Fault-tolerant control method and system for open-circuit fault of parallel three-level ANPC inverter

Abstract

The invention discloses a fault-tolerant control method and system for an open-circuit fault of a parallel three-level ANPC inverter, and relates to the technical field of inverter fault control. According to the method, on the basis of sequence model predictive control, a multi-control-target value function is constructed in combination with fault-tolerant control, circulating current suppression and neutral-point potential balance. In order to realize synchronous preferential selection of an approximate priority control target, a series-parallel hybrid structure is established, so that the problem of complexity of weight factor design and control priority distribution in sequence model prediction is avoided. For fault conditions, a hybrid logic dynamic model considering open-circuit faults is established, and the problem of mathematical model mismatch under the fault conditions is corrected on the basis of SPH-SMPC. Under the condition of open-circuit fault of the single inverter of the parallel system, the existing sequence model prediction control method is improved to perform multi-target control, efficient fault-tolerant control is realized, circulating current is effectively suppressed, and grid-connected current output is recovered.

Description

Technical Field

[0001] The present invention relates to the technical field of inverter fault control, and in particular to a parallel three-level ANPC inverter open circuit fault tolerance control method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] As photovoltaic power generation gradually increases its share in the energy mix, 1500V string-type photovoltaic power plants, designed to both reduce costs and increase efficiency, are a future trend in photovoltaic development, aiming to improve the efficiency of photovoltaic power generation systems. However, traditional two-level topologies cannot withstand such high voltage stresses. Three-level topologies, with their advantages of a large number of voltage levels and reduced switching stress, have attracted considerable attention in the high-voltage, high-power segment. Currently, the main three-level photovoltaic inverter topologies in use are the T-type and the NPC topology. Because the vertical switching voltage stress of the current in the T-type topology is equal to the DC bus voltage, as the bus voltage increases to 1500V, the vertical switches require high-voltage switching diodes with higher switching losses, significantly increasing switching losses and reducing power density. The uneven distribution of voltage stress and losses in NPC switching diodes severely limits the increase in inverter capacity and switching frequency. To address this issue, researchers have proposed the Active Neutral Point Clamped (ANPC) topology. By replacing the clamping diodes in the NPC topology with switching devices, redundancy is achieved. By rationally switching switching states, power device losses can be balanced and voltage stress can be evened out. Existing research has demonstrated that the redundant state inherently provides ANPC with fault tolerance, resulting in superior reliability compared to the NPC topology in the event of an open-circuit fault. In summary, the ANPC topology, due to its superior performance, holds great promise for application in high-voltage, high-power applications such as 1500V photovoltaic power generation systems.

[0004] In high-voltage, high-power applications, limited by the power-carrying capacity of a single inverter, parallel inverter systems have attracted widespread attention as a method for increasing system capacity. However, because parallel systems lack electrical isolation measures such as transformers or isolated DC buses, some output harmonic currents can become zero-sequence circulating currents (ZSCCs). This distorts the grid current and seriously affects system reliability. While numerous studies have investigated the ZSCC issue, most have focused on analyzing parallel systems under normal operating conditions. In the event of a switch failure, many methods are no longer applicable due to model mismatch.

[0005] On the other hand, the substantial increase in the number of power switches in parallel systems, coupled with their inherent aging, has significantly increased the failure rate of power switch devices. Fault-tolerant control is a key approach to improving system reliability. Short-circuit faults are typically converted to open-circuit faults by hardware circuits or fuse protection. Therefore, much research has focused on fault-tolerant control strategies for open-circuit faults.

[0006] Currently, numerous fault-tolerant control methods have been developed for open-circuit faults in ANPC three-level converters. These technologies are primarily categorized into two main categories: hardware fault tolerance and software fault tolerance. Hardware fault tolerance adds redundant hardware to power devices, bridge arms, or submodules, allowing backup components to immediately take over in the event of a fault. However, these hardware-based fault tolerance methods inevitably increase system complexity, cost, and size. Furthermore, midpoint potential balance is fundamental to the stable operation of three-level converters and a prerequisite for circulating current suppression. Some research has addressed the issue of fault-tolerant control and midpoint potential balance in three-level ANPC inverters by modifying modulation algorithms.

[0007] However, many fault-tolerant control methods fail to consider parallel systems. If an open-circuit fault occurs in a parallel system, not only will the grid current be severely distorted, but the ZSCC will also increase significantly, seriously threatening the safe and stable operation of the system. Therefore, fault-tolerant control for ZSCC suppression in parallel systems is urgently needed. Current research on software-based fault-tolerant control for parallel systems has proposed reconstructing the switching sequence based on the area equivalence principle to achieve fault-tolerant control. A PI+feedforward zero-sequence circulating current controller was proposed to adjust the zero-sequence voltage duty cycle in real time. However, these studies only investigated horizontal switch faults in T-type converters and did not consider vertical switches. However, in parallel systems, software can be used to control the current of other healthy converters to compensate for open-circuit faults in the vertical switches of a single inverter. Furthermore, if ZSCC can be effectively suppressed and midpoint potential balance can be achieved, high-quality grid-connected circuit output can be achieved. With the development of microprocessors, model predictive control (MPC) has become a popular inverter control method due to its advantages such as fast dynamic response, nonlinear multivariable optimization, and flexibility. Regardless of the modulation method used in a parallel system, the modulation depth and precise control of the vector action time must be considered. However, the coupling between circulating current control and midpoint potential balance control makes it difficult to meet multi-objective control requirements. Multi-processor control (MPC) avoids the modulation step and eliminates the need to consider the complex decoupling of different control objectives. Therefore, using MPC to solve multi-objective control problems has attracted particular attention.

[0008] Some MPC methods use low common mode vectors to achieve circulating current suppression. However, when an open circuit fault occurs, some switch states cannot be output, so the method of using low common mode vectors to achieve circulating current suppression is no longer applicable. If three or more value functions are used for control, the problem of value function sorting or complex weight factor parameter adjustment is faced. The control effect of each control objective of traditional sequential model predictive control (SMPC) is related to the calculation order of the value function. The calculation order of low-priority objectives is later, and its control performance cannot be guaranteed. In addition, since the control objectives of circulating current suppression and midpoint potential balance have similar priorities, SMPC cannot solve the problem of equal weight control of similar priority control objectives. Based on this, how to improve the existing sequential model predictive control method to perform multi-objective control in the case of an open circuit fault in a single inverter in a parallel system, realize efficient fault-tolerant control and effectively suppress circulating current, and restore grid-connected current output has become a technical problem that needs to be solved urgently in the existing technology. Summary of the Invention

[0009] In response to the shortcomings of the prior art, the present invention aims to provide a method and system for fault-tolerant control of open-circuit faults in parallel three-level ANPC inverters. This method takes into account the control requirements of midpoint potential balance, circulating current suppression, and grid-connected control, designs a value function for multiple control objectives, selects the optimal vector through a sequence model prediction method for a series-parallel hybrid structure, and establishes a hybrid logic dynamic model that takes open-circuit faults into account for fault conditions, thereby achieving efficient fault-tolerant control of the ANPC inverter under open-circuit faults.

[0010] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0011] A first aspect of the present invention provides a method for controlling an open-circuit fault of a parallel three-level ANPC inverter, comprising the following steps:

[0012] A fault-free prediction model is constructed based on the topology of the parallel three-level ANPC inverter system;

[0013] Perform open circuit fault analysis on the fault-free prediction model to determine the cause of periodic grid-side current distortion caused by the open circuit fault;

[0014] Design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and establish a series-parallel hybrid structure based on the ranking of the multi-objective value function;

[0015] Determine the sequential model predictive control strategy for healthy inverters based on a series-parallel hybrid structure;

[0016] According to the series-parallel hybrid structure and the cause of the periodic distortion of the grid-side current caused by the open circuit fault, the sequence model prediction phase-loss fault-tolerant control strategy of the faulty inverter is determined;

[0017] Based on the sequence model predictive control strategy of the healthy inverter, the sequence model predictive phase loss fault-tolerant control strategy of the faulty inverter and the signal interaction between the inverters, the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system is determined.

[0018] A second aspect of the present invention provides a parallel three-level ANPC inverter open circuit fault tolerant control system, comprising:

[0019] a health status analysis module configured to construct a fault-free prediction model based on a topology of a parallel three-level ANPC inverter system;

[0020] a fault status analysis module configured to perform an open circuit fault analysis on the no-fault prediction model to determine the cause of the periodic distortion of the grid-side current caused by the open circuit fault;

[0021] a hybrid structure design module configured to design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and to establish a series-parallel hybrid structure based on the ranking of the multi-objective value functions;

[0022] a healthy inverter control module configured to determine a sequential model predictive control strategy for the healthy inverter based on a series-parallel hybrid structure;

[0023] a faulty inverter control module configured to determine a sequence model prediction phase-loss fault-tolerant control strategy for the faulty inverter based on a series-parallel hybrid structure and a cause of periodic distortion of grid-side current caused by an open-circuit fault;

[0024] The overall control module is configured to determine the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system based on the sequence model prediction control strategy of the healthy inverter, the sequence model prediction phase loss fault-tolerant control strategy of the faulty inverter, and the signal interaction between the inverters.

[0025] A third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps of the open circuit fault tolerance control method of the parallel three-level ANPC inverter as described in the first aspect of the present invention.

[0026] A fourth aspect of the present invention provides a device comprising a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of the open-circuit fault-tolerant control method of the parallel three-level ANPC inverter as described in the first aspect of the present invention are implemented.

[0027] One or more of the above technical solutions have the following beneficial effects:

[0028] The present invention discloses a fault-tolerant control method and system for open-circuit faults in parallel three-level ANPC inverters. This method addresses the problem that, in parallel ANPC three-level systems, open-circuit faults significantly increase zero-sequence circulating currents, severely distorting the grid-connected current, and posing a serious threat to the safe and stable operation of the system. While traditional model predictive control (MPC) can achieve multi-objective control under open-circuit faults, the design of multi-objective control weighting factors is difficult, making it difficult to achieve ideal control results. This invention proposes a series-parallel hybrid sequential model predictive control (SPH-SMPC) for three-level ANPC parallel systems. Based on SPH, this method considers midpoint potential balance, circulating current suppression, and grid-connected control requirements, designs value functions for multiple control objectives, and selects the optimal vector using the SPH-SMPC method for the series-parallel hybrid. This strategy fully considers the priorities of different control objectives, ensuring simultaneous optimization of similar-priority objectives. Furthermore, switching states are grouped and simplified according to the value function, avoiding complex weighting factor design and significantly reducing the computational complexity of the SPH-SMPC algorithm.

[0029] To achieve simultaneous optimization of approximate priority control objectives, this invention establishes a series-parallel hybrid structure, thus avoiding the complexities of weight factor design and control priority allocation in sequence model prediction. For fault conditions, this invention establishes a hybrid logic dynamic model that accounts for open-circuit faults. Based on SPH-SMPC, this method corrects the mathematical model mismatch problem under fault conditions and reconstructs the circulating current control value function, improving the accuracy of the mathematical relationship under fault conditions. This method ensures the accuracy of the mathematical model, improves the grid current quality, and enhances the circulating current suppression effect under fault conditions.

[0030] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0032] Figure 1 1 is a topological diagram of a parallel three-level ANPC inverter system in Embodiment 1 of the present invention;

[0033] Figure 2 1 is a conventional and improved LCL circulating current equivalent circuit diagram in Example 1 of the present invention;

[0034] Figure 3 This is a current path transformation diagram after an open circuit fault in the first embodiment of the present invention;

[0035] Figure 4 This is a schematic diagram of the SMPC principle in Example 1 of the present invention;

[0036] Figure 5 Schematic diagram of the SPH-SMPC structure in Example 1 of the present invention;

[0037] Figure 6 This is a control block diagram of the health status SPH-SMPC method in Example 1 of the present invention;

[0038] Figure 7 The sequence model prediction phase loss fault tolerant control strategy of the faulty inverter in the first embodiment of the present invention;

[0039] Figure 8 This is a block diagram of the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system in Example 1 of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0041] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations;

[0042] Example 1:

[0043] A first embodiment of the present invention provides a method for controlling an open-circuit fault of a parallel three-level ANPC inverter, comprising the following steps:

[0044] Step 1: Construct a fault-free prediction model based on the topology of the parallel three-level ANPC inverter system.

[0045] The fault-free prediction model specifically includes the output current model, the midpoint voltage model and the generator-side circulating current model.

[0046] In a specific embodiment, the topology of the parallel three-level ANPC inverter system is as follows: Figure 1 Each phase has six active switches S x1 、Sx2 、S x3 、S x4 、S x5 、S x6 , where x = a, b, c. (1) and (2) are the serial numbers of the inverters, (1) represents the parameters of ANPC three-level inverter-1, and (2) represents the parameters of ANPC three-level inverter-2; two inverters in parallel share the DC bus, AC bus and midpoint, and are directly connected without isolation equipment on the AC side. The DC bus side of the inverter contains two capacitors C in series. dc1 with C dc2 , whose voltages are u p and u n To suppress high-frequency circulating current, a new LCL filter with a return capacitor loop is used. Compared with the traditional LCL filter, the common point of the AC side filter capacitor of each inverter using the new LCL filter is returned to the DC side neutral point, providing a low-impedance path for high-frequency circulating current. Its main parameters include the machine-side inductance L m , filter capacitor C f , grid-side inductance L g . R m and R g L m and L g The parasitic resistance of np ,i 1abc ,i 2abc ,i gabc ,i zcf ,i zg They are the current flowing through the midpoint, the machine side current, the grid side current, the PCC current, the filter capacitor back-connected circulating current, and the grid side circulating current. ia 、U gabc and U C The voltages are the generator-side voltage (the voltage from the inverter output to the neutral point), the grid voltage, and the voltage across the filter capacitor and active damping resistor Rc. Udc is the DC bus voltage. Parallel three-level ANPC inverters have three switching states: [P], [O], and [N]. Table 1 summarizes the relationship between these switching states, input voltage, and switching transistors. Parallel three-level ANPC inverters have 27 different basic voltage vectors, which can be categorized as large, medium, small, and zero based on their voltage amplitudes.

[0047] Specifically, in vector control of a three-level inverter, the 27 spatial voltage vectors are typically categorized based on their vector modulus. Large vectors have the longest modulus and are located at the vertices of the hexagon, with an amplitude of 2 / 3*Udc. Medium vectors have an amplitude of 1 / √3*Udc. Small vectors have an amplitude of 1 / 3*Udc, and zero vectors have an amplitude of 0.

[0048] The 27 vectors can be divided into 7 groups according to their corresponding common-mode voltages, as shown in Table 1.

[0049] Table 1 Common mode voltage corresponding to voltage vector

[0050]

[0051] And define the switch state as follows:

[0052]

[0053] s a ,S b ,S c ∈{1,0,-1}y=1,2 (1).

[0054] Among them, S a,b,c The values ​​are 1, 0, and -1, representing the [P], [O], and [N] states respectively; the values ​​of y are 1 and 2, representing the inverter number.

[0055] In addition, the inverter side voltage can be expressed as:

[0056]

[0057] where u ia,b,c is the three-phase output voltage [u ia ,u ib ,u ic ] T .

[0058] The three-level inverter using an LCL filter with a loop capacitor has four important control objectives: tracking the set current, balancing the NP voltage, and suppressing the loop current of the filter capacitor loop and the grid-side loop current. According to Kirchhoff's law, the circuit equation of the inverter system is established as follows:

[0059]

[0060] Where i 1a,b,c For [i 1a ,i 1b ,i 1c ] T ;i 2a,b,c =[i 2a ,i 2b ,i 2c ] T ;U Ca,b,c =[U Ca , U Cb , U Cc ] TDiscretize according to the first-order forward Euler formula, perform Clark transformation to decouple the three-phase current control in the αβ0 coordinate system, and establish the output current model as follows:

[0061]

[0062] Midpoint voltage fluctuation is an inherent problem of the three-level topology compared to the traditional two-level topology. It is necessary to ensure that the DC bus capacitors evenly distribute the DC bus voltage. Assuming that the DC capacitors have the same capacitance, that is, C dc1 =C dc2 =C dc The midpoint current is defined as the sum of the three-phase midpoint currents. The midpoint current is equal to the difference between the currents flowing through the upper and lower capacitors of the DC bus. The relationship is as follows:

[0063]

[0064] Where i np_a 、i np_b 、i np_c are the three-phase midpoint currents, i Cdc1 、i Cdc2 They are respectively the DC side capacitor current. Among them, only when the switch state is [O], the switch tube S x2 , S x5 Simultaneously conducting or S x3 , S x6 At the same time, the current flows through the midpoint, otherwise i np_x = 0. Therefore, according to Kirchhoff's law, the relationship between the switch state and the midpoint current can be established as follows:

[0065]

[0066] The first-order forward Euler formula is also used to approximate the midpoint (Neutral-Point, NP) voltage model:

[0067]

[0068] In the formula, k represents the kth moment, and k+1 represents the k+1th moment, which is the next moment after the kth moment. np =u n -u p , is the voltage difference between the two capacitors on the DC side.

[0069] The above midpoint voltage model represents u np The value at time k+1 is determined by the value of the variable at time k.

[0070] In the inverter parallel system, the ZSCC problem caused by inconsistent parameters and switching states after parallel connection cannot be ignored. For the parallel system with traditional LCL filter, the circulating current transfer function of ZSCC is as follows:

[0071]

[0072] Where G zg1 (s) represents the circulation transfer function of ZSCC.

[0073] To suppress the high-frequency circulating current in the system, a new LCL filter with a return capacitor loop is used to provide a low-impedance loop for the circulating current. The transfer function of the improved LCL's ZSCC changes from first-order to third-order as follows:

[0074]

[0075] Where G zg2 (s) represents the circulation transfer function of the improved ZSCC.

[0076] A Bode plot comparison of the improved LCL zero-sequence circulating current transfer function and the traditional LCL zero-sequence circulating current transfer function is drawn. It is found that the traditional LCL filter and the improved LCL filter have the same circulating current suppression capability in the low-frequency band, both of which are first-order systems; in the high-frequency band, the circulating current suppression capability of the improved LCL filter is significantly higher than that of the traditional LCL filter, because a path of filter capacitor Cf is added to the circulating current equivalent circuit, which provides a low-impedance path for the high-frequency circulating current, thereby providing the improved LCL filter with a stronger high-frequency circulating current attenuation capability.

[0077] The circulating current equivalent circuits of the traditional LCL filter and the improved LCL filter are as follows: Figure 2 As shown, Figure 2 (a) is the traditional LCL circulating current equivalent circuit diagram. Figure 2 (b) is the improved LCL circulating current equivalent circuit diagram. Figure 2 It can be seen that although the improved LCL filter significantly improves the ability to suppress high-frequency circulating currents, it also brings about a new problem of internal capacitor circulating currents, making the circulating current problem more complicated.

[0078] Combine Figure 1 and Figure 2 For the parallel system in (b), the DC side midpoint is selected as the reference point. According to Kirchhoff's law, the following equivalent loop voltage equation is obtained:

[0079]

[0080] The capacitor circulating current is defined as the sum of the three-phase filter capacitor currents, the grid-side circulating current is defined as the sum of the three-phase grid-side currents, and the generator-side circulating current is defined as the sum of the capacitor circulating current and the grid-side circulating current, as shown in formula (11):

[0081] i z_inv =i zcf +i zg (11).

[0082] Where i z_inv is the machine side circulation, i zcf is the circulating current of the filter capacitor, i zg It is the grid side circulation.

[0083] according to Figure 2 (b) in the figure establishes the equivalent capacitance circulation equation as follows:

[0084]

[0085] The equivalent grid-side circulation equation is established as follows:

[0086]

[0087] It can be seen that the only excitation source of the inverter circulating current is the difference in common mode voltage, where U cmv is the common-mode voltage, defined as follows:

[0088]

[0089] like Figure 2 As shown in (b), the generator-side circulating current is the sum of the capacitor circulating current and the grid-side circulating current. Therefore, the two circulating current suppression goals can be achieved indirectly by directly suppressing the generator-side circulating current. The generator-side circulating current model is established by performing a first-order forward Euler approximation as follows:

[0090]

[0091] It is worth noting that, as shown in Equation (12), the only excitation source for the capacitor circulating current is the common-mode voltage, and the two inverters share the neutral line, that is, the neutral points O of the parallel inverters are connected together, eliminating the impact of inconsistent midpoint potentials on the zero-sequence circulating current. As shown in Equation (13), the only excitation source for the grid-side circulating current is the difference in the common-mode voltages of the two inverters. If the zero common-mode vector in Table 1 is used, there are only 7 available vectors, which cannot achieve a good control effect. If only low common-mode vectors are used, although the excitation source of the capacitor circulating current can be reduced, it is still impossible to maintain a small common-mode voltage difference. Subsequent experiments also verified this point. The method of using only low common-mode vectors cannot maintain a good control effect.

[0092] Step 2: Perform open circuit fault analysis on the fault-free prediction model to determine the cause of the periodic distortion of the grid-side current caused by the open circuit fault.

[0093] In a specific embodiment, Figure 1As shown in the figure, the three-level ANPC topology is highly symmetrical, with each phase consisting of six power switches and their freewheeling diodes. Considering the device losses and current direction according to the operating state, there are four main paths: Figure 3 In the figure, the blue solid line is the current path in a healthy state, where Figure 3 (a) in the equation is S a1 Fault path change diagram, Figure 3 (b) in the equation is S a2 Fault path change diagram, Figure 3 (c) in the equation is S a6 Fault path change diagram. When i x >0, the bridge arm state is switching between [P] and [O+]. [P] state switch tube S x1 、S x2 and S x6 On, O+ state switch tube S x1 、S x3 and S x6 conduction; i x <0, the bridge arm state is switching between [O-] and [N] states. x2 、S x4 and S x5 On, [N] state is represented by the switch tube S x3 、S x4 and S x5 By selecting this modulation method, S x1 、S x4 、S x5 、S x6 Four switching tubes switch at the base frequency, S x2 、S x3 The two switching tubes switch at high frequency, minimizing the overall switching loss.

[0094] The open circuit fault in this embodiment refers to the open circuit fault of any single switch device of two parallel ANPC three-level inverters, and it is assumed that the freewheeling diode is not affected by the fault. In this embodiment, D represents the freewheeling diode, D a 、D b 、D c Respectively represent the freewheeling diodes of phases a, b, and c, with D a For example, D a1 、D a2 、D a3 、D a4 、D a5 、D a6 Respectively represent a corresponding S a1 、S a2 、S a3 、S a4 、S a5 、Sa6 The serial number of the freewheeling diodes of the six switching tubes, D b 、D c Same thing.

[0095] According to the S of the first inverter a1 、S a2 、S a6 The reasons for the waveform change before and after the open circuit fault are further analyzed. Due to the symmetry of the three-level ANPC topology, S a1 、S a2 、S a6 When the circuit is open, i a >0 when the current path is affected, i a <0 has no effect on the current path. Similarly, S a3 、S a4 、S a5 Open circuit fault only affects i a <0 when the current path, i a >0 has no effect on the current path. a1 、S a2 、S a6 Take open circuit fault as an example for analysis. a1 、S a2 、S a6 The current path during an open circuit fault is as follows Figure 3 As shown, the fault analysis at other locations is similar. Figure 3 (a) is when S a1 When an open circuit fault occurs, the fault bridge arm cannot be in the switching state [P]. a When >0, large vectors [PPN], [PNN], and [PNP] cannot be realized, medium vectors [PON] and [PNO] cannot be realized, and small vectors [PPO], [POO], and [POP] cannot be realized. At this time, phase a cannot be connected to the positive polarity DC bus, but is connected to the neutral point of the DC bus, and the current path is changed to flow through the conducting S a6 and freewheeling diode D a3 , or flows through D a5 and the conducting S a2 , the [P] state is forced to convert to the [O] state, resulting in an unbalanced output current; when i a <0, S a1 If not triggered, the current direction remains unchanged, that is, the switch state [P] remains unchanged; Figure 3 (b) is when S a2 When an open circuit fault occurs, i a >0, the fault bridge arm cannot achieve the switching state [P], nor can it achieve the [O-] state, and the current path is changed to flow through the conducting S a6 and Da3 or flows through D a4 and D a3 ;i a <0, the current flows through the freewheeling diode D a2 , the switch state does not change. Figure 3 (c) in the equation is when S a6 When an open circuit fault occurs, i a When >0, the fault bridge arm cannot achieve the switching state [O-], and the current path changes to flow through D a4 and D a3 , [O] is forced to convert to [N] state; i a < 0, the switch state is unaffected. Similar analysis can be performed for faults in other locations. The conclusions are summarized in Table 2. It can be seen that the open-circuit fault device and the positive and negative current jointly determine the path of the current after the fault, which in turn changes the switch state after the fault. The number of output vectors decreases accordingly, and the original mathematical model is no longer accurate.

[0096] Table 2 Relationship between faulty devices and output status

[0097]

[0098] From the above analysis, we can see that S a1 or S a2 After the fault, the [P] state cannot be achieved, so the modulation range is reduced by half in each power frequency cycle. a5 or S a6 The fault does not affect the modulation range and has a minimal impact on the system. However, other switch failures can affect the modulation range within each half-cycle of the power frequency, preventing the system from outputting the required voltage level. This can cause significant distortion in the output phase current, leading to midpoint voltage imbalance and large circulating currents, affecting the normal operation of the three-level ANPC inverter. Therefore, it is necessary to establish appropriate methods to resolve system open-circuit faults.

[0099] Step 3: Design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and establish a series-parallel hybrid structure based on the ranking of the multi-objective value function.

[0100] Step 3.1: Design a multi-objective value function based on fault-tolerant control, circulating current suppression and midpoint potential balance. In a specific embodiment, the control objectives of the parallel three-level ANPC inverter system using an improved LCL filter are multiple and complex, with a total of four control objectives: internal capacitor circulating current suppression of the LCL filter, external grid-side circulating current suppression, midpoint voltage balance, and grid-side current quality. From the analysis of step 1, it can be seen that internal and external circulating current suppression can be achieved simultaneously by controlling and suppressing the machine-side circulating current. Therefore, three value functions are designed to achieve simultaneous optimization of the four objectives. Specifically, the machine-side circulating current value function J iz, midpoint voltage value function J np and grid-connected current value function J i .

[0101] (1) Generator-side circulation value function J iz .

[0102]

[0103] (2) Midpoint voltage value function J np .

[0104]

[0105] (3) Grid-connected current value function J i .

[0106]

[0107] The predicted value of the machine-side circulation at time k+1 is calculated by the discrete model formula (15) established above: z_inv (k+1), Equation (16) The machine side circulation value function J iz Function value J at time k iz (k) is the expected value of the generator side circulation current at the k+1 moment compared to the predicted value. The smaller the value of the generator side circulation current value function, the closer the generator side circulation current control effect at the next moment is to the expected value, that is, the better the control effect. The same applies to formulas (17) and (18). The midpoint voltage value function J in formula (17) is np Function value J at time k np (k) is the predicted value u of the midpoint voltage at time k+1 np (k+1) and the expected value of the midpoint voltage at time k+1. Formula (18) Grid current value function J i Function value J at time k i (k) is the predicted value of the grid-connected current at time k+1. Through the above three value functions, circulating current suppression, midpoint potential balance and high-quality output of grid-connected current can be achieved respectively. Traditional finite-control-set model predictive control (FCS-MPC) is a value function formed by weighted summation of the value functions of different control objectives. However, when there are three or more value functions, the design of the weight factor becomes complicated, while sequential model predictive control (SMPC) can be designed as three sub-value functions for hierarchical control, which has a smaller computational burden and difficulty in adjusting parameters. The working principle of SMPC is as follows: Figure 4As shown: The cost function J1 of the first layer evaluates all spatial voltage vectors, selects N1 optimal solutions, and inputs them into the cost function J2 of the second layer to evaluate N1 spatial voltage vectors, and selects N2 optimal solutions, and so on, and finally selects the optimal voltage vector. The series structure of the traditional SMPC determines that it can only complete the priority division and sub-cost function evaluation of each control target in a unique order. However, since it is difficult to distinguish the priorities of the three cost functions, it is necessary to consider how to perform hierarchical sorting; in addition, some control targets with similar priorities should be controlled with equal weights as much as possible, rather than forcibly dividing priorities. Therefore, the series sequence structure of the SMPC may not be absolutely reasonable. In summary, the traditional SMPC has two shortcomings:

[0108] (1) The fixed priority order of SMPC makes the control effect of low priority poor.

[0109] (2) Control objectives with similar priorities in SMPC are difficult to be controlled with equal weights.

[0110] Step 3.2: Establish a series-parallel hybrid structure based on the ranking of multi-objective value functions.

[0111] In a specific implementation, the cost functions need to be sorted. Since the three cost functions have no obvious priority, the three cost functions still need to be sorted. By observing the machine-side circulation sub-cost function formula (16), the formula contains the sum of the input variables, that is, the cost function values ​​of some input variables are the same, for example, [0-11] has the same control effect as [01-1] and

[000] . Therefore, the elements of the finite control set can be grouped according to the cost function. As shown in Table 1, J iz The 27 inputs can be divided into 7 groups, so this layer only needs to calculate the value function 7 times to get the optimal solution and the suboptimal solution, without having to calculate it 27 times. iz The grouping criteria can be expressed as follows:

[0112] J iz (S a,b,c (k))→J iz (S a (k)+S b (k)+S c (k)) (19).

[0113] Similarly, the absolute value of the input variable exists in the midpoint voltage value function (17), so the value function values ​​of some input variables are also consistent, such as [11-1], [1-1-1] and [1-11]. np The finite control set of is divided into 8 groups, as shown in Table 3. np The grouping method can be expressed as follows:

[0114] J np (S a,b,c (k))→J np (|S a,b,c (k)|) (20).

[0115] Table 3 Midpoint voltage value function grouping

[0116] Voltage Vector Group 000 I 001，00-1 Ⅱ 010，0-10 III 100，-100 Ⅳ 011，01-1，0-11，0-1-1 V 101，10-1，-101，-10-1 VI 110，1-10，-110，-1-10 VII 111，-1-1-1，11-1，-111，1-11,1-1-1，-11-1，-1-11 VIII

[0117] According to formula (18), we can know that J i Cannot be grouped, so the value function J needs to be iz With J np Place it in a relatively forward position and i Placed in the last layer to ensure that the controller finally outputs a single optimal solution. iz or J np Placed in the last layer, there may be several optimal solutions. It should be noted that this method only needs to ensure that J i Just place it in the last layer, and the outer layer is for midpoint voltage control and circulating current control.

[0118] Aiming at the limitations of SMPC in multi-objective collaborative control, this embodiment proposes an improved SPH-SMPC method, the principle diagram of which is shown in the following figure: Figure 5 As shown in the figure, the series-parallel hybrid structure consists of an outer control objective set and an inner control objective set in series. Within the outer control objective level, the generator-side circulating current value function and the midpoint voltage value function are used for midpoint voltage balancing control and circulating current suppression. A parallel structure is used to prioritize the generator-side circulating current value function and the midpoint voltage value function. As previously analyzed, both midpoint voltage balancing control and circulating current suppression can be reduced in computational effort by grouping input vectors, and they have similar control priorities. The grid current value function, however, is placed in the inner layer because its vectors cannot be grouped. Therefore, midpoint voltage balancing and circulating current suppression can be considered as the outer control objective set, while the grid current output can be considered as the inner control objective set. The SPH-SMPC is a series structure consisting of an outer control objective set and an inner control objective set, with the outer control objective set taking precedence over the inner control objective set. Within the same control objective set, multiple sub-objectives have similar priorities and are prioritized using a parallel structure. To eliminate the directional bias of priority sorting on control effects in traditional series structures, a set of series subsystems is constructed according to all possible priority permutations and combinations within the same control objective set. Each subsystem will produce output results that focus on different control objectives due to differences in priority sequences. Subsequently, by obtaining the intersection of the output results of each subsystem, this method mathematically achieves equal weighting of each priority permutation and combination, thereby ensuring that the system has a balanced control capability for all control objectives.

[0119] Step 4: Determine the sequential model predictive control strategy for the healthy inverter based on the series-parallel hybrid structure.

[0120] In a specific embodiment, Figure 6 The schematic diagram of the SPH-SMPC method for healthy operation is shown. The overall series structure control for the healthy state is divided into two layers of series structure: the outer layer is a series-parallel structure with generator-side circulating current control and midpoint voltage control, and the inner layer is the output current control target.

[0121] There are two control objectives with similar priorities in the outer control objective set. All possible priority orders within the set include circulating flow control-midpoint potential control and midpoint potential-circulating flow control. This is used to establish a sub-series structure for optimizing the value function. For circulating flow control-midpoint potential control, the 27 elements of the original finite set can be divided into 7 groups according to Equation 19. One input variable in each group is taken to reconstruct a simplified finite set. After calculating the simplified finite set, the calculation results are arranged in ascending order, which can be expressed as:

[0122] n iz =sort{J iz (S h (k+1),x(k))|h=1,2,…,7}(21).

[0123] Where S h (k+1) is the switch state corresponding to the hth vector. The k1 minimum suboptimal solutions that satisfy the circulating current control are obtained and the finite control set is reconstructed. The input value midpoint voltage sub-cost function is subjected to optimal control. The calculation result can be expressed as:

[0124]

[0125] The k3 minimum suboptimal solutions that satisfy the circulating current control can be obtained, and the voltage vector set S1 is obtained.

[0126]

[0127] Similarly, in another sub-series system, the midpoint voltage sub-cost function selects k2 vectors by preference, reconstructs the finite set and then inputs the circulating current control sub-cost function to obtain k3 minimum sub-optimal solutions, and obtains the voltage vector set S2 output by the midpoint potential-circulating current control series structure. The output voltage vector set M of the outer control target set is the intersection of S1 and S2

[0128] M=S1∩S2(24).

[0129] It should be noted that the number of vectors in the intersection is set to k4. If S1 and S2 have no intersection vectors or the number of intersection vectors is less than k4, in order to reduce the amount of calculation, the voltage vector set M is set to the union of S1 and S2.

[0130] Compared with the traditional MPC method and the SMPC method, the advantages of this method are as follows:

[0131] (1) Simplify the computational complexity: By adopting a grouping method, the computational complexity is reduced. Taking k1 = 3, k2 = 4, k3 = 9, k4 = 5 as an example, in the outer control target set, on the circulating current control - neutral point potential control path, the circulating current control only needs to be calculated 7 times, and the neutral point potential control needs to be calculated at most 19 times; similarly, on the neutral point potential - circulating current control path, the neutral point potential control only needs to be calculated 8 times, and the circulating current control needs to be calculated at most 16 times. In the inner control target level, only k3 times of calculation are required, that is, 9 times. Therefore, to achieve the control target, it needs to traverse at most 59 times. Compared with the traditional MPC control of three - target control which requires 81 times of calculation, the computational complexity is reduced by at least 27%.

[0132] (2) Simplify the tuning of weight factors: In the outer control target set, k1 minimum sub - optimal solutions are selected through the value function Jiz and enter the value function Jnp, where 1 < k1 < 7. The value function Jnp selects k3 minimum sub - optimal solutions for output to the inner layer. Through experimental testing, generally k3 = 9 can be selected. Compared with the traditional FSC - MPC method, the trial time of weight factors is greatly reduced.

[0133] (3) Multi - target precise control: This series - parallel hybrid - structure SMPC control method can achieve equal - weight control when the control target priorities are similar, avoiding the forced hierarchical sorting of the SMPC method. At the same time, it also retains the advantages of the SMPC method, such as simple structure, small computational complexity, and high efficiency.

[0134] Step 5: Determine the sequence - model - prediction open - circuit fault tolerant control strategy for the faulty inverter according to the series - parallel hybrid structure and the reason for the periodic distortion of the grid - side current caused by the open - circuit fault.

[0135] Table 4 shows the MATLAB / Simulink simulation parameters of the two - machine parallel ANPC three - level inverter. The simulation parameters and control strategies of the two inverters are exactly the same.

[0136] Table 4 Simulation Parameters of the Parallel Three - Level ANPC Inverter System

[0137]

[0138] Through theoretical analysis, it shows that the three - level ANPC topology has switch redundancy characteristics in the [O] operating state. When S x5 or S x6 has an open - circuit fault, the system can still maintain the original modulation ratio operation and the performance degradation is small. In contrast, the faults of other switching tubes will cause a non - negligible deterioration of the system performance. To effectively verify the engineering value of the fault - tolerant control strategy, in this paper, the first inverter S with typical damage characteristics is selecteda3 The open circuit fault is the research object. Under this fault condition, the system will also cause significant circulation phenomenon. The grid-side current of the two inverters before and after the open circuit fault occurs and grid current i gabc , DC bus midpoint voltage u p and u n Waveform, capacitor circulating current i zcf (1) 、i zcf (2) and the grid-side circulation i zg The waveforms of the output current and the related waveforms during SPH-SMPC control indicate that before the fault occurs, both the FCS-MPC method, which only controls the output current and midpoint voltage, and the healthy state SPH-SMPC method mentioned above, ensure high-quality grid-connected current, and the midpoint voltage and grid-side circulating current are also well controlled. After an open-circuit fault occurs, the grid-side current of the first inverter experiences significant distortion, causing the combined grid current to halve in amplitude during the negative half-cycle of phase A. Phases B and C also experience some distortion. Finally, under FCS-MPC control, a large capacitive circulating current is generated. However, under SPH-SMPC control, the grid-side circulating current is suppressed to a low level, and the midpoint voltage remains balanced, with only a slight increase during the fault half-cycle.

[0139] The open circuit fault analysis of the fault-free prediction model reveals two reasons for the periodic distortion of the grid-side current caused by open circuit faults and their corresponding solutions:

[0140] Reason 1: The modulation index is reduced. When the switch tube S x1 , S x2 , S x3 , S x4 After an open-circuit fault occurs, the system modulation index decreases mainly due to two factors: first, the fault causes the loss of effective switching state, which reduces the modulation range of the faulty phase inverter to 50% of the original within the fault half-cycle; second, under the constraint that the DC bus voltage cannot be adjusted, the faulty phase inverter cannot output the original reference voltage value, resulting in a decrease in the overall system modulation index.

[0141] Solution 1: For a single inverter system, switch tube S x1 , S x2 , S x3 , S x4After an open-circuit fault, derating operation fault tolerance will result in output power loss. While increasing the DC bus voltage can restore output, this is difficult to achieve in actual engineering and is therefore not suitable for 1500V parallel three-level ANPC inverter systems. In a parallel system, although the output capacity of the faulty inverter cannot be directly restored, the given reference value of the healthy inverter can be compensated within the modulation range of the normal inverter. However, the reference current of the normal inverter can be dynamically corrected, and its remaining modulation margin can be used to compensate for the voltage loss of the fault phase in the fault half-cycle. By superimposing and synthesizing the output waveforms of the two inverters, the grid current ig fault-tolerant control is ultimately achieved. The reference value correction under the fault-tolerant control of the healthy inverter is as follows:

[0142]

[0143] Reason 2: The switch state cannot be realized after the open circuit fault. When a fault occurs, the relationship in Equation (2) no longer holds true, and the output voltage on the generator side is no longer linearly related to the switch state, which affects the subsequent mathematical modeling. Therefore, it is necessary to correct the phase loss formula for the faulty inverter.

[0144] Solution 2: The effective switching vector set of the faulty inverter has been reduced from 27 to 18. The lack of small vectors causes periodic fluctuations in the midpoint voltage. Since the midpoint potentials of the two inverters in the parallel system are connected and the midpoint voltages of the two inverters are completely consistent, the midpoint potential balance can be achieved by controlling the normal inverter. After an open circuit fault occurs, the relationship in equation (2) no longer holds. Therefore, the mathematical model (3) for the output current under open circuit fault needs to be modified as follows, where S health1 With S health2 is the healthy two-phase switching state, u iOC is the fault phase voltage.

[0145]

[0146] Based on the fault tolerance mathematical model (26), Taking the case of an open circuit fault as an example, the original mathematical model (2) no longer holds true after an open circuit fault occurs, that is, the original functional relationship between the inverter output voltage and the switch state has become invalid, which will affect the output performance. However, since the inverter switch tube is damaged after the fault occurs, the mathematical model cannot be repaired. Therefore, in order to correct the circulating current control value function of the faulty inverter and realize the fault-tolerant control of the circulating current, the output relationship of formula (15) is rewritten as the machine-side voltage u ia , that is, the voltage signal of phase A in the original formula is changed to the sampling signal of the phase A voltage sensor, which is expressed as follows:

[0147]

[0148] The value function is also rewritten as:

[0149]

[0150] It can be seen from formula (27) that the input finite control set of the faulty inverter does not include the state of the faulty phase, that is, the value function formula (28) cannot control the conduction of the faulty phase. However, the circulating current suppression is only determined by the switching state of the healthy phases (phases B and C), and has nothing to do with the switching state of the faulty phase (phase A). Based on this characteristic, the phase-loss circulating current control is first placed in the outer control. The outer value function outputs several minimum suboptimal solutions (each minimum suboptimal solution only contains the states of phases B and C), and then supplements the phase A signal. Finally, the complete three-phase switching signal is input to the inner control for subsequent processing. It can be observed from formula 27 that the input variables of the phase-loss circulating current control can still be grouped. The grouping standard is changed from formula (19) to:

[0151]

[0152] In a healthy state, the 27 vectors are divided into seven groups based on their common-mode voltages. After a fault occurs on phase A, the phase A state is lost, and its phase-loss finite control set has nine elements. Based on the states of phases B and C, only five CMVs, ranging from -2Udc / 3 to 2Udc / 3, can be generated. Therefore, the vectors are divided into five groups based on their phase-loss CMVs.

[0153] The grouping rules are summarized in Table 5. After simplification, the number of variables in the finite control set is reduced from 9 to 5. Several minimum suboptimal solutions are selected using the value function (27). The state of the faulty phase is then restored to the three-phase finite control set and the next level of control is entered. This grouping method still helps reduce the number of value function calculations and thus the computational complexity.

[0154] Table 5 Phase-missing vector grouping table

[0155] Phase loss voltage vector Group Original three-phase vector -1-1 I(-1-1) 1-1-1，0-1-1，-1-1-1 -10，0-1 Ⅱ(-10) 1-10，0-10，-1-10，10-1，00-1，-10-1 00，1-1，-11 Ⅲ(00) 100，000，-100，11-1，01-1，-11-1，1-11,0-11，-1-11 10，01 Ⅳ(10) 110，010，-110,101，001，-101 11 V(11) 111，011，-111

[0156] Based on the above analysis, in order to solve the above two problems, the control strategy of the fault inverter is modified as follows: Figure 7 As shown, in order to keep the algorithm simple, the healthy inverter control strategy remains unchanged.

[0157] Step 6: Determine the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system based on the sequence model predictive control strategy of the healthy inverter, the sequence model predictive phase loss fault-tolerant control strategy of the faulty inverter, and the signal interaction between the inverters.

[0158] In a specific embodiment, the overall control block diagram of the parallel three-level ANPC inverter system is as follows: Figure 8As shown in the figure, an open-circuit fault occurs in the first inverter. The green part shows the SPH-SMPC method used by the healthy inverter, the red part shows the SMPC phase-loss fault-tolerant control strategy of the faulty inverter, and the purple part shows the signal interaction between the two inverters.

[0159] For the healthy inverter, the reference given during fault-tolerant control is compensated according to formula (25), and the rest of the control strategy still adopts the SPH-SMPC method.

[0160] In the SPH-SMPC control and measurement part of the healthy inverter, the outer control part of the healthy inverter is the machine-side circulating current value function J iz and midpoint voltage value function J np Specifically, by inputting the circulating current signal i z_inv , capacitor voltage signal U C , DC bus voltage signal U dc , output current signal i 1a,b,c , midpoint voltage signal u np Perform outer layer control. The first calculation path is to calculate the side circulation value function J iz Get the optimal k1 vectors that meet the requirements of the function, and then input these k1 vectors into the midpoint voltage value function J np Obtain the optimal k3 vectors that meet the requirements of the function; the second calculation path is to first calculate the midpoint voltage value function J np Get the optimal k2 vectors that meet the requirements of the function, and then input these k1 vectors into the generator side circulation value function J iz Obtain the optimal k3 vectors that meet the requirements of the function. The two paths each calculate a vector set containing k3 voltage vectors, and then take the intersection of the vector sets. If the number of vectors in the intersection is greater than or equal to k4, then these k4 vectors are input to the inner control; if the number of vectors in the intersection is less than k4, then take the union of the two vector sets and input it to the inner control. The inner control is responsible for the grid current quality control, and its input includes several vectors obtained by the outer control in the previous step and the capacitor voltage signal U required for the operation. C , DC bus voltage signal U dc , output current signal i 1a,b,c , and finally through the grid current value function J i The optimal switching sequence of the healthy inverter is obtained by selection.

[0161] exist Figure 8 In the signal interaction part of the two inverters, the deviation between the output current of the faulty inverter and the reference current signal is injected into the healthy inverter as a compensation signal, so that the healthy inverter can compensate for the power output shortfall of the faulty inverter.

[0162] For a faulty inverter, use Figure 7 The missing phase SMPC method shown in Table 5 is used as the missing phase vector set, which contains the vectors of phase B and phase C, and then the missing phase circulating current value function is used. The optimal k1 vectors are obtained by selection. However, since the obtained vector lacks the A-phase signal, the A-phase signals of the missing k1 vectors are supplemented according to Table 5, and the vectors are restored to a three-phase finite set, which is then input into the grid current value function J. i Perform optimal control of the faulty inverter.

[0163] For convenience, four methods are defined: Method 1 is a non-circulating current control method that uses the FCS-MPC method with midpoint potential balance control and output current control. Its candidate vectors include all 27 vectors. Method 1 is the control method used in traditional inverter control and does not consider parallel circulating current and fault-tolerant control issues. Method 2 is a simplified SMPC method (Sample Sequence Model Predictive Control, SSMPC). Its candidate vectors include 19 low common-mode voltage vectors, namely the three groups III, IV, and V in Table 1. Method 2 is a general approach to inverter parallel circulation control in existing research, which does not consider the impact of inverter faults on the system. Method 3 is the series-parallel hybrid sequential model predictive control (SPH-SMPC) proposed in this paper, whose candidate vectors are all 27 vectors. Method 4 is the fault-tolerant series-parallel hybrid sequential model predictive control (FT-SPH-SMPC), which adds the fault-tolerant phase loss control method proposed in step 4 to the series-parallel hybrid structure model predictive control. The outer control of the faulty inverter uses the phase loss control set. Methods 1, 2, 3, and 4 all achieve fault tolerance by output compensation using the normal inverter. Method 1 has no circulating current control. After an open circuit fault, the common-mode voltage difference between the two inverters is distorted, resulting in a large grid-side circulating current spike. Method 2 uses 19 low common-mode vectors to suppress the circulating current. Under normal conditions, the common-mode voltage of the inverter can be controlled. However, after the fault, the corresponding relationship between the output switching state and the output phase voltage (Equation (2)) has become invalid. In addition, Method 2 has fewer available vectors, which increases the common-mode voltage difference. Therefore, a higher grid-side circulating current spike is generated than in Method 1. Method 3 always uses the value function to suppress the circulating current. Therefore, the three circulating currents during the fault period are only slightly larger than those during normal operation, and no large spikes are generated, achieving good control of the three goals. Method 4 performs fault-tolerant correction on the basis of Method 3. On the basis of ensuring good control of the goals, it further reduces the grid-side circulating current and improves the THD of the grid-connected current. Compared with Method 3, the circulating current spike is reduced by 21%.

[0164] To further compare the effects of the four methods on circulating current suppression and power DC, the data of the four methods are summarized in Table 6. It can be seen that method 4 has the best circulating current suppression capability and the best THD.

[0165] Table 6 Simulation results of grid-side circulating current peak and grid-connected circuit THD during fault-tolerant operation of four methods

[0166] method <![CDATA[Fault-tolerant operation grid-side circulating current i zg Spike]]> Fault-tolerant operation THD Method 1 19.7A 3.43％ Method 2 43.2A 3.77％ Method 3 10.2A 3.28％ Method 4 8.1A 2.83％

[0167] Comparing the grid-side circulating current under the four control methods, Methods 3 and 4 both maintain low grid-side circulating current after a fault. During fault-tolerant operation, switching from Method 3 to Method 4 is defined as Switch 1, and switching from Method 1 to Method 4 is defined as Switch 2. The grid-side circulating current is significantly reduced before and after the switch, and the control methods exhibit good stability and speed.

[0168] The current given by inverter 2 is i g *(2) From the simulation waveforms of the circulating current before and after the sudden change to 40A, it can be seen that the current of inverter 1 is given by i g *(1) The value remains constant at 70A. Under conditions of given current imbalance, the grid-side circulating current under method 1 shows significantly increased amplitude fluctuations, with the peak amplitude increasing to 37.7A. Under method 3, the grid-side circulating current amplitude approaches that under balanced current conditions, with the peak amplitude being approximately 9.7A. Under method 4, the circulating current peak is essentially eliminated, further reducing the grid-side circulating current to 8.0A. Simulation results demonstrate that the proposed method maintains excellent circulating current suppression effectiveness under unbalanced current conditions, and the fault tolerance model is effectively established. Because method 2 has fewer available vectors and significantly increases the circulating current peak under dynamic switching, making comparison insignificant, we do not compare method 2 here. The data are summarized in Table 7, showing that method 4 achieves the best control effect under conditions of given current abrupt changes.

[0169] Table 7 Simulation results of grid-side circulating current peak and grid-connected circuit THD after the given current mutation during fault-tolerant operation

[0170]

[0171] The filter inductor Lm(1) on the side of inverter 1 is changed to 8mH, and the inductor Lm(2) on the side of inverter 2 remains unchanged. The grid-side circulating current under the condition of unbalanced inductance is compared. Under the condition of unbalanced inductance, the capacitor circulating current i zcf1 、i zcf2 and grid-side circulation i gThe phase A grid-connected current waveform significantly increases, exhibiting significant distortion and deteriorating THD. The SPH-SMPC controller effectively suppresses circulating current, maintaining the circulating current at a low value, but spikes still exist. Using method 4, the grid-side circulating current exhibits no spikes, and the phase A grid-connected current waveform quality is excellent. Comparing the grid-side circulating current under the four control methods, after fault-tolerant control, the maximum peak value of the grid-side circulating current is approximately 43.6A for method 1 and approximately 119.7A for method 2. Method 3, which employs a value function to suppress circulating current, achieves a maximum peak value of approximately 15.1A. Method 4, which further modifies the fault model, achieves a maximum peak value of approximately 8.9A, approaching the value of balanced inductance conditions. Therefore, the proposed method can effectively suppress circulating current under unbalanced inductance conditions. The data are summarized in Table 8, showing that method 4 achieves the best control effect under unbalanced inductance conditions.

[0172] Table 8 Simulation data of grid-side circulating current peak and grid-connected circuit THD under fault-tolerant operation with inductance unbalanced working conditions for four methods

[0173] method <![CDATA[Fault-tolerant operation grid-side circulating current i zg Peak]]> Fault-tolerant operation THD Method 1 43.6A 4.07％ Method 2 119.7A 4.52％ Method 3 15.1A 3.61％ Method 4 8.9A 3.40％

[0174] Example 2:

[0175] A second embodiment of the present invention provides a parallel three-level ANPC inverter open circuit fault tolerant control system, comprising:

[0176] a health status analysis module configured to construct a fault-free prediction model based on a topology of a parallel three-level ANPC inverter system;

[0177] a fault status analysis module configured to perform an open circuit fault analysis on the no-fault prediction model to determine the cause of the periodic distortion of the grid-side current caused by the open circuit fault;

[0178] a hybrid structure design module configured to design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and to establish a series-parallel hybrid structure based on the ranking of the multi-objective value functions;

[0179] a healthy inverter control module configured to determine a sequential model predictive control strategy for the healthy inverter based on a series-parallel hybrid structure;

[0180] a faulty inverter control module configured to determine a sequence model prediction phase-loss fault-tolerant control strategy for the faulty inverter based on a series-parallel hybrid structure and a cause of periodic distortion of grid-side current caused by an open-circuit fault;

[0181] The overall control module is configured to determine the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system based on the sequence model prediction control strategy of the healthy inverter, the sequence model prediction phase loss fault-tolerant control strategy of the faulty inverter, and the signal interaction between the inverters.

[0182] Example 3:

[0183] A third embodiment of the present invention provides a medium having a program stored thereon. When the program is executed by a processor, the program implements the steps of the open circuit fault tolerance control method of the parallel three-level ANPC inverter as described in the first embodiment of the present invention.

[0184] Example 4:

[0185] Embodiment 4 of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in the open-circuit fault-tolerant control method of the parallel three-level ANPC inverter as described in Embodiment 1 of the present invention are implemented.

[0186] The steps involved in the above embodiments 2, 3 and 4 correspond to those in the method embodiment 1. For the specific implementation methods, please refer to the relevant description part of the embodiment 1.

[0187] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computer device. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0188] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A parallel three-level ANPC inverter open circuit fault tolerance control method, characterized in that: The following steps are involved: A fault-free prediction model is constructed based on the topology of the parallel three-level ANPC inverter system; Perform open circuit fault analysis on the fault-free prediction model to determine the cause of periodic grid-side current distortion caused by the open circuit fault; Design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and establish a series-parallel hybrid structure based on the ranking of the multi-objective value function; Determine the sequential model predictive control strategy for healthy inverters based on a series-parallel hybrid structure; According to the series-parallel hybrid structure and the cause of the periodic distortion of the grid-side current caused by the open circuit fault, the sequence model prediction phase-loss fault-tolerant control strategy of the faulty inverter is determined; Based on the sequence model predictive control strategy of the healthy inverter, the sequence model predictive phase loss fault-tolerant control strategy of the faulty inverter and the signal interaction between the inverters, the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system is determined.

2. The open-circuit fault-tolerant control method for parallel three-level ANPC inverters according to claim 1, characterized in that: The open circuit fault refers to the open circuit fault of any single switch device of the two parallel ANPC three-level inverters, and it is assumed that the freewheeling diode is not affected by the fault.

3. The open-circuit fault-tolerant control method for parallel three-level ANPC inverters according to claim 1, characterized in that: Based on fault-tolerant control, circulating current suppression and midpoint potential balance, a multi-objective value function is designed, including the generator-side circulating current value function, the midpoint voltage value function and the grid-connected current value function.

4. The open-circuit fault-tolerant control method for parallel three-level ANPC inverters according to claim 3, characterized in that: The specific steps of sorting based on multi-objective value function are: The machine-side circulating current value function and the midpoint voltage value function are placed at a relatively forward position, and the grid-connected current value function is placed at the last layer to ensure that the controller finally outputs a single optimal solution.

5. The open circuit fault tolerant control method for parallel three-level ANPC inverters according to claim 1, characterized in that: The specific settings of the series-parallel hybrid structure are: The series-parallel hybrid structure is divided into a series structure of an outer control target set and an inner control target set. In the outer control target level, the machine-side circulating current value function and the midpoint voltage value function are used for midpoint voltage balance control and circulating current suppression, and a parallel structure is used to select the best between the machine-side circulating current value function and the midpoint voltage value function.

6. The open circuit fault tolerance control method for parallel three-level ANPC inverters according to claim 1, characterized in that: The reasons why the open circuit fault causes the periodic distortion of the grid-side current are the reduction of the modulation index and the inability to achieve the switching state after the open circuit fault.

7. The open circuit fault tolerance control method for parallel three-level ANPC inverters according to claim 1, characterized in that: The specific steps of signal interaction between inverters are: The deviation between the output current of the faulty inverter and the reference current signal is used as a compensation signal to be injected into the healthy inverter, so that the healthy inverter can compensate for the power output shortfall of the faulty inverter.

8. A parallel three-level ANPC inverter open circuit fault tolerant control system, characterized in that: include: a health status analysis module configured to construct a fault-free prediction model based on a topology of a parallel three-level ANPC inverter system; a fault status analysis module configured to perform an open circuit fault analysis on the no-fault prediction model to determine the cause of the periodic distortion of the grid-side current caused by the open circuit fault; a hybrid structure design module configured to design a multi-objective value function based on fault-tolerant control, circulating current suppression, and midpoint potential balance, and to establish a series-parallel hybrid structure based on the ranking of the multi-objective value functions; a healthy inverter control module configured to determine a sequential model predictive control strategy for the healthy inverter based on a series-parallel hybrid structure; a faulty inverter control module configured to determine a sequence model prediction phase-loss fault-tolerant control strategy for the faulty inverter based on a series-parallel hybrid structure and a cause of periodic distortion of grid-side current caused by an open-circuit fault; The overall control module is configured to determine the overall fault-tolerant control strategy of the parallel three-level ANPC inverter system based on the sequence model prediction control strategy of the healthy inverter, the sequence model prediction phase loss fault-tolerant control strategy of the faulty inverter, and the signal interaction between the inverters.

9. A computer-readable storage medium, characterized in that A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executed by the open circuit fault tolerance control method of a parallel three-level ANPC inverter according to any one of claims 1 to 7.

10. A terminal device, characterized in that: The invention comprises a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; and the computer-readable storage medium is used to store multiple instructions, wherein the instructions are suitable for being loaded by the processor and executing the open-circuit fault-tolerant control method of the parallel three-level ANPC inverter according to any one of claims 1 to 7.

Images