Discrete space vector model predictive control method and system for three-level active filter bridge arm fault

Abstract

A discrete space vector model predictive control method and system for three-level active power filter (APFilter) arm faults is proposed. Based on the fault-tolerant structure of the APF arm, the output voltage vector plane of the APF arm fault is iteratively divided and a virtual voltage vector is searched. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. After the search ends, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. According to the region of the voltage vector to be reconstructed in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector. Based on the final voltage vector, the switching state of the three-phase bridge arm is inferred to generate a pulse signal for discrete space vector model predictive control of the three-level APF arm faults. This achieves the control objectives of high precision, low computational load, and fault tolerance.

Description

Technical Field

[0001] This invention belongs to the field of power quality management and grid connection control, specifically relating to a discrete space vector model predictive control method and system for bridge arm faults of a three-level active filter. Background Technology

[0002] The widespread use of nonlinear loads in modern industry injects a large amount of harmonic current into the power grid, leading to a decline in power quality and threatening the safe and stable operation of the power system. Therefore, power quality management has become an urgent need. Active power filters are the core devices for power quality management. Among them, three-level active power filters have become the preferred solution for high-voltage, high-power scenarios due to their advantages such as low voltage resistance of switching devices, excellent output waveform, and low losses. Their operational stability and control accuracy directly determine the management effect.

[0003] In long-term operation of three-level active power filters, the bridge arm switching devices are prone to failure due to factors such as device aging and power grid impact. Without effective fault-tolerant control, this can lead to filter output degradation, increased harmonics, and even shutdown. Therefore, fault-tolerant control of the bridge arm is crucial for ensuring its reliable operation. Existing technologies mostly achieve fault-tolerant operation by reconstructing the topology after a fault, but the design of control strategies under fault conditions still has significant defects and shortcomings. On the one hand, traditional fault-tolerant control strategies are mostly based on conventional voltage vector sets for control, without optimizing the voltage vector configuration for the topological characteristics after a bridge arm fault. This results in insufficient voltage vector adjustment accuracy, difficulty in effectively suppressing current harmonics under fault conditions, and large control errors. On the other hand, existing fault-tolerant schemes based on model predictive control require online evaluation of all voltage vectors to select the optimal control vector, resulting in high computational complexity and poor real-time performance. This makes it difficult to meet the control response speed requirements of high-voltage, high-power scenarios. Furthermore, while some schemes attempt to simplify the search process, this further reduces control accuracy, failing to balance control performance and computational efficiency. Furthermore, existing fault-tolerant control technologies lack voltage vector discretization optimization strategies for bridge arm fault scenarios, making it difficult to improve control accuracy through flexible vector configuration. At the same time, the inefficiency of the optimal voltage vector search algorithm leads to excessive computational burden on the controller, which easily causes control delays and affects the stable operation of the fault-tolerant structure. It is impossible to achieve high-precision and high-stability control of the filter after a fault, and it is difficult to meet the industrial field's requirements for "non-stop operation, low harmonics, and high reliability" after a bridge arm fault of a three-level active filter.

[0004] In summary, current fault-tolerant control technology for three-level active power filters suffers from problems such as insufficient control accuracy, poor current harmonic suppression, heavy computational burden, and poor real-time performance, which restrict the long-term reliable application of three-level active power filters in high-voltage and high-power power quality management scenarios. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a discrete space vector model predictive control method and system for bridge arm faults of three-level active power filters (APFs). By integrating discrete space vector modulation (DSVM) and an efficient iterative search algorithm, this invention solves the problems of insufficient voltage vector regulation accuracy leading to poor harmonic suppression, high computational burden leading to poor real-time performance, and control delay in traditional model predictive control (MPC) methods when reconstructing a three-phase four-switch fault-tolerant structure of an existing three-level active power filter (APF) after a bridge arm fault. This enables power quality management and grid connection control.

[0006] The present invention adopts the following technical solution, including the following steps: This invention proposes a discrete space vector model predictive control method for bridge arm faults of a three-level active filter, comprising the following steps: Based on the fault-tolerant structure of the active filter bridge arm, the switching state of the three-phase bridge arm is obtained to obtain the output voltage vector of the active filter bridge arm fault. The output voltage vector plane of the active filter bridge arm fault is iteratively divided and a virtual voltage vector is searched. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. When the search ends, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector; Based on the final voltage vector, the switching state of the three-phase bridge arm is inferred to generate a pulse signal, and discrete space vector model predictive control of the three-level active filter bridge arm fault is performed.

[0007] Each phase arm of the three-level active filter includes four switching transistors. The upper arm has two switching transistors connected in series, and the lower arm has two switching transistors connected in series. The DC side consists of two capacitors connected in series, with the connection point of the two capacitors being the neutral point. Each phase output terminal is connected in series with the neutral point, which is used to switch the output terminal of the faulty phase to the neutral point when a bridge arm fails, thus obtaining a bridge arm fault-tolerant structure.

[0008] Based on the bridge arm fault-tolerant structure, the predicted value of the active filter output current based on the active filter output voltage is obtained, including: Establish a state equation with the output current of the active filter as the state variable under the fault-tolerant structure of the bridge arm; Based on the state equation, the predicted value of the active filter output current based on the active filter output voltage is obtained, as shown in the following equation:

[0009] In the formula, for Predicted output current of active filter in coordinate system for coordinate system The output current of the active filter at any given time. for coordinate system The output voltage vector of the active filter at time t. for coordinate system The grid voltage at any given time, To control the cycle, , These are the inductance and parasitic resistance of the active filter, respectively.

[0010] Calculate the reference value of the harmonic compensation current for the active filter based on the load current.

[0011] Based on the output current of the active filter, the voltage of the two series capacitors on the DC side, and the switching state of the bridge arm, the predicted voltage difference between the two series capacitors on the DC side is obtained, as shown in the following formula:

[0012] In the formula, , They are respectively The voltage of the two series capacitors on the DC side at any given moment; - This is the predicted voltage difference between the two series capacitors on the DC side. The capacitance values ​​are the two capacitors connected in series on the DC side. To control the cycle; , , for The output current of the three phases of the active filter at all times; , , These represent the switching states of the three-phase bridge arms, with values ​​of -1, 0, or 1. When a bridge arm fault occurs in a certain phase, the switching state of that phase bridge arm remains constant at 0.

[0013] A cost function is established using the deviation between the harmonic compensation current reference value and the predicted output current value of the active filter, and the predicted voltage difference between the two series capacitors on the DC side. As shown in the following formula:

[0014] In the formula, , , These are the reference values ​​for the compensation current of phases a, b, and c, respectively. , , These are the predicted output current values ​​for phases a, b, and c of the active filter. - This is the predicted voltage difference between the two series capacitors on the DC side. These are the weighting coefficients.

[0015] The cost function corresponding to the virtual voltage vector is obtained by replacing the output voltage of the active filter with the virtual voltage vector.

[0016] Based on the fault-tolerant structure of the active filter bridge arm, the switching states of the three-phase bridge arms are obtained to obtain the output voltage vector of the active filter bridge arm fault; the nine output voltage vectors of the active filter bridge arm fault are used to form a rhombus; the plane is iteratively divided and the virtual voltage vector is searched.

[0017] During the first search, the rhombic plane is divided into two congruent equilateral triangles. The geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. The virtual voltage vector is set during the first search. ,in, , =0,1 This refers to the DC bus voltage of the active filter. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the first search. Optimal voltage vector The equilateral triangle containing it will be the area for the second search.

[0018] During the second search, the region determined in the first search is divided into four congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. During the second search, if If = 0, then record the triangle orientation coefficient. Set to 1, and set the virtual voltage vector. , , =0,1,2; if =1, then record the triangle orientation coefficient. Set to -1 and set the virtual voltage vector. , , =0,1,2; The virtual voltage vector that minimizes the cost function is used as the optimal voltage vector obtained in the second search. Optimal voltage vector The equilateral triangle containing it will be the area for the third search.

[0019] No. During the second search, the first The region determined by the -1 search is divided into 4 congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector; Compare the first The optimal voltage vector obtained from -2 searches and the The optimal voltage vector obtained from -1 search ; like Then the triangle's orientation is the opposite of the number. =- And set virtual voltage vector , , =0,1,2; if Then the orientation of the triangle remains unchanged. = And set virtual voltage vector , , =0,1,2; The virtual voltage vector that minimizes the cost function is taken as the first... The optimal voltage vector obtained from the second search Optimal voltage vector It is the actual optimal voltage vector The closest value is used as the voltage vector to be reconstructed.

[0020] If the voltage vector to be reconstructed is within the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the three unit voltage vectors within the triangle containing the voltage vector to be reconstructed; if the voltage vector to be reconstructed is outside the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the middle vector, zero vector, and small vector within the triangle containing the voltage vector to be reconstructed.

[0021] In another aspect, this invention proposes a discrete space vector model predictive control system for a three-level active filter bridge arm fault, comprising: The discrete space vector module is used to obtain the switching state of the three-phase bridge arm based on the fault-tolerant structure of the active filter bridge arm to obtain the output voltage vector of the active filter bridge arm fault. It iteratively divides the output voltage vector plane of the active filter bridge arm fault and searches for virtual voltage vectors. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. At the end of the search, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector. The pulse modulation module is used to generate pulse signals by inferring the switching states of the three-phase bridge arms based on the final voltage vector, and to perform discrete space vector model predictive control of bridge arm faults in a three-level active filter.

[0022] The present invention is also a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.

[0023] The present invention is also a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0024] The beneficial effects of this invention, compared with the prior art, include at least the following: By reconstructing the fault-tolerant structure of the three-level active filter after a bridge arm fault, and taking the rhomboid plane enclosed by the nine basic voltage vectors under the fault-tolerant structure as the object, a massive number of uniformly distributed virtual voltage vectors are generated through a geometric iterative equal division method. An iterative search algorithm matching the geometric equal division method is designed, where each search evaluates only three new virtual vectors within the small triangle selected in the previous round. After n iterations, only 3n-1 vectors need to be evaluated to approximate the optimal solution. This not only significantly improves the degree of freedom of the voltage vector and reduces voltage control error, but also effectively suppresses current harmonics in the fault state of the three-level active filter bridge arm, ensuring stable fault-tolerant operation of the system, and greatly refines the adjustment resolution of the voltage vector, fundamentally solving the problem of insufficient accuracy, but also reduces the computational load from exponential growth with the number of iterations to linear growth, completely solving the computational burden problem. Therefore, while ensuring the real-time performance of the controller, the output current harmonics in the fault state are significantly reduced, and the three mutually constraining control objectives of high accuracy, low computational load, and fault tolerance are achieved simultaneously. Attached Figure Description

[0025] Figure 1 This is a flowchart of a discrete space vector model predictive control method for bridge arm faults of a three-level active filter proposed in this invention. Figure 2 This is a fault-tolerant structure diagram of a three-level active filter provided by the present invention.

[0026] Figure 3 This is a fault-tolerant structure diagram of a three-level active filter after a fault in phase A bridge arm provided in an embodiment of the present invention.

[0027] Figure 4 This is a voltage vector diagram of the fault-tolerant structure of the three-level active filter after a fault in phase A bridge arm in this embodiment of the invention. Figure 5 This is a voltage vector diagram of the fault-tolerant structure of the three-level active filter after a fault in phase B bridge arm in this embodiment of the invention. Figure 6This is a voltage vector diagram of the fault-tolerant structure of the three-level active filter after a C-phase bridge arm fault in this embodiment of the invention. Figure 7 This is a schematic diagram of the virtual voltage vector generated by the fault-tolerant structure of the three-level active filter after a fault in phase A bridge arm in this embodiment of the invention, and the first search step. Figure 8 This is a schematic diagram of the virtual voltage vector generated by the fault-tolerant structure of the three-level active filter after a fault in phase A bridge arm in this embodiment of the invention, and the second search step. Figure 9 This refers to the virtual voltage vector generated by the fault-tolerant structure of the three-level active filter after a fault in phase A bridge arm in this embodiment of the invention, and the virtual voltage vector and the first A diagram illustrating the search steps; Figure 10 This is a schematic diagram of how, in an embodiment of the present invention, the voltage vector to be reconstructed is located within a small vector hexagonal region, and the final voltage vector is reconstructed using three unit voltage vectors within the triangle containing the voltage vector to be reconstructed. Figure 11 This is a schematic diagram of an embodiment of the present invention in which the voltage vector to be reconstructed is outside the small vector hexagonal region, and the final voltage vector is obtained by using the middle vector, zero vector and small vector within the triangle where the voltage vector to be reconstructed is located. Figures 4 to 11 In the diagram, solid dots represent zero vectors, arrowed dotted lines represent small vectors, arrowed dashed lines represent medium vectors, solid hexagons represent the optimal voltage vector, solid circles represent virtual voltage vectors, shaded triangles represent the region where the optimal voltage vector is located, and solid hexagons represent the optimal virtual vector obtained in the current search. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0029] This invention proposes a discrete space vector model predictive control method for bridge arm faults of a three-level active filter, such as... Figure 1 As shown, the method includes the following steps: Step S1: When the active filter bridge arm fails, the bridge arm fault-tolerant structure is reconstructed.

[0030] Specifically, each phase arm of the three-level active filter includes four switching transistors, with two switching transistors in series in the upper arm and two switching transistors in series in the lower arm; the DC side consists of two series capacitors, with the connection point of the two capacitors being the neutral point; each phase output terminal is connected in series with the neutral point, which is used to switch the output terminal of the faulty phase to the neutral point when the bridge arm fails, thus obtaining a bridge arm fault-tolerant structure.

[0031] This invention analyzes the operating mechanism of a three-level active filter bridge arm when a fault occurs, reconstructs it into a bridge arm fault-tolerant structure after the fault occurs, and analyzes the switching state and voltage vector set of the remaining bridge arms.

[0032] The three-level active filter fault-tolerant structure provided by this invention, such as... Figure 2 As shown, the three-level active filter is connected to the power grid via a transmission line. , These represent the inductance and parasitic resistance of the active power filter, and the output current of the three phases of the active power filter. , , The output current of the active filter The current in the three phases of the power grid , , Constituting grid-connected current The three-phase current of the load , , Constituting load current The three-level active filter adopts a three-phase twelve-switch fault-tolerant structure. Taking phase A as an example, the phase A bridge arm includes four switching transistors. , , , Each phase arm is equipped with an additional IGBT device at both ends, and each phase output terminal and neutral point are equipped with an additional switching transistor. , , Both series capacitors on the DC side are The capacitor voltages are respectively , .

[0033] The fault-tolerant structure proposed in this invention incorporates an additional switching transistor between the three-phase output terminals and the neutral point of a conventional three-level active filter. , , When a three-level active filter bridge arm fails, an additional switching transistor is controlled. , , The circuit is switched on to isolate the faulty bridge arm, allowing the output of the faulty phase to be directly connected to the neutral point of the three-level active filter, thereby reconstructing the three-level active filter into a fault-tolerant bridge arm structure. The fault-tolerant structure of the three-level active filter after a fault in phase A will be reconstructed as follows: Figure 3 The topology shown, Figure 3 In the middle, phase B bridge arm includes four sets of switching transistors. , , , The load includes inductors ,resistance and equivalent impedance .

[0034] Step S2: Based on the bridge arm fault-tolerant structure, obtain the predicted value of the active filter output current based on the output voltage of the active filter.

[0035] Specifically, step 2 includes: Step 2.1: Based on the bridge arm fault-tolerant structure, establish the state equation with the output current of the active filter as the state variable, as shown in the following equation: (1) In the formula, for coordinate system The output voltage of the active filter at time . for coordinate system The output current of the active filter at any given time. , These are the inductance and parasitic resistance of the active filter, respectively. for coordinate system The grid voltage at any given time.

[0036] Step 2.2, based on the state equation, obtain the output voltage of the active filter. The predicted output current of the active filter at time +1 is shown in the following formula: (2) In the formula, To control the cycle.

[0037] Step 2.3, then... The output current of the active filter at time +1 Transform to the abc coordinate system to obtain The output current of the active filter at time +1 That is, the predicted output current of the three phases of the active filter. , , .

[0038] Step S3: Real-time acquisition of the active filter output current, load current, voltage of the two series capacitors on the DC side, and bridge arm switching status; calculation of the harmonic compensation current reference value based on the load current; and acquisition of the predicted voltage difference between the two series capacitors on the DC side using the active filter output current, the voltage of the two series capacitors on the DC side, and the bridge arm switching status.

[0039] Specifically, step S3 includes: Step S3.1: Sample using a current sensor and a voltage sensor respectively. Active filter output current at time 1 Load current Grid-connected current Grid voltage and DC side capacitor voltage , And the phase angle of the power grid is obtained using a phase-locked loop. Then, based on the phase angle, the collected current and voltage are converted to... In coordinate system, we obtain Active filter output current at time 1 Load current Grid-connected current Grid voltage .

[0040] Step S3.2: Calculate the reference value of the harmonic compensation current of the active filter based on the load current; Specifically, according to the power grid phase angle load current Transform to the pq coordinate system to obtain Then a low-pass filter is used to filter out... The high-frequency noise is then transformed to the abc coordinate system to obtain the corresponding fundamental component of the load current. Then use the load current Subtract fundamental component The corresponding three-phase harmonic compensation current reference value can then be obtained. .

[0041] Step S3.3: Using the output current of the active filter, the voltage of the two series capacitors on the DC side, and the switching state of the bridge arm, obtain the predicted value of the voltage difference between the two series capacitors on the DC side, as shown in the following formula: (3) In the formula, , They are respectively The voltage of the two series capacitors on the DC side at any given moment; - for The voltage difference between the two series capacitors on the DC side at time +1, i.e., the predicted value of the voltage difference between the two series capacitors on the DC side. The capacitance values ​​are the two capacitors connected in series on the DC side. To control the cycle; , , for The output current of the three phases of the active filter at all times; , , These represent the switching states of the three-phase bridge arms. A value of -1 indicates the "N" state, and for phase A, it represents... and Turn off and and When the circuit is on, a value of 0 indicates the "O" state. For phase A, this means... and Turn off and and When the signal is on, a value of 1 indicates the "P" state. and Conductive and and The switch is turned off, and when a bridge arm fault occurs in a certain phase, the switching state of that phase bridge arm remains constant at 0.

[0042] Step S4: Using the deviation between the harmonic compensation current reference value and the predicted output current value of the active filter, and the predicted voltage difference between the two series capacitors on the DC side, a cost function is established with harmonic compensation and capacitor midpoint voltage stabilization as the joint control objectives. As shown in the following formula: (4) In the formula, , , These are the reference values ​​for the compensation current of phases a, b, and c, respectively. , , These are the predicted output current values ​​for phases a, b, and c of the active filter. - This is the predicted voltage difference between the two series capacitors on the DC side. These are the weighting coefficients.

[0043] Step S5: Iteratively divide the voltage vector plane under the fault-tolerant structure of the active filter and search for virtual voltage vectors. In the current search, based on the virtual voltage vector, repeat steps S2, S3, and S4 to obtain the cost function corresponding to the virtual voltage vector. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search, and the region for the next search is determined based on the optimal voltage vector.

[0044] The number of available voltage vectors after a fault is extremely small (only 9), resulting in coarse control accuracy. The traditional solution is to enumerate all virtual vectors, but this involves too much computation. In order to achieve both high-precision control and reduced computation, this invention proposes to provide a rich selection of vectors through discretization to achieve high-precision control, and to use an intelligent search algorithm to avoid redundant calculations and reduce computation.

[0045] Specifically, step S5 includes: Step S5.1: Based on the fault-tolerant structure of the active filter bridge arm, obtain the switching state of the three-phase bridge arm to obtain the output voltage vector of the active filter bridge arm fault. The relationship between the switching states of the three-phase bridge arms and the output voltage vector of the active filter is shown in the following equation: (5) In the formula, , For voltage vector Quantity, This is the DC bus voltage of the active filter.

[0046] Step S5.2: Iteratively divide the output voltage vector plane of the active filter bridge arm fault and search for virtual voltage vectors; in the current search, the virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search, and the region for the next search is determined based on the optimal voltage vector; at the end of the search, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed.

[0047] In this process, the output voltage in equation (2) is replaced by the virtual voltage vector obtained from the search, and steps S2, S3, and S4 are repeated to obtain the cost function corresponding to the virtual voltage vector.

[0048] When an active filter experiences a bridge arm failure and forms a fault-tolerant structure, the number of available output voltage vectors decreases from 27 to 9, and these 9 output voltage vectors form a rhombus. Figure 4 The shaded area in the image represents the fault-tolerant structure of the three-level active filter after a fault in phase A bridge arm. Voltage vector in coordinate system Figure 5 The shaded area in the image represents the fault-tolerant structure of the three-level active filter after a fault in phase B bridge arm. Voltage vector in coordinate system Figure 6 The shaded area in the image represents the fault-tolerant structure of the three-level active filter after a fault in the C-phase bridge arm. Voltage vector in coordinate system.

[0049] Therefore, assuming It is the actual optimal voltage vector, such as Figure 7 , Figure 8 , Figure 9 The solid hexagon in the image is used to search for virtual voltage vectors after dividing the diamond-shaped plane formed by voltage vectors, including: 1) such as Figure 7 The first search shown: The rhomboid plane is divided into two congruent equilateral triangles. The geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. During the first search, the virtual voltage vector is set. ,in, , =0,1, such as Figure 7 As shown in the shaded area, the geometric center of an equilateral triangle is represented by a solid hexagon, corresponding to the virtual voltage vector. The geometric center of the other equilateral triangle is represented by a solid circle, corresponding to the virtual voltage vector. Repeat steps S2, S3, and S4 to calculate the cost function corresponding to these two virtual voltage vectors respectively. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the first search. , Figure 7 middle, The optimal voltage vector obtained from the first search The equilateral triangle containing it will be the area for the second search.

[0050] 2) For example Figure 8 The second search shown: The region determined by the first search is divided into four congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. During the second search, if If = 0, then record the triangle orientation coefficient. Set to 1, and set the virtual voltage vector. , , =0,1,2; if =1, then record the triangle orientation coefficient. Set to -1 and set the virtual voltage vector. , , =0,1,2; , =0,1,2 =0 =1 or =1 =-1.

[0051] like Figure 8 As shown, the geometric center of an equilateral triangle is represented by a solid hexagon, corresponding to the virtual voltage vector. The geometric centers of the other three equilateral triangles are represented by solid circles. One of these solid circles must coincide with the solid hexagon found in the first search, while the other two non-coincident geometric centers correspond to virtual voltage vectors. , ; Repeat steps S2, S3, and S4 to calculate the cost function for each of the three virtual voltage vectors. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the second search. , Figure 8 middle, The optimal voltage vector obtained from the second search The equilateral triangle containing it will be the area for the third search.

[0052] 3) For example Figure 9 The first one shown Second search: The first The region determined by the -1 search is divided into 4 congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector; Compare the first The optimal voltage vector obtained from -2 searches and the The optimal voltage vector obtained from -1 search ; like Then the triangle's orientation is the opposite of the number. =- And set virtual voltage vector , , =0,1,2; if Then the orientation of the triangle remains unchanged. = And set virtual voltage vector , , =0,1,2; Repeat steps S2, S3, and S4 to calculate the cost function corresponding to each of the three virtual voltage vectors, and use the virtual voltage vector that minimizes the cost function as the first virtual voltage vector. The optimal voltage vector obtained from the second search , Figure 9 middle, , No. The optimal voltage vector obtained from the second search It is the actual optimal voltage vector The closest value is used as the voltage vector to be reconstructed.

[0053] Based on the above-mentioned plane division and search... Secondary or cost function Less than or equal to the preset threshold The iteration ends when the voltage vector plane is partitioned after a bridge arm fault, and the total number of virtual voltage vectors obtainable is [number missing]. for:

[0054] In the formula, It is a positive integer, representing the number of times the output voltage vector plane of the active filter is divided, which is also the number of searches.

[0055] Traditional model predictive control requires sequentially substituting each discretely generated virtual voltage vector into the prediction model and cost function for rolling optimization, which significantly increases the computational load and consequently affects control performance. Therefore, this invention proposes an efficient optimal voltage vector search method.

[0056] The core principle of the search method proposed in this invention is that it can quickly and accurately locate the optimal voltage vector within the discrete voltage vector plane without traversing all virtual voltage vectors. This method not only effectively reduces the computational burden of the system but also ensures the effective operation of the current prediction model and cost function, optimizes control performance, achieves precise and stable control, and reduces current ripple. This is a significant improvement of the traditional model-predictive current control strategy. Furthermore, this application introduces the concept of discrete space vector modulation, which has not been addressed in the field of fault-tolerant control, for the specific fault topology of three-level circuits. It creates a massive number of more refined virtual vectors through geometric methods.

[0057] go through After this search process, only 3 voltage vectors need to be evaluated online. -1; Compared with the traditional traversal method, this iterative search strategy can significantly reduce the total number of voltage vectors that need to be evaluated online, effectively reducing the computational burden on the system. Furthermore, after... The optimal voltage vector obtained after successive iterations and fine-grained search. It is already able to match the actual optimal voltage vector of the system. The approximation is high, which can meet the control accuracy requirements of subsequent three-level active filters.

[0058] Step S5.3: Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, select the corresponding basic voltage vector combination for vector synthesis to obtain the final voltage vector; Specifically, if the voltage vector to be reconstructed is within the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the three unit voltage vectors within the triangle containing the voltage vector to be reconstructed; if the voltage vector to be reconstructed is outside the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the middle vector, zero vector, and small vector within the triangle containing the voltage vector to be reconstructed.

[0059] like Figure 10 As shown, if the optimal voltage vector obtained in the last search... Within the small vector hexagonal region, the optimal voltage vector is extracted. The final voltage vector is obtained by reconstructing the three unit voltage vectors within the triangle. If the optimal voltage vector obtained in the last search Outside the small vector hexagonal region, the optimal voltage vector is extracted. The final voltage vector is obtained by reconstructing the median vector, zero vector, and small vector within the triangle. .

[0060] This invention defines a mapping rule between virtual vectors and physical vectors, which matches different combinations of synthesized vectors (small vector groups, or combinations of medium, zero, and small vectors) based on the region where the vector to be reconstructed is located (inside / outside the small vector hexagon). This rule is a key link in the complete technical loop formed by DSVM and efficient search algorithms.

[0061] Step S6: Based on the final voltage vector, the switching state of the three-phase bridge arm is inferred to generate a pulse signal for discrete space vector model predictive control of the three-level active filter bridge arm fault.

[0062] Assuming a fault occurs in phase A bridge arm, the switching transistor will be activated. When phase A of the bridge arm is short-circuited, its output side is directly connected to the midpoint of the capacitor voltage on the DC bus. Assuming the motor operates with fault tolerance after a fault in phase A, the actual optimal voltage vector is: The DC bus voltage is 400 V. The specific search process is as follows: During the first search, a new virtual voltage vector is added. and The optimal voltage vector for the first search is selected based on the cost function. = .

[0063] During the second search, a new virtual voltage vector was added. , and The optimal voltage vector for the second search is selected based on the cost function. = .

[0064] After multiple searches, the final voltage vector selected in step n will be close to the optimal voltage vector. ≈ Assuming the final voltage vector =(10, 17), DC bus voltage 400 V, then it is composed of two small vectors (0, 0, -1) and (0, -1, -1) and zero vector (0, 0, 0). The αβ axis voltages corresponding to the small vectors (0, 0, -1) and (0, -1, -1) are (66.67, 115.47) and (133.33, 0) respectively, and the αβ axis voltage corresponding to the zero vector (0, 0, 0) is (0, 0). According to the principle of vector composition, the duty cycles of the small vectors (0, 0, -1), (0, -1, -1), and (0, 0, 0) are 0.1472, 0.0014, and 0.8514 respectively.

[0065] In another aspect, this invention proposes a discrete space vector model predictive control system for a three-level active filter bridge arm fault, comprising: The discrete space vector module is used to obtain the switching state of the three-phase bridge arm based on the fault-tolerant structure of the active filter bridge arm to obtain the output voltage vector of the active filter bridge arm fault. It iteratively divides the output voltage vector plane of the active filter bridge arm fault and searches for virtual voltage vectors. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. At the end of the search, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector. The pulse modulation module is used to generate pulse signals by inferring the switching states of the three-phase bridge arms based on the final voltage vector, and to perform discrete space vector model predictive control of bridge arm faults in a three-level active filter.

[0066] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0067] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0068] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0069] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A discrete space vector model predictive control method for bridge arm faults of a three-level active filter, characterized in that, Includes the following steps: Based on the fault-tolerant structure of the active filter bridge arm, the switching state of the three-phase bridge arm is obtained to obtain the output voltage vector of the active filter bridge arm fault. The output voltage vector plane of the active filter bridge arm fault is iteratively divided and a virtual voltage vector is searched. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. When the search ends, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector; Based on the final voltage vector, the switching state of the three-phase bridge arm is inferred to generate a pulse signal, and discrete space vector model predictive control of the three-level active filter bridge arm fault is performed.

2. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 1, characterized in that, Each phase arm of the three-level active filter includes four switching transistors. The upper arm has two switching transistors connected in series, and the lower arm has two switching transistors connected in series. The DC side consists of two capacitors connected in series, with the connection point of the two capacitors being the neutral point. Each phase output terminal is connected in series with the neutral point, which is used to switch the output terminal of the faulty phase to the neutral point when a bridge arm fails, thus obtaining a bridge arm fault-tolerant structure.

3. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 2, characterized in that, Based on the bridge arm fault-tolerant structure, the predicted value of the active filter output current based on the active filter output voltage is obtained, including: Establish a state equation with the output current of the active filter as the state variable under the fault-tolerant structure of the bridge arm; Based on the state equation, the predicted value of the active filter output current based on the active filter output voltage is obtained, as shown in the following equation: In the formula, for Predicted output current of active filter in coordinate system for coordinate system The output current of the active filter at any given time. for coordinate system The output voltage vector of the active filter at time t. for coordinate system The grid voltage at any given time, To control the cycle, , These are the inductance and parasitic resistance of the active filter, respectively.

4. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 3, characterized in that, Calculate the reference value of the harmonic compensation current for the active filter based on the load current.

5. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 4, characterized in that, Based on the output current of the active filter, the voltage of the two series capacitors on the DC side, and the switching state of the bridge arm, the predicted voltage difference between the two series capacitors on the DC side is obtained, as shown in the following formula: In the formula, , They are respectively The voltage of the two series capacitors on the DC side at any given moment; - This is the predicted voltage difference between the two series capacitors on the DC side. The capacitance values ​​are the two capacitors connected in series on the DC side. To control the cycle; , , for The output current of the three phases of the active filter at all times; , , These represent the switching states of the three-phase bridge arms, with values ​​of -1, 0, or 1. When a bridge arm fault occurs in a certain phase, the switching state of that phase bridge arm remains constant at 0.

6. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 5, characterized in that, A cost function is established using the deviation between the harmonic compensation current reference value and the predicted output current value of the active filter, and the predicted voltage difference between the two series capacitors on the DC side. As shown in the following formula: In the formula, , , These are the reference values ​​for the compensation current of phases a, b, and c, respectively. , , These are the predicted output current values ​​for phases a, b, and c of the active filter. - This is the predicted voltage difference between the two series capacitors on the DC side. These are the weighting coefficients.

7. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 6, characterized in that, The cost function corresponding to the virtual voltage vector is obtained by replacing the output voltage of the active filter with the virtual voltage vector obtained by the search.

8. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 1, characterized in that, Based on the fault-tolerant structure of the active filter bridge arm, the switching states of the three-phase bridge arms are obtained to obtain the output voltage vector of the active filter bridge arm fault; the nine output voltage vectors of the active filter bridge arm fault are used to form a rhombus; the plane is iteratively divided and the virtual voltage vector is searched.

9. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 8, characterized in that, During the first search, the rhombic plane is divided into two congruent equilateral triangles. The geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. The virtual voltage vector is set during the first search. ,in, , =0,1 This refers to the DC bus voltage of the active filter. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the first search. Optimal voltage vector The equilateral triangle containing it will be the area for the second search.

10. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 9, characterized in that, During the second search, the region determined in the first search is divided into four congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector. During the second search, if If = 0, then record the triangle orientation coefficient. Set to 1, and set the virtual voltage vector. , , =0,1,2; if =1, then record the triangle orientation coefficient. Set to -1 and set the virtual voltage vector. , , =0,1,2; The virtual voltage vector that minimizes the cost function is used as the optimal voltage vector obtained in the second search. Optimal voltage vector The equilateral triangle containing it will be the area for the third search.

11. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 10, characterized in that, No. During the second search, the first The region determined by the -1 search is divided into 4 congruent equilateral triangles, and the geometric center of each equilateral triangle is taken as the newly added virtual voltage vector; Compare the first The optimal voltage vector obtained from -2 searches and the The optimal voltage vector obtained from -1 search ; like Then the triangle's orientation is the opposite of the number. =- And set virtual voltage vector , , =0,1,2; if Then the orientation of the triangle remains unchanged. = And set virtual voltage vector , , =0,1,2; The virtual voltage vector that minimizes the cost function is taken as the first... The optimal voltage vector obtained from the second search Optimal voltage vector It is the actual optimal voltage vector The closest value is used as the voltage vector to be reconstructed.

12. The discrete space vector model predictive control method for bridge arm faults of a three-level active filter according to claim 1, characterized in that, If the voltage vector to be reconstructed is within the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the three unit voltage vectors within the triangle containing the voltage vector to be reconstructed; if the voltage vector to be reconstructed is outside the small vector hexagonal region, the final voltage vector is obtained by reconstructing it using the middle vector, zero vector, and small vector within the triangle containing the voltage vector to be reconstructed.

13. A discrete space vector model predictive control system for bridge arm faults of a three-level active filter, used to implement the discrete space vector model predictive control method for bridge arm faults of a three-level active filter as described in any one of claims 1 to 12, characterized in that, include: The discrete space vector module is used to obtain the switching state of the three-phase bridge arm based on the fault-tolerant structure of the active filter bridge arm to obtain the output voltage vector of the active filter bridge arm fault. It iteratively divides the output voltage vector plane of the active filter bridge arm fault and searches for virtual voltage vectors. The virtual voltage vector that minimizes the cost function is taken as the optimal voltage vector obtained in the current search. The region for the next search is determined based on the optimal voltage vector. At the end of the search, the optimal voltage vector obtained in the last search is extracted as the voltage vector to be reconstructed. Based on the region where the voltage vector to be reconstructed is located in the voltage vector plane, the corresponding basic voltage vector combination is selected for vector synthesis to obtain the final voltage vector. The pulse modulation module is used to generate pulse signals by inferring the switching states of the three-phase bridge arms based on the final voltage vector, and to perform discrete space vector model predictive control of bridge arm faults in a three-level active filter.

14. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-12.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-12.

Images