0086] Figure 18 The current transient change diagram of the method proposed in embodiment L of the present invention when L changes from 50%L0 to L0;

[0087] Figure 19 The current dynamic response diagram of the proposed method in this embodiment of the invention when R changes from 50%R0 to R0;

[0088] Figure 20 A transient current change diagram when the reference current amplitude changes from 20A to 10A in an embodiment of the present invention;

[0089] Figure 21 System power loss diagram of the model predictive voltage control (MPVC) method according to an embodiment of the present invention;

[0090] Figure 22 System power loss diagram of this invention embodiment; Detailed Implementation

[0091] The present invention will be further described below with reference to the accompanying drawings and embodiments;

[0092] The method for balancing the neutral point potential of a three-level T-type inverter based on MPCC-DSVM specifically includes the following steps:

[0093] Step 1: Model the three-level T-type inverter topology system according to Kirchhoff's voltage law, then perform coordinate system transformation and discretization on the modeling results, and finally write the objective function equation of FCS-MPC according to the principle of FCS-MPC.

[0094] The topology diagram of a three-level T-type inverter is as follows: Figure 1 As shown, V dc V represents the DC-side voltage of the inverter. up and V low These represent the voltages of the upper and lower capacitors on the DC side, i c1 and i c2i represents the current flowing through the upper and lower capacitors, respectively. O It is the midpoint current, L is the load inductance, and R is the load resistance, such as Figure 1 As shown.

[0095] Step 1.1: The differential form of the output voltage of the three-level T-type inverter in the abc coordinate system is as follows:

[0096]

[0097] Where V xn (x = a, b, c) represents the voltage of phase x relative to the neutral point, V xO (x = a, b, c) represents the voltage at the x-phase output point relative to the DC side midpoint O; V On It is the voltage at point O on the DC side relative to the neutral point n; i ox (x = a, b, c) represents the load current of phase x; L is the load inductance;

[0098] Step 1.2: Transform Equation 1 from the abc coordinate system to the αβ coordinate system to obtain:

[0099]

[0100] Among them, V α and V β i represents the output voltage of the inverter in the α and β coordinate systems. oα and i oβ These are the inverter's output currents in the α and β coordinate systems, respectively; R is the load resistance.

[0101] Step 1.3: Each arm of the inverter generates three different output voltages, which are VDC voltages relative to the DC-side midpoint potential of the inverter topology. dc / 2, 0 and -V dc / 2, the inverter generates 27 switching states, where "1", "0" and "-1" correspond to the three voltage levels output by the inverter, i.e., V. dc / 2, 0, -V dc / 2; Set the switching function T i Defined as:

[0102]

[0103] Where T i1 and T i4 T represents the upper and lower switching transistors of a T-type inverter. i2 and T i3 Represents the two middle switching transistors;

[0104] The output voltage of the inverter is then expressed as:

[0105]

[0106] Where T a T b T c These represent the switching states of phases a, b, and c of the T-type inverter, respectively; V dc It is the DC side voltage.

[0107] Step 1.4: Predict the current value i at time (k+1) by applying the forward Euler formula to Equation 2. oα (k+1) and i oβ (k+1) are respectively:

[0108]

[0109] Where i oα (k) and i oβ (k) represents the output current of the inverter at time k in the αβ coordinate system, T S The sampling period;

[0110] The current flowing through the DC-side capacitor of a three-level T-type inverter topology is expressed as:

[0111]

[0112] Where i c1 and i c2 These represent the current flowing through the upper capacitor and the lower capacitor on the DC side, respectively; V up It is the voltage across the capacitor, V low C is the voltage across the capacitor; C is the DC-side voltage regulator capacitor value.

[0113] Step 1.5: Applying the forward Euler formula to Equation 6, the predicted values ​​of the DC-side upper and lower capacitor voltages are:

[0114]

[0115] The current flowing through the upper and lower capacitors at time k can be expressed as:

[0116]

[0117] Where i dc (k) represents the DC current at time k, i c,a (k), i c,b (k), i c,c (k) represents the branch current flowing through the three-phase bridge arms a, b, and c at time k;

[0118] If T i ="1",H 1x =1, otherwise H 1x =0

[0119] If T i ="-1",H 2x =1, otherwise H 2x =0.

[0120] x = {a, b, c}.

[0121] To track the reference current and simultaneously balance the voltage across the capacitor, the objective function of the traditional FCS-MPC is defined as:

[0122]

[0123] i oα (k+1) and i oβ (k+1) represent the predicted current at time k+1 in the αβ coordinate system; ΔV dc (k+1) is the difference in voltage between the upper and lower capacitors on the DC side at time k+1, expressed as V. up (k+1)-V low (k+1), λ np It is a weighting factor; and Let be the reference current values ​​of the inverter at time k+1 in the αβ coordinate system, which are calculated using the third-order Lagrange extrapolation equation as follows:

[0124]

[0125] in and These are the reference current values ​​of the inverter at time k in the αβ coordinate system; and These are the reference current values ​​of the inverter at time k-1; and These are the reference current values ​​of the inverter at time k-2; and These are the reference current values ​​of the inverter at time k-3.

[0126] Step 2: Based on the discretized model of the predicted current and the objective function of FCS-MPC obtained in Step 1, vector synthesis and MPCC-DSVM strategy are carried out according to the construction principle of virtual vector. Finally, the optimal switching state at the next moment is determined by the objective function without weight factors.

[0127] Traditional MPC (Multi-Current Control) generates high harmonic content and variable switching frequencies because it operates on only one vector per cycle. To reduce current harmonics while maintaining a constant switching frequency;

[0128] Step 2.1: Using MPC combined with DSVM, assuming the sampling period is divided into N equal parts, then the virtual vector V vir From real vector Synthesis, represented as follows:

[0129]

[0130] t1+t2+...+t N =T S (33)

[0131]

[0132] The new virtual vector needs to satisfy the condition that the sum of the currents flowing into and out of the midpoint of the DC side in the inverter topology is zero within one sampling period;

[0133] Normally, if i O If the average value of the midpoint current is not zero within a period, it will cause midpoint potential oscillation or even shift, because it will lead to unbalanced charging and discharging of the upper and lower capacitors. Therefore, if the average value of the midpoint current within the sampling period is zero, midpoint potential oscillation can be avoided. Virtual vector V vir Through multiple basic vectors Linear combinations result in an average midpoint current of zero. A key feature of this paper is that it considers all factors leading to i... O Non-zero original vectors are replaced by equivalent virtual vectors with different magnitudes and angles, but the resulting i O It is zero; this avoids using the DC-side midpoint voltage balance term in the objective function.

[0134] Step 2.2: As Figure 2 The new spatial voltage vector diagram is shown, and the synthesis of each virtual vector in the diagram and its impact on the voltage balance of the DC-side capacitor are analyzed.

[0135] V1 is (0,0,0), and the corresponding i O A value of 0 does not affect the midpoint potential. Considering the principle of achieving a fixed switching frequency and minimizing the number of switching operations, only (0,0,0) is considered when selecting the zero vector. Secondly, from the perspective of reducing common-mode voltage, (0,0,0) does not generate common-mode voltage.

[0136] New virtual small vectors V8~V 13 ,correspond Figure 2 V8~V 13 Each is acted upon by its own positive and negative small vectors, respectively. S Synthesizing at equal time intervals (e.g., V8) follows the following principle:

[0137]

[0138] Where V S1 [i a [] is a small positive vector with a switching state of (1,0,0), producing a midpoint current of i. a , and V S1 [-i a ] is a negative small vector with a switching state of (0, -1, -1), and produces a midpoint current of -i. a Each vector is applied 0.5T. S , then i O The value will be zero, and will not affect the midpoint potential;

[0139] The newly synthesized virtual vectors V2 to V7 have vector magnitudes corresponding to their respective virtual subvectors V8 to V7. 13 Half of the, producing i O It is also zero; taking V2 as an example:

[0140]

[0141] New vector V 14 ~V 19 Each vector is synthesized from a pair of positive and negative small vectors and the fundamental medium vector in each sector, with each vector acting for an equal time, i.e., each vector acts for 1 / 3T. S The sector is a regular hexagon in the space vector diagram of a three-level T-type inverter, divided into 6 equal sectors, each 60° apart, denoted as sectors I to VI; the composition is as follows:

[0142]

[0143] Step 2.3: A novel medium vector V 20 ~V 25 The midpoint currents generated by the equal-time synthesis of the fundamental midpoint vector of the current sector and the midpoint vectors of the two adjacent sectors are i. a i b and i c Therefore, the average midpoint current generated by the synthesized virtual vector is 0; with sector I V 20 For example, vector synthesis is as follows:

[0144]

[0145] Where V M1 V M2 and V M6 These are the basic mid-vectors of sectors I, II, and VI, respectively, corresponding to the switching states of the three phases of the inverter as (1, 0, -1), (0, 1, -1), and (1, -1, 0).

[0146] V 26 ~V 31 Located at the position where the large vector and the small vector are equally divided, with V of sector I. 26 For example, the synthesis is as follows:

[0147]

[0148] Step 2.4: A new type of vector V, located within the inscribed circle with the magnitude of the median vector as its radius, and symmetrically distributed around the line containing the median vector as its axis. 32 ~V 43 V of sector I 32 V 33 For example, vector synthesis is shown in Equations 19 and 20; where V in sector I 33 The vector synthesis principle diagram is as follows Figure 3 As shown.

[0149]

[0150]

[0151] Virtual large vector V 44 ~V 49 For basic large vectors, such as Figure 2 V 44 ~V 49 The resulting midpoint current is zero and does not affect the midpoint potential;

[0152] The switching state of the optimal vector is compared with the carrier waveform, the period of which is the sampling period T of the controller. S , Figure 4 (a) and (b) respectively generate the virtual voltage vector V 16 V 28 A schematic diagram of the gate signal.

[0153] Tables 1 and 2 list V8 to V8 respectively. 13 and V 32 ~V 43 The vectors involved in the synthesis and their corresponding switch states.

[0154] Table 1 V8~V 13 Vector synthesis and switch state table;

[0155]

[0156] Table 2 V 32 ~V 43 Vector synthesis and switch status table;

[0157]

[0158]

[0159] Step 2.5: In Equation 9, due to the presence of a midpoint potential balance term in the traditional MPC objective function, adjusting the weighting factor is difficult. To avoid this process, an objective function without weighting factors is used, as shown in Equation 21. The control block diagram of the entire system is as follows. Figure 5 As shown.

[0160]

[0161] Simulation verification;

[0162] The effectiveness and feasibility of the proposed method were verified through simulation studies using Matlab / Simulink. The parameters of the control system are shown in Table 3. The main purpose of the simulation verification is to demonstrate that the MPCC-DSVM method, which introduces newly constructed virtual space vectors (VSVs) in this paper, has better steady-state performance and similar dynamic characteristics than the model predictive voltage control (MPVC) method and the traditional MPC method.

[0163] Table 3 Control System Parameters

[0164]

[0165] Steady-state analysis;

[0166] Figure 6 , 7 Figure 8 shows the DC-side capacitor voltage diagrams during the steady-state operation of the inverter using the traditional MPC, Model Predictive Voltage Control (MPVC) method, and the proposed MPCC-DSVM method. The upper and lower capacitor voltage diagrams of the traditional MPC method are shown below. Figure 6 As shown, applying the traditional MPC strategy will result in a peak capacitor voltage ΔV. C,MPC The voltage fluctuations are significant. The voltage difference across the capacitor in the DC link is 4.2V, and the voltage difference across the capacitor in the lower link is 4.5V. The midpoint voltage fluctuation amplitude is large, indicating a midpoint potential imbalance. The DC-side capacitor voltage predicted by the model-predicted voltage-controlled MPVC method is as follows: Figure 7 As shown, the black line represents the voltage across the upper capacitor in the model predicted voltage control (MPVC) method, and the gray line represents the voltage across the lower capacitor. It can be seen that the switching frequency oscillation amplitude of the midpoint voltage is significantly reduced, the midpoint potential is basically horizontal, and there is only a small amount of midpoint voltage ripple. Figure 8 This is a waveform diagram of the midpoint potential of the method proposed in this paper. The black line represents the voltage V across the upper capacitor of the MPCC-DSVM method proposed in this paper. C1 The gray line represents the voltage V across the lower capacitor. C2It can be seen that the proposed method can completely eliminate the low-frequency oscillation of the midpoint potential, and the amplitude of the midpoint voltage fluctuation is significantly reduced compared with the traditional MPC, but not much different from the model predictive voltage control MPVC method. At the same time, both the proposed method and the model predictive voltage control MPVC method have a midpoint potential shift phenomenon, but compared with the model predictive voltage control MPVC method, the voltage shift speed of the upper and lower capacitors in the proposed method is slower.

[0167] An analysis of the enlarged voltage diagram of the DC-side capacitor in the proposed method is performed, such as... Figure 9 As shown, the black line represents the upper capacitor voltage switching frequency oscillation amplitude of 0.29V in the proposed MPCC-DSVM method, and the gray line represents the lower capacitor voltage switching frequency oscillation amplitude of 0.31V. It can be concluded that the peak capacitor voltage fluctuation of the proposed method is reduced by more than 90% compared with the traditional method.

[0168] Secondly, the line voltage V generated by the inverter... c,ab The voltage levels were compared and analyzed. The line voltage diagrams of the proposed method and the model-predicted voltage control MPVC method are shown below. Figure 10 and Figure 11 As shown. Figure 10 The diagram shows the inverter's output line voltage when the proposed method is applied to a reference current of 25A. It can be seen that the output line voltage is normal, without distortion, and the level is 5. Figure 11 The line voltage diagram for the Model Predictive Voltage Control (MPVC) method has three output levels, meaning three line voltages. The line voltage waveform is severely distorted, degenerating into a two-level output state, which negatively impacts system stability. The reason for this is that when the tracking reference current amplitude is large, it can be analogous to a large radius of the reference vector circle in modulation; the outermost hexagonal output vector in the original spatial vector diagram lacks a zero state (0 state).

[0169] When the tracking reference current is 20A, the output current of a traditional MPC is as follows: Figure 12 As shown, the black solid line 1 represents the output current i of phase a in the traditional MPC method. oa The black dashed line 2 represents the output current i of phase b. ob The black dashed line 3 represents the output current i of phase c. oc The current waveform exhibits some distortion, but overall presents a sinusoidal shape. Its current THD diagram is shown below. Figure 13 As shown, the black solid line 1 represents the output current i of phase a in the model predictive voltage control (MPVC) method. oa The black dashed line 2 represents the output current i of phase b. ob The black dashed line 3 represents the output current i of phase c. ocTraditional MPC exhibits a relatively high THD, with dispersed and irregular harmonic distribution and an unstable switching frequency. The output current diagrams of MPCC-DSVM and the proposed method are shown below. Figure 14 and 15 As shown, Figure 14 The solid black line 1 represents the output current i of phase a in the MPCC-DSVM method proposed in this paper. oa The black dashed line 2 represents the output current i of phase b. ob The black dashed line 3 represents the output current i of phase c. oc It can be seen that the output current has good characteristics, with little distortion and is very close to a sine wave. Figure 16 The output current THD diagram of the model predicts voltage-controlled MPVC method. Compared with traditional methods, it has a lower THD, but the harmonic spectrum distribution is wide and irregular, mainly concentrated in the low-frequency band, and the switching frequency is still not fixed. The current THD diagram of the proposed MPCC-DSVM method is shown below. Figure 17 As shown, analysis reveals that its current THD is lower, and the harmonic frequencies surround the switching frequency (10kHz) and its integer multiples, with the switching frequency being fixed.

[0170] Dynamic analysis;

[0171] Table 4 Transient Changes

[0172]

[0173] The purpose of dynamic verification is to ensure that the proposed method retains the good dynamic characteristics of traditional MPC. The three transient change scenarios are listed in Table 4. The transient changes of current during load abrupt changes in the proposed method are as follows: Figure 18 and 19 As shown. Figure 18 The graph shows the transient change of current when L changes from 50%L0 to L0. A sudden change is set at 0.1s, resulting in a short transient response time and no large dynamic overshoot. Furthermore, the THD decreases after the current change because the increased inductance enhances the filtering effect, reducing current harmonics to some extent. Figure 19 The diagram shows the current dynamic response when R changes from 50%R0 to R0. It can be seen that the current THD does not change much before and after the change in current amplitude, and it has a short dynamic response time and no large overshoot. Figure 20 The graph shows the transient change of the reference current amplitude when it changes from 20A to 10A at 0.1s. It can be seen from the graph that the dynamic response time of the transient change of the current is short, the dynamic overshoot is small, and the THD of the current before and after the sudden change is not obvious.

[0174] Power loss analysis;

[0175] Since the proposed method uses a larger number of virtual vectors, it increases the switching losses of the inverter. Therefore, it is necessary to detect the power loss of the system and ensure that the efficiency of the inverter is not seriously affected. Figure 21 and 22 The power loss waveforms of the model predictive voltage control (MPVC) method and the proposed method are shown respectively. It can be seen that the power loss of the proposed method is nearly 500W higher than that of the model predictive voltage control (MPVC) method. However, considering that the rated power of the inverter is close to 100kW, the efficiency reduction is only 0.5%.

[0176] This paper proposes a novel MPCC-DSVM method for application in a three-level T-type inverter. Compared with the traditional MPC method, the proposed method can completely eliminate low-frequency oscillations in the midpoint potential, while avoiding the difficulty of selecting weight factors in multi-objective functions, thus improving controller performance. Furthermore, the proposed method improves the quality of the output current, solves the problem of the inverter's output voltage degenerating to two levels when the tracking reference current amplitude is large in the Model Predictive Voltage Control (MPVC) method, and achieves a fixed switching frequency. However, drawbacks include higher switching losses, increased power losses, and the existence of midpoint potential offset, which require further investigation and resolution in the future.

[0177] From the simulation waveforms of the Model Predictive Voltage Controlled MPVC method, it can be seen that the output level is approximately a two-level state when the modulation coefficient is high. The reason for this issue is analyzed below. While the virtual midpoint vector of the outermost hexagon in the vector diagram satisfies the condition that the average midpoint current is zero when the modulation coefficient is large, this virtual vector is synthesized by the equal time of the action of the basic large vector in the sector. The fundamental reason is that there is no zero state among the switching states participating in the synthesis vector. To solve this problem, the newly constructed virtual vector in this paper includes a small vector, i.e., it has the action of a zero-state switching state. Simulation verification shows that when the reference current amplitude i... ref At 21A to 31A, the output line voltage of the model-predicted voltage-controlled MPVC method is approximately a two-level state, while the simulation results of the proposed method show a normal three-level state.

[0178] In MPVC, when the radius of the reference vector circle is different, theoretical analysis of the switching state switching rules of the vector within and between sectors of each reference vector circle reveals that its switching frequency is not fixed. Simulation verification shows that the harmonics in its THD spectrum are dispersed and exhibit an irregular distribution. However, theoretical analysis and simulation results verify that the harmonic frequencies in the THD spectrum of the proposed method surround the switching frequency, thus achieving a fixed switching frequency.

Claims

1. A method for balancing the neutral point potential of a three-level T-type inverter based on MPCC-DSVM, characterized in that, The specific steps are as follows: Step 1: Model the three-level T-type inverter topology system according to Kirchhoff's voltage law, then perform coordinate system transformation and discretization on the modeling results, and finally write the objective function equation of FCS-MPC according to the principle of FCS-MPC. Step 2: Based on the discretized model of the predicted current and the objective function of FCS-MPC obtained in Step 1, vector synthesis and MPCC-DSVM strategy are carried out according to the construction principle of virtual vector. Finally, the optimal switching state at the next moment is determined by the objective function without weight factors. Step 2 is as follows: Step 2.1: Using MPC combined with DSVM, the sampling period is divided into N equal parts, then the virtual vector... From real vector Synthesis, represented as follows: The new virtual vector needs to satisfy the condition that the sum of the currents flowing into and out of the midpoint of the DC side in the inverter topology is zero within one sampling period; All of the consequences Non-zero original vectors are replaced by equivalent virtual vectors; Step 2.2: Analyze the synthesis principle of each constructed virtual vector and its impact on the voltage balance of the DC-side capacitor; Considering the principle of achieving a fixed switching frequency and minimizing the number of switching operations, only (0,0,0) is considered when selecting the zero vector. Secondly, from the perspective of reducing common-mode voltage, (0,0,0) does not generate common-mode voltage. New virtual small vectors ~ Each is acted upon by its own positive and negative small vectors at its respective position. Synthesized over equal time intervals, with For example, its synthesis principle is as follows: in It is a small positive vector with a switching state of (1,0,0), and the resulting midpoint current is ,and It is a negative small vector with a switching state of (0, -1, -1), and the resulting midpoint current is Each virtual vector is applied with 0.

5. ,but The value will be zero, and will not affect the midpoint potential; Newly synthesized virtual vector ~ The vector magnitudes are respectively the corresponding virtual small vectors, i.e. ~ Half of, produced Also zero; with For example: New vector ~ Each pair of positive and negative small vectors and the fundamental medium vector in each sector act for equal time, that is, each pair of positive and negative small vectors and the fundamental medium vector act... The sector is a regular hexagon in the space vector diagram of a three-level T-type inverter, divided into 6 equal sectors of 60° each, denoted as sectors I to VI; the composition is as follows: Step 2.3: Novel Medium Vector ~ The midpoint currents generated by the equal-time synthesis of the fundamental midpoint vector of the current sector and the midpoint vectors of the two adjacent sectors are respectively , and Therefore, the average midpoint current generated by the synthesized virtual vector is 0; taking sector I as an example. For example, the novel vector synthesis is as follows: in , and These are the basic mid-vectors of sectors I, II, and VI, respectively, corresponding to the switching states of the three phases of the inverter as (1, 0, -1), (0, 1, -1), and (1, -1, 0). ~ Located at the position where the large vector and the small vector are equally divided, with sector I For example, the synthesis is as follows: Step 2.4: A new type of vector located within the inscribed circle with the magnitude of the median vector as its radius, and symmetrically distributed around the line containing the median vector as its axis. ~ With sector I , For example, the novel vector synthesis is shown in Equations 19 and 20; Virtual Large Vector ~ As a fundamental large vector, the resulting midpoint current is zero and does not affect the midpoint potential; Step 2.5: Use the objective function without weighting factors as shown in Equation 21: 。 2. The method for balancing the neutral point potential of a three-level T-type inverter based on MPCC-DSVM according to claim 1, characterized in that, Step 1 is as follows: Step 1.1: The differential form of the output voltage of the three-level T-type inverter in the abc coordinate system is as follows: in This represents the voltage of phase x relative to the neutral point. This indicates the potential of the x-phase output point relative to the DC side midpoint. The voltage; DC side Point relative to neutral point The voltage; This represents the load current of phase x; It is the load inductance; Step 1.2: Subtract Equation 1 from... Coordinate system transformation to In the coordinate system, we get: in, and Indicates that the inverter is in and Output voltage in coordinate system and The inverter is in , Output current in coordinate system; It is the load resistor; Step 1.3: Each arm of the inverter generates three different output voltages, which are respectively... 0 and The inverter generates 27 switching states, with "1", "0", and "-1" corresponding to three different voltage levels output by the inverter. ,0, ; switch function Defined as: in and This represents the upper and lower switching transistors of a T-type inverter. and Representing the two switching transistors in the middle; The output voltage of the inverter is then expressed as: in , , These represent T-type inverters. Three-phase switch status; It is the DC side voltage; Step 1.4: Predict the current value at time (k+1) by applying the forward Euler formula to Equation 2. and They are respectively: in and They are respectively The inverter's output current at time k in the coordinate system. The sampling period; The current flowing through the DC-side capacitor of a three-level T-type inverter topology is expressed as: in and These represent the current flowing through the upper capacitor and the lower capacitor on the DC side, respectively. It is the voltage across the capacitor. It is the voltage across the lower capacitor; This is the value of the DC-side voltage regulator capacitor; Step 1.5: Applying the forward Euler formula to Equation 6, the predicted values ​​of the DC-side upper and lower capacitor voltages are: The current flowing through the upper and lower capacitors at time k can be expressed as: in Let k be the DC-side current. , , These are the branch currents flowing through the three-phase bridge arms a, b, and c at time k, respectively. To track the reference current and simultaneously balance the voltage across the capacitor, the objective function of the traditional FCS-MPC is defined as: and Represent Predicted current at time k+1 in the coordinate system; The difference between the voltages of the upper and lower capacitors on the DC side at time k+1 is expressed as: , It is a weighting factor; and They are respectively The reference current value of the inverter at time k+1 in the coordinate system is calculated using the third-order Lagrange extrapolation equation as follows: in and They are respectively The reference current value of the inverter at time k in the coordinate system; and These are the reference current values ​​of the inverter at time k-1; and These are the reference current values ​​of the inverter at time k-2; and These are the reference current values ​​of the inverter at time k-3.

Images