Model predictive control method for bridge arm fault tolerance of NPC three-level virtual synchronous machine

By reconstructing the VSG structure and using the current prediction model for fault tolerance control, the problem of failure of control strategy after the bridge arm of the virtual synchronous machine is solved, and the fault tolerance operation and grid stability of NPC three-level VSG are achieved.

CN114421794BActive Publication Date: 2024-11-12ZHENGZHOU UNIVERSITY OF LIGHT INDUSTRY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210097681.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-01-27
Publication Date
2024-11-12
Estimated Expiration
2042-01-27

AI Technical Summary

Technical Problem

The existing virtual synchronous machine (VSG) fails after the bridge arm failure, resulting in the failure of the grid-connected control strategy, affecting the stable operation of the power grid, and lacking fault-tolerant design.

Method used

A method for predicting and controlling fault tolerance model of NPC three-level virtual synchronous machine bridge arm is proposed. By reconstructing the VSG structure, using the current prediction model and cost function to perform fault tolerance control, considering the problem of capacitance voltage imbalance on the DC side, and improving fault tolerance capabilities.

Benefits of technology

The VSG is realized continuously after the bridge arm failure, which improves operating reliability and fault tolerance, and verifies the effectiveness of the control strategy through experiments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114421794B_ABST
    Figure CN114421794B_ABST
Patent Text Reader

Abstract

The present invention proposes a NPC three-level virtual synchronous machine bridge arm fault tolerance model prediction control method, which is used to solve the technical problem of control strategy failure after the NPC three-level VSG bridge arm fault; the steps are: first, reconstruct the NPC three-level VSG structure into a VSG bridge arm fault tolerance structure, and establish a prediction current model under fault conditions and a voltage vector of the VSG bridge arm fault tolerance structure; second, use a sensor to collect the DC side capacitor voltage at the current moment, and predict the DC capacitor voltage difference at the next moment based on the DC side capacitor voltage at the current moment; finally, construct a cost function with a delay compensation function, and select the voltage vector corresponding to the minimum value of the cost function as the optimal voltage vector to act at the next moment, so as to realize the VSG bridge arm fault tolerance operation. After the switch device fails, the NPC three-level VSG can operate continuously with fault tolerance, which verifies the effectiveness of the method of the present invention and improves the reliability of VSG operation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of power electronics, and in particular to a fault tolerance model predictive control method for a bridge arm of an NPC three-level virtual synchronous machine. Background Art

[0002] Wind power and photovoltaic power generation are connected to the power grid through power electronic converters, but their randomness and volatility pose challenges to the stable operation of the power grid. Virtual Synchronous Generator (VSG) technology overcomes the shortcomings of low inertia and no damping of renewable energy generation and improves the power grid's ability to absorb renewable energy. However, the reliability of VSG in energy conversion and transmission directly affects the stable operation of the power grid, so the research on VSG bridge arm fault tolerance technology has attracted the attention of more and more experts and scholars.

[0003] Professor Beck proposed the Virtual Synchronous Generator (VSG) technology based on the characteristics of synchronous generators, which provides control reference variables for the grid-connected system according to the preset power and grid parameters, strengthens the grid disturbance response capability, and enables the grid-connected system to have self-recovery capabilities. References [Yang Jian, Liu Yong, Pan Hongbin, et al. Frequency-difference-free control of microgrid inverters based on virtual synchronous machines [J]. Power System Technology, 2016, 40(07): 2001-2008.] Based on the VSG algorithm, a frequency-difference-free active-frequency control algorithm is proposed, which uses frequency proportional integral feedback instead of the traditional damping link to eliminate the frequency offset in the isolated grid mode. For unbalanced power grids, the literature [Wan Xiaofeng, Hu Hailin, Nie Xiaoyi, et al. Improved virtual synchronous machine control strategy for unbalanced power grid voltage [J]. Power System Technology, 2017, 41(11): 3573-3581.] and the literature [Sun Jun, Jiang Tianlong, Wang Yangming, et al. Virtual synchronous machine control optimization strategy for double-fed induction generator under unbalanced power grid [J]. Automation of Electric Power Systems, 2020, 44(10): 135-144.] proposed an improved VSG control strategy using balanced current VSG. The control obtains the final modulated voltage signal by synthesizing the reference positive sequence current command and the positive and negative sequence current commands, thus achieving the three-phase balance of the output current; the literature [Wang Yangming, Sun Dan, Nian Heng, et al. Research on the extended power control strategy of self-synchronous virtual synchronous machine under unbalanced power grid [J]. Power System Technology, 2020, 44(09): 3368-3375.] analyzes the feasibility of combining extended power theory with VSG, and proposes a self-synchronous VSG control strategy based on extended power to achieve accurate self-synchronous grid connection of VSG under unbalanced power grid. However, the above VSG control method requires PWM wave modulation and PI control, and the parameter design is relatively complex.

[0004] Model Predictive Control (MPC) is applied to the field of nonlinear control due to its strong robustness, simple control, and no need for PWM modulation. Reference [Zhang Zhenyan. Research on novel three-phase three-level rectifier based on improved finite set model predictive control under weak power grid conditions [J]. Journal of Power Supply, 2021, 19(01): 59-68.] In view of the large amount of calculation of conventional MPC, an improved finite set model predictive control method is proposed to achieve fast grid current tracking. Reference [Chen Laijun, Wang Ren, Zheng Tianwen, et al. Virtual synchronous generator model predictive control to improve the frequency dynamic characteristics of independent microgrids [J]. Automation of Electric Power Systems, 2018, 42(03): 40-47.] A VSG control method based on MPC is proposed. The power increment of the prediction model is calculated according to the system frequency and VSG output, and the VSG set power is changed to reduce the system frequency oscillation caused by the fluctuation of new energy output. Reference [Li Yanyan, Zhao Yaqian, Guo Leilei, et al. Current model predictive control strategy for virtual synchronous generator [J / OL]. Journal of Power Supply: 1-13.] Combining MPC and VSG, a VSG grid-connected model predictive control method is proposed. The reference current and predicted current are generated by VSG and the current prediction model, which not only provides inertia and damping for the power grid, but also provides active and reactive power support. However, the above method does not consider the problem that the converter switching device may fail during operation, lacks fault-tolerant design, and the grid-connected control strategy fails after the VSG fails, affecting the normal operation of the grid-connected system. Summary of the invention

[0005] Aiming at the technical problem that the control strategy fails after a neutral point clamped (NPC) three-level VSG bridge arm fault, the present invention proposes a NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method. After a switch device failure occurs in the VSG, the conventional NPC three-level is reconstructed into a VSG fault-tolerant structure; the synchronous machine characteristics are simulated by the VSG fault-tolerant structure, and the grid-connected reference current is adjusted according to the system preset power; the current prediction model is used to predict the current at the next moment, the optimal voltage vector is selected, and the advance prediction is used to reduce the system delay; considering the DC side capacitor voltage imbalance problem in the VSG fault-tolerant structure, the DC side capacitor voltage prediction term is added to the cost function to balance the capacitor voltage, thereby improving the fault tolerance and reliability of the NPC three-level VSG.

[0006] The technical solution of the present invention is achieved in this way:

[0007] A NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method, the steps of which are as follows:

[0008] Step 1: Analyze the operating mechanism of the NPC three-level VSG single-phase bridge arm when a fault occurs, and reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure;

[0009] Step 2: Establish a prediction current model when a single-phase bridge arm of the NPC three-level VSG fails, and construct a voltage vector of the fault-tolerant structure of the VSG bridge arm;

[0010] Step 3: Use the sensor to collect the DC side capacitor voltage at time k, and use the DC side capacitor voltage prediction model at time k+2 to predict the DC capacitor voltage difference;

[0011] Step 4: Construct a cost function with delay compensation function based on the predicted current model and the DC capacitor voltage difference at time k+2, and select the voltage vector corresponding to the minimum value of the cost function as the optimal voltage vector to act at time k+1 to achieve fault-tolerant operation of the VSG bridge arm.

[0012] Preferably, the operating mechanism of the NPC three-level VSG single-phase bridge arm when a fault occurs is: when the NPC three-level grid-connected converter bridge arm fails, the fast fuse F n1 and F n2 Fuse the fault bridge arm and turn on the corresponding bidirectional thyristor T rn , connect the faulty phase to the midpoint of the DC side capacitor through a filter inductor, and then reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure; where n = a, b, c.

[0013] Preferably, the functions of the VSG bridge arm fault-tolerant structure include three functional modules: active-frequency modulation, reactive-voltage regulation, and output reference current; regulating the mechanical torque T m and electromagnetic torque T e Realize active power-frequency modulation; given reactive power Q set , the grid reactive power Q and the voltage difference constitute reactive voltage regulation; the output reference current is generated according to the VSG virtual potential and the terminal voltage difference; the expression is as follows:

[0014]

[0015] Where, J represents the VSG moment of inertia; represents the VSG angular frequency; represents the VSG angular acceleration; D p Indicates the voltage droop coefficient; P set Indicates the given active power; M f represents the maximum mutual inductance between the excitation winding and the three-phase stator coil; θ represents the rotation angle of the VSG; i f represents the rotor excitation current, i represents the three-phase current, i aβ_ref represents the reference current, uabc Represents the three-phase voltage, R represents the parasitic resistance of the filter inductor, ω represents the angular frequency, L represents the filter inductor, and j represents the sign of the imaginary part.

[0016] Preferably, the method for constructing the predicted current model is:

[0017] From the NPC three-level VSG structure, we can know the voltage equation of the NPC three-level grid-connected converter:

[0018]

[0019] In the formula, u aβ =[u α ,u β ] T ,u α and u β represents the output voltage of the converter in the aβ coordinate system; i aβ =[i α ,i β ] T ,i α and i β Indicates the converter output current; e aβ =[e α , e β ] T , e α and e β Represents the grid voltage; R represents the parasitic resistance of the filter inductor, and L represents the filter inductor;

[0020] By performing Euler discretization on the voltage equation of the NPC three-level grid-connected converter, we can obtain:

[0021]

[0022] In the formula, T represents the sampling period; X(k) represents the value of variable X at time k; X(k+1) represents the value of variable X at time (k+1);

[0023] Simplify the discretized expression to get the predicted current at time k+1:

[0024]

[0025] Preferably, the voltage vector of the VSG bridge arm fault-tolerant structure is constructed as follows: the VSG bridge arm fault-tolerant structure has eight switching devices, the fault phase connected to the midpoint of the DC side capacitor is in the "0" state, and the other phases have three states each, for a total of 9 switching states; the VSG switch state is defined as S i , the expression is as follows:

[0026]

[0027] When a bridge arm fault occurs in the VSG, the voltage component u aβ It is expressed as:

[0028]

[0029] The voltage vector of the VSG bridge arm fault-tolerant structure corresponding to the a-phase bridge arm fault is shown in Table 1:

[0030] Table 1 Voltage vector of VSG bridge arm fault tolerance structure when phase a bridge arm fails

[0031]

[0032] Preferably, the method for obtaining the DC capacitor voltage difference at time k+2 is:

[0033] Predict the DC capacitor voltage difference at time k+1:

[0034]

[0035] Among them, u c1 (k+1) represents the capacitor voltage on the DC side at time k+1, u c2 (k+1) represents the capacitor voltage on the DC side at time k+1, u c1 (k) represents the capacitor voltage on the DC side at time k+1, u c2 (k) represents the capacitor voltage on the DC side at time k+1, and C represents capacitance;

[0036] The DC capacitor voltage difference at time k+2 can be obtained from the DC capacitor voltage difference at time k+1.

[0037] Preferably, the method for constructing the cost function with delay compensation function is:

[0038] According to the predicted current model, the predicted current at time k+1 can be obtained, and the cost function is constructed using the predicted current at time k+1, the reference current and the DC capacitor voltage difference at time k+1, which is expressed as:

[0039] g=|i αref -i α (k+1)|+|i βref -i β (k+1)|+λu c1 (k+1)-u c2 (k+1)|;

[0040] In the formula, i αref represents the α-axis component of the virtual synchronous machine output reference current, i βref represents the β-axis component of the virtual synchronous machine output reference current, i α (k+1) represents the α-axis component of the predicted current at time k+1, iβ (k+1) represents the β-axis component of the predicted current at time k+1, and λ represents the voltage balancing weight coefficient;

[0041] According to the predicted current at time k+1, the optimal voltage vector at time k+2 can be further predicted, and the predicted current at time k+2 can be obtained according to the optimal voltage vector; therefore, according to the expression of the cost function and the predicted current at time k+2, a cost function with delay compensation function can be constructed:

[0042] g'=|i αref -i α (k+2)|+|i βref -i β (k+2)|+λu c1 (k+2)-u c2 (k+2)|;

[0043] Among them, i α (k+2) represents the α-axis component of the predicted current at time k+2, i β (k+2) represents the β-axis component of the predicted current at time k+2, u c1 (k+2) represents the capacitor voltage on the DC side at time k+2, u c2 (k+2) represents the capacitor voltage on the DC side at time k+2.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1) The operation mechanism of the NPC three-level VSG single-phase bridge arm after a fault is analyzed. After the switching device fails, the DC side capacitor is used to form a virtual bridge arm and reconstruct it into a VSG bridge arm fault-tolerant structure.

[0046] 2) Establish a predictive current model under fault conditions, reconstruct the fault state space voltage vector, introduce the DC side midpoint capacitor voltage into the fault tolerance model cost function, reduce the capacitor voltage fluctuation, and realize the fault-tolerant operation of the VSG bridge arm.

[0047] 3) After the switching device fails, the NPC three-level VSG can operate continuously with fault tolerance. The effectiveness of the proposed model prediction fault-tolerant control strategy is verified through experiments, which improves the reliability of VSG operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 This is the NPC three-level VSG structure diagram.

[0050] Figure 2 This is the fault tolerance structure diagram of the VSG bridge arm.

[0051] Figure 3 This is the spatial voltage vector distribution diagram of the VSG fault-tolerant structure.

[0052] Figure 4 Predict the current control structure diagram for the VSG bridge arm fault tolerance model.

[0053] Figure 5 It is a hardware-in-the-loop semi-physical experimental platform for NPC three-level VSG fault-tolerant structure.

[0054] Figure 6 The frequency change waveform of VSG without midpoint voltage balance after phase A fault; (a) voltage change, (b) frequency change.

[0055] Figure 7 Dynamic waveform with midpoint voltage balance after phase A fault; (a) voltage change, (b) frequency change.

[0056] Figure 8 Comparison diagram of SVPWM control and MPC control after phase A fault; (a) voltage change, (b) frequency change.

[0057] Fig. 9 The power and current waveforms are as follows when the frequency drops from 50 Hz to 49.95 Hz.

[0058] Fig.10 The power and current waveforms are as the frequency increases from 49.95Hz to 50Hz.

[0059] Fig.11 The power and current waveforms are as the frequency increases from 50Hz to 50.05Hz.

[0060] Fig.12 The power and current waveforms are as follows when the frequency drops from 50.05 Hz to 50 Hz.

[0061] Fig.13 The power and current waveforms are as follows when the voltage drops from 110V to 104.5V.

[0062] Fig.14 The power and current waveforms are as follows when the voltage drops from 104.5V to 110V.

[0063] Fig.15 The power and current waveforms are as follows when the voltage rises from 110V to 115.5V.

[0064] Fig.16 The power and current waveforms are as follows when the voltage drops from 115.5V to 110V.

[0065] Fig.17 The current and power waveforms are shown below when the voltage is 110V and the frequency is 50Hz. DETAILED DESCRIPTION

[0066] 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.

[0067] An embodiment of the present invention provides a NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method, the steps of which are as follows:

[0068] Step 1: Analyze the operating mechanism of the NPC three-level VSG single-phase bridge arm when a fault occurs, and reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure; the NPC three-level VSG structure is as follows: Figure 1 , where the switch components may be damaged due to spike voltage and overcurrent, thereby reducing the reliability of the VSG and affecting the safe and stable operation of the power grid. When the NPC three-level grid-connected converter bridge arm (such as the A phase bridge arm) fails, the fast fuse F n1 and F n2 Fuse the fault bridge arm and turn on the corresponding bidirectional thyristor T rn , connect the faulty phase to the midpoint of the DC side capacitor through the filter inductor, and then reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure, such as Figure 2 ; Where n = a, b, c.

[0069] The functions of the VSG bridge arm fault-tolerant structure include three functional modules: active-frequency regulation, reactive-voltage regulation, and output reference current; regulating the mechanical torque T m and electromagnetic torque T e Realize active power-frequency modulation; given reactive power Q set , the grid reactive power Q and the voltage difference (VSG virtual potential and terminal voltage difference) constitute reactive voltage regulation; the output reference current is generated according to the VSG virtual potential and the terminal voltage difference; the expression is as follows:

[0070]

[0071] Where, J represents the VSG moment of inertia; represents the VSG angular frequency; represents the VSG angular acceleration; Dp Indicates the voltage droop coefficient; P set Indicates the given active power; M f represents the maximum mutual inductance between the excitation winding and the three-phase stator coil; θ represents the rotation angle of the VSG; i f represents the rotor excitation current, i represents the three-phase current, i aβ_ref represents the reference current, u abc Represents the three-phase voltage, R represents the parasitic resistance of the filter inductor, ω represents the angular frequency, L represents the filter inductor, and j represents the sign of the imaginary part.

[0072] Step 2: Establish a prediction current model when a single-phase bridge arm of the NPC three-level VSG fails, and construct a voltage vector of the fault-tolerant structure of the VSG bridge arm;

[0073] Depend on Figure 1 It can be seen that the voltage equation of the NPC three-level grid-connected converter is as follows:

[0074]

[0075] In the formula, u aβ =[u α ,u β ] T ,u α and u β represents the output voltage of the converter in the aβ coordinate system; i aβ =[i α ,i β ] T ,i α and i β Indicates the converter output current; e aβ =[e α , e β ] T , e α and e β Represents the grid voltage; R represents the parasitic resistance of the filter inductor, and L represents the filter inductor.

[0076] By performing Euler discretization on the voltage equation of the NPC three-level grid-connected converter, we can obtain:

[0077]

[0078] In the formula, T represents the sampling period; X(k) represents the value of variable X at time k; X(k+1) represents the value of variable X at time (k+1);

[0079] Simplify the discretized expression to get the predicted current at time k+1:

[0080]

[0081] Figure 2 In the fault-tolerant structure of the VSG bridge arm, there are eight switching devices. The fault phase is connected to the midpoint of the DC side capacitor in the "0" state, and the other phases have three states each, for a total of 9 switching states, such as Figure 3 ; Define VSG switch state as S i , the expression is as follows:

[0082]

[0083] When a bridge arm fault occurs in the VSG, the voltage component u aβ It is expressed as:

[0084]

[0085] Taking the a-phase bridge arm fault as an example, the voltage vector of the VSG bridge arm fault-tolerant structure is shown in Table 1:

[0086] Table 1 Voltage vector of VSG bridge arm fault tolerance structure when phase a bridge arm fails

[0087]

[0088]

[0089] Step 3: Use the sensor to collect the DC capacitor voltage at time k, and predict the DC capacitor voltage difference at time k+2 based on the DC capacitor voltage at time k; Since MPC has the advantages of no control delay, fast dynamic response, and strong ability to handle multi-constraint problems, a VSG fault-tolerant model predictive control strategy is proposed. The control structure is as follows: Figure 4 First, the sensor sampling converter outputs the three-phase grid-connected current i a 、i b 、i c and three-phase grid voltage e a 、e b 、e c , and generate reference current according to preset active power and reactive power. Collect the DC side capacitor voltage at the current moment, and predict the DC capacitor voltage difference at the next moment according to the capacitor voltage prediction model, as shown in formula (7).

[0090] Predict the DC capacitor voltage difference at time k+1:

[0091]

[0092] Among them, u c1 (k+1) represents the capacitor voltage on the DC side at time k+1, u c2 (k+1) represents the capacitor voltage on the DC side at time k+1, u c1 (k) represents the capacitor voltage on the DC side at time k+1, uc2 (k) represents the capacitor voltage on the DC side at time k+1, and C represents the capacitance.

[0093] The DC capacitor voltage difference at time k+2 can be obtained from the DC capacitor voltage difference at time k+1.

[0094] Step 4: Construct a cost function with delay compensation function based on the predicted current model and the DC capacitor voltage difference at time k+2, and select the voltage vector corresponding to the minimum value of the cost function as the optimal voltage vector to act at time k+1 to achieve fault-tolerant operation of the VSG bridge arm.

[0095] Using the grid voltage e α , e β and the current grid current i α ,i β The grid-connected current at the next moment is predicted. The sum of the absolute value of the error between the predicted current and the reference current and the absolute value of the predicted DC side capacitor voltage difference is defined as the cost function, as shown in formula (8), and the voltage vector corresponding to the minimum value is selected as the optimal voltage vector to act on the next moment.

[0096] According to the predicted current model, the predicted current at time k+1 can be obtained, and the cost function is constructed using the predicted current at time k+1, the reference current and the DC capacitor voltage difference at time k+1, which is expressed as:

[0097] g=|i αref -i α (k+1)|+|i βref -i β (k+1)|+λu c1 (k+1)-u c2 (k+1)| (8);

[0098] In the formula, i αref represents the α-axis component of the virtual synchronous machine output reference current, i βref represents the β-axis component of the virtual synchronous machine output reference current, i α (k+1) represents the α-axis component of the predicted current at time k+1, i β (k+1) represents the β-axis component of the predicted current at time k+1, and λ represents the voltage balancing weight coefficient.

[0099] The model predictive control strategy has a time delay. According to the predicted current at time k+1, the optimal voltage vector at time k+2 can be further predicted, and the predicted current at time k+2 can be obtained according to the optimal voltage vector. Therefore, according to the expression of the cost function and the predicted current at time k+2, a cost function with delay compensation function can be constructed:

[0100] g'=|i αref -iα (k+2)|+|i βref -i β (k+2)|+λ|u c1 (k+2)-u c2 (k+2)| (9);

[0101] Among them, i α (k+2) represents the α-axis component of the predicted current at time k+2, i β (k+2) represents the β-axis component of the predicted current at time k+2, u c1 (k+2) represents the capacitor voltage on the DC side at time k+2, u c2 (k+2) represents the capacitor voltage on the DC side at time k+2.

[0102] Experimental example verification

[0103] In order to verify the effectiveness of the fault-tolerant control strategy of the VSG prediction model, a MATLAB / Simulink simulation platform and an NPC three-level VSG fault-tolerant structure hardware-in-the-loop semi-physical experimental platform were built. Figure 5 The experimental platform is composed of Typhoon602+ and PE-Expert4. The control algorithm is executed on the PE-Expert4 processor board composed of DSP and FPGA control chips. The experimental parameters are shown in Table 2.

[0104] Table 2 Simulation and experimental parameters

[0105]

[0106] 1) Midpoint voltage balancing capability and comparison

[0107] In order to verify the midpoint voltage balancing ability after the midpoint voltage balancing term is introduced, the control effects without and with midpoint voltage balancing after the A phase bridge arm fails are compared. Figure 6 and Figure 7 . Figure 6 (a) Shows the power and voltage fluctuations when the VSG voltage fluctuates without midpoint voltage balancing after the A-phase bridge arm fails: 2.5s ago, the grid voltage was 110V with midpoint voltage balancing capability, the system operated stably, the DC side midpoint voltage was relatively balanced, and the voltage fluctuation was about 5V; within 3-4s, the midpoint voltage balancing item was removed, the grid voltage dropped to 104.5V, the reactive power increased, the grid current increased, and the DC side capacitor voltage fluctuation increased sharply and became uncontrollable; within 4-5s, the grid voltage recovered to 110V, the reactive power and grid current returned to their original values, and the DC side capacitor voltage fluctuation was completely out of control.

[0108] Figure 6(b) Shows the power and voltage fluctuations when the VSG frequency fluctuates without midpoint voltage balance after the A-phase bridge arm fails: before 2.5s, the grid-connected frequency is 50Hz, the system operates stably with midpoint voltage balance, the DC side midpoint voltage is relatively balanced, and the voltage fluctuation is about 5V; at 3-4s, the grid frequency drops to 45.95Hz, the active power increases, the grid current increases, and the DC side capacitor voltage fluctuation increases rapidly and becomes uncontrollable; at 4-5s, the grid frequency returns to 50Hz, the active power and grid current return to their original values, and the DC side capacitor voltage fluctuation becomes uncontrollable.

[0109] Figure 7 (a) is the power and voltage fluctuations when the midpoint voltage is balanced after the A-phase bridge arm fails and the VSG voltage fluctuates: before 2.5s, the grid voltage is 110V, the system is stable, the DC side midpoint voltage is relatively balanced, and the voltage fluctuation is about 5V; at 3-4s, the grid voltage drops to 104.5V, the reactive power increases, and the grid current increases. At this time, the DC side capacitor voltage fluctuation first increases to 10V and then stabilizes to 5V with a certain coefficient; at 4-5s, the grid voltage returns to 110V, the reactive power and grid current return to their original values, and the DC side capacitor voltage fluctuation first increases to 10V and then stabilizes to 5V.

[0110] Figure 7 (b) is the power and voltage fluctuations of the VSG with midpoint voltage balance after the A-phase bridge arm fails: before 2.5s, the grid frequency is 50Hz, the system works stably, the DC side midpoint voltage is relatively balanced, and the voltage fluctuation is about 5V; at 3-4s, the grid frequency drops to 45.95Hz, the active power increases, the grid current increases, and the DC side capacitor voltage fluctuation increases by about 10V; at 4-5s, the grid frequency returns to 50Hz, the active power and grid current return to the original values, and the DC side capacitor voltage fluctuation is reduced to 8V. From the comparison of the two groups, it can be seen that the DC side midpoint voltage balance term can effectively suppress the electric voltage fluctuation.

[0111] 2) Dynamic response speed comparison

[0112] Comparison of the effects of using MPC and SVPWM control strategies after a phase A bridge arm fault Figure 8 . Figure 8(a) is the response of VSG during voltage fluctuation. Before 2.5s, the grid voltage was 110V, the system was stable, the DC side midpoint voltage was relatively balanced, and the voltage fluctuation was about 5V. At 3-4s, the grid voltage dropped to 104.5V, and the reactive power rose to the peak value in about 0.43s under SVPWM control, and the MPC control took 0.28s. The grid current increased, and the DC side capacitor voltage fluctuation first increased to 10V and then stabilized to 5V. At 4-5s, the grid voltage recovered to 110V, and it took about 0.39s for the reactive power to recover to the original value under SVPWM control, and the MPC control took 0.25s. The DC side capacitor voltage fluctuation first increased to 10V and then stabilized to 5V.

[0113] Figure 8 (b) is the response of VSG when the frequency fluctuates. Before 2.5s, the grid frequency is 50Hz, the system works stably, the DC side midpoint voltage is relatively balanced, and the voltage fluctuation is about 5V; at 3-4s, the grid frequency drops to 45.95Hz, and the active power rises to the peak value in about 0.78s under SVPWM control, which is the same time as MPC control; at 4-5s, the grid frequency returns to 50Hz, and the active power change trend is also very similar, and the DC side capacitor voltage fluctuation is reduced to 8V. The results show that compared with SVPWM, MPC has a faster response speed in the "reactive-voltage" function and a comparable response speed in the "active-frequency" function.

[0114] 3) Active power-frequency regulation capability verification

[0115] Given the initial grid frequency f = 50Hz, active power P = 1000W, reactive power Q = 0Var. Fig. 9 When the power is stable, the frequency f drops from 50Hz to 49.95Hz, the active power P increases from 1000W to 1400W and then stabilizes to 1200W, the reactive power Q fluctuates around 0Var, the three-phase current increases from 4.9A and then stabilizes at 6A, the DC side capacitor voltage difference is about 5V, and the fluctuation amplitude is relatively stable, with no increasing trend, because the midpoint voltage balance item makes the optimal voltage vector select the voltage vector that produces a low midpoint current. When the power is stable, the frequency f rises from 49.95Hz to 50Hz, such as Fig.10, the active power P drops from 1200W to 1000W, the reactive power Q first rises to 100Var and then stabilizes to near 0Var, the grid-connected current drops from 6A to 4.9A, and the DC side capacitor voltage difference is about 4V, which is slightly reduced. When the grid frequency is lower than the rated frequency, the power is adjusted by calculating the active-frequency characteristics, and the frequency of the reference current is generated together with the difference between the given power and the output active power. Finally, according to the generated reference current, the MPC is used to select the optimal voltage vector to output more active power and achieve a new balance. The experimental results show that when the frequency drops, the proposed VSG fault-tolerant control strategy can achieve active-frequency regulation, and the NPC three-level grid-connected converter VSG has self-recovery ability.

[0116] like Fig.11 When the power is stable, the frequency f rises from 50Hz to 50.05Hz, the active power P drops from 1000W to 500W and then stabilizes to around 700W, the reactive power Q is around 0Var, the three-phase current drops from 4.9A first and then stabilizes to around 3.6A, and the DC side capacitor voltage difference is about 4V, which is slightly smaller than when it is stable; when the power is stable, the frequency f drops from 50.05Hz to 50Hz, such as Fig.12 , the active power P first rises from 700W and then stabilizes to around 1000W, the reactive power Q first decreases and then stabilizes to around 0Var, the grid-connected current rises from 3.6A to 4.9A, and the DC side capacitor voltage difference is about 5V, which is slightly increased compared to the stable state. When the grid frequency is higher than the rated frequency, the power is adjusted by calculating the active-frequency characteristics, and the amplitude of the reference current is generated together with the difference between the given power and the output reactive power. Finally, the MPC is used to select the optimal voltage vector based on the generated reference current to reduce the active power output. The experimental results show that the proposed VSG fault-tolerant control strategy can achieve active-frequency regulation when the frequency rises, which further proves the effectiveness of the strategy.

[0117] 4) Reactive power-voltage regulation capability verification

[0118] Given the initial grid voltage e = 110V, the grid voltage drops from 110V to 104.5V when the power is stable, such as Fig.13 , the reactive power Q increases from 600Var to 800Var and then stabilizes to around 700Var, the active power P decreases first and then stabilizes to around 800W, and the grid-connected current increases from 4.9A to around 5.8A. Fig.14When the power is stable, the grid voltage rises from 104.5V to 110V, the reactive power Q gradually stabilizes from 700Var to 600Var, the active power P fluctuates with the grid voltage and then returns to its original state, and the three-phase current rises from 5.8A to 4.9A. When the grid voltage is lower than the rated voltage, the power is adjusted by calculating the reactive-voltage characteristic, and then the current reference amplitude is generated through the regulator and integration. Finally, the MPC is used to select the optimal voltage vector according to the generated reference current to increase the reactive power output. The experimental results show that when the grid voltage drops, the reactive power will increase; when the grid voltage returns to the rated value from the drop, the reactive power will also return to the rated value.

[0119] When the power is stable, the grid voltage rises from 110V to 115.5V. Fig.15 , the reactive power Q drops from 600Var to 300Var and then stabilizes to around 500Var, the active power P first drops and then stabilizes to around 800W, and the grid-connected current rises from 4.9A to around 3.8A. Fig.16 , when the power is stable, the grid voltage drops from 115.5V to 110V, and the reactive power Q gradually stabilizes from 600Var to 800Var. The active power P fluctuates with the grid voltage and then returns to its original state, and the three-phase current increases from 3.8A to 4.9A. When the grid voltage is higher than the rated voltage, the power is adjusted by calculating the reactive-voltage characteristics, and then the current reference amplitude is generated through the regulator and integration to reduce the reactive power output. The experimental results show that when the grid voltage rises, the reactive power will decrease appropriately to adapt to the grid changes; when the grid voltage returns to the rated value from the rise, the reactive power will also return to the rated value, verifying that the proposed VSG fault-tolerant control strategy can automatically adjust the reactive power-voltage changes.

[0120] When the grid voltage is 110V and the frequency is 50Hz, the grid current steady-state waveform is as follows: Fig.17 In the fault-tolerant state, the three-phase grid current works stably, the THD is 3.6%, the active power and reactive power are stable, and the DC side capacitor voltage difference fluctuates around 1V, with a maximum fluctuation of 5V, which is 0.83% of the DC voltage source.

[0121] VSG provides inertia and damping for the renewable energy grid-connected system. However, VSG arm failure will cause the conventional grid-connected control strategy to fail, resulting in severe distortion of the output current waveform and affecting the normal operation of the grid-connected system. In order to enable fault-tolerant continuous operation after VSG arm failure, a fault-tolerant model predictive control method is proposed for the NPC three-level VSG arm failure problem.

[0122] a) After a switch device failure occurs in the VSG, the proposed control strategy reconstructs the NPC three-level VSG into a VSG bridge arm fault-tolerant structure, establishes a VSG bridge arm fault-tolerant model, and analyzes the voltage vector and working principle of the VSG bridge arm fault-tolerant structure.

[0123] b) Based on the model predictive control theory, a NPC three-level VSG bridge arm fault-tolerant model predictive control strategy is proposed to make the VSG operate stably after the bridge arm fault. Compared with the conventional SVPWM control strategy, the proposed control strategy shortens the time in the active power-frequency regulation link and has a fast dynamic response speed.

[0124] c) A DC side capacitor voltage balancing strategy is proposed, and the DC side capacitor voltage prediction term is added to the cost function. When a bridge arm fault occurs in the VSG and active-frequency and reactive-voltage regulation is performed, the DC side capacitor voltage is unbalanced under the conventional control strategy. The proposed control strategy can stabilize the DC side capacitor voltage fluctuation within 10V.

[0125] The experimental results show that the proposed control strategy can regulate the active power-frequency and reactive power-voltage after the VSG bridge arm fails, suppress the voltage fluctuation of the DC side capacitor, and effectively improve the operation reliability of VSG.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method, characterized in that: The steps are as follows: Step 1: Analyze the operating mechanism of the NPC three-level VSG single-phase bridge arm when a fault occurs, and reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure; Step 2: Establish a prediction current model when a single-phase bridge arm of the NPC three-level VSG fails, and construct a voltage vector of the fault-tolerant structure of the VSG bridge arm; The method for constructing the predicted current model is: From the NPC three-level VSG structure, we can know the voltage equation of the NPC three-level grid-connected converter: In the formula, u aβ =[u α ,u β ] T ,u α and u β represents the output voltage of the converter in the aβ coordinate system; i aβ =[i α ,i β ] T ,i α and i β Indicates the converter output current; e aβ =[e a , e β ] T , e α and e β Represents the grid voltage; R represents the parasitic resistance of the filter inductor, and L represents the filter inductor; By performing Euler discretization on the voltage equation of the NPC three-level grid-connected converter, we can obtain: In the formula, T represents the sampling period; X(k) represents the value of variable X at time k; X(k+1) represents the value of variable X at time (k+1); Simplify the discretized expression to get the predicted current at time k+1: The voltage vector of the VSG bridge arm fault-tolerant structure is constructed as follows: the VSG bridge arm fault-tolerant structure has eight switching devices, the fault phase is connected to the midpoint of the DC side capacitor in the "0" state, and the other phases have three states each, a total of 9 switch states; the VSG switch state is defined as S i , the expression is as follows: When a bridge arm fault occurs in the VSG, the voltage component u aβ It is expressed as: The voltage vector of the VSG bridge arm fault-tolerant structure corresponding to the a-phase bridge arm fault is shown in Table 1: Table 1 Voltage vector of VSG bridge arm fault tolerance structure when phase a bridge arm fails Step 3: Use the sensor to collect the DC side capacitor voltage at time k, and use the DC side capacitor voltage prediction model at time k+2 to predict the DC capacitor voltage difference; Step 4: Construct a cost function with delay compensation function based on the predicted current model and the DC capacitor voltage difference at time k+2, and select the voltage vector corresponding to the minimum value of the cost function as the optimal voltage vector to act at time k+1 to achieve fault-tolerant operation of the VSG bridge arm.

2. The NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method according to claim 1, characterized in that: The operating mechanism of the NPC three-level VSG single-phase bridge arm when a fault occurs is as follows: when the NPC three-level grid-connected converter bridge arm fails, the fast fuse F n1 and F n2 Fuse the fault bridge arm and turn on the corresponding bidirectional thyristor T rn , connect the faulty phase to the midpoint of the DC side capacitor through a filter inductor, and then reconstruct the NPC three-level VSG structure into a VSG bridge arm fault-tolerant structure; where n = a, b, c.

3. The NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method according to claim 1, characterized in that: The functions of the VSG bridge arm fault-tolerant structure include three functional modules: active-frequency regulation, reactive-voltage regulation, and output reference current; regulating the mechanical torque T m and electromagnetic torque T e Realize active power-frequency modulation; given reactive power Q set , the grid reactive power Q and the voltage difference constitute reactive voltage regulation; the output reference current is generated according to the VSG virtual potential and the terminal voltage difference; the expression is as follows: Where, J represents the VSG moment of inertia; represents the VSG angular frequency; represents the VSG angular acceleration; D p Indicates the voltage droop coefficient; P set Indicates the given active power; M f represents the maximum mutual inductance between the excitation winding and the three-phase stator coil; θ represents the rotation angle of the VSG; i f represents the rotor excitation current, i represents the three-phase current, i aβ_ref represents the reference current, u abc Represents the three-phase voltage, R represents the parasitic resistance of the filter inductor, ω represents the angular frequency, L represents the filter inductor, and j represents the sign of the imaginary part.

4. The NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method according to claim 1, characterized in that: The method for obtaining the DC capacitor voltage difference at time k+2 is: Predict the DC capacitor voltage difference at time k+1: Among them, u c1 (k+1) represents the capacitor voltage on the DC side at time k+1, u c2 (k+1) represents the capacitor voltage on the DC side at time k+1, u c1 (k) represents the capacitor voltage on the DC side at time k+1, u c2 (k) represents the capacitor voltage on the DC side at time k+1, and C represents capacitance; The DC capacitor voltage difference at time k+2 can be obtained from the DC capacitor voltage difference at time k+1.

5. The NPC three-level virtual synchronous machine bridge arm fault tolerance model predictive control method according to claim 4, characterized in that: The method for constructing the cost function with delay compensation function is: According to the predicted current model, the predicted current at time k+1 can be obtained, and the cost function is constructed using the predicted current at time k+1, the reference current and the DC capacitor voltage difference at time k+1, which is expressed as: g=|i αref -i α (k+1)|+|i βref -i β (k+1)|+λ|u c1 (k+1)-u c2 (k+1)|; In the formula, i αref represents the α-axis component of the virtual synchronous machine output reference current, i βref represents the β-axis component of the virtual synchronous machine output reference current, i α (k+1) represents the α-axis component of the predicted current at time k+1, i β (k+1) represents the β-axis component of the predicted current at time k+1, and λ represents the voltage balancing weight coefficient; According to the predicted current at time k+1, the optimal voltage vector at time k+2 can be further predicted, and the predicted current at time k+2 can be obtained according to the optimal voltage vector; therefore, according to the expression of the cost function and the predicted current at time k+2, a cost function with delay compensation function can be constructed: g'=|i αref -i α (k+2)|+|i βref -i β (k+2)|+λ|u c1 (k+2)-u c2 (k+2)|; Among them, i α (k+2) represents the α-axis component of the predicted current at time k+2, i β (k+2) represents the β-axis component of the predicted current at time k+2, u c1 (k+2) represents the capacitor voltage on the DC side at time k+2, u c2 (k+2) represents the capacitor voltage on the DC side at time k+2.