阅读说明：本技术 基于自适应svpwm的逆变器开关管故障容错控制方法 (Fault-tolerant control method for inverter switching tube based on self-adaptive SVPWM ) 是由 朱琴跃 魏伟 谭喜堂 李爱华 李姚霖 于 2021-09-01 设计创作，主要内容包括：本发明涉及一种基于自适应SVPWM的逆变器开关管故障容错控制方法,包括以下步骤：1)基于NPC五电平逆变器所对应的空间电压矢量平面,获取每一个节点所分布的空间电压矢量的数量n；2)根据故障信息获取NPC五电平逆变器在开关管故障后仍然能够继续输出的空间电压矢量集S-(n)；3)根据空间电压矢量的数量n以及空间电压矢量集S-(n)计算在发生故障后的最大线性调制系数m；4)根据最大线性调制系数m实时调整目标参考电流的幅值,以故障下的实际电流输出与参考电流之差作为目标控制函数,自适应地选择满足目标控制函数最小的电压矢量序列S-(opt)进而生成下一时刻的PWM控制信号。与现有技术相比,本发明具有适用性强、实用安全、实现难度低等优点。(The invention relates to a fault-tolerant control method for a switching tube fault of an inverter based on self-adaptive SVPWM (space vector pulse width modulation), which comprises the following steps of: 1) acquiring the number n of space voltage vectors distributed by each node based on a space voltage vector plane corresponding to the NPC five-level inverter; 2) space voltage vector set S which is obtained according to fault information and can still be continuously output by NPC five-level inverter after switching tube fault n (ii) a 3) According to the number n of space voltage vectors and the space voltage vector set S n Calculating the maximum linear modulation coefficient m after the fault occurs; 4) adjusting the amplitude of the target reference current in real time according to the maximum linear modulation coefficient m, and taking the difference between the actual current output under the fault and the reference current as the differenceA target control function, adaptively selecting a voltage vector sequence S which satisfies the minimum of the target control function opt And further generates a PWM control signal at the next time. Compared with the prior art, the invention has the advantages of strong applicability, practicability, safety, low realization difficulty and the like.)

1. A fault-tolerant control method for inverter switching tube faults based on self-adaptive SVPWM is characterized by comprising the following steps:

1) acquiring the number n of space voltage vectors distributed by each node based on a space voltage vector plane corresponding to the NPC five-level inverter;

2) space voltage vector set S which is obtained according to fault information and can still be continuously output by NPC five-level inverter after switching tube fault_n；

3) According to the number n of space voltage vectors and the space voltage vector set S_nCalculating the maximum linear modulation coefficient m after the fault occurs;

4) adjusting the amplitude of the target reference current in real time according to the maximum linear modulation coefficient m, taking the difference between the actual current output under the fault and the reference current as a target control function, and performing a spatial voltage vector set S_nAdaptively selecting a voltage vector sequence S which satisfies a minimum target control function_optAnd further generating a PWM control signal at the next moment, controlling the output of the NPC five-level inverter and finishing the whole fault-tolerant process.

2. The method for fault-tolerant control of the switching tube fault of the inverter based on the adaptive SVPWM according to claim 1, wherein the step 1) specifically comprises the following steps:

101) space voltage vector (S) with minimum a-phase component on each node_a,S_b,S_c) Converting the continuous voltage variable into alpha and beta components of the continuous voltage variable under a two-phase static coordinate system, and then:

wherein, U_dcIs the DC side voltage, U, of the NPC five-level inverter_α、U_βRespectively are alpha and beta components of continuous voltage variable;

102) will U_α、U_βConverting the space voltage vector into an amplitude form to obtain a space voltage vector amplitude U and an included angle delta between the space voltage vector amplitude U and an alpha axis, wherein the included angle delta comprises the following components:

103) calculating the number n of the space voltage vectors distributed by the node according to the space voltage vector amplitude U, and then:

wherein ceil () is a ceiling function.

3. The method for fault-tolerant control of inverter switching tube based on adaptive SVPWM (space vector pulse width modulation) according to claim 2, characterized in that for space voltage vector (S)_a,S_b,S_c) After the number n of the space voltage vectors distributed on the node is determined, the space voltage vector (S) can be obtained_a,S_b,S_c) The redundant space voltage vector of (a) is:

(S_a+1,S_b+1,S_c+1)，(S_a+2,S_b+2,S_c+2)，…，+S_a+n-1,S_b+n-1,S_c+n-1)。

4. the method according to claim 1, wherein the step 2) specifically comprises the following steps:

201) generating 5 of NPC five-level topology under normal operation³A space voltage vector (S)_a,S_b,S_c) The formed vector set S;

202) according to the input fault information matrix F, combining the loss conditions of the electrical levels when different switching tubes generate different types of faults, obtaining a fault vector set S which cannot be normally output under the influence of the fault tubes under the set fault type_f；

203) Rejecting fault vector set S in vector set S_fObtaining a vector set S which is influenced by the fault tube and can still continuously output_n。

5. The method as claimed in claim 4, wherein the fault information matrix F is a matrix with i rows and j columns, where i is 1,2,3, j is 1, …,8, each element in the matrix represents the state of each switching tube, 0 represents normal operation, 1 represents open-circuit fault, and 2 represents short-circuit fault.

6. The method according to claim 1, wherein the step 3) specifically includes the following steps:

301) when the reference voltage vector is positioned in the x-th sector in the space vector plane, the maximum vector which can be output by the system in the x-th sector is (S)_a0,S_b0,S_c0)∈S_nCalculating a maximum linear modulation coefficient m, and respectively drawing a symmetrical hexagonal area corresponding to the maximum linear modulation coefficient m in a change process from large to small on a five-level space voltage vector plane;

302) taking a space voltage vector which is smaller than the maximum vector capable of being output in the x-th sector as a starting point, and setting a vector set S_nThe remaining 5 vectors forming the vertices of the hexagon are searched, and when two hexagons formed by different vectors have the same maximum linear modulation coefficient, the vectors have the same search priority, if the vector set S_nVector in (1) can not be output due to faultAnd when the fault occurs, searching the redundant vector, and if the redundant vector still can be output, calculating the maximum linear modulation coefficient m of the system after various faults occur in real time according to the space voltage vector coordinate with the minimum a-phase component in the maximum space voltage vector in the x-th sector.

7. The method according to claim 6, wherein in step 301), the maximum linear modulation coefficient m is calculated as:

wherein, N represents the number of the levels that each phase can output when the system is in normal operation, and for NPC five-level topology, the value of N is 5.

8. The method for fault-tolerant control of the switching tube of the inverter based on the adaptive SVPWM according to claim 1, wherein the step 4) specifically comprises the following steps:

401) three-phase output current i of NPC five-level inverter is obtained by sampling at k moment_a(k)，i_b(k) And i_c(k)；

402) Clark conversion is carried out on the three-phase output current to obtain controlled variables under a two-phase static coordinate system, and the Clark conversion comprises the following steps:

wherein i_α(k)、i_β(k) Current components of an alpha axis and a beta axis under a two-phase static coordinate system are respectively;

403) obtaining a predicted value of the controlled variable at the k +1 moment according to a prediction model of the NPC five-level inverter;

404) the reference current value at the time of k +1 is calculated by a second-order Lagrange extrapolation method, and then:

wherein the content of the first and second substances,reference current values of an alpha axis and a beta axis at the time of k +1, k-1 and k-2 respectively;

405) the amplitude of the target reference current is adjusted in real time through the maximum linear modulation coefficient m, and the following steps are performed:

wherein the content of the first and second substances,respectively the amplitude of the adjusted target reference current;

406) set of vectors S_nAll voltage vector states in the voltage vector model are sequentially substituted into the cost function to be calculated, and the space vector S which enables the cost function to be minimum is selected_optThe method is applied to the NPC five-level inverter at the next moment, and the square error is selected as a value function of a current tracking model, so that the following steps are provided:

where g represents the square error.

9. The method according to claim 8, wherein in the step 403), the process of establishing the five-level inverter predictive control model specifically includes the following steps:

4031) assuming that the three-phase output load of the NPC five-level inverter is a symmetrical inductance resistance load, the relationship between the three-phase voltage and the three-phase current is obtained by kirchhoff voltage and current law, and the following steps are provided:

wherein u is_a(k)、u_b(k)、u_c(k) Three-phase output voltages are respectively adopted, L is a load inductor, and R is a load resistor;

4032) clark transformation is carried out to obtain an output model under a two-phase static alpha beta coordinate system, and the output model comprises the following components:

wherein u is_α(k)、u_β(k) Output voltages of an alpha axis and a beta axis under a two-phase static coordinate system are respectively;

4033) the sampling forward difference method is used for discretizing, converting and sorting the continuous output model to obtain a discrete prediction model of the NPC five-level inverter under the two-phase static coordinate system, and the discrete prediction model comprises the following steps:

wherein, T_sFor the sampling period, R is the load resistance and L is the load inductance.

10. The method of claim 8, wherein the method is implemented by an inverter switching tube fault-tolerant control system, the inverter switching tube fault-tolerant control system comprising:

a space voltage vector redundancy calculation module: the method comprises the steps that the number n of space voltage vectors distributed by each node is calculated based on a space voltage vector plane corresponding to the NPC five-level inverter;

a voltage vector set calculation module: space voltage vector set S used for calculating and obtaining space voltage vector set S capable of being continuously output after fault according to fault information_n；

A maximum linear modulation coefficient calculation module: combining the output results of the space voltage vector redundancy calculation module and the voltage vector set calculation module, and calculating the maximum linear modulation coefficient m of the system after the system fails;

a current tracking control module: adjusting the amplitude of the target reference current in real time through the output result of the maximum linear modulation coefficient calculation module, taking the difference between the actual current output under the fault and the reference current as a target control function, and obtaining a vector set S by the voltage vector set calculation module_nAdaptively selecting a voltage vector sequence S which satisfies a minimum target control function_optAnd generating a PWM control signal at the next moment, controlling the output of the NPC five-level inverter and finishing the whole fault-tolerant process.

Technical Field

The invention relates to the field of multi-level inverter control, in particular to a fault tolerance control method for a switching tube of an NPC five-level inverter based on self-adaptive SVPWM.

Background

The multilevel inverter has the advantages of reducing the withstand voltage value of a power device, reducing the output harmonic distortion rate, improving the output waveform quality and the like, and is widely applied to medium-high voltage and high-power occasions, so that the requirements on the safety and the reliability of the multilevel inverter are higher and higher. How to adopt an effective fault-tolerant control method after the multi-level inverter fails is a great concern of experts and scholars. At present, the main fault-tolerant control methods can be divided into two categories, namely fault-tolerant control based on hardware topology reconstruction and software fault-tolerant control based on redundant space vectors.

For hardware fault tolerant control, since the nineties of the last century, various inverter fault tolerant topologies have been proposed, which can be broadly classified into a switch-redundancy type fault tolerant topology, a phase-redundancy type fault tolerant topology, and an active neutral point clamped type fault tolerant topology. The method can tolerate various fault types, but the hardware cost and the control complexity of the multilevel inverter are greatly increased due to the need of adding additional devices or bridge arms.

The basic idea of software fault-tolerant control is to replace a spatial voltage vector which cannot be output by a redundant spatial voltage vector when a power device fails, so as to perform fault-tolerant control. The method does not increase the system volume and the system cost, but different modulation strategies need to be formulated according to different fault types, a vector action sequence needs to be adjusted offline according to the fault types and fault-tolerant schemes, the calculated amount is complex, the response time of the system to fault-tolerant control is delayed greatly, the control effect and the output performance of the system under the fault condition do not reach the expected requirement, and further research needs to be carried out.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provide a fault-tolerant control method for an inverter switching tube based on self-adaptive SVPWM.

The purpose of the invention can be realized by the following technical scheme:

a fault tolerance control method for an inverter switching tube based on self-adaptive SVPWM comprises the following steps:

1) acquiring the number n of space voltage vectors distributed by each node based on a space voltage vector plane corresponding to the NPC five-level inverter;

2) space voltage vector set S which is obtained according to fault information and can still be continuously output by NPC five-level inverter after switching tube fault_n；

3) According to the number n of space voltage vectors and the space voltage vector set S_nCalculating the maximum linear modulation coefficient m after the fault occurs;

4) adjusting the amplitude of the target reference current in real time according to the maximum linear modulation coefficient m, taking the difference between the actual current output under the fault and the reference current as a target control function, and performing a spatial voltage vector set S_nAdaptively selecting a voltage vector sequence S which satisfies a minimum target control function_optAnd further generating a PWM control signal at the next moment, controlling the output of the NPC five-level inverter and finishing the whole fault-tolerant process.

The step 1) specifically comprises the following steps:

101) space voltage vector (S) with minimum a-phase component on each node_a，S_b，S_c) Converting the continuous voltage variable into alpha and beta components of the continuous voltage variable under a two-phase static coordinate system, and then:

wherein, U_dcIs the DC side voltage, U, of the NPC five-level inverter_α、U_βRespectively are alpha and beta components of continuous voltage variable;

102) will U_α、U_βConverting the space voltage vector into an amplitude form to obtain a space voltage vector amplitude U and an included angle delta between the space voltage vector amplitude U and an alpha axis, wherein the included angle delta comprises the following components:

103) calculating the number n of the space voltage vectors distributed by the node according to the space voltage vector amplitude U, and then:

wherein ceil () is a ceiling function.

For space voltage vector (S)_a，S_b，S_c) After the number n of the space voltage vectors distributed on the node is determined, the space voltage vector (S) can be obtained_a，S_b，S_c) The redundant space voltage vector of (a) is:

(S_a+1，S_b+1，S_c+1)，(S_a+2，S_b+2，S_c+2)，…，(S_a+n-1，S_b+n-1，S_c+n-1)。

the step 2) specifically comprises the following steps:

201) generating 5 of NPC five-level topology under normal operation³A space voltage vector (S)_a，S_b，S_c) The formed vector set S;

202) according to the input fault information matrix F, combining the loss conditions of the electrical levels when different switching tubes generate different types of faults, obtaining a fault vector set S which cannot be normally output under the influence of the fault tubes under the set fault type_f；

203) Rejecting fault vector set S in vector set S_fObtaining a vector set S which is influenced by the fault tube and can still continuously output_n。

The failure information matrix F is a matrix with i rows and j columns, wherein i is 1,2,3, and j is 1, …,8, each element in the matrix represents the state of each switching tube, 0 represents normal operation, 1 represents open-circuit failure, and 2 represents short-circuit failure.

The step 3) specifically comprises the following steps:

301) when the reference voltage vector is positioned in the x-th sector in the space vector plane, the maximum vector which can be output by the system in the x-th sector is (S)_a0，S_b0，S_c0)∈S_nCalculating a maximum linear modulation coefficient m, and respectively drawing a symmetrical hexagonal area corresponding to the maximum linear modulation coefficient m in a change process from large to small on a five-level space voltage vector plane;

302) taking a space voltage vector which is smaller than the maximum vector capable of being output in the x-th sector as a starting point, and setting a vector set S_nThe remaining 5 vectors forming the vertices of the hexagon are searched, and when two hexagons formed by different vectors have the same maximum linear modulation coefficient, the vectors have the same search priority, if the vector set S_nAnd when the vector in the x sector can not be output due to faults, searching the redundant vector, and if the redundant vector still can be output, calculating the maximum linear modulation coefficient m of the system after various faults occur in real time according to the space voltage vector coordinate with the minimum a-phase component in the maximum space voltage vector in the x sector.

In the step 301), the maximum linear modulation coefficient m is calculated as:

wherein, N represents the number of the levels that each phase can output when the system is in normal operation, and for NPC five-level topology, the value of N is 5.

The step 4) specifically comprises the following steps:

401) three-phase output current i of NPC five-level inverter is obtained by sampling at k moment_a(k)，i_b(k) And i_c(k)；

402) Clark conversion is carried out on the three-phase output current to obtain controlled variables under a two-phase static coordinate system, and the Clark conversion comprises the following steps:

wherein i_α(k)、i_β(k) Current components of an alpha axis and a beta axis under a two-phase static coordinate system are respectively;

403) obtaining a predicted value of the controlled variable at the k +1 moment according to a prediction model of the NPC five-level inverter;

404) the reference current value at the time of k +1 is calculated by a second-order Lagrange extrapolation method, and then:

wherein the content of the first and second substances,reference current values of an alpha axis and a beta axis at the time of k +1, k-1 and k-2 respectively;

405) the amplitude of the target reference current is adjusted in real time through the maximum linear modulation coefficient m, and the following steps are performed:

wherein the content of the first and second substances,respectively the amplitude of the adjusted target reference current;

406) set of vectors S_nAll voltage vector states in the voltage vector model are sequentially substituted into the cost function to be calculated, and the space vector S which enables the cost function to be minimum is selected_optThe method is applied to the NPC five-level inverter at the next moment, and the square error is selected as a value function of a current tracking model, so that the following steps are provided:

where g represents the square error.

In the step 403), the process of establishing the five-level inverter predictive control model specifically includes the following steps:

4031) assuming that the three-phase output load of the NPC five-level inverter is a symmetrical inductance resistance load, the relationship between the three-phase voltage and the three-phase current is obtained by kirchhoff voltage and current law, and the following steps are provided:

wherein u is_a(k)、u_b(k)、u_c(k) Three-phase output voltages are respectively adopted, L is a load inductor, and R is a load resistor;

4032) clark transformation is carried out to obtain an output model under a two-phase static alpha beta coordinate system, and the output model comprises the following components:

wherein u is_α(k)、u_β(k) Output voltages of an alpha axis and a beta axis under a two-phase static coordinate system are respectively;

4033) the sampling forward difference method is used for discretizing, converting and sorting the continuous output model to obtain a discrete prediction model of the NPC five-level inverter under the two-phase static coordinate system, and the discrete prediction model comprises the following steps:

wherein, T_sFor the sampling period, R is the load resistance and L is the load inductance.

The method is realized by a fault-tolerant control system of the inverter switching tube, and the fault-tolerant control system of the inverter switching tube comprises the following steps:

a space voltage vector redundancy calculation module: the method comprises the steps that the number n of space voltage vectors distributed by each node is calculated based on a space voltage vector plane corresponding to the NPC five-level inverter;

a voltage vector set calculation module: space voltage vector set S used for calculating and obtaining space voltage vector set S capable of being continuously output after fault according to fault information_n；

A maximum linear modulation coefficient calculation module: combining the output results of the space voltage vector redundancy calculation module and the voltage vector set calculation module, and calculating the maximum linear modulation coefficient m of the system after the system fails;

a current tracking control module: adjusting the amplitude of the target reference current in real time through the output result of the maximum linear modulation coefficient calculation module, taking the difference between the actual current output under the fault and the reference current as a target control function, and obtaining a vector set S by the voltage vector set calculation module_nAdaptively selecting a voltage vector sequence S which satisfies a minimum target control function_optAnd generating a PWM control signal at the next moment, controlling the output of the NPC five-level inverter and finishing the whole fault-tolerant process.

Compared with the prior art, the invention has the following advantages:

firstly, the requirement of the method for controlling the object is simple, and the provided design method only requires to know the corresponding parameters of the controlled object, namely the five-level inverter, the fault information of the switching tube, the sampling current/voltage, the space voltage vector parameters and the like, so that the requirement on the object is greatly relaxed, and the applicability of the method is enhanced.

In the design control method process, the constraint conditions of the NPC five-level inverter in practical application are fully considered, so that the output current harmonic wave can be effectively inhibited, effective fault-tolerant control can be performed when the inverter has open-circuit and short-circuit faults, the dynamic and steady-state performance of the inverter during operation is improved, and the practicability and the safety are enhanced.

When the composite faults of different positions and types of the inverter are faced, a corresponding PWM mapping table does not need to be established, the real-time delineation of an available space voltage vector set is carried out through the fault types, the corresponding modulation ratio is calculated, and a grid signal of the switching tube is generated by matching with a model predictive control algorithm, so that the difficulty in realizing the fault-tolerant control algorithm is reduced.

The method solves the problem that the existing off-line algorithm cannot tolerate the fault-tolerant unexpected fault through the self-adaptive SVPWM fault-tolerant control, has certain universality for the fault-tolerant control of the open-circuit and short-circuit faults of the switching tube of the NPC five-level inverter, and is flexible to use.

The invention does not need to increase the number of other devices, does not increase the control cost, does not increase the system volume, has high cost performance, is easy to realize, is convenient to apply, improves the utilization rate of power devices, and has higher practical application value.

Drawings

Fig. 1 is a schematic diagram of signal transmission between a controlled object and the present invention.

Fig. 2 is a diagram of a NPC five-level inverter topology.

FIG. 3 is a block diagram of the system of the present invention.

FIG. 4 is a schematic diagram of the spatial voltage vector redundancy calculation.

FIG. 5 is a vector set S_nA flow chart is calculated.

Fig. 6 shows different m and their corresponding hexagonal regions.

Fig. 7 is a current tracking control block diagram of the NPC five-level inverter.

FIG. 8 is S_a1And S_b1Three-phase output current waveform when open circuit fault.

FIG. 9 is S_a1And S_b1Three-phase output voltage waveform when open circuit fault.

FIG. 10 is S_a1And S_a3Three-phase output current waveform when short circuit fault occurs.

FIG. 11 is S_a1And S_a3Three-phase output voltage waveform when short circuit fault occurs.

Detailed Description

The invention is described in detail below with reference to the figures and specific embodiments.

Examples

The invention provides a fault-tolerant control system and a fault-tolerant control method for a switching tube of an NPC (neutral point clamped) five-level inverter based on self-adaptive SVPWM (space vector pulse width modulation). The signal transmission condition among the NPC five-level inverters is shown in figure 1, an NPC five-level inverter module is a controlled object, and the topological structure of the NPC five-level inverter module is shown in figure 2. The self-adaptive fault-tolerant control module contains a program for realizing the fault-tolerant control method, and the two modules complete fault-tolerant control by transmitting sampling current/voltage values, fault information and switching signals.

As shown in fig. 3, the adaptive SVPWM-based fault-tolerant control system mainly includes a space voltage vector redundancy calculation module, a voltage vector set calculation module, a maximum linear modulation coefficient calculation module, and a current tracking control module. The space voltage vector redundancy calculation module calculates the number n of space voltage vectors distributed by each node based on a space voltage vector plane corresponding to the NPC five-level inverter; the voltage vector set calculation module calculates a space voltage vector set S which can still be continuously output after the fault is obtained according to the fault information_n(ii) a The maximum linear modulation coefficient calculation module integrates the results of the two modules to realize the calculation of the maximum linear modulation coefficient m of the system after the fault occurs; the current tracking control module adjusts the amplitude of the target reference current in real time through the output result of the maximum linear modulation coefficient calculation module, takes the difference between the actual current output under the fault and the reference current as a target control function, and obtains a vector set S through the voltage vector set calculation module_nAdaptively selecting a voltage vector sequence S which satisfies the minimum target control function_optAnd generating a PWM control signal at the next moment, and controlling the output of the NPC five-level inverter, thereby completing the whole fault-tolerant process.

The invention provides a fault-tolerant control method based on a self-adaptive SVPWM fault-tolerant control system, which comprises the following steps:

(1) establishing a space voltage vector redundancy calculation model

According to the basic principle of the SVPWM modulation strategy, the side length of the space vector plane hexagon is knownIs 2U_dcAnd/3, for a space voltage vector plane of a five-level topology, the side length of each small triangle is U_dcAnd/6, as shown in FIG. 4. The more the space voltage vectors distributed on each node are, the more the space voltage vectors are closer to the inner side, so that the model mainly calculates the number of the space voltage vectors corresponding to each node on line based on the geometrical relation of the space vector planes and the basic principle of SVPWM.

The method specifically comprises the following substeps:

(101) space voltage vector (S) with minimum a-phase component on each node_a，S_b，S_c) Converting the continuous voltage into alpha and beta components of continuous voltage variable under a two-phase static coordinate system by the formula (1), wherein U_dcThe dc side voltage of the NPC five-level inverter includes:

(102) will U_α、U_βConverting the space voltage vector into an amplitude form to obtain a space voltage vector amplitude U and an included angle delta between the space voltage vector amplitude U and an alpha axis, wherein the formula (2) is as follows:

(103) as can be seen from fig. 4, for each space vector, the distance from U to zero vector determines the redundancy of the space voltage vector with respect to the size of each side of the small triangle, and therefore, after U is obtained, the number n of space voltage vectors distributed at the node can be obtained by using equation (3), where ceil () is an upward integer function.

(104) For space voltage vector (S)_a，S_b，S_c) After determining the nodeThe number n of the space voltage vectors of the cloth can be obtained (S)_a，S_b，S_c) The redundant space voltage vector of (a) is: (S)_a+1，S_b+1，S_c+1)，(S_a+2，S_b+2，S_c+2)，…，(S_a+n-1，S_b+n-1，S_c+n-1) The vector and the obtained redundant vector are used as input quantities to participate in the calculation of the maximum linear modulation coefficient of the system after the fault occurs.

(2) Establishing a voltage vector set calculation model

The function of the voltage vector set calculation module is 5 corresponding to NPC five-level topology³In a vector set S formed by space voltage vectors, the vector set S which can still continuously output under the condition of no fault pipe influence under the fault type is calculated in real time according to fault information_n。

The method specifically comprises the following substeps:

(201) generating 5 of NPC five-level topology under normal operation³A space voltage vector (S)_a，S_b，S_c) The formed vector set S;

(202) according to the input fault information F, combining the loss conditions of the electrical levels when different switch tubes generate different types of faults, obtaining a fault vector set S which cannot be normally output under the influence of the fault tubes under the fault type_f. F is a matrix of i rows and j columns, wherein i is 1,2,3, j is 1, …,8, each element in the matrix represents the state of each switching tube, "0" represents normal operation, "1" represents open-circuit fault, and "2" represents short-circuit fault;

(203) subtracting the fault vector set S from the vector set S_fObtaining the vector set S which is influenced by the fault tube and can still continuously output_nVector set S_nThe calculation flow chart of (2) is shown in fig. 5.

(3) Establishing a maximum linear modulation coefficient calculation model

The maximum linear modulation coefficient calculation model is obtained in a vector set S obtained by a voltage vector set calculation module_nAccording to a pre-designed search sequence, the amplitude is maximumAnd the vector is taken as the start, and is continuously changed to the inner layer vector, and finally, the maximum linear modulation coefficient of the system after the fault occurs is calculated on line.

The method specifically comprises the following substeps:

(301) when the reference voltage vector is positioned in the x-th sector in the space vector plane, the maximum vector which can be output by the system in the x-th sector is (S)_a0，S_b0，S_c0) Here (S)_a0，S_b0，S_c0)∈S_nThe maximum linear modulation coefficient m can be calculated according to equation (4), where N represents the number of levels that can be output by each phase in normal operation of the system, i.e., N is 5 for NPC five-level topology. According to the above calculation method, on the five-level space voltage vector plane, the symmetric hexagonal regions corresponding to the maximum linear modulation coefficient m in the process of changing from large to small are respectively drawn, as shown in fig. 6.

(302) Starting from a space voltage vector smaller than the maximum vector which can be output in the x-th sector at S_nThe remaining 5 vectors that constitute the vertices of the hexagon are found. It should be noted that for different vectors, the two hexagons that they form may have the same maximum linear modulation coefficient, when they have the same search priority.

Whereby from sector 1 to sector 6, from the large vector to the zero vector, are set in the vector set S_nAnd sequentially searching each vector according to the priority sequence in the searching sequence of the medium priority from large to small. If the vector in the table can not be output due to faults, the redundant vector obtained by the space voltage vector redundancy calculation module needs to be searched, if the redundant vector still can be output, the coordinate of the space voltage vector with the minimum a-phase component in the maximum space voltage vector in the x-th sector is substituted into formula (3), and the maximum linear modulation coefficient m of the system after various faults occur can be obtained in real time.

(4) Establishing a current tracking control model

The current tracking control model is based on the idea of modern model predictive control, the difference between the actual three-phase output current and the target reference current is acquired at each sampling moment through acquiring the difference between the actual three-phase output current and the target reference current when the fault occurs, so that a target control function, namely a cost function, is obtained, the voltage vector sequence which meets the minimum of the cost function is solved on line and acts on a controlled object at the next moment, namely the measured value at each moment is adopted to predict the output performance of the system at the next moment, and the current tracking control under the fault condition is realized. The control flow of the current tracking control model is shown in fig. 7, and mainly includes the following steps:

(401) three-phase output current i to inverter at time k_a(k)，i_b(k) And i_c(k) Sampling is performed.

(402) And (5) performing Clark conversion on the sampled current value to obtain a controlled variable under a two-phase static coordinate system.

(403) Calculating the predicted value of the controlled variable at the moment k +1 according to the prediction model of the five-level inverter, wherein the establishment process of the five-level inverter prediction control model is as follows:

(4031) assuming that the three-phase output load of the NPC five-level inverter is a symmetrical inductance resistance load, the relation between the three-phase voltage and the three-phase current obtained by kirchhoff voltage and current law is shown as formula (6).

(4032) Clark transformation is carried out on the above formula, and an output model under a two-phase static alpha beta coordinate system can be obtained as shown in a formula (7).

(4033) Selecting forward difference method to pair continuous modelsAnd (3) carrying out discretization treatment, wherein a calculation formula of forward difference is shown as a formula (8), and an NPC five-level inverter discrete prediction model under an alpha beta coordinate system obtained by conversion and arrangement according to a forward difference method is shown as a formula (9). Wherein, T_sFor the sampling period, R is the load resistance and L is the load inductance.

(404) And calculating to obtain the reference current value at the k +1 moment by a second-order Lagrange extrapolation method, wherein the expression is shown as a formula (10).

(405) And adjusting the amplitude of the target reference current in real time according to the output result of the maximum linear modulation coefficient calculation model, as shown in the following formula (11).

(406) A vector set S is obtained by a voltage vector set calculation module_nAll voltage vector states in the voltage vector model are sequentially substituted into the cost function to be calculated, and the space vector S which enables the cost function to be minimum is selected_optAnd (3) acting on the NPC five-level inverter at the next moment, and selecting the square error as a basic form of the value function of the current tracking model, as shown in a formula (12).

The effectiveness of the present invention is illustrated by the simulation operation result of the NPC five-level inverter connected to the three-phase resistive-inductive symmetric load.

(501) Assume that the parameters are set as follows: the three-phase resistive-inductive load is formed by a resistor R of 10 omega and an inductor L of 10 mH. DC side voltage U_dc1500V, voltage dividing capacitor C_i2.125mF (i is 1,2,3, 4), and the inverter parameter is the fundamental frequency f₁50Hz, modulation ratio m 0.8, carrier frequency f_c＝1250Hz。

(502)S_a1And S_b1The output current and voltage simulation waveforms of the switching tube after the interphase open circuit fault are respectively shown in fig. 8 and 9. Because the fault-tolerant control method provided by the invention does not change the original topological structure of the NPC five-level inverter, the system is S_x4The open-circuit fault of the switch tube does not have fault-tolerant capability. Therefore, the fault-tolerant control method provided by the invention does not contain S for all the objects_x4The single-tube open-circuit fault, the single-phase double-tube open-circuit fault and the interphase double-tube open-circuit fault, which include the switch tube open-circuit fault, have better fault-tolerant capability.

(503)S_a1And S_a3The output current and voltage simulation waveforms after the single-phase short-circuit fault occurs in the switching tube are respectively shown in fig. 10 and 11. Although from the fault tolerance point of view, the short-circuit fault with the same fault tube position and the same fault tube number has better fault tolerance capability relative to the open-circuit fault, the influence on the system is much more serious than the open-circuit fault. The fault-tolerant control method provided by the invention has better fault-tolerant capability for single-tube short-circuit faults and single-phase double-tube short-circuit faults, and for interphase double-tube short-circuit faults and more complex short-circuit fault types, the condition of the system after the fault becomes more complex, and the fault-tolerant control method needs to be considered by combining an actual topological structure and a current circulation path when the fault-tolerant control is carried out on the system, so that the fault-tolerant control method provided by the invention can only carry out fault-tolerant control on the two short-circuit fault types of the single-tube short-circuit faults and the single-phase double-tube short-circuit faults.

In conclusion, the fault-tolerant control method for the power switching tube of the NPC five-level inverter is designed aiming at the open-circuit and short-circuit faults of the power switching tube of the NPC five-level inverter and based on the self-adaptive SVPWM. When facing composite faults of different positions and types, real-time selection of an available space voltage vector set is carried out through the fault type without establishing a corresponding PWM mapping table, a corresponding modulation ratio is calculated, and a switching signal is generated by matching with a model predictive control algorithm. The invention reduces the implementation difficulty of the fault-tolerant control algorithm, solves the problem that the existing off-line algorithm cannot tolerate the fault of the unexpected fault, and improves the fault-tolerant capability of the NPC five-level inverter on the complex fault.