Midpoint potential fluctuation control factor-based grid-connected inverter model prediction control method

Abstract

The invention relates to the technical field of grid-connected inverter control, and discloses a grid-connected inverter model predictive control method based on a neutral-point potential fluctuation control factor, in the process of controlling a grid-connected inverter by using model predictive control, firstly, an SVPWM method and the model predictive control method are combined, the design of an objective function is simplified, and the model predictive control factor is optimized; the optimal switch sequence can be selected only by calculating the midpoint potential deviation; secondly, neutral-point potential balance control is realized in spatial modulation, a large sector is further divided into a plurality of small sectors, the position of the small sector where a reference voltage vector is located is determined, a switch sequence is selected from the determined small sectors for traversal, the traversal frequency is reduced, and the calculation efficiency is improved; and finally, designing a midpoint potential control factor, and greatly reducing the content of high-amplitude common-mode voltage on the premise of ensuring the balance of the midpoint potential. Moreover, the overall control structure is clear and easy to implement, and the dynamic performance and the steady-state tracking performance of the controller are both improved.

Description

Technical Field

[0001] The present invention relates to the technical field of grid-connected inverter control, and in particular to a grid-connected inverter model predictive control method based on a midpoint potential fluctuation control factor. Background Art

[0002] As one of the main technologies for renewable energy generation, distributed power generation technology has the characteristics of low transmission loss, safety, and high efficiency. Photovoltaic power generation directly converts solar energy into electricity. Compared with traditional coal-fired power generation, it has less impact on the environment, is green and low-carbon, has the characteristics of large total volume and wide distribution, and its application field is developing rapidly.

[0003] As a key link in photovoltaic power generation, grid-connected inverter plays a key role in power conversion, and its working performance directly determines the quality of grid-connected power. In low-power photovoltaic power generation, two-level inverters are widely used due to their simple structure, low cost, and easy control and implementation. Due to the withstand voltage and overcurrent limitations of a single semiconductor device, multi-level inverters are often used in series and parallel combinations in medium and high power photovoltaic power generation.

[0004] However, this system has a complex structure, high design and maintenance costs, and is difficult to control. In order to solve the above problems, high-power multi-level inverters came into being. Compared with two-level inverters, three-level inverters have the advantages of large output power and high withstand voltage, and are more suitable for medium and high capacity photovoltaic grid connection. And three-level inverters include NPC three-level inverters.

[0005] Model predictive control is an optimization control method based on the prediction model of the controlled object. Its control concept is intuitive, easy to model, and does not require complex control parameter design. In most power electronic control application fields, through the ergodic method, according to the discrete mathematical model of the converter, the output state of the converter under all switching states can be calculated. By setting an appropriate objective function, the optimal switching state is selected to control the converter to achieve the desired control effect.

[0006] For the NPC three-phase three-level inverter circuit structure, in order to solve the midpoint potential balance problem of the NPC grid-connected inverter, it is necessary to add a midpoint potential control constraint term to the objective function. The error weight of the grid-connected current following and the weight selection of the midpoint potential deviation often rely on practical experience. Improper design will affect the control accuracy and effect of the grid-connected inverter, which undoubtedly makes the design of the objective function more difficult. Moreover, the traditional finite control set model predictive control needs to traverse 27 switch states. This method has a large amount of calculation and low calculation efficiency, resulting in a long overall control response time and poor dynamic performance of the controller. Summary of the invention

[0007] The present invention provides a model predictive control method for a grid-connected inverter based on a midpoint potential fluctuation control factor, which solves the problem of large amount of calculation and long control response time when the model predictive control in the prior art is directly applied to a grid-connected inverter.

[0008] The present invention provides the following technical solution: a model predictive control method for a grid-connected inverter based on a midpoint potential fluctuation control factor, wherein the DC end of the grid-connected inverter is used to connect a DC power supply, and the AC end of the grid-connected inverter is connected to a power grid through an LCL filter, and the capacitor in the LCL filter is called a filter capacitor. The method comprises the following steps:

[0009] Step S1: First, sample and obtain the switching state S of the inverter at the previous moment opt , sample and obtain the actual value i of the inverter output current at time k 1(k) , the actual value of the filter capacitor voltage at time k is u C(k) , and the voltage difference ΔV between the two clamping capacitors on the DC side (k) ; The grid-side current given value i at time k+1 * 1(k+1) , determine the reference voltage vector u at time k+1 * (k+1) ;

[0010] Step S2: transform the sampled quantities from the three-phase rotating coordinate system to the two-phase stationary coordinate system, and use the switching state of the inverter at the previous moment to predict the output current i of the inverter side at the next moment 1(k+1) And the voltage difference V between the two capacitors on the DC side (k+1) , and delay compensation to obtain i * 1(k+2) 、u C(k+1) ;

[0011] Step S3: Calculate the reference voltage vector u * (k+1) , and judge the large sector and small sector in the space voltage vector diagram, where each large sector in the space voltage vector diagram is divided into four small sectors of equal size; then predict the midpoint potential deviation ΔV that will be generated by the three groups of switch sequences corresponding to the small sector where the reference voltage vector is located at time k+2 (k+2) , compare this deviation with the control factor μ, and select a set of optimal switching sequences to act on the inverter.

[0012] Preferably, the tracking target of the system is converted from traditional current tracking to voltage tracking, and the switch sequence is selected according to the predicted spatial position of the reference voltage. The predicted reference voltage at time k+1 is:

[0013]

[0014] Where i * 1(k+2) 、u C(k+1) are the reference current of the inverter side at time k+2 obtained by delay compensation and the filter capacitor voltage at time k+1, i 1(k+1) is the current outflow from the inverter predicted at time k+1.

[0015] Preferably, three groups of three-vector switching sequences are designed to achieve a zero weight factor; according to the distribution of 27 basic spatial voltage vectors, they can be divided into 6 large sectors, and the large sectors are further divided into 4 small sectors. The three vertices of each small sector are the end points of the basic voltage vectors, and each small sector contains at least one group of redundant small vectors. By analyzing the midpoint current flow direction of the small vectors, it can be seen that the redundant small vectors have opposite effects on the midpoint potential. Therefore, the small vectors with the same effect on the midpoint potential can be put into a group of switching sequences. Therefore, two groups of switching sequences with opposite effects on the midpoint potential can be listed. At the same time, in order to fix the switching frequency, the designed switching sequence is three-vector, and the switching between the vectors is smooth; at the same time, in order to minimize the adverse effects of common-mode voltage on the system, all basic voltage vectors with high common-mode voltage are eliminated, and then a group of low common-mode voltage switching sequences are designed in each small sector according to the principle of smooth switching.

[0016] Preferably, a midpoint potential control factor μ is proposed, the amplitude of which can be changed according to the needs of specific occasions. The factor represents the tolerance of the system to the midpoint potential fluctuation. When the midpoint potential fluctuation range is within μ, a low common mode voltage switching sequence is selected as the optimal switching sequence output; when the midpoint potential fluctuation range exceeds μ, a group of low midpoint potential difference is selected from the remaining two groups of switching sequences for controlling the midpoint potential as the optimal switching sequence output, so as to pull the midpoint potential back to the equilibrium position;

[0017] Preferably, the midpoint potential difference ΔV of the three switch sequences in the small sector where the reference voltage is located at time k+2 is required. (k+2) To make a prediction, ΔV (k+2) Compared with the midpoint potential control factor, the optimal switching sequence of the final output is determined. The cost function of the optimal sequence is:

[0018]

[0019] Among them, t i (i=1,2,3) is the action time of the three basic voltage vectors, S a,b,c(k+1) is the switching state of the inverter at time k+1, and the three groups of switching sequences include two groups of switching sequences of balanced midpoint potential and one group of low common mode voltage switching sequence.

[0020] Preferably, the actual inverter output current at time k is i 1(k)value, actual value of filter capacitor voltage at time k u C(k) , grid-side current given value i at time k+1 * 1(k+1) , the reference voltage vector u at time k+1 * (k+1) 、Filter capacitor voltage u at time k+1 C(k+1) They each include a component of the α axis and a component of the β axis.

[0021] The present invention has the following beneficial effects:

[0022] 1. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor, in the process of controlling the grid-connected inverter by using model predictive control, firstly combines the SVPWM method with the model predictive control method, simplifies the design of the objective function, and only needs to calculate the midpoint potential deviation to select the optimal switching sequence; secondly, the midpoint potential balance control is implemented in spatial modulation, the large sector is further divided into multiple small sectors, the position of the small sector where the reference voltage vector is located is determined, and the switching sequence is selected from the determined small sector for traversal, which reduces the number of traversals and improves the calculation efficiency; finally, the midpoint potential control factor is designed, which greatly reduces the high-amplitude common-mode voltage content under the premise of ensuring the midpoint potential balance, and the overall control structure is clear and simple to implement, and the dynamic performance and steady-state tracking performance of the controller are improved.

[0023] 2. The model predictive control method of the grid-connected inverter based on the midpoint potential fluctuation control factor, through the design of the switching sequence according to the redundant small vector having opposite effects on the midpoint potential, designs two groups of switching sequences in each small sector, when the midpoint potential difference changes, selects one group of switching sequences to act on the inverter to restore the midpoint potential to a balanced position; the other group of switching sequences is designed according to the low common-mode voltage principle, when the midpoint potential deviation is within the tolerance range of the midpoint potential control factor μ, this group of switching sequences acts on the inverter as the optimal sequence; the optimal switching sequence acting on the inverter is adjusted according to the midpoint potential deviation of the grid-connected inverter, the cost function is simplified, the midpoint potential balance is achieved with a zero weight factor, and the content of high-amplitude common-mode voltage is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the model predictive control method of the grid-connected inverter based on the midpoint potential fluctuation control factor of the present invention;

[0025] Figure 2 It is a schematic diagram of the structure of the grid-connected system of the present invention;

[0026] Figure 3 A control block diagram of the general control concept of the model predictive control of the present invention;

[0027] Figure 4 It is a specific control block diagram of the model predictive control of the present invention;

[0028] Figure 5 The spatial voltage vector diagram of the present invention;

[0029] Figure 6 It is the space vector diagram of the large sector I of the present invention;

[0030] FIG. 7( a ) is a diagram showing THD analysis results of the present invention;

[0031] FIG7( b ) is a grid-connected current waveform diagram of the present invention;

[0032] FIG. 7( c ) is a diagram showing the simulation result of the midpoint potential of the present invention;

[0033] FIG. 7( d ) is a diagram showing the common mode voltage simulation result of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] An embodiment of a model predictive control method for a grid-connected inverter based on a midpoint potential fluctuation control factor:

[0036] The structural diagram of the grid-connected system of the present invention is as follows: Figure 1 As shown, it is arranged between the photovoltaic power station and the public power grid.

[0037] The grid-connected system includes a diode-clamped three-phase three-level inverter circuit (i.e., the grid-connected inverter of the present application) and an LCL filter. After the photovoltaic power station is converted by a DC converter, it is connected to the DC end of the diode-clamped three-phase three-level inverter circuit, and the AC end of the diode-clamped three-phase three-level inverter circuit is connected to the public power grid through the LCL filter.

[0038] In addition, the grid-connected system also includes a controller, a sensor and a drive circuit.

[0039] The diode clamped three-phase three-level inverter circuit comprises an inverter bridge arm and two voltage balancing capacitors.

[0040] The voltage-equalizing capacitor includes a first voltage-equalizing capacitor C1 and a second voltage-equalizing capacitor C2; the positive polarity end of the first voltage-equalizing capacitor C1 is connected to the positive output terminal of the DC end of the diode-clamped three-phase three-level inverter circuit, and the negative polarity end is connected to the neutral line; the positive polarity end of the second voltage-equalizing capacitor C2 is connected to the neutral line, and the negative polarity end is connected to the negative output terminal of the DC end of the diode-clamped three-phase three-level inverter circuit.

[0041] The inverter bridge arm includes 6 clamping diodes and 12 fully controlled open devices. The 6 clamping diodes are clamping diode D1, clamping diode D2, clamping diode D3, clamping diode D4, clamping diode D5, and clamping diode D6.

[0042] The 12 fully controlled switch devices are fully controlled switch devices S a1 , Fully controlled switching device S a2 , Fully controlled switching device S a3 , Fully controlled switching device S a4 , Fully controlled switching device S b1 , Fully controlled switching device S b2 , Fully controlled switching device S b3 , Fully controlled switching device S b4 , Fully controlled switching device S c1 , Fully controlled switching device S c2 , Fully controlled switching device S c3 , Fully controlled switching device S c4 .

[0043] The cathode terminals of clamping diodes D1, D3, and D5 are connected to fully controlled switching devices S a1 , S b1 , S c1 The emitter of the clamping diodes D1, D3, and D5 is connected to the neutral line, the cathode ends of the clamping diodes D2, D4, and D6 are connected to the neutral line, and the anode ends of the clamping diodes D2, D4, and D6 are connected to the fully controlled switching device S a3 , S b3 , S c3 of the emitter.

[0044] The LCL filter includes a filter inductor L1 and a filter inductor L2.

[0045] The filter inductor L1 is an inverter-side filter inductor, whose input end is connected to the AC end of the diode-clamped three-phase three-level inverter circuit. The output end of the filter inductor L1 is connected to the upper end of the filter capacitor C and is also connected to the input end of the filter inductor L2. The filter inductor L2 is a grid-side filter inductor, and the output end of the filter inductor L2 is connected to the public power grid.

[0046] The input end of the phase-locked loop is connected to the public power grid, and the output end of the phase-locked loop is connected to the input end of the controller.

[0047] The sensor module includes a plurality of voltage sensors and current sensors for detecting the grid-side voltage, grid-side current, filter capacitor voltage and inverter output current, and also detects the midpoint potential and inverter DC terminal voltage.

[0048] The output end of the sensor module is connected to the input end of the controller to send the collected data to the controller for the controller to perform logical judgment and processing based on the collected data.

[0049] The output end of the controller is connected to the input end of the driving circuit to control each clamping diode and the fully controlled switch device in the diode clamping type three-phase three-level inverter circuit.

[0050] Based on the grid-connected system introduced above, a grid-connected inverter model predictive control method based on the midpoint potential fluctuation control factor of the present invention can be implemented. The control block diagram and specific block diagram are shown as follows: Figure 2 and Figure 3 As shown, the control flow of the entire method is as follows Figure 4 As shown;

[0051] The main features of model predictive control are: under the premise of ensuring the quality of grid-connected power, reducing the number of optimization traversals of model predictive control, ensuring the midpoint potential balance under zero weight coefficient, and reducing the proportion of high-amplitude common-mode voltage content.

[0052] The following first introduces the establishment of the LCL type grid-connected model predictive control model.

[0053] 1. The switching function of the inverter is defined as:

[0054]

[0055] 2. Based on the system structure and Kirchhoff's voltage law, the expression of the inverter side output voltage in the static abc coordinate system can be listed:

[0056]

[0057] where u a 、u b 、u c is the inverter output voltage, L1 is the inverter side filter inductor, i 1a 、i 1b 、i 1c is the three-phase current output by the inverter, u ca 、u cb 、u cc is the three-phase filter capacitor voltage.

[0058] 3. Clark transformation of both sides of equation (2) can obtain the expression of the inverter side output voltage in the αβ coordinate system:

[0059]

[0060] 4. When establishing a predictive control model for a three-level NPC inverter, the mathematical model needs to be discretized. Commonly used discretization methods include forward difference method and backward difference method. This application adopts the forward difference method, and its expression is:

[0061]

[0062] Where T S is the switching period, x is the variable, and k is the kth moment.

[0063] 5. Combining equations (3) and (4), the output current of the inverter at time k+1 can be obtained as:

[0064]

[0065] 6. Ideally, C=C1=C2, and the expression of midpoint current is:

[0066]

[0067] The forward difference of equation (10) is discretized to obtain the predicted value of the midpoint potential deviation at time k+1:

[0068]

[0069] 7. Control target conversion, improve current tracking effect, the control target can be converted from current to voltage. Assuming that the inverter side current tracks the reference current at time k+1, the current at time k+1 in equation (5) can be replaced by the reference current, then:

[0070]

[0071] where u * αk 、u * βk is the reference voltage vector output by the inverter at time k. (The traditional model predictive control adds equation (5) and equation (7) as the cost function to traverse 27 basic voltage vectors to select the optimal vector. This method not only has poor current tracking effect but also requires the determination of weight factors.)

[0072] 8. The reference current flowing out of the inverter side can be calculated based on the grid-connected reference current. In order to achieve unity power factor grid connection, the grid-connected current is given on the dq axis. According to Kirchhoff's voltage law on the grid side, the reference voltage equation of the filter capacitor voltage dq axis can be listed, that is:

[0073]

[0074] Where w is the grid angular frequency, L2 is the grid side filter inductance, e d ,e q ,i 2d * ,i 2q * are the given values ​​of grid voltage and grid-connected current in the dq coordinate system respectively.

[0075] Similarly, the reference current flowing out of the inverter side can be written according to Kirchhoff's current law, that is:

[0076]

[0077] Where C is the filter capacitor value, the obtained dq axis reference current is transformed by inverse park to obtain i 1αβ * .

[0078] 9. Formula (8) is obtained by transforming formula (5). The optimal switch state at time k+1 predicted at time k is output immediately. However, considering the delay caused by sampling and system calculation in practice, this paper adopts a two-step predictive control algorithm. If you want to get the reference voltage at time k+1, you need to push formula (8) back one moment, that is:

[0079]

[0080] 10.i * 1α(k+2) 、i * 1β(k+2) is the value of the reference current on the inverter side at time k+2, k+2, u cβ(k+1) is the value of the filter capacitor voltage at time k+1. These four values ​​are obtained by delay compensation based on time k:

[0081]

[0082] Where w is the fundamental angular frequency of the grid voltage and Ts is the switching period.

[0083] 11. Establish the objective function, the expression is shown in formula (14)

[0084]

[0085] Based on the LCL grid-connected model predictive control model introduced above, the implementation process of the method of the present invention is introduced below.

[0086] Step 1: construct a spatial voltage vector diagram, such as Figure 5 shown.

[0087] The space vector diagram contains 6 large sectors, namely large sector I, large sector II, large sector III, large sector IV, large sector V and large sector VI.

[0088] Step 2: Determine the reference voltage vector u at time k+1 * α / β (k+1).

[0089] The optimized model predictive control of the present invention can obtain the reference voltage vector u at time k+1 by bringing the reference current value into the model for calculation. * α / β (k+1):

[0090]

[0091] Step 3: Determine the reference voltage vector u at time k+1 * α / β (k+1), determine which large or small sector it is in the space voltage vector diagram. The angle of each large sector in the space voltage vector diagram is 60°, so the large sector it is in can be determined based on the angle of the reference voltage, that is:

[0092]

[0093] Wherein, N is the large sector number, ranging from 1 to 6; θ is the angle of the reference voltage vector.

[0094] In order to reduce the amount of calculation, the rectangular (αβ) coordinate system where the reference voltage vector is located is transformed into the 60° (gh) coordinate system, and the remaining large sectors are rotated to large sector I through rotation change, and the reference voltage coordinates are obtained through standardization. The coordinate range can determine the small sector where the reference voltage is located. Each large sector is further divided into 4 small sectors of equal size, namely small sector 1, small sector 2, small sector 3, and small sector 4. Figure 6 It is the space vector diagram of large sector I. Figure 6 It can be seen that each basic voltage vector has a corresponding coordinate, so the small sector where the reference voltage is located can be determined by the coordinate, and the judgment rules are shown in Table 1. * g(k+1) 、V * h(k+1) is the coordinate of the reference voltage in the gh coordinate system;

[0095] Table 1 Small sector judgment rules

[0096]

[0097] According to the predicted reference voltage, the values ​​of the sector and the small sector are determined. The principle is as follows: SVPWM selects the basic voltage vector for synthesis according to the sector where the reference voltage vector is located, and the model predictive control will traverse and optimize the 27 switch states. The optimization calculation is huge, so the optimal sector is determined in combination with vector control to reduce the number of optimization traversals required for each control. Figure 4 It can be seen that by combining the model predictive control and SVPWM method, combined with the sector where the reference voltage vector is located, the number of optimization traversals is reduced from 27 to 3 times, and vector pulse width modulation can be used to balance the midpoint potential.

[0098] Step 4: Determine the sector where the reference voltage is located, V * g(k+1) 、V * h(k+1) The three vertex basic voltage vectors of the small sector can be synthesized according to the volt-second balance principle, combined with Figure 6 , taking the first small sector of the first large sector as an example:

[0099]

[0100] Among them, V0, V3, and V5 are the three basic voltage vectors of the first small sector, and t1, t2, and t3 are the action times of the three basic voltage vectors. The vector action times are shown in Table 2:

[0101] Table 2 Vector action time

[0102]

[0103]

[0104] Step 5: Design of three-vector switching sequence. The three-level NPC inverter has 27 switching states, of which only the large vector and the zero vector do not affect the midpoint potential. The remaining vectors have different effects on the midpoint potential. Since the small voltage vector has redundant vectors, the effects on the midpoint potential are opposite. For example, the redundant vectors POO and ONN, when the switching state of the inverter is POO, the generated midpoint current is i 1a ; When the switching state of the inverter is ONN, the midpoint current generated is -i 1a , so the midpoint potential can be controlled by selecting different redundant small vectors. Different types of basic voltage vectors generate different common-mode voltages. The common-mode voltage generated by zero vectors PPP and NNN is the largest, and its amplitude is V dc / 2, in order to reduce the impact of common-mode voltage on inverter performance, these two zero vectors are removed and only the OOO state is used. In order to reduce the common-mode voltage, the voltage with amplitude V dcBy eliminating the small vector of the common mode voltage / 3, the third set of switching sequences can be obtained. At the same time, the switching order of the switching sequence should satisfy the requirement that only one phase bridge arm switching state is changed each time and the switching state is prohibited from changing from P to N. Taking large sector I as an example, its switching sequence output is shown in Table 3:

[0105] Table 3 Large sector I switch sequence

[0106]

[0107]

[0108] Step 6: Design of midpoint potential control factor. When only considering low common mode voltage, since the midpoint potential cannot be controlled, once a voltage difference occurs between the two capacitors, the voltage difference will continue to increase, and eventually the grid-connected current will be distorted. If only considering the low midpoint potential deviation, the common mode voltage amplitude is V dc / 3 will greatly increase, and the high-frequency common-mode leakage current generated will generate electromagnetic interference when it flows back to the power grid through the ground wire, affecting the normal operation of the system; in order to balance the two indicators of midpoint potential and common-mode voltage, the hysteresis loop idea is introduced, and the midpoint potential deviation limit value μ is proposed. When the midpoint potential deviation is within μ, the low common-mode voltage switch sequence output is selected, and the amplitude is reduced to V in one cycle. dc / 3 common mode voltage at the output.

[0109] Step 7: Calculate the absolute value of the midpoint potential deviation of the three switching sequences in the sector where the reference voltage is located at time k+2 as the cost function, compare it with the midpoint potential control factor, and select the optimal switching sequence. The optimal switching sequence cost function is:

[0110]

[0111] Taking large sector I as an example, the switch sequence output rules are shown in Table 4:

[0112] Table 4 Large sector I switch sequence output rules

[0113]

[0114]

[0115] In summary, this method simplifies the design of the objective function. The voltage-balanced control (i.e., midpoint potential balance control) is controlled by the designed switch sequence. By detecting the position of the reference voltage, three groups of switch sequences in the sector are selected for optimal traversal, which narrows the traversal optimization range and reduces the number of traversals of the model predictive control, effectively improving the computational efficiency. Combining the space vector modulation algorithm with the model predictive control can make it easier to find the optimal switch state combination. Moreover, the control structure is clear and the implementation is simple, and the controller dynamic characteristics and steady-state tracking performance can meet the requirements.

[0116] The following simulation experiments are carried out to illustrate the effectiveness of this method. Figure 2 , 3 is the simulation block diagram of the optimization model predictive control algorithm. The simulation parameters are shown in Table 4. Figure 7(a) to Figure 7(d) This is the system grid-connected simulation result diagram.

[0117] Table 5 System simulation parameters

[0118]

[0119] Depend on Figure 7(a) , 7(b) It can be seen that the grid-connected current waveform presents a relatively good sinusoidal shape, with a total harmonic content of 1.14%, which meets the grid-connected requirements and has good dynamic characteristics. As shown in Figure 7(c), the midpoint potential waveform is controlled within 8V, the midpoint potential fluctuation is within the midpoint potential control factor range, and the voltage balancing speed is only 1.7ms, with good dynamic response performance. As can be seen from Figure 7(d), the high-amplitude common-mode voltage content is low.

[0120] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0121] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A model predictive control method for a grid-connected inverter based on a midpoint potential fluctuation control factor, characterized in that: The DC end of the grid-connected inverter is used to connect a DC power supply, and the AC end of the grid-connected inverter is connected to the power grid through an LCL filter. The capacitor in the LCL filter is called a filter capacitor. The method comprises the following steps: Step S1: First, sample and obtain the switching state S of the inverter at time k opt , sample and obtain the actual value i of the inverter output current at time k 1(k) , the actual value of the filter capacitor voltage at time k is u C(k) , and the voltage difference ΔV between the two clamping capacitors on the DC side (k) ; The grid-side current given value i at time k+1 * 1(k+1) , determine the reference voltage vector u at time k+1 * (k+1) ; Step S2: transform the sampled quantities from the three-phase rotating coordinate system to the two-phase stationary coordinate system, and use the switching state of the inverter at the previous moment to predict the output current i of the inverter side at the next moment 1(k+1) And the voltage difference ΔV between the two capacitors on the DC side (k+1) , and delay compensation to obtain i * 1(k+2) and u C(k+1) ; Step S3: Calculate the reference voltage vector u * (k+1) , and judge the large sector and small sector in the space voltage vector diagram, where each large sector in the space voltage vector diagram is divided into four small sectors of equal size; then predict the midpoint potential deviation ΔV that will be generated by the three groups of switch sequences corresponding to the small sector where the reference voltage vector is located at time k+2 (k+2) , compare this deviation with the control factor μ, and select a set of optimal switching sequences to act on the inverter.

2. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor according to claim 1, characterized in that: The tracking target of the system is converted from traditional current tracking to voltage tracking. The switch sequence is selected according to the predicted spatial position of the reference voltage. The predicted reference voltage at time k+1 is: Where i * 1(k+2) 、u C(k+1) are the reference current of the inverter side at time k+2 obtained by delay compensation and the filter capacitor voltage at time k+1, i 1(k+1) is the current outflow from the inverter predicted at time k+1.

3. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor according to claim 2, characterized in that: Three groups of three-vector switching sequences are designed to achieve zero weight factor; 27 basic spatial voltage vectors are divided into 6 large sectors according to their distribution, and the large sectors are further divided into 4 small sectors. The three vertices of each small sector are the end points of the basic voltage vectors, and each small sector contains at least one group of redundant small vectors. Analysis of the midpoint current flow direction of the small vectors shows that the redundant small vectors have opposite effects on the midpoint potential. Therefore, the small vectors with the same effect on the midpoint potential can be put into a group of switching sequences. Therefore, two groups of switching sequences with opposite effects on the midpoint potential can be listed. At the same time, in order to fix the switching frequency, the designed switching sequence is three-vector, and the switching between vectors is smooth; at the same time, in order to minimize the adverse effects of common-mode voltage on the system, all basic voltage vectors with high common-mode voltage are eliminated, and then a group of low common-mode voltage switching sequences are designed in each small sector according to the principle of smooth switching.

4. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor according to claim 3 is characterized in that: The midpoint potential control factor μ is proposed. The amplitude of this factor can be changed according to the needs of specific occasions. This factor represents the tolerance of the system to the midpoint potential fluctuation. When the midpoint potential fluctuation range is within μ, the low common mode voltage switching sequence will be selected as the optimal switching sequence output. When the midpoint potential fluctuation range exceeds μ, a set of low midpoint potential difference is selected from the remaining two sets of switching sequences controlling the midpoint potential as the optimal switching sequence output to pull the midpoint potential back to the equilibrium position.

5. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor according to claim 4 is characterized in that: The midpoint potential difference ΔV of the three switch sequences in the small sector where the reference voltage is located at time k+2 needs to be (k+2) To make a prediction, ΔV (k+2) Compared with the midpoint potential control factor, the optimal switching sequence of the final output is determined. The cost function of the optimal sequence is: Among them, t i (i=1,2,3) is the action time of the three basic voltage vectors, S a,b,c(k+1) is the switching state of the inverter at time k+1, and the three groups of switching sequences include two groups of switching sequences of balanced midpoint potential and one group of low common mode voltage switching sequence.

6. The model predictive control method for grid-connected inverter based on midpoint potential fluctuation control factor according to claim 5, characterized in that: The actual inverter output current at time k is i 1(k) value, actual value of filter capacitor voltage at time k u C(k) , grid-side current given value i at time k+1 * 1(k+1) , the reference voltage vector u at time k+1 * (k+1) 、Filter capacitor voltage u at time k+1 C(k+1) They each include a component of the α axis and a component of the β axis.

Images