A Protection Method for Modular Multilevel Converter Based on Digital Twin

Abstract

Aiming at the problem that the existing differential protection of modular multilevel converters (MMCs) cannot identify the inter-turn short-circuit faults of arm reactors and sub-module faults, a protection method for modular multilevel converters is proposed based on the digital twin concept. The details are as follows: The converter is modeled as a whole by using the physical laws that the MMC conforms to and combining with the control method, and the digital twin concept is integrated into the relay protection through dynamic state estimation; and two sets of protection criteria are proposed according to the protection action types. Since the twin model contains all the physical laws of the MMC, based on the differential protection, the proposed method can additionally identify sub-module open-circuit faults, inter-turn short-circuit faults of arm reactors, and phase-to-phase short-circuit faults of arm.

Description

Technical Field

[0001] The present invention relates to the technical field of relay protection in power systems, and particularly to a protection strategy for modular multilevel converters based on digital twins. Background Art

[0002] Modular multilevel converters (MMCs) have advantages such as easy voltage equalization, large capacity, low loss, low step voltage, and good waveform quality, and have received attention at home and abroad. When a fault occurs in the converter area of a flexible DC transmission system, it will cause components of the MMC to be damaged by overcurrent or overvoltage impacts. As the key to the safe and stable operation of the MMC, relay protection has become a research hotspot in the field of flexible DC transmission.

[0003] At present, the protection of the converter area in a flexible DC transmission system mainly draws on the protection of conventional high-voltage DC transmission systems. The main protection uses differential protection, which has good rapidity, but it is difficult to reflect inter-phase short-circuit faults of the bridge arm, open-circuit faults of sub-modules, and inter-turn short-circuit faults of the bridge arm reactor. Power semiconductor devices are one of the most vulnerable components in power electronic converters, and the MMC contains a large number of IGBTs, which makes the IGBT the component with the highest failure rate in the MMC. An open-circuit fault in a sub-module will cause an increase in inter-phase circulating current and AC side harmonics, deteriorating the operating performance of the MMC; the bridge arm inductor is a key energy storage component of the MMC. When an inter-turn short-circuit fault occurs in the bridge arm reactor of the MMC, it will cause an increase in the complexity of the circulating current, resulting in fundamental frequency oscillations of the DC side voltage, current, and transmitted power.

[0004] In order to improve the protection performance, scholars at home and abroad have conducted a large number of studies. For the open-circuit fault of sub-modules, the existing technical solutions mainly adopt hardware-based methods, model-based methods, and machine learning-based methods for fault diagnosis; for the inter-turn short-circuit fault of the arm reactor, the existing technical solutions generally monitor it by observing the temperature rise and noise of the arm reactor, or configure a 100Hz protection for the arm reactor that only issues alarm signals. These protection schemes are all based on some physical laws followed by the object, and it is difficult to completely describe the protected object, so they cannot make up for the deficiencies of differential protection in being unable to identify the open-circuit fault of sub-modules and the inter-turn short-circuit fault of the arm reactor. The digital twin concept points out a direction for improving the performance of relay protection, that is, making full use of various physical laws followed by the object. The Georgia Institute of Technology in the United States proposed a protection principle based on dynamic state estimation in the AC system, that is, performing dynamic state estimation based on the mathematical model of the physical entity, providing an idea for the integration of digital twins and relay protection. The present invention proposes a modular multilevel converter (MMC) protection principle based on digital twins, establishes a digital twin model by combining MMC control information, performs dynamic state estimation according to real-time data, discriminates faults using the residuals between the measured values and the estimated values, and proposes two sets of protection criteria for starting and operating according to the protection action type. Since the twin model contains all the physical laws of the MMC, the proposed protection method can identify the inter-turn short-circuit fault of the arm reactor, the open-circuit fault of sub-modules, and the phase-to-phase short-circuit fault of the arm compared with the existing differential protection. The present invention theoretically has the ability to withstand a certain amount of transition resistance and measurement noise. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technology, the present invention first establishes a digital twin model according to all the physical laws followed by the MMC and combines the MMC control information, then establishes a measurement equation based on the MMC twin model, uses the least squares method to perform dynamic state estimation to calculate the measurement error, and proposes two sets of protection criteria for starting and operating according to the protection action type. Since the twin model contains all the physical laws of the MMC, the invention aims to overcome the problem that the existing differential protection cannot identify the open-circuit fault of sub-modules and the inter-turn short-circuit fault of the arm reactor.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is:

[0007] A modular multilevel converter protection strategy based on digital twins, comprising the following steps:

[0008] Establish a digital twin model according to all the physical laws followed by the MMC and combine the MMC control information;

[0009] Establish a measurement equation based on the MMC twin model;

[0010] Use the least squares method to perform dynamic state estimation to calculate the measurement error;

[0011] Two sets of protection criteria for starting and operating are proposed according to the protection action type.

[0012] Specifically, each arm of the MMC is composed of an arm reactor L connected in series with N sub-modules (SM). The upper and lower arms of the same phase form a phase unit.

[0013] The sub-module topology structure adopts a half-bridge structure. Through the mutual cooperation of switching devices VT1, VT2, VD1, and VD2, the switching operation of the sub-module is realized. C is the sub-module capacitor.

[0014] Based on the above scheme, the establishment of the digital twin model according to all the physical laws followed by the MMC and combined with the MMC control information specifically includes:

[0015] At any moment, the arm voltage formed by the sub-modules of the same arm u rj (r = p, n, respectively representing the upper and lower arms; j = a, b, c, representing the three phases of abc) is expressed by the sub-module capacitor voltage and its switching function, that is:

[0016] ;

[0017] In formula (1), u SM,rj_i represents the voltage coupled into the arm by the i-th sub-module of the r-arm of the j-phase; S rj_i represents the switching function of the i-th sub-module of the r-arm of the j-phase; when the sub-module is inserted, S rj_i = 1, when the sub-module is removed, S rj_i = 0; u c,rj_i represents the capacitor voltage of the i-th sub-module of the r-arm of the j-phase. The arm resistance R is used to equivalent the arm loss;

[0018] According to KCL, KVL and the relationship between the sub-module capacitor current and the arm current, the state equation of the MMC is written to obtain the digital twin model of the MMC in the time domain:

[0019] ;

[0020] In formula (2), i rj is the current flowing through the arm; i j is the current fed into the AC side; i dc , i dcn are the DC side currents; uc,rj_total is the cumulative value of the capacitor voltage of the bridge arm sub-module; S rj_total is the cumulative value of the switching function of the bridge arm sub-module;

[0021] In Equation (2), the differential term is eliminated by the integration method and converted into a historical value for calculation;

[0022] The first-order integration method and the second-order integration method are shown in Equations (3) and (4-5) respectively:

[0023] ;

[0024] ;

[0025] ;

[0026] where Δ t is the sampling interval.

[0027] Based on the above scheme, the establishment of the measurement equation based on the MMC twin model specifically includes:

[0028] Taking the currents of each bridge arm i pa ( t ), i pb ( t ), i pc ( t ), i na ( t ), i nb ( t ), i nc ( t )] as state variables; the measured variables are divided into real measured variables and virtual measured variables; the real measured variables include the MMC AC-side fed-in current i a ( t ), i b ( t ), i c ( t )], the DC-side current i dc ( t ), i dcn ( t )] and the voltage drops across the arm reactors and arm resistors u 1( t ), u2( t ), u 3( t ), u 4( t ), u 5( t ), u 6( t )] and the cumulative value of the capacitor voltage of the bridge arm sub-module u c,pa_total ( t ), u c,pb_total ( t ), u c,pc_total ( t ), u c,na_total ( t ), u c,nb_total ( t ), u c,nc_total ( t )]; The virtual measurement quantity is represented by 0, representing the relationship satisfied by the twin model; where:

[0029] ;

[0030] In Equation (6), k = 1, 2 corresponds to phase a; k = 3, 4 corresponds to phase b; k = 5, 6 corresponds to phase c;

[0031] The differential term d u c,rj_total / d t is expanded using quadratic integration;

[0032] Based on the above measurement quantities and state quantities, a measurement equation is established:

[0033] ;

[0034] In Equation (7), z is the measurement column vector; x is the state column vector; Y is the relationship matrix, obtained by calculating the parameters of the MMC and is a known quantity; C is the historical value matrix of the measurement quantities and state quantities; v represents the measurement error column vector.

[0035] On the basis of the above scheme, the specific steps of using the least squares method for dynamic state estimation to calculate the measurement error include:

[0036] Finding the optimal solution of the data by minimizing the sum of the squares of the errors:

[0037] ;

[0038] In Equation (8), W is the weight matrix, which reflects the noise estimation level of each measurement quantity; the weight matrix W is a diagonal matrix, and the diagonal elements are the standard deviations of the corresponding measurement quantities The reciprocal of the squared value;

[0039] According to the Karush-Kuhn-Tucker conditions, the necessary condition for the minimum of the above objective function is:

[0040] ;

[0041] The optimal estimate of the state quantity is:

[0042] ;

[0043] In the formula, H is the Jacobian matrix of h(x).

[0044] Based on the above scheme, the two sets of protection criteria for startup and operation proposed according to the protection action type specifically include:

[0045] The operating startup criterion module;

[0046] If it is the true value, the protection issues an alarm signal;

[0047] The operating action criterion module;

[0048] If it is the true value, the protection issues a trip signal.

[0049] Based on the above scheme, the startup criterion is established based on formula (2) and is responsible for determining whether a fault occurs within the zone;

[0050] The action criterion is established based on formula (11) and is responsible for determining the type of action signal;

[0051] ;

[0052] Construct a protection criterion for in-zone faults based on residuals;

[0053] ;

[0054] Furthermore, calculate the normalized sum of squares of all residuals as shown in Equation (13);

[0055] ;

[0056] The probability density of the chi-square distribution is shown in Equation (14), and Γ represents the gamma function;

[0057] ;

[0058] Set the protection criterion as follows:

[0059] ;

[0060] In the formula,

[0061] ;

[0062] In formula (15-16), P(t)=1 indicates that the MMC may have a fault; P(t)=0 indicates that the MMC is operating normally; Signal(t) represents the action signal given by the twin model at time t; is the residual threshold, obtained from the chi-square distribution critical value table; Tset is the criterion time window; Nset is the number of consecutive samples.

[0063] Based on the above solution, a power electronic system is provided, including a modular multilevel converter protection strategy based on digital twin as described in any one of claims 1-7.

[0064] The beneficial technical effects of the present invention are as follows:

[0065] Based on differential protection, the present invention can additionally detect inter-turn short circuit faults of the arm reactor, open circuit faults of the sub-module, and inter-phase short circuit faults of the arm. Description of the Drawings

[0066] The present invention has the following drawings:

[0067] Figure 1 A flowchart of a modular multilevel converter protection strategy

[0068] Figure 2 MM topology and sub-module topology diagram

[0069] Figure 3 Schematic diagram of the main faults of the converter

[0070] Figure 4 Probability density curve of the chi-square distribution

[0071] Figure 5 Flowchart of the protection principle

[0072] Figure 6 Schematic diagram of the simulation of a four-terminal pseudo-bipolar flexible DC transmission system

[0073] Figure 7 Protection discrimination result of Fault I

[0074] Figure 8 Protection discrimination result of Fault II

[0075] Figure 9 Protection discrimination result of out-of-zone fault Detailed Implementation Manner

[0076] The following combines the attachedFigures 1-9 A further detailed description of the present invention will be given below.

[0077] As Figure 1 , a protection strategy for a modular multilevel converter based on digital twin includes the following steps:

[0078] Establish a digital twin model according to all the physical laws followed by the MMC and in combination with the MMC control information;

[0079] Establish a measurement equation based on the MMC twin model;

[0080] Adopt the least squares method to perform dynamic state estimation to calculate the measurement error;

[0081] Propose two sets of protection criteria for starting and acting according to the protection action type.

[0082] As shown in the figure, each arm of the MMC is composed of an arm reactor L in series with N sub-modules (SM), and the upper and lower two arms of the same phase form a phase unit;

[0083] The topology structure of the sub-module adopts a half-bridge structure, and the switching operation of the sub-module is realized through the mutual cooperation of the switching devices VT1, VT2, VD1, and VD2. C is the sub-module capacitor.

[0084] On the basis of the above scheme, the establishment of the digital twin model according to all the physical laws followed by the MMC and in combination with the MMC control information specifically includes:

[0085] At any moment, the arm voltage urj (r = p, n, respectively representing the upper and lower arms; j = a, b, c, representing the three phases of abc) formed by the sub-modules of the same arm is expressed by the sub-module capacitor voltage and its switching function, that is:

[0086] ;

[0087] In formula (1), u SM,rj_i represents the voltage coupled into the arm by the i-th sub-module of the r-arm of the j-phase; S rj_i represents the switching function of the i-th sub-module of the r-arm of the j-phase; when the sub-module is put in, S rj_i = 1, when the sub-module is cut out, S rj_i = 0; u c,rj_i represents the capacitor voltage of the i-th sub-module of the r-arm of the j-phase, and the arm resistance R is used to equivalently represent the arm loss;

[0088] Based on KCL, KVL, and the relationship between the capacitor current of the sub-module and the arm current, the state equation of the MMC is written to obtain the digital twin model of the MMC in the time domain:

[0089] ;

[0090] In Equation (2), i rj is the current flowing through the arm; i j is the current fed into the AC side; i dc , i dcn are the DC side currents; u c,rj_total is the accumulated value of the capacitor voltage of the arm sub-module; S rj_total is the accumulated value of the switching function of the arm sub-module;

[0091] Equation (2) eliminates the differential term through the integration method and converts it into historical values for calculation;

[0092] The first-order integration method and the second-order integration method are shown in Equations (3) and (4 - 5) respectively:

[0093] ;

[0094] ;

[0095] ;

[0096] where Δt is the sampling interval. Compared with the first-order integration, the second-order integration improves the accuracy and numerical stability, but increases the measurement redundancy, resulting in an increase in the computational amount and a decrease in the protection sensitivity.

[0097] Based on the above scheme, the establishment of the measurement equation based on the MMC twin model specifically includes:

[0098] Taking the arm currents i pa ( t ), i pb ( t ), i pc ( t ), i na ( t ), i nb ( t ), i nc ( t)]As a state quantity; the measured quantities are divided into real measured quantities and virtual measured quantities; the real measured quantities include the MMC AC-side feeding current i a ( t ), i b ( t ), i c ( t )], the DC-side current i dc ( t ), i dcn ( t )], and the voltage drops across the arm reactors and arm resistors u 1( t ), u 2( t ), u 3( t ), u 4( t ), u 5( t ), u 6( t )], and the cumulative value of the arm sub-module capacitor voltages u c,pa_total ( t ), u c,pb_total ( t ), u c,pc_total ( t ), u c,na_total ( t ), u c,nb_total ( t ), u c,nc_total ( t )]; the virtual measured quantity is represented by 0, representing the relationship satisfied by the twin model; where:

[0099] ;

[0100] In Equation (6), k = 1, 2 corresponds to phase a; k = 3, 4 corresponds to phase b; k = 5, 6 corresponds to phase c;

[0101] Considering both the protection sensitivity and the calculation accuracy, only the differential term d u c,rj_total / d t is expanded using double integration.

[0102] Based on the above measured quantities and state quantities, a measurement equation is established:

[0103] ;

[0104] In Equation (7), z is the measurement column vector; x is the state column vector; Y is the relationship matrix, which is obtained by calculating the parameters of the MMC and is a known quantity; C is the historical value matrix of the measurement quantity and the state quantity; v represents the measurement error column vector.

[0105] The number of measurement quantities in the measurement equation is 24, and the number of state quantities is 12. Therefore, the redundancy is 200%.

[0106] On the basis of the above scheme, the present invention uses the least squares method for dynamic state estimation, and its basic principle is: to find the optimal solution of the data by minimizing the sum of the squares of the errors:

[0107] ;

[0108] In Equation (8), W is the weight matrix, which reflects the noise estimation level of each measurement quantity; the weight matrix W is a diagonal matrix, and the diagonal elements are the reciprocals of the standard deviations of the corresponding measurement quantities The reciprocal of the square value;

[0109] According to the Karush-Kuhn-Tucker conditions, the necessary condition for the minimum of the above objective function is:

[0110] ;

[0111] The optimal estimate of the state quantity is:

[0112] ;

[0113] In the formula, H is the Jacobian matrix of h(x).

[0114] Using the state quantity estimated by Equation (10), the discrimination of internal faults can be carried out.

[0115] On the basis of the above scheme, the two sets of protection criteria for starting and operating proposed according to the protection action type specifically include:

[0116] (1) Fault classification

[0117] Such as Figure 3, The twin model includes all the physical laws of the MMC. Therefore, when a fault occurs in the converter section, that is, an AC-side fault, a valve-section fault, or a DC-side fault occurs, the relationship matrix Y will be damaged, and the proposed protection algorithm can reliably identify it. A full investigation has been carried out on the transient characteristics of various faults in the converter section. Among them, after an AC-side fault, a valve short-circuit fault, an inter-phase short-circuit fault of the bridge arm, and a DC-side fault occur, the protection actions are consistent, which are to block the converter and simultaneously trip the AC circuit breaker; when an inter-turn short-circuit fault of the bridge arm reactor or an open-circuit fault of the sub-module occurs in the converter, the protection device needs to send out an alarm signal and automatically start the fault recording.

[0118] In this paper, two sets of protection criteria, namely startup and action, are established according to the protection action conditions. The startup criterion is established based on Equation (2) and is responsible for determining whether a fault occurs within the zone; the action criterion is established based on Equation (11) and is responsible for determining the type of action signal.

[0119] ;

[0120] (2) Construction of protection criteria

[0121] The residual probability distribution check is a classic algorithm for power system state estimation. Its basic principle is: assuming that the measurement is reliable, when there is no fault in the actual model, the residuals of the state estimation will conform to a certain probability distribution model. Based on this, a protection criterion for in-zone faults based on residuals is constructed in this paper. The residual is the difference between the estimated value of the measured quantity after dynamic state estimation and the measured quantity.

[0122] ;

[0123] Furthermore, the normalized sum of squares of all residuals is calculated.

[0124] ;

[0125] When the MMC is operating normally, the twin model matches the actual model at this time, and the residuals r of all measured quantities conform to the Gaussian distribution. Therefore, the normalized sum of squares of the residuals conforms to the chi-square distribution χ2(K). K is the degree of freedom of the chi-square distribution, which is equal to the number of measured quantities minus the number of state quantities. The probability density of the chi-square distribution is shown in Equation (14), Γ represents the gamma function, and its probability density curve is as Figure 4 shown.

[0126] ;

[0127] According to Figure 4 it can be known that the probability that it is greater than a certain large threshold is close to 0. It can be understood that after a fault occurs in the MMC, the twin model no longer matches the actual model, It increases correspondingly with the change of the residual. Therefore, the threshold of the normalized sum of squares of the residual can be reasonably set to detect faults in the MMC area. Therefore, the protection criterion is set as follows:

[0128] ;

[0129] In the formula,

[0130] ;

[0131] In formulas (15-16), P(t)=1 indicates that a fault may occur in the MMC; P(t)=0 indicates that the MMC is operating normally; Signal(t) represents the action signal given by the twin model at time t; is the residual threshold, obtained from the chi-square distribution critical value table; Tset is the criterion time window; Nset is the number of consecutive samples.

[0132] Considering the coordination of the two sets of protection criteria and the requirements of the MMC for the reliability and sensitivity of the protection, this paper sets ; Tset,1 = 0.1ms, Tset,2 = 1ms; Nset,1 = 1, Nset,2 = 3.

[0133] The protection principle flow chart is as Figure 5 shown.

[0134] To verify the effectiveness of the proposed invention, a ±250kV four-terminal pseudo-bipolar flexible DC transmission system as shown in Figure 6 is built. The actual parameters of the Zhangbei four-terminal flexible DC transmission system are used, and the voltage level is half of that of the actual project. Its main parameters are shown in Table 1. The protection sampling frequency is set to 50kHZ, and the proposed protection principle is verified with MMC1 as the protection object. The present invention classifies the faults in the MMC area into Fault Ⅰ (protection tripping) and Ⅱ (protection alarm) according to the protection action type.

[0135] When the system operates stably to 0ms, Fault Ⅰ occurs in MMC1, and the protection discrimination result is as shown in Figure 7 shown, and its fault setting and protection duration are shown in Table 2. The simulation results show that the normalized sum of squares of the residual is approximately 0 before the fault, and the relationship matrix is damaged after the fault, resulting in a rapid increase in the normalized sum of squares of the residual. Through the cooperation of the startup criterion and the action criterion, the protection algorithm can send out a tripping signal within 2ms after the fault. Since the valve-side winding of the MMC in the built pseudo-bipolar flexible DC transmission system is connected in a triangle, which belongs to a small-current grounding system, single-phase grounding faults on the AC side and valve group grounding faults will not damage the relationship matrix, so they are not considered in this paper.

[0136] When the system operates stably to 0ms, an inter-turn short circuit fault (the most extreme case) occurs in the upper bridge arm of phase A of MMC1, and the protection discrimination result is as shown in Figure 8(c). After 1.56 ms, the sum of the squared residuals after normalization is greater than the threshold of the starting criterion, and the protection issues an alarm signal. The operating criterion will not malfunction.

[0137] When the system operates stably until 0 ms, an open-circuit fault occurs in the first sub-module VT1 of the upper arm of phase A of MMC1. The protection discrimination result is as Figure 8 (a). After 0.04 ms, the sum of the squared residuals after normalization is greater than the threshold of the starting criterion, and the protection issues an alarm signal. The operating criterion will not malfunction; it is assumed that an open-circuit fault occurs in the first sub-module VT2 of the upper arm of phase A of MMC1 at 0 ms, and the protection discrimination result is as Figure 8 (b). After 13.76 ms, the sum of the squared residuals after normalization is greater than the threshold of the starting criterion, and the protection issues an alarm signal. The operating criterion will not malfunction.

[0138] The present invention conducts simulation verification on the reliability of the protection algorithm for external DC-side bipolar short-circuit faults and external AC-side three-phase short-circuit faults. It is assumed that an external DC-side bipolar short-circuit fault occurs at 0 ms, and the protection discrimination result is as Figure 9 (a). After the fault, the sum of the squared residuals after normalization fluctuates slightly but is much smaller than the threshold, and the protection will not malfunction. It is assumed that an external AC-side three-phase short-circuit fault occurs at 0 ms, and the protection discrimination result is as Figure 9 (b). The protection will not malfunction.

[0139] Table 1 Main parameters of MMC1

[0140]

[0141] Table 2 Fault settings and protection durations

[0142] Fault type Fault setting Protection operation duration / ms Valve short-circuit fault Upper arm of phase A 1 Inter-phase short-circuit fault on the AC side Short circuit between phases B and C 1.2 DC side single-pole grounding fault Positive-pole metallic grounding on the DC side 0.4 DC side bipolar short-circuit fault DC side bipolar short circuit 0.9 Inter-phase short-circuit fault of bridge arm Short circuit between bridge arms A and B 1.7

[0143] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

Claims

1. A protection method for a modular multilevel converter based on digital twin, characterized in that, It includes the following steps: Establish a digital twin model according to all the physical laws followed by the MMC and in combination with the MMC control information; Establish a measurement equation based on the MMC twin model; Use the least squares method for dynamic state estimation to calculate the measurement error; Propose two sets of protection criteria for startup and operation according to the protection action type; The specific steps of establishing a digital twin model according to all the physical laws followed by the MMC and in combination with the MMC control information include: At any moment, the arm voltage formed by the sub-modules of the same arm u rj ; r = p, n, representing the upper and lower arms respectively; j = a, b, c, representing the three phases of abc; expressed in terms of the capacitor voltage of the sub-module and its switching function, that is: ; In formula (1), u SM,rj_i represents the voltage coupled to the arm by the i-th sub-module of the r arm of the j-phase; S rj_i represents the switching function of the i-th sub-module of the r arm of the j-phase; when the sub-module is put in, S rj_i = 1, when the sub-module is cut out, S rj_i = 0; u c,rj_i represents the capacitor voltage of the i-th sub-module of the r arm of the j-phase, and the arm resistance R is used to equivalently represent the arm loss; Write the state equation for the MMC according to KCL, KVL and the relationship between the capacitor current of the sub-module and the arm current, and obtain the digital twin model of the MMC in the time domain: ; In formula (2), i rj is the current flowing through the arm; i j is the current fed into the AC side; i dc 、 i dcn is the DC side current; u c,rj_total is the cumulative value of the capacitor voltages of the arm sub-modules; S rj_total is the cumulative value of the switching functions of the arm sub-modules; In Equation (2), the differential term is eliminated by the integration method and converted into historical values for calculation; The first integration method and the second integration method are shown in Equations (3), (4) and (5) respectively: ; ; ; where Δ t is the sampling interval.

2. The modular multilevel converter protection method based on digital twin according to claim 1, wherein Each arm of the MMC is composed of an arm reactor L connected in series with N sub-modules (SM), and the upper and lower two arms of the same phase form a phase unit; The sub-module topology structure adopts a half-bridge structure, and the switching operation of the sub-module is realized through the mutual cooperation of the switching devices VT1, VT2, VD1 and VD2. C is the sub-module capacitor.

3. The protection method for a modular multilevel converter based on digital twin according to claim 2, characterized in that, The specific steps of establishing a measurement equation based on the MMC twin model include: Take the currents of each bridge arm i pa ( t ), i pb ( t ), i pc ( t ), i na ( t ), i nb ( t ), i nc ( t )] as state variables; the measured variables are divided into real measured variables and virtual measured variables; the real measured variables include the MMC AC-side feeding current i a ( t ), i b ( t ), i c ( t )], the DC-side current i dc ( t ), i dcn ( t )], the voltage drops across the arm reactors and arm resistors u 1( t ), u 2( t ), u 3( t ), u 4( t ), u 5( t ), u 6( t )], and the accumulated values of the capacitor voltages of the arm sub-modules u c,pa_total ( t ), u c,pb_total ( t ), u c,pc_total ( t ), u c,na_total ( t ), u c,nb_total ( t ), u c,nc_total ( t )]; The virtual measurement quantity is represented by 0, which represents the relationship satisfied by the twin model; where: ; In formula (6), k = 1, 2 correspond to phase a; k = 3, 4 correspond to phase b; k = 5, 6 correspond to phase c; Differential term d u c,rj_total / d t Expand using double integral; Establish a measurement equation according to the above measurement quantities and state quantities: ; In Equation (7), z is the measurement column vector; x is the state column vector; Y is the relationship matrix, which is obtained by calculating the parameters of the MMC and is a known quantity; C is the historical value matrix of the measurement quantity and the state quantity; v represents the measurement error column vector.

4. The protection method for a modular multilevel converter based on digital twin according to claim 2, wherein The specific steps of using the least squares method for dynamic state estimation to calculate the measurement error include: Find the optimal solution of the data by minimizing the sum of the squares of the errors: ; In Equation (8), W is the weight matrix, which reflects the noise estimation level of each measurement; the weight matrix W is a diagonal matrix, and the diagonal elements are the standard deviations of the corresponding measurements the reciprocal of the squared value; According to the Karush-Kuhn-Tucker conditions, the necessary conditions for the above optimal solution are: ; The optimal estimate of the state quantity is: ; In the formula, H is the Jacobian matrix of h(x).

5. The protection method for a modular multilevel converter based on digital twin according to claim 4, characterized in that, The specific steps of proposing two sets of protection criteria for startup and operation according to the protection action type include: Run the startup criterion module; If it is true, the protection issues an alarm signal; Run the operation criterion module; If it is true, the protection issues a tripping signal.

6. A protection method for a modular multilevel converter based on digital twin according to claim 5, characterized in that The startup criterion is established based on Formula (2) and is responsible for discriminating whether an internal fault occurs; The operation criterion is established based on Formula (11) and is responsible for discriminating the type of operation signal; ; Construct an internal fault protection criterion based on the residual; ; Further, calculate the normalized sum of squares of all residuals as shown in Equation (13); ; The probability density of the chi-square distribution is shown in Equation (14), and Γ represents the gamma function; ; Set the protection criterion as follows: ; In the formula, ; In Equation (15-16), P(t)=1 indicates that the MMC may malfunction; P(t)=0 indicates that the MMC is operating normally; Signal(t) represents the action signal given by the twin model at time t; is the residual threshold, obtained from the chi-square distribution critical value table; Tset is the criterion time window; Nset is the number of consecutive samples.

7. A power electronic system includes a protection method for a modular multilevel converter based on digital twin according to any one of claims 1-6.

Images