Method for improving reactive power support ability of MMC considering dynamic boundary under power grid fault

Abstract

The present invention relates to a method for improving the reactive power support capability of an MMC considering the dynamic boundary under grid faults, belonging to the field of flexible DC transmission control, and includes: constructing an MMC dynamic boundary calculation unit under grid faults; constructing a fitting function for a zero-sequence signal calculation unit to improve the reactive power support capability of the MMC; based on the reactive power Q <subgt; 0.max < / subgt;, Q <subgt; 1.max < / subgt; and the reactive power Q <subgt; det

Description

Technical Field

[0001] The present invention relates to a method for improving the reactive power support capability of an MMC considering the dynamic boundary under grid faults, belonging to the technical field of flexible DC transmission control. Background Art

[0002] As one of the core devices of a flexible DC transmission system, the Modular Multilevel Converter (MMC) has the advantages of excellent output waveform quality, high modularity and scalability, high reliability, and flexible control, and is widely used in fields such as large-capacity long-distance power transmission, offshore wind power grid connection, and power system regulation and optimization. Compared with traditional two-level or three-level converters, the MMC has significant differences in topological structure and control methods, making the system operation more reliable and flexible. However, when a grid-side fault causes a voltage drop, the grid-forming MMC needs to increase reactive power output in a short time to support the grid-side voltage. If the MMC lacks sufficient reactive power support capability, it will cause a serious drop in grid voltage, which may lead to voltage collapse. At the same time, the converter itself may also be overloaded or shut down due to the inability to provide reactive power support, affecting the energy transmission of the grid and causing large-scale power outages.

[0003] However, the traditional methods for improving the reactive power support capability of converter stations mainly include increasing the converter capacity, adding reactive power compensation devices such as Static Var Compensators (SVCs), adjusting control strategies, allowing short-term overload operation, and adding filters or energy storage systems. However, these methods generally have defects such as high cost, increased equipment complexity and maintenance difficulty, compensation capacity limited by the rated capacity, difficulty in long-term full-load operation, and insufficient dynamic adaptability, and it is difficult to balance economy, reliability, and flexibility. Therefore, it is of great significance to develop a method for economically and flexibly improving the reactive power support capability of MMCs under grid faults. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a method for improving the reactive power support capability of an MMC considering the dynamic boundary under grid faults, which can expand the operation range of the MMC, improve the reactive power support capability of the converter, reduce the risks of grid collapse and system oscillation, and overcome the defects of high cost, large equipment complexity and maintenance difficulty, and insufficient dynamic adaptability of traditional methods. By considering the dynamic boundary of the converter under grid faults, it has the characteristics of flexibility, real-time performance, and economy.

[0005] The present invention adopts the following technical solutions:

[0006] A method for improving the reactive power support capability of an MMC considering the dynamic boundary under grid faults, comprising the following steps:

[0007] S1: Constructing the MMC dynamic boundary calculation unit under power grid fault;

[0008] S2: Constructing the fitting function of the zero-sequence signal calculation unit to improve the reactive power support capability of the MMC;

[0009] S3: Reactive power with and without zero-sequence signal injection output based on the constructed dynamic boundary calculation unit 、 The reactive power detected in real time in the system Q det Determine whether to start the zero-sequence signal calculation unit;

[0010] S4: If the zero-sequence signal calculation unit is started, the zero-sequence signal of the expanded operating domain under the current working condition of the MMC is calculated in real time and injected into the MMC modulation module to improve the reactive power support capability of the MMC.

[0011] Preferably, the implementation process of step S1 is:

[0012] S11: Considering the voltage drop on the grid side under the power grid fault, the minimum allowable voltage of the PCC point is Us.min =0.7* Us.rated , the maximum allowable voltage is Us.max =1.0* Us.rated ,in Us.rated The rated voltage of the PPC point is divided by the AC side voltage U s Calculate the point and calculate the AC side voltage U s ;

[0013] S12: Calculate the voltages on each AC side when no zero-sequence signal is injected Us i Command modulation signal margin S margin The maximum active power output by MMC when it is 0 P 0 , maximum reactive power Q 0 , modulation signal margin when there is no zero-sequence signal injection S margin Considered as Us i 、 P 0 、 Q 0 function, each U si and U si The corresponding calculated P 0 、Q 0 Form a data set G no.A3 ;

[0014] Then calculate the AC side voltages when injecting zero-sequence signals Us i Order the modulation signal margin S margin The maximum active power output by the MMC when the modulation signal margin is 0 P 1 and the maximum reactive power Q 1 , and the modulation signal margin when injecting zero-sequence signals S margin is regarded as Us i , P 1 , Q 1 , A 3 , α 3 functions of, and for each U si and U si the corresponding calculated P 1 , Q 1 , A 3 , α 3 form a data set G A3 , where A 3 is the amplitude of the third harmonic component, α 3 is the phase angle of the third harmonic component

[0015] Preferably, the implementation process of step S2 is as follows:

[0016] S21: Using the AC side voltages G no.A3 and the maximum active power U s obtained in step S12 as independent variables, and the maximum reactive power P 0 as the dependent variable, obtain the fitting function A as follows: Q 0 ;

[0017] ;

[0018] Where represents a certain maximum reactive power when no zero-sequence signal is injected Q 0 can be expressed as a series of numbers composed of

[0019] Taking the AC-side voltage G A3 in the data set obtained in step S12 U s and the maximum active power P 1 as independent variables, and the maximum reactive power Q 1 as the dependent variable, a fitting function B is obtained as follows:

[0020] ;

[0021] where represents a certain maximum reactive power when a zero-sequence signal is injected Q 1 can be expressed as a series of numbers composed of

[0022] Input any AC-side voltage U s and active power P , then the maximum reactive power that the MMC can output without zero-sequence signal injection and the maximum reactive power that the MMC can output when a zero-sequence signal is injected can be obtained;

[0023] S22: Divide the calculation points of the AC-side voltage U s , active power P , and reactive power Q . Each set of U s , P , Q constitutes an operating condition point of the MMC. At each operating condition point, there corresponds an amplitude A 3 and phase angle α 3 of the 3rd harmonic component of the zero-sequence signal to maximize the modulation signal margin;

[0024] Calculate the U s , P , Q under which the modulation signal margin is maximized A 3 and α 3 , and store the data to form a data setG A3α3 ;

[0025] S23: Use the data set obtained in step S22 G A3α3 in U s , P , Q as independent variables, and use A 3 and α 3 as dependent variables for function fitting respectively to obtain fitting function C and fitting function D:

[0026] Fitting function C: ;

[0027] Fitting function D: ;

[0028] wherein, A 3.Cal and the phase angle α 3.Cal are respectively the amplitude and phase angle of the 3rd harmonic component that maximizes the modulation signal margin after fitting;

[0029] Input any AC side voltage U s , active power P , reactive power Q to obtain the amplitude A 3.Cal and phase angle α 3.Cal of the 3rd harmonic component of the zero-sequence signal that enables the MMC to maximize the modulation signal margin at this time.

[0030] Preferably, the implementation process of step S3 is as follows:

[0031] S31: Real-time detect the AC side voltage U s.det , the active power P det output by the MMC, and the reactive power Q det , and substitute U s.det , P det into fitting function A and fitting function B respectively to obtain the maximum reactive power that the MMC can output without zero-sequence signal injection and the maximum reactive power that the MMC can output with zero-sequence signal injection;

[0032] Substitute the real-time detected Us.det , P det and Q det are respectively substituted into the fitting function C and the fitting function D to obtain the amplitude A 3.Cal and the phase angle α 3.Cal of the 3rd harmonic component of the zero-sequence signal injected into the MMC at this time;

[0033] S32: Compare the reactive power Q det calculated in step S6, and . When Q det ≤ , A 3 = 0, α 3 = 0, no alarm; when <Q det < , A 3 = A 3.Cal , α 3 = α 3.Cal , no alarm; when Q det ≥ , A 3 = A 3.Cal , α 3 = α 3.Cal , alarm, indicating that the current operating condition of the MMC has exceeded the limit operating condition;

[0034] When there is no alarm, start the zero-sequence signal calculation unit, and inject the A 3 and α 3 under the condition of no alarm into the MMC modulation module to improve the reactive power support ability of the MMC without overmodulation under linear control.

[0035] Preferably, in step S11, ;

[0036] Among them, U si represents the iThe AC-side voltage of the calculation point n is the calculation interval.

[0037] Preferably, in step S12, when there is no zero-sequence signal injection, the modulation signal margin S margin is regarded as Us i , P 0 , Q 0 The function of is:

[0038] ;

[0039] When a zero-sequence signal is injected, the modulation signal margin S margin is regarded as Us i , P 1 , Q 1 , A 3 , α 3 The function of is:

[0040] ;

[0041] Taking the upper bridge arm of phase A as an example, the modulation signal margin S margin The calculation formula is as follows:

[0042] ;

[0043] ;

[0044] Among them, S ap ( t ) is the modulation signal of the upper bridge arm of phase A, A 0 , A 1 , A 2 and A 3 are the amplitudes of the DC, fundamental, second, and third harmonic components respectively, A 3 is also called the zero-sequence signal amplitude; α 1 , α 2 , α 3 represent the phase angles of the fundamental, second, and third harmonic components; ω is the fundamental angular frequency.

[0045] Preferably, in step S22, the AC-side voltage U s , active power P , reactive power Q calculation points are provided, where the active power P , reactive power Q calculation points are adjusted according to the MMC operation parameters, and their expressions are as follows:

[0046] ;

[0047] wherein, i represents the i th calculation point, j represents the maximum number of active power calculations, k represents the maximum number of reactive power calculations; P max , Q max are the maximum active power and reactive power recorded in the data sets G no.A3 and G A3 , and their expressions are as follows:

[0048] ;

[0049] represents all the active power data in the data set G no.A3 obtained without zero-sequence signal injection, represents all the reactive power data in the data set G no.A3 obtained without zero-sequence signal injection, represents all the active power data in the data set G A3 obtained with zero-sequence signal injection, represents all the reactive power data in the data set G A3 obtained with zero-sequence signal injection.

[0050] For details not elaborated in the present invention, reference may be made to the prior art.

[0051] The beneficial effects of the present invention are as follows:

[0052] The present invention proposes a method for improving the reactive power support ability of an MMC considering the dynamic boundary under grid faults, which can expand the operation range of the MMC, enhance the reactive power support ability of the converter, and reduce the risks of grid collapse and system oscillation. The method proposed by the present invention overcomes the defects of traditional methods such as high cost, large equipment complexity and maintenance difficulty, and insufficient dynamic adaptability. By considering the dynamic boundary of the converter under grid faults, it has flexibility, real-time performance and economy. The present invention can realize adjustment based on the real-time working conditions of the converter, improve the reactive power support ability of the MMC, ensure effective voltage regulation and improve the system operation performance when grid-side faults or voltage fluctuations occur, providing a solid technical support for the safe and reliable operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation of this application.

[0054] Figure 1 It is an overall schematic diagram of a method for improving the reactive power support ability of an MMC considering the dynamic boundary under grid faults;

[0055] Figure 2 It is a flow chart for constructing a fitting function of a zero-sequence signal calculation unit for improving the reactive power support ability of an MMC;

[0056] Figure 3 It is a diagram of MMC dynamic boundary data and fitting;

[0057] Figure 4 It is a comparison diagram of the reactive power output by a grid-forming MMC and the maximum available reactive power output during voltage sag;

[0058] Figure 5 It is a schematic diagram of the amplitude of the output zero-sequence signal;

[0059] Figure 6 It is a schematic diagram of the phase angle of the output zero-sequence signal;

[0060] Figure 7 It is a modulation signal diagram of the upper arm of phase A after adopting the method proposed by the present invention;

[0061] Figure 8 It is a modulation signal diagram of the upper arm of phase A before adopting the method proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] In order to enable those skilled in the art to better understand the technical solutions in this specification, the following clearly and completely describes the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of this specification, but not limited thereto. Those not elaborated in detail in the present invention are all conventional technologies in the art.

[0063] Embodiment 1

[0064] A method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults, as Figure 1 shown, as Figure 1 shown, includes the following steps:

[0065] S1: Construct a dynamic boundary calculation unit of MMC under grid faults;

[0066] S2: Construct a fitting function of a zero-sequence signal calculation unit for improving the reactive power support ability of MMC;

[0067] S3: Based on the reactive power with and without zero-sequence signal injection output by the constructed dynamic boundary calculation unit 、 and the reactive power detected in real time in the system Q det judge whether to start the zero-sequence signal calculation unit;

[0068] S4: If the zero-sequence signal calculation unit is started, then calculate in real time the zero-sequence signal for expanding the operation domain under the current working condition of MMC and inject it into the MMC modulation module to realize the improvement of the reactive power support ability of MMC.

[0069] Embodiment 2

[0070] A method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults, as shown in Embodiment 1. The difference is that, as Figure 2 shown, the implementation process of step S1 is:

[0071] S11: Considering the voltage dip on the grid side under grid faults, when the converter station does not trip off the grid, the lowest allowable voltage at the PCC point is Us.min =0.7* Us.rated and the highest allowable voltage is Us.max =1.0* Us.rated where Us.rated is the rated voltage of the PPC point, divide the AC side voltage U s calculation points, and calculate the AC side voltage U s as follows:

[0072] ;

[0073] where, U si represents the AC side voltage at the i th calculation point in the calculation interval of the lowest allowable voltage and the highest allowable voltage at the PPC point, n is the calculation interval, which can be determined by the required accuracy of the calculation. If n =10, then Usi For U s0 ~U s10 There are 11 in total. Divide Us.min to Us.max into 11 calculation points.

[0074] S12: Calculate the AC side voltages when no zero-sequence signal is injected Us i Order the modulation signal margin S margin When it is 0, the maximum active power output by the MMC P 0 and the maximum reactive power Q 0 . In the construction of the dynamic boundary calculation unit of the present invention, when no zero-sequence signal is injected, the modulation signal margin S margin is regarded as Us i and P 0 and Q 0 function:

[0075] ;

[0076] In this function, P 0 and Q 0 are all unknowns. Model the MMC. When other parameters and Us i are determined, the current modulation signal margin S margin when it is 0 P 0 and Q 0 can be calculated through the established model. For each Us i , multiple P 0 and Q 0 are calculated. For example, Us 0 、Us 1 each corresponds to 20 P 0 and Q 0 ( P 0 and Q 0 determine one point).

[0077] In this embodiment, the modulation signal of the MMC is obtained by superimposing the output signals of the power controller, the circulating current controller, and the zero-sequence signal controller in the substation-level controller. Among them, Us i 、 P 0 、 Q 0 will affect the modulation signal S ap ( t ) in A 1 and α 1 values, and further affect the modulation signal margin S margin . Therefore, S margin can be regarded as Us i 、 P 0 、 Q 0 function. Regarding Us i 、 P 0 、 Q 0 as independent variables and the signal margin S margin as the dependent variable.

[0078] For each U si and U si calculate the corresponding P 0 、 Q 0 to form a data set G no.A3 ;

[0079] Then calculate the maximum active power Us i commanded by the modulation signal margin S margin when the zero-sequence signal is injected and the modulation signal margin P 1 is 0, and the maximum reactive power Q 1 output by the MMC. In the construction of the dynamic boundary calculation unit of the present invention, the modulation signal margin S margin is regarded as Us i 、 P 1 、 Q1 , A 3 , α 3 functions:

[0080] ;

[0081] In this function, the modulation signal margin S margin is the dependent variable, Us i , P 1 , Q 1 , A 3 , α 3 are the independent variables. That is to say, given Us i , P 1 , Q 1 , A 3 , α 3 values can obtain a S margin . By modeling the MMC, when other parameters and Us i are determined, the modulation signal margin S margin equal to 0 of P 1 , Q 1 , A 3 , α 3 values can be calculated.

[0082] Each U si and U si corresponding calculated P 1 , Q 1 , A 3 , α 3 form a data set G A3 , where A 3 is the amplitude of the third harmonic component, α 3is the phase angle of the 3rd harmonic component.

[0083] Taking the upper arm of phase A as an example, the modulation signal margin S margin The calculation formula is as follows:

[0084] ;

[0085] ;

[0086] Among them, S ap ( t ) is the modulation signal of the upper arm of phase A, which is generated by the MMC station-level control: power controller, circulating current controller, average capacitor voltage controller, zero-sequence signal injection controller, as Figure 1 , S ap generally refers to the modulation signal of the upper arm of phase A, S an represents the modulation signal of the lower arm of phase A. When no zero-sequence signal is injected, the modulation signal margin S margin and the modulation signal margin S margin when a zero-sequence signal is injected can both be calculated using this formula. When no zero-sequence signal is injected, A 3 , α 3 have values of 0; when a zero-sequence signal is injected, A 3 , α 3 have non-zero values. A 0 , A 1 , A 2 and A 3 are the amplitudes of the DC, 1st, 2nd, and 3rd harmonic components respectively, A 3 is also called the zero-sequence signal amplitude; α 1 , α 2 , α 3 represent the phase angles of the 1st, 2nd, and 3rd harmonic components; ω is the fundamental angular frequency. Different U s , P , Q will affect the magnitudes of the modulation signals of each frequency through each station-level controller, thereby affecting the magnitude of the modulation signal margin.

[0087] Embodiment 3

[0088] A method for improving the reactive power support ability of an MMC considering the dynamic boundary under grid faults is as shown in Embodiment 2. The difference is that the implementation process of step S2 is as follows:

[0089] S21: Take the AC-side voltage G no.A3 and the maximum active power U s obtained in step S12 as independent variables, and the maximum reactive power P 0 as the dependent variable to obtain the fitting function A as follows: Q 0 ;

[0090]

[0091] Among them, represents a certain maximum reactive power when no zero-sequence signal is injected, Q 0 can represent a series of numbers composed of;

[0092] f b0 ( ) represents a function, indicating the boundary function without zero-sequence signal. This function can be obtained by fitting with an existing neural network, and the specific function expression cannot be given. Enter the independent variable into the black box, and the dependent variable can be output. Us i represents Us the i th calculation point of, and finally the AC-side voltage U s is used as the independent variable for function fitting.

[0093] Take the AC-side voltage G A3 and the maximum active power U s in the data set obtained in step S12 as independent variables, and the maximum reactive power P 1 as the dependent variable to obtain the fitting function B as follows: Q 1 ;

[0094]

[0095] Among them, represents a certain maximum reactive power when a zero-sequence signal is injected, Q 1 can represent A series of numbers;

[0096] f b1 ( ) represents a function, which is the boundary function when there is a zero-sequence signal. This function can be obtained by fitting with an existing neural network. Since the specific function expression cannot be given, when the independent variable is input into the black box, the dependent variable can be output.

[0097] Input any AC-side voltage U s and active power P , and then the maximum reactive power that the MMC can output without zero-sequence signal injection can be obtained and the maximum reactive power that the MMC can output with zero-sequence signal injection ;

[0098] S22: Divide the AC-side voltage U s , active power P , reactive power Q calculation points, where the calculation points of the AC-side voltage U s are divided as shown in the formula of the above step S11. The calculation points of the active power P , reactive power Q are adjusted according to the MMC operating parameters, and their expressions are as follows:

[0099] ;

[0100] Among them, i represents the i th calculation point, j represents the maximum number of calculations of the active power, k represents the maximum number of calculations of the reactive power; P max , Q <s max are the maximum values of the active power and reactive power recorded in the data sets G no.A3 and G A3 , and their expressions are as follows:

[0101] ;

[0102] represents all the active power data in the data set G no.A3 obtained without zero-sequence signal injection, represents all the reactive power data in the data set G no.A3 obtained without zero-sequence signal injection, Denote the data set obtained under the injection of zero-sequence signals G A3 All the active power data in Denote the data set obtained under the injection of zero-sequence signals G A3 All the reactive power data in

[0103] Each U s 、 P 、 Q constitutes an operating condition point of the MMC operation, P i Denote P The i th calculation point, Q i Denote Q The i th calculation point. At each operating condition point, there corresponds an amplitude A 3 and phase angle α 3 of the 3rd harmonic component of the zero-sequence signal to maximize the modulation signal margin;

[0104] Through modeling and calculation, given U s 、 P 、 Q to maximize the modulation signal margin of A 3 and α 3 , and store the data to form a data set G A3α3 ; Since U s 、 P 、 Q and the zero-sequence signal will affect the magnitudes of the modulation signals at each frequency through each station-level controller, thereby affecting the modulation signal margin. Therefore, in the construction of the zero-sequence signal calculation unit of the present invention, the modulation signal margin S margin can be regarded as Us i 、 P i 、 Q i 、 A 3 、 α 3 、 a function of

[0105] S23: The data set obtained in step S22 G A3α3 In Us , P , Q As independent variables, A 3 and α 3 are respectively loaded as dependent variables into the neural network for training and function fitting, and the trained fitting functions C and D are obtained:

[0106] Fitting function C: ;

[0107] Fitting function D: ;

[0108] Among them, A 3.Cal and the phase angle α 3.Cal are respectively the amplitude and phase angle of the 3rd harmonic component that makes the modulation signal margin maximum after fitting;

[0109] The neural network is a tool that can be used to fit data to generate a fitting function. By inputting the dependent variable and independent variable through the neural network package built in MATLAB, the data can be fitted and trained to obtain the fitting function. This part can be obtained by using the existing technology.

[0110] The above-mentioned fitting functions C and D are assembled into the zero-sequence signal calculation unit, and any AC side voltage U s , active power P , reactive power Q are input into the zero-sequence signal calculation unit, and the amplitude A 3.Cal and phase angle α 3.Cal of the 3rd harmonic component of the zero-sequence signal that enables the MMC to have the maximum modulation signal margin at this time can be obtained.

[0111] Embodiment 4

[0112] A method for improving the reactive power support ability of an MMC considering the dynamic boundary under grid faults is as shown in Embodiment 3. The difference is that the implementation process of step S3 is as follows:

[0113] S31: Real-time detect the AC side voltage U s.det , the active power P det output by the MMC and the reactive power Q det . U s.det , P detSubstitute into fitting function A and fitting function B respectively to obtain the maximum reactive power that the MMC can output without zero-sequence signal injection and the maximum reactive power that the MMC can output with zero-sequence signal injection ;

[0114] Substitute the U s.det 、 P det and Q det obtained by real-time detection into fitting function C and fitting function D respectively to obtain the amplitude A 3.Cal and phase angle α 3.Cal of the 3rd harmonic component of the zero-sequence signal injected into the MMC at this time;

[0115] S32: Compare the reactive power Q det obtained in step S6, and . When Q det ≤ , A 3 = 0, α 3 = 0, no alarm; when <Q det < , A 3 = A 3.Cal , α 3 = α 3.Cal , no alarm; when Q det ≥ , A 3 = A 3.Cal , α 3 = α 3.Cal , alarm, indicating that the current operating condition of the MMC has exceeded the limit condition. If the MMC is to continue running, other methods need to be taken by the staff. Specifically, as shown in Table 1:

[0116] Table 1 Zero-sequence signal output comparison table

[0117]

[0118] When there is no alarm, the zero-sequence signal calculation unit is started, and the A 3 and α 3 are injected into the MMC modulation module to improve the reactive power support ability of the MMC without overmodulation under linear control.

[0119] The present invention inputs the real-time collected AC-side voltage and the active and reactive powers output by the converter, and calculates the amplitude and phase angle of the zero-sequence signal that maximizes the modulation signal margin of the MMC under the current working condition. Injecting this zero-sequence signal can enable the MMC to improve its reactive power support ability and operate in the linear control region without overmodulation.

[0120] In order to verify the method for improving the reactive power support ability of the MMC considering the dynamic boundary under grid faults proposed in this embodiment, the following is verified in combination with an example. The main circuit parameters of the MMC of the flexible DC converter station in the example are shown in Table 2.

[0121] Table 2 MMC main circuit parameters

[0122]

[0123] First, a dynamic boundary calculation unit is constructed. When a fault occurs on the grid side, resulting in a drop in the grid-side voltage U s When it occurs, considering the grid-side voltage margin standard and the data in Table 2, select Us.min =148.75 kV, take n= 12 pairs U s of calculation points are divided. Calculate the active power U s and reactive power S margin when the modulation signal margin P 0 is 0 for each grid-side voltage Q 0 , and save the data obtained from each calculation to form a data set G no.A3 .

[0124] Then calculate the active power Us i output by the MMC and the reactive power S margin when the modulation signal margin P 1 is 0 for each AC-side voltage Q 1 when the zero-sequence signal is injected, and save the data obtained from each calculation to form a data set G A3 .

[0125] Among them, the data set G no.A3 The operating points in are as Figure 3 shown by the blue solid dots in the data set G A3 The operating points in are as Figure 3 shown by the red hollow dots. The data set G no.A3 The AC-side voltage in U s , active power P 0 are used as independent variables, and the reactive power Q 0 is used as the dependent variable, and function fitting is performed through a neural network ; the data set G A3 The AC-side voltage in U s , active power P 1 are used as independent variables, and the reactive power Q 1 is used as the dependent variable, and function fitting is performed through a neural network . The two fitted function surfaces are as Figure 3 shown

[0126] Combine the above two fitted functions into the dynamic boundary calculation unit. At this time, the MMC dynamic boundary calculation unit under the MMC main circuit parameters in Table 2 is obtained. When the following are input into this unit U s , P the maximum reactive power that can be output with or without zero-sequence signal injection under the current voltage and active power output of the MMC can be obtained and .

[0127] Then, a zero-sequence signal calculation unit is constructed. Calculate the zero-sequence signal amplitude U s and phase angle P that maximize the modulation signal margin under the given AC-side voltage Q , MMC output active power A 3 α 3 , select P max , Q max as the maximum values of the active power and reactive power recorded in the data set G A3 , select n= 7, j= ​22 ，k= 11. Solve the zero-sequence signal amplitude and phase angle that maximize the modulation signal margin at each operating point. The solution process is as follows: By modeling the MMC, taking the maximum modulation signal margin as the optimization objective function, regarding the zero-sequence signal amplitude A 3 and the phase angle α 3 as independent variables and imposing constraint conditions to ensure the final convergence of the objective function. Using algorithms such as the interior-point method to solve the constrained nonlinear optimization problem, the solution of the zero-sequence signal amplitude and phase angle can be completed. This is the prior art and will not be elaborated here. In this embodiment, the zero-sequence signal amplitude and phase angle at a total of 1694 operating points are solved, and the data is stored to form a data set G A3α3 .

[0128] Regarding the AC-side voltage G A3α3 , active power U s , and reactive power P in the data set Q as independent variables, and the zero-sequence signal amplitude A 3 and the phase angle α 3 as dependent variables, and through neural network training function fitting, obtain A 3.Cal =f A3 (U s , P , Q) . α 3.Cal =f A3 (U s , P , Q) . Combine the above two fitting functions into the zero-sequence signal calculation unit. At this time, the zero-sequence signal calculation unit under the MMC main circuit parameters in Table 2 is obtained. When inputting U s , P , Q to it, the zero-sequence signal amplitude A 3.Cal and the phase angle α 3.Cal of the MMC that maximize the modulation signal margin under the current voltage and active and reactive power outputs can be obtained

[0129] Taking the grid-forming MMC operating normally at U s = 212.5 kV,P = 1500 MW, Q = 100 MVar operating condition, when T = 1 s, a fault occurs on the grid side, resulting in the grid side voltage U s dropping to 0.75 pu, that is, 159.375 kV as an example. Due to the characteristics of the grid-forming control, in order to maintain the grid side voltage from dropping, the MMC increases the reactive power output. At this time, the maximum reactive power that the MMC can output under the current voltage and active power output of the dynamic boundary calculation unit 、 and the real-time detection value of the reactive power output by the MMC Q det such as Figure 4 shown. As can be seen from Figure 4 , when T < 1 s, Q det < , at this time the output A 3 = 0, α 3 = 0, such as Figure 5 、 Figure 6 shown. When T = 1 s, the MMC increases the reactive power output, making <Q det < , at this time the zero-sequence signal calculation unit is started.

[0130] After the zero-sequence signal calculation unit is started, the grid side voltage U s , the real-time detection value of the active power output by the MMC P det , the real-time detection value of the reactive power Q det are input into the zero-sequence signal calculation unit, and the amplitude and phase angle of the output zero-sequence signal are respectively as Figure 5 、 Figure 6 shown. Injecting the calculated zero-sequence signal into the modulation module, the waveform diagram of the modulation signal after injecting the zero-sequence signal using the proposed method can be obtained as Figure 7 shown. As can be seen from the figure, at this time the MMC avoids the overmodulation risk and is still in the linear control region (when the modulation signal is between 0 and 1, it is in the linear control region). At the same time, without using the method proposed in the present invention, when the voltage drops and the grid-forming MMC increases the reactive power output, the waveform diagram of the modulation signal is as Figure 8 shown. As can be seen from the figure, at this time the modulation signal of the MMC is overmodulated. Due to the existence of the amplitude limiting link in the modulation module, the signal exceeding 0 to 1 of the modulation signal is clipped, so that the MMC enters the non-linear control region.

[0131] By adopting the method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults proposed by the present invention, the real-time improvement of the reactive power support ability of MMC can be achieved economically and efficiently according to the current operating conditions, enabling the grid-forming MMC to provide greater reactive power support during voltage dips, significantly improving the voltage stability and disturbance resistance of the system, while avoiding the harmonic distortion and operation risks brought by overmodulation, and improving the overall performance and reliability of the system.

[0132] The above is the preferred embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for improving the reactive power support ability of MMC considering the dynamic boundary under power grid faults, characterized in that, It includes the following steps: S1: Construct a dynamic boundary calculation unit of the MMC under grid faults; S2: Construct a fitting function of a zero-sequence signal calculation unit for enhancing the reactive power support ability of the MMC; S3: Calculate the reactive power with and without zero-sequence signal injection output by the constructed dynamic boundary calculation unit and the reactive power detected in real time in the system Q det to determine whether to start the zero-sequence signal calculation unit; S4: If the zero-sequence signal calculation unit is started, calculate in real time the zero-sequence signal of the expanded operation area under the current working condition of the MMC and inject it into the MMC modulation module to achieve the improvement of the reactive power support ability of the MMC.

2. The method for improving the reactive power support ability of MMC considering dynamic boundaries under power grid faults according to claim 1, characterized in that, The implementation process of step S1 is as follows: S11: Considering the grid-side voltage dip under grid faults, when the converter station operates without disconnecting from the grid, the minimum allowable voltage at the PCC point is Us.min = 0.7 * Us.rated , and the maximum allowable voltage is Us.max = 1.0 * Us.rated , where Us.rated is the rated voltage of the PPC point, divide the AC-side voltage U s calculation point, and calculate the AC-side voltage U s ; S12: Calculate the AC side voltages when no zero-sequence signal is injected Us i Order the modulation signal margin S margin The maximum active power output by the MMC when it is 0 P 0 and the maximum reactive power Q 0 , the modulation signal margin when no zero-sequence signal is injected S margin is regarded as Us i , P 0 , Q 0 's function, and for each U si and U si calculate the corresponding P 0 , Q 0 to form a data set G no.A3 ; Then calculate the AC-side voltages when injecting zero-sequence signals Us i Order the modulation signal margin S margin The maximum active power output by the MMC when it is 0 P 1 and the maximum reactive power Q 1 When injecting zero-sequence signals, the modulation signal margin S margin is regarded as Us i and P 1 and Q 1 and A 3 and α 3 functions of, and for each U si and U si the corresponding calculated P 1 and Q 1 and A 3 and α 3 form a data set G A3 where A 3 is the amplitude of the 3rd harmonic component, α 3 is the phase angle of the 3rd harmonic component.

3. The method for improving the reactive power support ability of MMC considering the dynamic boundary under grid fault according to claim 2, characterized in that, The implementation process of step S2 is as follows: S21: Use the data set obtained in step S12 G no.A3 AC side voltage U s , maximum active power P 0 as independent variables, and maximum reactive power Q 0 as the dependent variable to obtain the fitting function A as follows: ； Among them, represents a certain maximum reactive power when no zero-sequence signal is injected; The data set obtained in step S12 G A3 The AC side voltage in U s , the maximum active power P 1 As independent variables, the maximum reactive power Q 1 As the dependent variable, the fitting function B is obtained as follows: ； Among them, represents a certain maximum reactive power when injecting a zero-sequence signal; Input any AC-side voltage U s and active power P , and then the maximum reactive power that the MMC can output without zero-sequence signal injection and the maximum reactive power that the MMC can output with zero-sequence signal injection ; S22: Divide the AC-side voltage U s , active power P , reactive power Q 's calculation points, each U s , P , Q constitutes an operating condition point of the MMC operation. At each operating condition point, there corresponds an amplitude A 3 and phase angle α 3 of the 3rd harmonic component of the zero-sequence signal to maximize the modulation signal margin; Calculate the given U s 、 P 、 Q such that the modulation signal margin is maximized A 3 and α 3 , and store the data to form a data set G A3α3 ; S23: Take the data set obtained in step S22 G A3α3 in U s 、 P 、 Q as independent variables, and take A 3 and α 3 as dependent variables respectively for function fitting to obtain fitting function C and fitting function D: Fitting function C: ; Fitting function D: ; Among them, A 3.Cal and the phase angle α 3.Cal are respectively the amplitude and phase angle of the third harmonic component that maximizes the modulation signal margin after fitting; Input any AC side voltage U s , active power P , reactive power Q to obtain the amplitude of the 3rd harmonic component of the zero-sequence signal that can maximize the modulation signal margin of the MMC at this time A 3.Cal and phase angle α 3.Cal .

4. The method for improving the reactive power support capability of MMC considering the dynamic boundary under grid faults according to claim 3, wherein The implementation process of step S3 is as follows: S31: Detect the AC-side voltage in the real-time detection system U s.det , the active power output by the MMC P det and the reactive power Q det . Substitute U s.det and P det into the fitting function A and the fitting function B respectively, and obtain the maximum reactive power that the MMC can output without zero-sequence signal injection and the maximum reactive power that the MMC can output with zero-sequence signal injection ; Substitute the real-time detected U s.det , P det and Q det into the fitting function C and the fitting function D respectively, to obtain the amplitude A 3.Cal and the phase angle α 3.Cal ; S32: Compare the reactive power Q det , and the and calculated in step S6, when Q det ≤ , then A 3 = 0, α 3 = 0, no alarm; when <Q det < , then A 3 = A 3.Cal , α 3 = α 3.Cal , no alarm; When Q det ≥ , A 3 = A 3.Cal , α 3 = α 3.Cal Alarm to indicate that the current operating condition of the MMC has exceeded the limit condition; When there is no alarm, start the zero-sequence signal calculation unit, and under the condition of no alarm, A 3 and α 3 inject into the MMC modulation module to improve the reactive power support ability of the MMC without overmodulation under linear control.

5. The method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults according to claim 2, wherein In step S11, ; Among them, U si represents the AC-side voltage at the i th calculation point within the calculation range of the minimum allowable voltage and the maximum allowable voltage of the PPC point, n is the calculation interval.

6. The method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults according to claim 2, wherein In step S12, the modulation signal margin without zero-sequence signal injection S margin is regarded as Us i , P 0 , Q 0 The function of is: ； Modulation signal margin during zero-sequence signal injection S margin Regarded as Us i 、 P 1 、 Q 1 、A 3 、α 3 The function is as follows: ； Taking the upper arm of phase A as an example, the modulation signal margin S margin The calculation formula is as follows: ； ； Among them, S ap ( t ) is the modulation signal of the upper bridge arm of phase A, A 0 , A 1 , A 2 and A 3 are the amplitudes of the DC, fundamental, second, and third harmonic components respectively, A 3 which is also called the amplitude of the zero-sequence signal; α 1 , α 2 , α 3 represent the phase angles of the fundamental, second, and third harmonic components; ω is the fundamental angular frequency.

7. The method for improving the reactive power support ability of MMC considering the dynamic boundary under grid faults according to claim 3, characterized in that In step S22, the calculation points of the AC side voltage U s , active power P , and reactive power Q are determined. Among them, the calculation points of the active power P and reactive power Q are adjusted according to the MMC operating parameters, and their expressions are as follows: ； Among them, i represents the i th calculation point, j represents the maximum number of calculations of active power, k represents the maximum number of calculations of reactive power; P max , Q max are the G no.A3 and G A3 maximum values of active power and reactive power recorded in the data set, and the expression is as follows: ； Indicates all active power data in the dataset obtained without zero-sequence signal injection G no.A3 in Indicates all active power data in the dataset obtained without zero-sequence signal injection G no.A3 in Indicates all reactive power data in the dataset obtained with zero-sequence signal injection G A3 in Indicates all active power data in the dataset obtained with zero-sequence signal injection G A3 in

Images