CN116073432A - NPC three-level VSG limited control set model prediction control method - Google Patents

Abstract

The invention provides an NPC three-level VSG limited control set model prediction control method, which comprises the following steps: establishing a rotor motion equation and a prime motor adjustment equation, and realizing active frequency control based on the rotor motion equation and the prime motor adjustment equation to determine a power angle; establishing a reactive voltage droop characteristic equation, and realizing reactive voltage control based on the reactive voltage droop characteristic equation to determine the amplitude; determining a three-phase voltage reference value according to the power angle and the amplitude, introducing a virtual impedance link, and determining VSG output reference voltage according to the three-phase voltage reference value; establishing an LC filter model, and determining a capacitor voltage predicted value based on the LC filter model; and constructing a cost function according to the VSG output reference voltage, the capacitor voltage predicted value and the direct-current side midpoint voltage difference at the corresponding moment, and screening and outputting an optimal voltage vector according to the cost function to control the VSG. The invention solves the problems of poor frequency regulation performance and the like in the traditional SVPWM control method.

Description

NPC three-level VSG limited control set model prediction control method

Technical Field

The invention relates to the technical field of virtual synchronous motors, in particular to an NPC three-level VSG limited control set model predictive control method.

Background

The power electronic equipment generally has the characteristics of high response speed, low inertia and the like, and is exemplified by the power electronic equipment represented by a power electronic inverter, the inertia and the damping cannot be provided for the system due to the fact that the mechanical inertia and the damping characteristics of a traditional synchronous generator are not available, the inertia level of the system is obviously reduced due to excessive access, the disturbance rejection capability is weakened, and even misoperation of the relay equipment can be caused under serious conditions. The virtual synchronous generator (Virtual Synchronous Generator, VSG) technology combines the motion equation and damping characteristic of the synchronous generator rotor on the basis of droop control, realizes the frequency modulation and voltage regulation characteristic similar to the synchronous generator, provides inertia and damping for the system, ensures that the inverter has similar functions as a traditional generator in mechanism and external characteristics, and is beneficial to maintaining the running stability of the power system.

For three-level VSG, conventional SVPWM (Space VectorPulse WidthModulation ) control requires complex double loop control, which is difficult to set internal PI parameters, and requires more sensors. When the island system is switched by a larger power load, the fixed parameter VSG control can provide inertia and damping, but cannot guarantee the frequency adjustment performance.

Disclosure of Invention

The problem to be solved by the invention is that the traditional SVPWM control requires at least one of complex double loop control, difficult PI parameter setting, more sensors and poor frequency adjustment performance.

In order to solve the above problems, the present invention provides a NPC three-level VSG finite control set model predictive control method, including:

establishing a rotor motion equation and a prime motor regulation equation, and realizing active frequency control based on the rotor motion equation and the prime motor regulation equation to determine a power angle, wherein virtual inertia and damping coefficients in the rotor motion equation are related to angular velocity deviation and angular velocity change rate, the virtual inertia and the damping coefficients are selected according to an adaptive control strategy, and the angular velocity change rate is extracted by a fastest differentiator;

establishing a reactive voltage droop characteristic equation, and realizing reactive voltage control based on the reactive voltage droop characteristic equation to determine an amplitude;

determining a three-phase voltage reference value according to the power angle and the amplitude, introducing a virtual impedance link, and determining a VSG output reference voltage according to the three-phase voltage reference value;

establishing an LC filter model, and determining a capacitor voltage predicted value based on the LC filter model;

and constructing a cost function according to the VSG output reference voltage, the capacitor voltage predicted value and the direct-current side midpoint voltage difference at the corresponding moment, and screening and outputting an optimal voltage vector according to the cost function to control the VSG.

Alternatively, the rotor equation of motion is expressed as:

wherein J represents the virtual inertia, D represents the damping coefficient, P _m Representing mechanical power, P _e Represents electromagnetic power, ω represents actual angular velocity, dω/dt represents the rate of change of the angular velocity, ω ₀ Represents the nominal angular velocity, omega-omega ₀ And represents the angular velocity deviation, and θ represents the work angle.

Alternatively, the prime mover adjustment equation is expressed as:

P _m ＝P _ref +m(ω ₀ -ω)；

wherein P is _m Representing mechanical power, P _ref Representing an active input reference value, m representing an active droop control coefficient;

the implementing active frequency control based on the rotor motion equation and the prime mover adjustment equation to determine a power angle includes: substituting the mechanical power determined by the prime motor regulation equation into the rotor motion equation, and simultaneously solving and determining the power angle.

Alternatively, the reactive voltage droop characteristic equation is expressed as:

E＝U _N +n(Q _ref -Q)；

wherein E represents reactive voltage regulation output voltage, U _N Represents rated voltage, n represents reactive voltage droop coefficient, Q _ref Representing a reactive input reference value, Q representing VSG output reactive;

the implementing reactive voltage control based on the reactive voltage droop characteristic equation to determine an amplitude includes: and taking the maximum absolute value of the reactive voltage regulation output voltage as the amplitude value.

Optionally, the three-phase voltage reference value is expressed as:

wherein v is ^* And representing the three-phase voltage reference value, E represents the reactive voltage regulation output voltage, and theta represents the power angle.

Optionally, the virtual impedance element is expressed as:

wherein v is _{d_ref} Representing the d-axis component, v, of the VSG output reference voltage _d * And v _q * Respectively representing d-axis component and q-axis component of the three-phase voltage reference value, wherein R represents resistance component in virtual impedance, L represents inductance component in virtual impedance, and omega represents fundamental wave angular velocity;

the determining the VSG output reference voltage from the three-phase voltage reference value includes:

determining the VSG output reference voltage under a dq coordinate system according to the three-phase voltage reference value;

the VSG output reference voltage is converted from a dq coordinate system to an αβ coordinate system.

Optionally, the LC filter model is expressed as:

wherein U is _αβ Representing the inverter side voltage, i _fαβ Representing the inverter output current, i _αβ Indicating the load current, v _αβ Representing the capacitance voltage, i _Cαβ Representing the capacitive current;

the determining a capacitor voltage prediction value based on the LC filter model includes:

discretizing the LC filter model by adopting a first-order Euler equation to obtain an inductance current prediction formula and a capacitance voltage prediction formula;

and obtaining the capacitor voltage predicted value by adopting the capacitor voltage prediction formula.

Optionally, the NPC three-level VSG finite control set model predictive control method further includes: analyzing the phenomenon of neutral point voltage unbalance of a direct current side capacitor, and obtaining the neutral point voltage difference of the direct current side at a corresponding moment by combining a kirchhoff current law;

the dc side midpoint voltage difference is expressed as:

wherein Deltau (k+1) and Deltau (k) respectively represent the voltage difference between the upper capacitor and the lower capacitor at the time of k+1 and the time of k, T _s Representing the sampling period of the signal, C ₁ Represent the upper capacitance, C ₂ Represent lower capacitance, C ₁ ＝C ₂ ，i ₀ (k) In the k timePoint current.

Optionally, the cost function is expressed as:

g＝|v _{α_ref} -v _α (k+1)|+|v _{β_ref} -v _β (k+1)|+λΔu(k+1)；

wherein v is _{α_ref} And v _{β_ref} Representing the VSG output reference voltage, v _α (k+1) and v _β (k+1) represents the capacitor voltage at time k+1, λ represents the midpoint voltage weight coefficient at the dc side, and Δu (k+1) represents the voltage difference between the upper and lower capacitors at time k+1 at the dc side;

the filtering and outputting the optimal voltage vector according to the cost function to control the VSG comprises the following steps: and outputting the voltage vector corresponding to the minimum cost function as the optimal voltage vector.

Optionally, the adaptive control strategy is expressed as:

wherein J represents adaptive virtual inertia, J ₀ Represents virtual inertia at VSG steady state, D represents adaptive damping coefficient, D ₀ Represents the damping coefficient at VSG steady state, Δω represents the angular velocity deviation, k ₁ 、k ₂ Representing the virtual inertia adjustment coefficient, k ₃ 、k ₄ Represents the damping adjustment coefficient, v ₂ (t) represents the extraction of the angular velocity change rate using a fastest differentiator.

According to the NPC three-level VSG limited control set model prediction control method, MPC (Model Predictive Control, model prediction control) is adopted to replace traditional double-loop control, so that the problem of poor control effect caused by difficult PI parameter setting in a voltage-current double-loop can be effectively avoided, and the defect of complex PI parameter modulation is overcome; meanwhile, compared with the traditional SVPWM control which requires more sensors, the MPC is adopted to reduce the use of a DC side current sensor, and has smaller DC side voltage fluctuation and faster active power response speed; in addition, angular speed deviation and change rate are introduced in parameter self-adaptive control, the angular speed change rate is extracted by adopting a fastest differentiator, and virtual inertia and damping coefficient of the VSG system are adaptively adjusted, so that setting of a piecewise function and a threshold value is omitted, frequent fluctuation of parameters can be avoided, and frequency stability of the island system in power mutation is improved.

Drawings

FIG. 1 is a schematic flow chart of an NPC three-level VSG finite control set model predictive control method according to an embodiment of the invention;

FIG. 2 is a schematic diagram of a VSG topology according to an embodiment of the present invention;

FIG. 3 is a conventional VSG control block diagram;

FIG. 4 is a block diagram of active frequency control according to an embodiment of the present invention;

FIG. 5 is a reactive voltage control block diagram of an embodiment of the present invention;

FIG. 6 is a block diagram of a virtual impedance element according to an embodiment of the present invention;

FIG. 7 is a graph of an inverter output voltage vector according to an embodiment of the present invention;

FIG. 8 is a graph of a neutral point potential imbalance analysis according to an embodiment of the present invention;

FIG. 9 is a block diagram of an MPC based VSG control of an embodiment of the present invention;

FIG. 10 is a graph of power angle and angular velocity of a virtual synchronous generator according to an embodiment of the present invention;

FIG. 11 is a comparison of MPC control and SVPWM control according to an embodiment of the present invention;

FIG. 12 is a comparison of MPC control and SVPWM control according to an embodiment of the present invention;

FIG. 13 is a comparison plot of MPC control versus SVPWM control in accordance with an embodiment of the present invention;

FIG. 14 is a comparison of MPC control and SVPWM control according to an embodiment of the present invention;

FIG. 15 is a fifth comparison of MPC control and SVPWM control according to an embodiment of the present invention;

FIG. 16 is a waveform diagram of a VSG control frequency according to an embodiment of the invention;

FIG. 17 is a second waveform diagram of a VSG control frequency according to an embodiment of the invention;

FIG. 18 is a third waveform diagram of a VSG control frequency according to an embodiment of the invention;

fig. 19 is a graph showing the comparison of the effect of the fastest differentiator according to the embodiment of the present invention and the conventional differentiator.

Detailed Description

In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

As shown in fig. 1, an embodiment of the present invention provides a NPC three-level VSG finite control set model prediction control method, including:

establishing a rotor motion equation and a prime motor regulation equation, and realizing active frequency control based on the rotor motion equation and the prime motor regulation equation to determine a power angle, wherein virtual inertia and damping coefficients in the rotor motion equation are related to angular velocity deviation and angular velocity change rate, the virtual inertia and the damping coefficients are selected according to an adaptive control strategy, and the angular velocity change rate is extracted by a fastest differentiator;

establishing a reactive voltage droop characteristic equation, and realizing reactive voltage control based on the reactive voltage droop characteristic equation to determine an amplitude;

determining a three-phase voltage reference value according to the power angle and the amplitude, introducing a virtual impedance link, and determining a VSG output reference voltage according to the three-phase voltage reference value;

establishing an LC filter model, and determining a capacitor voltage predicted value based on the LC filter model;

and constructing a cost function according to the VSG output reference voltage, the capacitor voltage predicted value and the direct-current side midpoint voltage difference at the corresponding moment, and screening and outputting an optimal voltage vector according to the cost function to control the VSG.

Specifically, as shown in fig. 2, the NPC (Neutral Point Clamped ) three-level VSG system mainly includes a dc power supply, a voltage dividing capacitor, an NPC three-level converter, a filter inductor, a line impedance, a filter capacitor, a load, and the like, where U _dc The DC side voltage is C1 and C2 are DC side capacitors, L is a filter inductance, R is a line resistance, C is a filter capacitance, and ifabc isInductor current, vabc is capacitor voltage, and iabc is output current of VSG after LC filtering.

VSG frequency control is achieved by the rotor motion equation and the prime mover adjustment equation. Assuming a pole pair number of 1, where the mechanical angular velocity is equal to the electrical angular velocity, the rotor equation of motion can be expressed as:

wherein J represents virtual inertia, D represents damping coefficient, and P _m Representing mechanical power, P _e Represents electromagnetic power, ω represents actual angular velocity, dω/dt represents the rate of change of the angular velocity, ω ₀ Represents the nominal angular velocity, omega-omega ₀ And represents the angular velocity deviation, and θ represents the virtual work angle.

In conjunction with the illustration of fig. 4, to more accurately simulate synchronous generator characteristics, a prime mover adjustment equation is added to VSG frequency control, and the P-f droop characteristics are available:

P _m ＝P _ref +m(ω ₀ -ω)；(2)

wherein P is _m Representing mechanical power, P _ref Representing the active input reference value, m representing the active droop control coefficient. And solving and determining a work angle by the simultaneous formula (1) and the formula (2).

As shown in connection with fig. 5, the reactive voltage regulated output voltage E is obtained by Q-U droop characteristics:

E＝U _N +n(Q _ref -Q)；(3)

wherein E represents reactive voltage regulation output voltage, U _N Represents rated voltage, n represents reactive voltage droop coefficient, Q _ref Representing the reactive input reference value, Q represents the VSG output reactive. And taking the maximum absolute value of the reactive voltage regulation output voltage E as the amplitude.

After the work angle and the amplitude are respectively obtained from the active loop and the reactive loop of the VSG, a three-phase voltage reference value v is obtained:

wherein v is ^* And the reference value of the three-phase voltage is represented, E represents the reactive voltage regulation output voltage, and theta represents the virtual power angle.

In order to more approximately simulate quasi-static characteristics of the synchronous machine, a virtual impedance link is added before the three-phase voltage reference value is sent into the double-loop control, and the virtual impedance link is added, so that the output characteristics of the inverter power supply can be more similar to that of a traditional synchronous machine, meanwhile, power decoupling is facilitated, accurate power distribution is realized, circulation is inhibited, and the parallel operation stability of the virtual synchronous machine is improved. The virtual impedance element can be described as:

wherein v is _{d_ref} Representing the d-axis component, v, of the VSG output reference voltage _d * And v _q * The d-axis component and the q-axis component of the three-phase voltage reference value are respectively represented, R represents a resistance component in the virtual impedance, L represents an inductance component in the virtual impedance, and ω represents the fundamental angular velocity. Referring to fig. 9, a virtual impedance element is added to determine the VSG output reference voltage in the dq coordinate system, and then the VSG output reference voltage is converted into the αβ coordinate system.

With reference to fig. 9, the MPC (Model Predictive Control ) has advantages of good dynamic performance, simple control concept, easy implementation of multi-objective optimization control, and the like, and the adoption of the MPC instead of the conventional dual-loop control can effectively avoid the problem of poor control effect caused by difficult PI parameter tuning in the voltage-current dual-loop.

The LC filtered mathematical model based on model predictive VSG can be expressed as:

wherein U is _αβ Representing the inverter side voltage, i _fαβ Representing the inverter output current, i _αβ Indicating the load current, v _αβ Representing the capacitance voltage, i _Cαβ Representing the capacitive current.

When the NPC three-level inverter works normally, each bridge arm has 3 different working modes, and the switching state S of each bridge arm of the inverter is defined _i (i=a, b, c), expressed as a function:

as shown in connection with fig. 7, considering that the NPC three-level inverter has 27 different switching states, 27 voltage vectors can be generated accordingly.

In order to predict the filter capacitor voltage, let the signal sampling period be T _s Discretizing the formula (6) by adopting a first-order Euler equation to obtain:

wherein i is _fαβ (k) Representing the current at the current transformer side of the kth sampling period, i _αβ (k) Represents the load current, v, of the kth sampling period _αβ (k) Represents the filter capacitor voltage of the kth sampling period, U _αβ (k) Representing the kth sample period voltage vector, i _fαβ (k+1) represents the (k+1) th sampling period inductor current predictive value, v _αβ (k+1) represents the (k+1) th sampling period capacitance voltage prediction value.

The preparation method comprises the following steps of (1) finishing the formula (8):

by combining equation (8) and equation (9), the inductor current i _fαβ (k) And capacitance voltage v _αβ (k) The inductor current i at time k can be used to predict the inductor current at time k+1 _fαβ (k+1). By sampling the load current i _αβ (k) The capacitance voltage v at time k+1 can be further predicted _αβ (k+1)。

NPC three-level inverseThe transformer has the defect of unbalanced midpoint potential of the direct-current side capacitors, which is mainly characterized by uneven voltage division of the two capacitors and larger voltage deviation, and the unbalanced midpoint potential can lead to the degradation of the waveform quality of the output voltage. Therefore, the voltage of the two capacitors at the DC side needs to be correspondingly controlled. The direction of the current flowing through the upper and lower voltage dividing capacitors on the dc side and the direction of the midpoint current are shown in fig. 8. Suppose C ₁ ＝C ₂ (wherein C ₁ Is an upper capacitor; c (C) ₂ Lower capacitance), u _C1 、u _C2 Voltages respectively born by two capacitors, i ₀ Is the midpoint current, i _C1 、i _C2 Can be expressed as:

let the signal sampling period be T _s Discretizing the formula (10) by adopting a first-order Euler equation to obtain:

the preparation method comprises the following steps of (1) finishing:

from kirchhoff's current law, we can obtain:

there is also c1=c2. From formulae (12) and (13):

wherein Deltau (k+1) and Deltau (k) represent the voltage difference between the upper and lower capacitors on the DC side at time k+1 and time k, respectively.

And setting a cost function according to the formula, and selecting an optimal voltage vector to control VSG. The cost function g is expressed as:

g＝|v _{α_ref} -v _α (k+1)|+|v _{β_ref} -v _β (k+1)|+λΔu(k+1)；(15)

wherein v is _{α_ref} And v _{β_ref} Represents the VSG output reference voltage, v _α (k+1) and v _β (k+1) represents the capacitor voltage at the time k+1, λ represents the midpoint voltage weight coefficient at the dc side, and 0.8 is conventionally taken, and Δu (k+1) represents the voltage difference between the upper and lower capacitors at the time k+1.

The VSG output reference voltage value and the capacitor voltage predicted value are calculated by equations (5) and (9), respectively, the dc side midpoint voltage difference at time k+1 is predicted by equation (14), and the optimum voltage vector is screened out by equation (15) and output (for example, the voltage vector corresponding to the minimum cost function is output as the optimum voltage vector).

When the island system is disturbed by larger power, the virtual synchronous generator control with fixed parameters can provide inertia and damping for the system, but the frequency regulation and control flexibility are poor, so that the embodiment provides an improved self-adaptive virtual synchronous generator control strategy, frequency deviation and frequency change rate are introduced into VSG virtual parameters, and the virtual inertia and damping coefficient of the VSG system are adaptively adjusted, so that the frequency stability of the island system in power abrupt change is improved.

In combination with the change process of the angular characteristic curve, the rotor angular velocity change rate dω/dt and the rotor angular velocity deviation Δω when the system is disturbed as shown in fig. 10, in one oscillation period, the change process can be divided into 4 stages (i.e., (1) - (4)), and in stage (1), the VSG rotor angular frequency ω is greater than the grid angular frequency ω ₀ And dω/dt>0, where ω of VSG continues to increase, requiring an increase in J to decrease Δω, where Δω (dω/dt)>0. In stage (2), ω is still greater than ω ₀ But dω/dt<0, J should be reduced so that dω/dt| becomes larger, thereby bringing ω closer to ω faster ₀ At this time Δω (dω/dt)<0. Stage (3) and stage (4) are the same. At the same time, the damping coefficient D can be controlled to adjustThe frequency deviation of the system is reduced as D increases, so that the damping coefficient D can be appropriately adjusted when Δω changes. The selection criteria for the different stages J, D are shown in table 1.

Table 1-selection principle at different stages J, D

Based on the above analysis, the values of the adaptive moment of inertia J and the damping coefficient D should be correlated with Δω, dω/dt. According to the selection principle of the virtual inertia and the damping coefficient described in table 1, the proposed adaptive control strategy is expressed as follows:

wherein J is ₀ Representing virtual inertia at VSG steady state, D ₀ Represents the damping coefficient at VSG steady state, Δω represents the angular velocity deviation, dω/dt represents the angular velocity change rate, k ₁ 、k ₂ Representing the virtual inertia adjustment coefficient, k ₃ 、k ₄ Representing the damping adjustment coefficient.

Since in industrial practice the differentiator is physically unrealizable, it is generally replaced by an approximation method, but is easily submerged by amplified noise components. When the VSG parameter is designed to be self-adaptive, if the measured signal omega is interfered by noise, the differential value of the measured signal omega is affected by the noise, so that the self-adaptive parameter is inaccurate, and the optimal control cannot be realized. The tracking differentiator is a single-input multi-output dynamic structure which is provided for eliminating or weakening the noise amplification effect of the classical differentiator. Second order tracking differentiator for input signal v ₀ (t) will give two output signals v ₁ (t) and v ₂ (t)，v ₁ (t) tracking the input signal v ₀ (t), and v ₂ (t) is in fact v ₀ (t) generalized differentiation to overcome noise interference.

Constructing a second-order fastest discrete tracking differentiator as follows:

the expression of fhan () in the formula (17) is:

wherein h represents an integral step length, and reducing h has a remarkable effect of suppressing noise amplification; r represents a speed factor, and the larger r is, the faster the tracking speed is; h is a ₀ Representing the filtering factor, increasing the filtering factor may enhance the filtering effect.

From the above analysis, formula (16) can be expressed as:

where v2 (t) represents the extraction of the angular velocity change rate using a fastest differentiator.

The D omega/dt can be extracted by the differentiator, so that the problem of abrupt change of the angular velocity change rate caused by tiny disturbance of the angular velocity omega can be effectively reduced, the constraint of a piecewise function and a threshold value is not needed, the adaptive control of the parameters J and D can be completed by adopting the formula (19), and the method has the advantages of simple control strategy, strong robustness and the like.

The embodiment describes an NPC three-level VSG limited control set model prediction control strategy with parameter self-adaption, simplifies double-loop control and overcomes the defect of complex PI parameter modulation; meanwhile, the use of a direct current side current sensor can be reduced, and the direct current side voltage fluctuation and the active power response speed are smaller; in addition, angular speed deviation and change rate are introduced in parameter self-adaptive control, and the angular speed change rate is extracted by adopting a fastest differentiator, so that setting of a piecewise function and a threshold value is omitted, frequent fluctuation of parameters is avoided, and the frequency stability of the island system in power mutation is improved.

Wherein, as shown in connection with FIG. 3, the conventional VSG closed-loop control mainly comprises active frequency control and reactive voltage controlThe load side current i is usually used for the production _abc Voltage v _abc The VSG output active power P is obtained after power calculation _e And reactive power Q. The voltage frequency and the amplitude are obtained through the VSG active loop and the VSG reactive loop, and the voltage reference value of the double-loop control of the inverter is obtained through the calculation of the virtual impedance link. And (3) regulating the voltage reference value through double-loop control internal Proportional Integral (PI) to obtain two-phase voltage, and finally feeding back a signal to the gate of the NPC inverter through space vector modulation (SVPWM) to complete the whole set of closed-loop system control. However, this method has the drawbacks of complex double loop control and difficult PI parameter modulation. Meanwhile, when the island system is switched by a large power load, the fixed parameter VSG control can provide inertia and damping, but cannot guarantee the frequency adjustment performance.

The present embodiment is verified as follows.

The Matlab/simulink simulation software is used for carrying out simulation verification on the built mathematical model, and the obtained result can play a role in optimizing frequency support in the island system, so that the stability of the system is effectively improved.

The system load was initially 20kw, 10kw increased at 0.2s and shut down at 0.5 s.

11-15, the system works stably before 0.2s, the output active power is 20kW, the midpoint voltage on the direct current side is relatively balanced, the voltage fluctuation under SVPWM control is about 4V, and the voltage fluctuation under MPC control is about 1.6V; 10kW load is added in a 0.2-0.5s system, load current is increased, at the moment, voltage fluctuation of a direct current side under SVPWM control is about 6V, and voltage fluctuation under MPC control is about 2.6V; the active power rises to a stable value of about 0.016s under SVPWM control, and 0.015s under MPC control; cutting off the added load after 0.6s, and stabilizing the direct-current side voltage fluctuation of the two control methods to 4V and 1.6V respectively; the active power drops to a steady value of about 0.031s under SVPWM control and 0.026s under MPC control. The results indicate that MPC responds faster in power regulation and less dc-side voltage ripple than SVPWM.

With the adoption of the J adaptive control strategy, as shown in fig. 16, when the power of the system deviates from the rated value by t=0.2s, the J value of the adaptive parameter is increased, more inertia is provided for the system, and the dynamic response of the system frequency deviating from the rated value is slowed down; when t=0.5 s system power returns to the nominal value, the adaptive parameter J value decreases, at which time the dynamic response becomes faster.

With the adoption of the D self-adaptive control strategy, as shown in fig. 17, when the power of the system changes, the self-adaptive damping coefficient D changes, so that the frequency deviation is reduced from 0.23Hz to 0.2Hz while the proper damping is provided for the system, and the stability of the system is improved.

In combination with the illustration of fig. 18, a J, D adaptive control strategy is adopted, and the strategy can simultaneously optimize the dynamic response speed and the frequency deviation, so that the frequency adjustment performance is ensured, the control is flexible, and the frequency stability of the island system in power mutation is improved.

As shown in fig. 19, the VSG angular velocity change rate was calculated using a conventional differentiator and a fastest differentiator, and applied to the adaptive control strategy of the present embodiment, resulting in a corresponding J, D value versus graph. The angular velocity change rate calculated by the fastest differentiator in the embodiment does not cause abrupt change of the differential value due to fine angular velocity jitter during stable operation, and the application of the differential value to the calculation of the adaptive parameter J, D can prevent frequent parameter change due to abrupt change of the angular velocity change rate, thereby playing a certain filtering effect. As the load varies in magnitude, the adaptive parameter J increases or decreases depending on the frequency state of the system. When the frequency deviates from the normal value, the adaptive parameter D increases to reduce the frequency deviation.

Although the present disclosure is disclosed above, the scope of the present disclosure is not limited thereto. Various changes and modifications may be made by one skilled in the art without departing from the spirit and scope of the disclosure, and these changes and modifications will fall within the scope of the disclosure.

Claims (10)

1. The NPC three-level VSG limited control set model prediction control method is characterized by comprising the following steps of:

establishing a rotor motion equation and a prime motor regulation equation, and realizing active frequency control based on the rotor motion equation and the prime motor regulation equation to determine a power angle, wherein virtual inertia and damping coefficients in the rotor motion equation are related to angular velocity deviation and angular velocity change rate, the virtual inertia and the damping coefficients are selected according to an adaptive control strategy, and the angular velocity change rate is extracted by a fastest differentiator;

establishing a reactive voltage droop characteristic equation, and realizing reactive voltage control based on the reactive voltage droop characteristic equation to determine an amplitude;

determining a three-phase voltage reference value according to the power angle and the amplitude, introducing a virtual impedance link, and determining a VSG output reference voltage according to the three-phase voltage reference value;

establishing an LC filter model, and determining a capacitor voltage predicted value based on the LC filter model;

and constructing a cost function according to the VSG output reference voltage, the capacitor voltage predicted value and the direct-current side midpoint voltage difference at the corresponding moment, and screening and outputting an optimal voltage vector according to the cost function to control the VSG.

2. The NPC three-level VSG finite control set model predictive control method of claim 1, wherein the rotor motion equation is expressed as:

wherein J represents the virtual inertia, D represents the damping coefficient, P _m Representing mechanical power, P _e Represents electromagnetic power, ω represents actual angular velocity, dω/dt represents the rate of change of the angular velocity, ω ₀ Represents the nominal angular velocity, omega-omega ₀ And represents the angular velocity deviation, and θ represents the work angle.

3. The NPC three-level VSG limited control set model predictive control method of claim 2, wherein the prime mover adjustment equation is expressed as:

P _m ＝P _ref +m(ω ₀ -ω)；

wherein P is _m Representing mechanical power, P _ref Representing an active input reference value, m representing an active droop control coefficient;

the implementing active frequency control based on the rotor motion equation and the prime mover adjustment equation to determine a power angle includes: substituting the mechanical power determined by the prime motor regulation equation into the rotor motion equation, and simultaneously solving and determining the power angle.

4. The NPC three-level VSG limited control set model predictive control method of claim 1, wherein the reactive voltage droop characteristic equation is expressed as:

E＝U _N +n(Q _ref -Q)；

wherein E represents reactive voltage regulation output voltage, U _N Represents rated voltage, n represents reactive voltage droop coefficient, Q _ref Representing a reactive input reference value, Q representing VSG output reactive;

the implementing reactive voltage control based on the reactive voltage droop characteristic equation to determine an amplitude includes: and taking the maximum absolute value of the reactive voltage regulation output voltage as the amplitude value.

5. The NPC three-level VSG limited control set model predictive control method of claim 4, wherein the three-phase voltage reference value is expressed as:

wherein v is ^* And representing the three-phase voltage reference value, E represents the reactive voltage regulation output voltage, and theta represents the power angle.

6. The NPC three-level VSG finite control set model predictive control method of claim 5, wherein the virtual impedance element is expressed as:

wherein v is _{d_ref} Representing the d-axis component, v, of the VSG output reference voltage _d * And v _q * Respectively representing d-axis component and q-axis component of the three-phase voltage reference value, wherein R represents resistance component in virtual impedance, L represents inductance component in virtual impedance, and omega represents fundamental wave angular velocity;

the determining the VSG output reference voltage from the three-phase voltage reference value includes:

determining the VSG output reference voltage under a dq coordinate system according to the three-phase voltage reference value;

the VSG output reference voltage is converted from a dq coordinate system to an αβ coordinate system.

7. The NPC three-level VSG finite control set model predictive control method of claim 1, wherein the LC filter model is expressed as:

wherein U is _αβ Representing the inverter side voltage, i _fαβ Representing the inverter output current, i _αβ Indicating the load current, v _αβ Representing the capacitance voltage, i _Cαβ Representing the capacitive current;

the determining a capacitor voltage prediction value based on the LC filter model includes:

discretizing the LC filter model by adopting a first-order Euler equation to obtain an inductance current prediction formula and a capacitance voltage prediction formula;

and obtaining the capacitor voltage predicted value by adopting the capacitor voltage prediction formula.

8. The NPC three-level VSG limited control set model predictive control method of claim 1, further comprising: analyzing the phenomenon of neutral point voltage unbalance of a direct current side capacitor, and obtaining the neutral point voltage difference of the direct current side at a corresponding moment by combining a kirchhoff current law;

the dc side midpoint voltage difference is expressed as:

wherein Deltau (k+1) and Deltau (k) respectively represent the voltage difference between the upper capacitor and the lower capacitor at the time of k+1 and the time of k, T _s Representing the sampling period of the signal, C ₁ Represent the upper capacitance, C ₂ Represent lower capacitance, C ₁ ＝C ₂ ，i ₀ (k) Indicating the midpoint current at time k.

9. The NPC three-level VSG finite control set model predictive control method of claim 1, wherein the cost function is expressed as:

g＝v _{α_ref} -v _α (k+1)+v _{β_ref} -v _β (k+1)+λΔu(k+1)；

wherein v is _{α_ref} And v _{β_ref} Representing the VSG output reference voltage, v _α (k+1) and v _β (k+1) represents the capacitor voltage at time k+1, λ represents the midpoint voltage weight coefficient at the dc side, and Δu (k+1) represents the voltage difference between the upper and lower capacitors at time k+1 at the dc side;

the filtering and outputting the optimal voltage vector according to the cost function to control the VSG comprises the following steps: and outputting the voltage vector corresponding to the minimum cost function as the optimal voltage vector.

10. The NPC three-level VSG limited control set model predictive control method of any one of claims 1 to 9, wherein the adaptive control strategy is expressed as:

wherein J represents adaptive virtual inertia, J ₀ Represents virtual inertia at VSG steady state, D represents adaptive damping coefficient, D ₀ Represents the damping coefficient at VSG steady state, Δω represents the angular velocity deviation, k ₁ 、k ₂ Representing the virtual inertia adjustment coefficient, k ₃ 、k ₄ Represents the damping adjustment coefficient, v ₂ (t) represents the extraction of the angular velocity change rate using a fastest differentiator.

Images