Method for mid-high frequency resonance suppression based on impedance sensitivity analysis

Abstract

The application provides a kind of high-frequency resonance suppression method based on impedance sensitivity analysis, first based on harmonic linearization method establishes the fine model of equivalent impedance of flexible direct current grid-connected system.Second, the factors that may affect the equivalent resistance are summarized and the initial interval of each influencing factor is given, and the dominant factor ranking chart is calculated and analyzed;Then, for the resonance phenomenon, based on the dominant factor ranking, the dominant influencing factor affecting the equivalent resistance at this frequency can be located, and the resonance tracing is realized.Again, parameter optimization is carried out in the way of two-by-two joint adjustment of dominant influencing factors, and the parameter stable domain of dominant influencing factors at resonance frequency band can be obtained through three-dimensional joint adjustment diagram;Finally, the dominant influencing factors are optimized to the stable domain, and the resonance of flexible direct current grid-connected system is suppressed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system control, and particularly relates to a medium-high frequency resonance suppression method based on impedance sensitivity analysis. BACKGROUND

[0002] Flexible DC transmission technology has the advantages of flexible control, low harmonic content, power supply to passive systems, no commutation failure, etc., and is widely used in long-distance power transmission, offshore wind power DC transmission, new energy grid connection, asynchronous grid interconnection, and island drilling platform power supply. However, with the operation of multiple flexible DC projects, new stability problems have been brought.

[0003] The mechanism of harmonic resonance of the flexible DC system connected to the AC power grid is that the complex and fast control characteristics of the converter make the equivalent resistance at the harmonic frequency less than 0, showing negative damping characteristics, thereby interacting with the AC power grid to cause resonance. The modular topology and hierarchical control architecture of the MMC converter valve make it have multi-time scale dynamic characteristics and interact with the AC power grid, further exacerbating the resonance risk of the AC / DC system. At present, the medium-high frequency resonance suppression methods generally include active filtering method and passive damping method. The above methods are designed for damping parameters based on the analysis of the resonance reason of a single actual project. However, in actual flexible DC transmission projects, the system resonance is usually not caused by a single resonance source, and the resonance influencing factors of the AC / DC power grid are complex and interwoven. Although the current methods can effectively deal with the harmonic oscillation problem in some scenarios, they lack mechanism tracing and dominant factor discrimination, and often lack transferability. Especially when facing different topological structures, operation modes and control parameters of the flexible DC project, mechanism analysis and suppression measures need to be redeveloped, which increases the complexity of resonance suppression and makes it difficult to achieve rapid and effective suppression of system resonance in different scenarios and operation conditions.

[0004] In summary, in order to realize resonance tracing and make the suppression strategy universal, a resonance suppression method that can quantitatively analyze and discriminate the dominant factors and is applicable in different frequencies, different parameters and different operation conditions is needed. SUMMARY

[0005] The purpose of the present application is to overcome the shortcomings of the existing methods, and a medium-high frequency resonance suppression method based on MMC impedance sensitivity analysis is proposed, which can quantitatively and intuitively reflect the influence of various parameter changes on the system impedance, and the optimal stability domain of the MMC dominant influencing factors is proposed to suppress the medium-high frequency resonance in the flexible DC grid-connected system.

[0006] The present invention is implemented as follows: first, based on the harmonic linearization method, an analytical expression of the equivalent impedance fine model of the flexible direct current grid-connected system considering the control links such as the power outer loop, the current inner loop, and the phase-locked loop is established. Secondly, the factors that may affect the equivalent resistance and the initial interval are summarized, and the Sobol global sensitivity algorithm is used to calculate and analyze the influence index of the PI control link parameters and the delay link parameters on the MMC equivalent resistance and generate a ranking diagram of the dominant factors in the full frequency band; then, based on the ranking of the dominant factors in the full frequency band, the dominant influencing factors affecting the equivalent resistance at the resonant frequency are located to achieve resonance tracing, and then the parameters are optimized by the method of joint parameter adjustment between the dominant influencing factors in pairs. The parameter stability domain of the dominant influencing factors at the resonant frequency can be obtained through the joint parameter adjustment three-dimensional diagram; finally, the dominant influencing factors are optimized to the stable domain to achieve the suppression of the resonance of the flexible direct current grid-connected system.

[0007] The present invention specifically adopts the following technical solutions:

[0008] A method for suppressing mid- and high-frequency resonance based on impedance sensitivity analysis, characterized by:

[0009] Firstly, based on the harmonic linearization method, a detailed equivalent impedance model of the flexible DC grid-connected system is established, which takes into account the control links including the power outer loop, current inner loop, phase-locked loop, and delay loop.

[0010] Secondly, the factors that may affect the equivalent resistance are summarized and the initial range of each influencing factor is given. The Sobol global sensitivity algorithm is used to calculate and analyze the influence index of the PI control link parameters and the delay link parameters on the MMC equivalent resistance and generate a ranking diagram of the dominant factors at the resonant frequency.

[0011] Then, for the resonance phenomenon that occurs, the dominant factors that affect the equivalent resistance at that frequency are sorted based on the dominant factors, realizing resonance tracing. Parameter optimization is then performed by using a method of joint parameter adjustment between the dominant factors in pairs. The parameter stability domain of the dominant factors in the resonance frequency band is obtained through a three-dimensional joint parameter adjustment graph.

[0012] Finally, the dominant influencing factors are optimized to the stable domain to suppress the resonance of the flexible DC grid-connected system.

[0013] It specifically includes the following steps:

[0014] Step S1: Based on the impedance analytical expression of the MMC converter station of the flexible DC grid-connected system, the PI parameters of each control link and the delay link parameters are summarized and the fuzzy initial interval of each influencing factor is given;

[0015] Step S2: Calculate and analyze the first-order influence index, global influence index, and interaction influence index δ of the PI control link parameters and delay link parameters in the MMC impedance on the MMC equivalent resistance using a global sensitive algorithm s_iAccording to the size of the influence of each parameter on the equivalent resistance, a dominant factor ranking graph at the resonance frequency is drawn;

[0016] Step S3: When the HVDC system resonates, the resonance frequency is analyzed by FFT, and the dominant influencing factor of the equivalent resistance at the resonance frequency is obtained by using step S2 to realize resonance tracing; Considering the interaction between the dominant parameters, parameter optimization is further carried out in the way of two-by-two joint parameter adjustment;

[0017] Step S4: When parameter optimization is carried out in the way of two-by-two joint parameter adjustment, a three-dimensional joint parameter adjustment graph is drawn to obtain the parameter stability domain of the dominant influencing factor at the resonance frequency, and the initially given parameter fuzzy interval is optimized into the stability domain to realize the resonance suppression of the system.

[0018] Further, in step S1, the PI parameters of each control link and the delay link parameters are considered and the initial fuzzy interval is given:

[0019] Through the derived explicit expression of impedance, the real part R MMC The factors that may have an impact are preliminarily summarized as two types of control system PI control parameters and system delay parameters;

[0020] Specifically as shown in the following table:

[0021] Impedance influencing factors

[0022]

[0023] Based on the model and simulation model of R MMC , while considering the response speed and stability margin constraints of the control system, the fuzzy initial range of each influencing factor is given, which is used to give the stable domain subset after parameter adjustment in the subsequent steps.

[0024] Further, in step S2, the MMC impedance parameter sensitivity is analyzed by using the global sensitivity algorithm, which is specifically: by calculating the first-order influence index, global influence index and interaction influence index δ s_i of the PI control link parameters and the delay link parameters on the MMC equivalent resistance, according to the size of the influence of each parameter on the equivalent resistance, a dominant factor ranking graph at the resonance frequency is drawn.

[0025] Further, the calculation process of impedance sensitivity includes:

[0026] Interaction influence index δ s_i As shown in formula (1), it is used to measure the influence of the interaction of a certain parameter with other parameters on the output, the greater the value, the more intense the interaction of the parameter with other parameters, the greater the influence on the output; The priority in joint parameter adjustment is also higher;

[0027] Cross-impact index:

[0028]

[0029] Wherein, N represents the total number of PI control parameters and delay link parameters that may have an impact.

[0030] Furthermore, in step S3, considering the interaction between the dominant factors, the parameters are optimized by pairwise joint parameter adjustment:

[0031] If the system oscillates, the oscillation frequency of the system is extracted through FFT analysis, and the R in the oscillation frequency band is located according to the calculation results of the global sensitivity algorithm and the dominant factors. MMC The dominant influencing factors of the characteristics x1, x2, ... x n , where x1 is the most dominant factor, x2 is the second most dominant factor, and so on; according to the order of the dominant influence, first compare x1 with x2, ... x n If the value of x1 is not in the stable domain, the value is discarded, and the x1 in the stable domain is retained and a parameter stable domain [c, d] of x1 is generated, where c ≥ a, d ≤ b; each parameter adjustment generates a stable domain, and n-1 stable domains of x1 can be obtained. The intersection of each stable domain is used as the final value range of x1, and a value x1' is selected from them as the optimized value of x1 and substituted into R MMC , eliminating the influence of x1; the remaining main influencing parameters are x2,…x n As the parameter group to be optimized, the final value range of each influencing factor is obtained according to the above steps and the values ​​are taken in turn, that is, the R after the optimized parameters are satisfied MMC It is always greater than 0 in this frequency band, thus eliminating the negative damping range of MMC and improving the stability margin of the system.

[0032] Furthermore, during the optimization process, the parameter stability domain of any two dominant factors at the resonant frequency is obtained by drawing a three-dimensional joint parameter adjustment diagram, and the initial given parameter fuzzy interval is optimized to the stable domain to achieve the resonance suppression of the system:

[0033] The X and Y axes of the three-dimensional graph are defined as the fuzzy initial intervals of the two tuning parameters, and the Z axis is the equivalent resistance value obtained based on impedance calculation; the intersection line of the equivalent resistance three-dimensional graph and the zero plane is the stable boundary line. It can be seen that there are some areas in the dual-parameter plane with equivalent resistance values ​​less than 0, which means that when the system is disturbed, if the parameters take values ​​in this area, oscillation may occur; therefore, this area is divided into an unstable area; conversely, the area with an equivalent resistance value greater than 0 is divided into a stable area; the purpose of parameter optimization is to adjust and optimize the parameters originally in the unstable area to the stable area when resonance occurs, thereby suppressing resonance.

[0034] Furthermore, the impedance model of the flexible DC grid-connected MMC at least includes the MMC control links and delay links, namely:

[0035] Z MMC (s)=k1(s) / k2(s) (1)

[0036]

[0037]

[0038] Where: k1(s) is the numerator of the MMC AC measurement equivalent impedance expression; k2(s) is the denominator of the impedance expression; V1 is the AC voltage fundamental amplitude; is the phase angle difference between the fundamental frequency voltage and current on the AC side; i d0 and i q0 are the d-axis and q-axis steady-state currents respectively; L is the equivalent inductance of the commutation transformer and MMC, which is equal to the MMC bridge arm reactor L arm 1 / 2 of the connected transformer leakage inductance L t The sum of ω1 is the natural angular frequency, i.e. the fundamental angular velocity of the AC voltage; Gsv and Gsi are the sampling delay link transfer functions of the three-phase voltage and three-phase current respectively; G pll is the proportional-integral (PI) link of the phase-locked loop; G p is the PI link of the active power outer loop, G q For the PI link of the reactive power outer loop, further simplify the model so that G p =G q , that is, the default PI link parameters of the active power outer loop and the reactive power outer loop are equal; G i is the PI link of the inner current loop; K d is the current inner loop decoupling coefficient; K c is the decoupling coefficient of the circulation suppression link; G Td It is the system delay link;

[0039] MMC AC measurement of equivalent resistance R MMC and equivalent reactance X MMCis defined as follows:

[0040]

[0041] Furthermore, the dominant factors affecting the equivalent resistance of MMC should at least consider the PI parameters of each control link and the parameter T of the system delay link. d , and then use the Sobol algorithm to calculate the first-order impact index and the global impact index, and then calculate the cross-impact index based on this, that is:

[0042] First-order impact index:

[0043]

[0044] Global Impact Index:

[0045]

[0046] Monte Carlo sampling estimation is used to calculate the above two indices:

[0047]

[0048]

[0049] Among them, N is the number of sampling points, A, B, is the sample matrix, and its calculation process is as follows:

[0050] Using the Monte Carlo method or its variant, first generate an N*2d initial sample matrix, that is, N rows and 2d columns. Set the first d columns of the matrix to matrix A and the last d columns to matrix B. Use the i-th column in matrix B to replace the i-th column in matrix A to construct the matrix i=1,2,…,d; Since each row of the sample matrices A, B, and Ai B is a set of inputs, f(A) represents the output matrix corresponding to substituting A into the analytical expression. The representation of other sample matrices is the same; f(A) j Represents the j-th row of the matrix f(A), and the representation of other sample matrices is the same;

[0051] The cross-influence index is:

[0052]

[0053] Wherein, N represents the total number of PI control parameters and delay link parameters that may have an impact;

[0054] Interaction influence index δ s_iIndicates the impact of the interaction between a parameter and other parameters on the output. The larger the value, the more intense the interaction between the parameter and other parameters, and the greater the impact on the output; the higher the priority in joint parameter adjustment.

[0055] Furthermore, in order to improve the MMC impedance stability margin and realize parameter optimization to achieve the purpose of resonance suppression, the resonant frequency is first obtained by FFT. Based on the sensitivity index sorting and cross-influence factor, the factors that have a greater impact on the MMC equivalent resistance at the resonant frequency and the factors with a larger cross-factor are comprehensively considered; multiple parameters are jointly adjusted in pairs and a three-dimensional joint adjustment diagram is drawn. The optimized stable domain is calculated and generated in turn, and their intersection is calculated, and finally the multi-parameter coordinated optimization parameter stable domain is obtained.

[0056] Compared with the existing technology, the present invention and its preferred solution can quantitatively and intuitively reflect the degree of influence of various parameter changes on the system impedance, and specifically propose an optimized stability domain of the MMC dominant influencing factors to achieve the suppression of high-frequency resonance in the flexible direct current grid-connected system. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0058] Figure 1 This is a structural diagram of a flexible direct current grid-connected system according to an embodiment of the present invention;

[0059] Figure 2 This is a block diagram of the grid-connected MMC control according to an embodiment of the present invention;

[0060] Figure 3 This is a flow chart of a method for suppressing medium and high frequency resonance based on impedance sensitivity analysis according to an embodiment of the present invention;

[0061] Figure 4 This is the FFT analysis diagram of the resonant wave according to the embodiment of the present invention;

[0062] Figure 5 Schematic diagram of the first-order influence index of resonance at 465 Hz according to an embodiment of the present invention;

[0063] Figure 6 Schematic diagram of the global resonance impact index at 465 Hz according to an embodiment of the present invention;

[0064] Figure 7 This is a three-dimensional diagram of joint parameter adjustment of the leading parameters in an embodiment of the present invention;

[0065] Figure 8 This is an example of the parameter stability domain of the embodiment of the present invention. Figure 1 ;

[0066] Figure 9 This is an example of the parameter stability domain of the embodiment of the present invention. Figure 2 ;

[0067] Figure 10 This is a voltage comparison diagram before and after 465Hz resonance parameter optimization in an embodiment of the present invention;

[0068] Figure 11 This is a comparison diagram of current before and after 465Hz resonance parameter optimization in an embodiment of the present invention. DETAILED DESCRIPTION

[0069] Hereinafter, specific embodiments of the present application will be described in detail with reference to the accompanying drawings. Based on these detailed descriptions, those skilled in the art will be able to clearly understand the present application and implement the present application. Without violating the principles of the present application, the features of different embodiments may be combined to obtain new implementations, or certain features of certain embodiments may be substituted to obtain other preferred implementations.

[0070] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0071] To make the features and advantages of this patent more clearly understood, the following embodiments are specifically described in detail as follows:

[0072] This embodiment uses Figure 1 The flexible direct current grid-connected system structure shown in the figure and Figure 2 The MMC control block diagram shown in FIG. 1 is used as an example to illustrate the medium and high frequency resonance suppression method based on impedance sensitivity analysis proposed in the present invention ( Figure 3 ) to verify its validity.

[0073] The specific implementation is as follows:

[0074] 1. Establish the impedance model of flexible DC grid-connected MMC

[0075] The impedance model of the flexible DC grid-connected MMC must include the various control links and delay links of the MMC, namely:

[0076] Z MMC (s)=k1(s) / k2(s)(10)

[0077]

[0078]

[0079] Where: k1(s) is the numerator of the MMC AC measurement equivalent impedance expression; k2(s) is the denominator of the impedance expression; V1 is the AC voltage fundamental amplitude; is the phase angle difference between the fundamental frequency voltage and current on the AC side; i d0 and i q0 are the d-axis and q-axis steady-state currents respectively; L is the equivalent inductance of the commutation transformer and MMC, which is equal to the MMC bridge arm reactor L arm 1 / 2 of the connected transformer leakage inductance L t The sum of ω1 is the natural angular frequency, i.e. the fundamental angular velocity of the AC voltage. Gsv and Gsi are the sampling delay link transfer functions of the three-phase voltage and three-phase current respectively; G pll is the proportional-integral (PI) link of the phase-locked loop; G p is the PI link of the active power outer loop, G q For the PI link of the reactive power outer loop, further simplify the model so that G p =G q , that is, the default PI link parameters of the active power outer loop and the reactive power outer loop are equal; G i is the PI link of the inner current loop; K d is the current inner loop decoupling coefficient; K c is the decoupling coefficient of the circulation suppression link; G Td It is the system delay link.

[0080] 2. Use FFT to determine the resonant frequency and analyze the dominant factors affecting the MMC equivalent resistance at this frequency through impedance sensitivity.

[0081] By changing the RLC parameters of the AC system, the structure or parameters of the AC system in the actual project are simulated to make the system reach the impedance matching condition and thus resonate. The FFT resonance is shown in the attached figure. Figure 4 As shown, the resonance is found to be a mid-frequency resonance of 465 Hz.

[0082] To determine the dominant influencing factors of the MMC equivalent resistance at the resonant frequency, it is necessary to fully consider the PI parameters of each control link and the parameter T of the system delay link. d , and then use the Sobol algorithm to calculate the first-order influence index and the global influence index, and then calculate the interactive influence index based on this, that is:

[0083] First-order impact index:

[0084]

[0085] Global Impact Index:

[0086]

[0087] Monte Carlo sampling estimation is used to calculate the above two indices:

[0088]

[0089]

[0090] Among them, N is the number of sampling points, A, B, The calculation process is as follows: Using the Monte Carlo method or its variant method, first generate an initial sample matrix of N*2d (N rows and 2d columns), set the first d columns of the matrix to matrix A, and the last d columns to matrix B. Use the i-th column in matrix B to replace the i-th column in matrix A to construct the matrix (i=1,2,…,d). Since each row of the sample matrices A, B, and Ai B is a set of inputs, f(A) represents the output matrix corresponding to the substitution of A into the analytical expression, and the rest is similar; f(A) j represents the j-th row of the matrix f(A), and the rest are similar.

[0091] The interaction impact index is:

[0092]

[0093] Taking the sensitivity calculation of various influencing factors of impedance at the frequency of 465 Hz as an example, the factors that may have an impact are first summarized based on the analytical expression of impedance and the fuzzy initial interval is given.

[0094] The Sobol algorithm can be used to calculate the influence of each input factor on R MMC The first-order response index and global response index of Figure 5 、 Figure 6 As shown. The horizontal axis represents the number of sampling points. When the number of sampling points is too small, the sensitivity index calculation result may be unprepared. It can be found that when the number of sampling points is greater than 2500, the sensitivity calculation result and ranking are relatively stable. Therefore, the number of sampling points is set to 5000, and the sensitivity calculation value at 5000 sampling points is used as the final impact index value. Figure 5 、 Figure 6 The calculation results can be summarized as shown in Table 1:

[0095] Table 1 Ranking of response indices at 465Hz resonant frequency

[0096]

[0097]

[0098] Based on Table 1, we can see that the first-order influence index, global sensitivity index, and interaction influence index are ranked in the same order. Therefore, for the 465Hz resonance generated by the system, the dominant influencing factor is the communication control delay T d , power outer loop proportional coefficient kp_p , the inner loop proportional coefficient of current k i_p .

[0099] 3. The final parameter stable region is obtained by joint parameter adjustment optimization of the dominant influencing factors

[0100] Considering the above first-order influence index, global sensitivity index, and interaction index, the communication control delay T d , the outer loop proportional coefficient of power k p_p , the inner loop proportional coefficient of current k i_p Not only the change of itself, but also the interaction between the dominant factors MMC changes dramatically.

[0101] Therefore, the coordination and optimization between the dominant influencing factors are carried out, and the joint parameter adjustment three-dimensional graph of the three is shown in the accompanying Figure 7 The intersection of the medium equivalent resistance surface and the zero plane can be known that when the parameters change, there are some regions in the parameter plane corresponding to the equivalent resistance value less than 0, which represents that when the AC system structure and parameters change or are disturbed, if the parameter value is in this region, resonance may occur, so this region is divided into an unstable region. Conversely, the region corresponding to the equivalent resistance value greater than 0 is divided into a stable region. The purpose of parameter optimization is to suppress resonance by jointly adjusting and optimizing the parameters in the unstable region to the stable region when resonance occurs.

[0102] Therefore, according to the parameter adjustment three-dimensional graph of Figure 7 , the parameter stable region is drawn as shown in Figure 8 , Figure 9 According to the stable region division, the stable region A of the joint parameter adjustment of T d and k i_p requires T d ≤ 303 μs, the stable region C of the joint parameter adjustment of T d and k p_p requires T d ≤ 202 μs, so the value range of T d is the intersection of the range of T d in the stable region A and the range of T d in the stable region C, that is, 0 ≤ T d ≤ 202 μs, so T d ’ = 180 μs is selected. The stable boundary line of the stable region of the joint parameter adjustment of k p_p and k i_p only has an intersection with the k p_p axis, that is, k p_p = 0.16, as long as 0.16 ≤ k p_p ≤ 0.6, no matter k i_p in the initial interval 0.5 ≤ k i_p ≤ 0.6, the system is stable.i_p How does it change between ≤1.5? mmc It can always be guaranteed to be greater than 0. Therefore, the stable value k of the power outer loop proportional coefficient is selected in the stable region. p_p '=0.3, current inner loop proportional coefficient k i_p '=0.9,as attached Figure 7 As shown by the midpoint A. The parameter interval of the dominant factor after optimization avoids R MMC <0 to avoid negative damping of the system.

[0103] After the simulation verifies that the 2s system resonance occurs, the optimized parameters are set at 2.2s, and the electrical quantities of the flexible DC system gradually recover to a stable operating state. Figure 10 、 Figure 11 shown.

[0104] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

[0105] This patent is not limited to the above-mentioned optimal implementation method. Anyone can derive various other forms of medium and high frequency resonance suppression methods based on impedance sensitivity analysis under the inspiration of this patent. All equal changes and modifications made according to the scope of the patent application of this invention should be covered by this patent.

Claims

1. A method for suppressing mid- and high-frequency resonance based on impedance sensitivity analysis, characterized by: Firstly, based on the harmonic linearization method, a detailed equivalent impedance model of the flexible DC grid-connected system is established, which takes into account the control links including the power outer loop, current inner loop, phase-locked loop, and delay loop. Secondly, the factors that can affect the equivalent resistance are summarized and the initial range of each influencing factor is given. The Sobol global sensitivity algorithm is used to calculate and analyze the influence index of the PI control link parameters and the delay link parameters on the MMC equivalent resistance and generate a ranking diagram of the dominant factors at the resonant frequency. Then, for the resonance phenomenon that occurs, the dominant factors that affect the equivalent resistance at that frequency are sorted based on the dominant factors, realizing resonance tracing. Parameter optimization is then performed by using a method of joint parameter adjustment between the dominant factors in pairs. The parameter stability domain of the dominant factors in the resonance frequency band is obtained through a three-dimensional joint parameter adjustment graph. Finally, the dominant influencing factors are optimized to the stable domain to suppress the resonance of the flexible DC grid-connected system; The specific steps include: Step S1: Based on the impedance analytical expression of the MMC converter station of the flexible DC grid-connected system, the PI parameters of each control link and the delay link parameters are summarized and the fuzzy initial interval of each influencing factor is given; Step S2: Calculate and analyze the first-order influence index, global influence index, and interaction influence index δ of the PI control link parameters and delay link parameters in the MMC impedance on the MMC equivalent resistance using a global sensitive algorithm s_i , based on which we can measure the influence of each parameter on the equivalent resistance and draw a ranking diagram of the dominant factors at the resonant frequency; Step S3: When the flexible DC system resonates, the resonant frequency is analyzed through FFT, and the dominant factors affecting the equivalent resistance at the resonant frequency are obtained using step S2 to achieve resonance tracing. Considering the interaction between the dominant parameters, the parameters are further optimized by pairwise joint parameter adjustment. Step S4: When performing parameter optimization by means of pairwise joint parameter adjustment, a three-dimensional joint parameter adjustment diagram is drawn to obtain the parameter stability domain of the dominant influencing factors at the resonant frequency, and the initially given parameter fuzzy interval is optimized to the stable domain to achieve resonance suppression of the system; In step S1, the PI parameters of each control link and the delay link parameters are considered and the initial fuzzy interval is given: By deriving the explicit impedance expression, the equivalent resistance R of the MMC AC in the converter station is measured. MMC The influencing factors are initially summarized into two types: control system PI control parameters and system delay parameters; Based on R MMC The model and simulation model of the control system are considered at the same time, and the response speed and stability margin constraints of the control system are considered to provide the fuzzy initial range of each influencing factor, which is used to provide the stable domain subset after parameter adjustment in the subsequent steps; In step S2, the sensitivity of the MMC impedance parameters is analyzed using a global sensitivity algorithm. Specifically, the first-order influence index, global influence index, and interaction influence index δ of the PI control link parameters and the delay link parameters on the MMC equivalent resistance are calculated and analyzed. s_i , based on which we can measure the influence of each parameter on the equivalent resistance and draw a ranking diagram of the dominant factors at the resonant frequency; The calculation process of impedance sensitivity includes: Interaction influence index δ s_i As shown below, it is used to measure the impact of the interaction between a parameter and other parameters on the output. The larger the value, the more intense the interaction between the parameter and other parameters, and the greater the impact on the output; the higher the priority in joint parameter adjustment; Cross-impact index: Where M represents the total number of PI control parameters and delay link parameters that can have an impact; S i is the first-order impact index; S Ti is the global impact index.

2. The method for suppressing medium and high frequency resonance based on impedance sensitivity analysis according to claim 1, characterized in that : During the optimization process, the parameter stability domain of any two dominant factors at the resonant frequency is obtained by drawing a three-dimensional joint parameter adjustment diagram, and the initial given parameter fuzzy interval is optimized to the stable domain to achieve the resonance suppression of the system: The X and Y axes of the three-dimensional graph are defined as the fuzzy initial intervals of the two tuning parameters, and the Z axis is the equivalent resistance value obtained based on impedance calculation; the intersection line of the equivalent resistance three-dimensional graph and the zero plane is the stable boundary line. It can be seen that there are some areas in the dual-parameter plane with equivalent resistance values ​​less than 0, which means that when the system is disturbed, if the parameters take values ​​in this area, oscillation may occur; therefore, this area is divided into an unstable area; conversely, the area with an equivalent resistance value greater than 0 is divided into a stable area; the purpose of parameter optimization is to adjust and optimize the parameters originally in the unstable area to the stable area when resonance occurs, thereby suppressing resonance.

3. The method for suppressing medium and high frequency resonance based on impedance sensitivity analysis according to claim 1, characterized in that : The impedance model of the flexible DC grid-connected MMC includes at least the MMC control links and delay links, namely: Where: k1(s) is the numerator of the MMC AC measurement equivalent impedance expression; k2(s) is the denominator of the impedance expression; V1 is the AC voltage fundamental amplitude; φ i is the phase angle difference between the fundamental frequency voltage and current on the AC side; i d0 and i q0 are the d-axis and q-axis steady-state currents respectively; L is the equivalent inductance of the commutation transformer and MMC, which is equal to the MMC bridge arm reactor L arm 1 / 2 of the connected transformer leakage inductance L t The sum of; ω1 is the natural angular frequency, that is, the fundamental angular velocity of the AC voltage; G sv and G si are the sampling delay link transfer functions of the three-phase voltage and three-phase current respectively; G pll is the PI link of the phase-locked loop; G p is the PI link of the active power outer loop, G q As the PI link of the reactive power outer loop, the model is further simplified to make G p =G q , that is, the default PI link parameters of the active power outer loop and the reactive power outer loop are equal; G i is the PI link of the current inner loop; K d is the current inner loop decoupling coefficient; G Td It is the system delay link; MMC AC measurement of equivalent resistance R MMC and equivalent reactance X MMC is defined as follows: 。 4. The method for suppressing medium and high frequency resonance based on impedance sensitivity analysis according to claim 3 is characterized in that : The dominant factors affecting the equivalent resistance of MMC should at least consider the PI parameters of each control link and the parameter T of the system delay link. d , and then use the Sobol algorithm to calculate the first-order impact index and the global impact index, and then calculate the cross-impact index based on this, that is: First-order impact index: Global Impact Index: Monte Carlo sampling estimation is used to calculate the above two indices: Among them, N is the number of sampling points, A, B, is the sample matrix, and its calculation process is as follows: Using the Monte Carlo method, we first generate an N×2d initial sample matrix, i.e., an N-row, 2d-column matrix. The first d columns of the matrix are set as matrix A, and the last d columns are set as matrix B. We then replace the i-th column of matrix A with the i-th column of matrix B to construct the matrix ,i = 1,2,…, d;Since the sample matrices A, B, Each row of is a set of inputs, f(A) represents the output matrix corresponding to substituting A into the analytical expression, and the representation of other sample matrices is the same; f(A) j Represents the j-th row of the matrix f(A), and the representation of other sample matrices is the same; The cross-influence index is: 。 5. The method for suppressing medium and high frequency resonance based on impedance sensitivity analysis according to claim 4 is characterized in that In order to improve the MMC impedance stability margin and realize parameter optimization to achieve the purpose of resonance suppression, the resonant frequency is first obtained by using FFT. Based on the sensitivity index sorting and cross-influence factor, the factors that have a greater impact on the MMC equivalent resistance at the resonant frequency and the factors with a larger cross-factor are comprehensively considered; multiple parameters are jointly adjusted in pairs and a three-dimensional joint adjustment diagram is drawn. The optimized stable domain is calculated in turn and their intersection is calculated, and finally the multi-parameter coordinated optimization parameter stable domain is obtained.

Images