Multi-infeed direct current system frequency emergency control method and device and electronic equipment

Abstract

The application provides a multi-infeed DC system frequency emergency control method, device and electronic equipment, which is applied to a multi-infeed DC system. The method comprises the following steps: acquiring a first state quantity of the multi-infeed DC system under a preset fault, and constructing a base function library based on the first state quantity; sampling the base function library to obtain a plurality of sub-function libraries; performing integrated learning based on the plurality of sub-function libraries to obtain a target Koopman eigenvalue set and a target Koopman characteristic function set; setting a DC line reference active power regulation amount of the multi-infeed DC system under the preset fault; determining a control matrix based on the DC line reference active power regulation amount; and determining a target active power regulation amount of the multi-infeed DC system based on the control matrix, the target Koopman eigenvalue set and the target Koopman characteristic function set, so that the system frequency of the multi-infeed DC system is stable. The application realizes multi-infeed DC system frequency emergency support.

Description

Technical Field

[0001] This invention relates to the field of high voltage direct current technology, and in particular to an emergency frequency control method, device, and electronic equipment for multi-infeed DC systems. Background Technology

[0002] With the continuous development of high-voltage direct current (HVDC) transmission technology, traditional AC power grids are gradually transforming into large-scale, complex AC / DC hybrid power grids. Currently, the implementation of numerous HVDC transmission projects based on grid commutation converters has resulted in asynchronous interconnection between regional power grids, forming multiple multi-infeed DC systems. In these multi-infeed DC systems, an AC system is connected to multiple HVDC transmission lines, and the dynamic complexity and diverse faults pose a significant threat to the stable operation of the system.

[0003] According to relevant technologies, in emergency situations with significant power imbalance, multi-feed DC systems require emergency frequency control strategies. Currently, emergency frequency control is mainly achieved through emergency generator tripping and load shedding operations, but these methods result in substantial economic losses. Summary of the Invention

[0004] This invention provides a method, apparatus, and electronic device for emergency frequency control of a multi-infeed DC system, which addresses the shortcomings of existing technologies that rely on emergency generator and load shedding operations to achieve emergency frequency control. It enables the effective determination of the target active power regulation of the multi-infeed DC system to stabilize the system frequency.

[0005] This invention provides an emergency frequency control method for a multi-infeed DC system, applicable to a multi-infeed DC system including an AC system and multiple high-voltage DC transmission lines. The method includes: acquiring a first state variable of the multi-infeed DC system under a preset fault, and constructing a basis function library based on the first state variable, wherein the first state variable is used to characterize a first local quantity measurement in a time-delay coordinate; sampling the basis function library to obtain multiple sub-function libraries, wherein each sub-function library includes at least the first state variable; performing ensemble learning based on the multiple sub-function libraries to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set; setting a reference active power adjustment amount for the DC lines of the multi-infeed DC system under the preset fault, wherein the reference active power adjustment amount for the DC lines is determined according to the transmission power of the high-voltage DC transmission lines; determining a control matrix based on the reference active power adjustment amount for the DC lines; and determining a target active power adjustment amount for the multi-infeed DC system based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, so as to stabilize the system frequency of the multi-infeed DC system.

[0006] According to the present invention, a frequency emergency control method for a multi-infeed DC system is provided, wherein the first state quantity is determined in the following manner: based on the first local quantity measurement, the first time series data of the first local quantity measurement at a preset number of time points are determined; based on the first time series data, a first time series matrix is ​​constructed in the time delay coordinate; based on the first time series matrix in the time delay coordinate, the first state quantity is determined.

[0007] According to the present invention, a method for emergency frequency control of a multi-infeed DC system, wherein constructing a basis function library based on the first state variable specifically includes: determining the trigonometric function and polynomial function of the first state variable based on the first state variable; and constructing a basis function library based on the first state variable, the trigonometric function, and the polynomial function.

[0008] According to the present invention, a frequency emergency control method for a multi-infeed DC system, wherein the step of ensemble learning based on multiple sub-function libraries to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set specifically includes: obtaining a Koopman eigenvalue set and a Koopman eigenfunction set corresponding to each sub-function library through a feature pair identification method based on the sub-function libraries; obtaining a candidate set of Koopman eigenvalues ​​based on each Koopman eigenvalue set; obtaining a candidate set of Koopman eigenfunctions based on each Koopman eigenfunction set; and filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions respectively to obtain the target Koopman eigenvalue set and the target Koopman eigenfunction set.

[0009] According to the present invention, a frequency emergency control method for a multi-infeed DC system, wherein the step of filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions specifically includes: determining feature pairs under a first state variable based on the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions; determining the prediction error based on the feature pairs under the first state variable; filtering out target feature pairs whose prediction error is less than an error threshold, and filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions based on the target feature pairs to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions.

[0010] According to the present invention, a frequency emergency control method for a multi-infeed DC system, wherein determining a control matrix based on the DC line reference active power adjustment specifically includes: determining a second state quantity corresponding to the DC line reference active power adjustment based on the DC line reference active power adjustment, wherein the second state quantity is used to characterize a second local quantity measurement corresponding to the DC line reference active power adjustment in a time-delay coordinate; determining timing data of the second state quantity at a preset number of time points based on the second state quantity; and determining the control matrix based on the timing data.

[0011] According to the present invention, a frequency emergency control method for a multi-infeed DC system, wherein determining a second state quantity corresponding to the DC line reference active power adjustment based on the DC line reference active power adjustment specifically includes: determining a second local quantity measurement based on the DC line reference active power adjustment; determining second time-series data of the second local quantity measurement at a preset number of time points based on the second local quantity measurement; constructing a second time-series matrix in the time-delay coordinate based on the second time-delay coordinate based on the second time-series data; and determining the second state quantity based on the second time-delay coordinate.

[0012] According to the present invention, a frequency emergency control method for a multi-infeed DC system, wherein determining the control matrix based on the timing data specifically includes: determining the control matrix based on the timing data, the target Koopman eigenvalue set, and the target Koopman eigenfunction set.

[0013] This invention also provides a frequency emergency control device for a multi-infeed DC system, applied to a multi-infeed DC system including an AC system and multiple high-voltage DC transmission lines. The device comprises: a first module for acquiring a first state variable of the multi-infeed DC system under a preset fault, and constructing a basis function library based on the first state variable, wherein the first state variable is used to characterize a first local quantity measurement in a time-delay coordinate system; a second module for sampling the basis function library to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state variable; and a third module for performing ensemble learning based on the multiple sub-function libraries to obtain a target Koop. The system consists of a set of Koopman eigenvalues ​​and a set of target Koopman eigenfunctions; a fourth module, used to set the reference active power adjustment of the DC lines in the multi-infeed DC system under the preset fault conditions, wherein the reference active power adjustment of the DC lines is determined based on the transmission power of the high-voltage DC transmission lines; a fifth module, used to determine the control matrix based on the reference active power adjustment of the DC lines; and a sixth module, used to determine the target active power adjustment of the multi-infeed DC system based on the control matrix, the set of target Koopman eigenvalues, and the set of target Koopman eigenfunctions, so as to stabilize the system frequency of the multi-infeed DC system.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the emergency frequency control method for a multi-feed DC system as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the emergency frequency control method for a multi-feed DC system as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the emergency frequency control method for a multi-feed DC system as described above.

[0017] The present invention provides a method, apparatus, and electronic device for emergency frequency control of multi-infeed DC systems. It constructs a basis function library based on the first state variables of the multi-infeed DC system under a preset fault condition, and then uses this basis function library to obtain multiple sub-function libraries. Based on the integrated learning of these sub-function libraries, a target Koopman eigenvalue set and a target Koopman eigenfunction set are obtained. This ensures the accuracy of the approximate representation of the Koopman eigenfunction by the basis function library, reducing computational costs during the identification process. Furthermore, based on the reference active power adjustment of the DC line, a control matrix is ​​determined. Then, based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, the target active power adjustment of the multi-infeed DC system is determined to stabilize the system frequency, thus achieving emergency frequency support for the multi-infeed DC system. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the emergency frequency control method for multi-infeed DC systems provided by the present invention.

[0020] Figure 2 This is a flowchart illustrating the process of constructing a basis function library based on a first state variable, as provided by the present invention.

[0021] Figure 3 This is a flowchart illustrating the process of obtaining the target Koopman feature set and the target Koopman feature function set through integrated learning based on multiple sub-function libraries, as provided by this invention.

[0022] Figure 4 This is a flowchart illustrating the process of filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions, as provided by the present invention.

[0023] Figure 5 This is a flowchart illustrating the process of determining the control matrix based on the DC line reference active power adjustment amount provided by the present invention.

[0024] Figure 6 This is a schematic diagram of the structure of the emergency frequency control device for multi-feed DC systems provided by the present invention;

[0025] Figure 7This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] The frequency emergency control method for multi-infeed DC systems provided by this invention can be applied to multi-infeed DC systems (also known as MIDC). A multi-infeed DC system may include an AC system and multiple high-voltage DC transmission lines (also known as LCC-HVDC).

[0028] Figure 1 This is a flowchart illustrating the emergency frequency control method for multi-infeed DC systems provided by this invention. The following will combine... Figure 1 The process of emergency frequency control for multi-infeed DC systems is explained.

[0029] In an exemplary embodiment of the present invention, combined with Figure 1 As can be seen, the frequency emergency control method for multi-infeed DC systems may include steps 110 to 160, and each step will be described below.

[0030] In step 110, the first state variables of the multi-infeed DC system under a preset fault are obtained, and a basis function library is constructed based on the first state variables.

[0031] In one embodiment, for a multi-infeed DC system, a first state variable is obtained by setting the multi-infeed DC system under a preset fault. The preset fault can be adjusted according to actual conditions; for example, it could be the fault-free tripping of one generator in the multi-infeed DC system. In this embodiment, the preset fault is not specifically limited.

[0032] In one example, the first state variable can be used to characterize the first local quantity measurement in time-delay coordinates. The first local quantity measurement can be expressed as: Where, ω i Indicates the frequency deviation relative to the rated frequency (50Hz or 60Hz); δ represents the transmission power of the i-th high-voltage direct current transmission line (>0 when the AC main system is the receiving end, <0 when it is the sending end); q This represents the phase angle at node q relative to the synchronously rotating coordinate system; This indicates the number of generator nodes in a multi-feed DC system. Indicates the DC landing point (sending end or receiving end) of the high-voltage direct current transmission line, the quantity is:

[0033] In yet another exemplary embodiment of the present invention, the first state quantity can be determined in the following manner:

[0034] Based on the first local quantity measurement, determine the first time series data of the first local quantity measurement at a preset number of time points;

[0035] Based on the first time series data, a first time series matrix is ​​constructed in time-delay coordinates;

[0036] The first state variable is determined based on the first time series matrix in the time-delay coordinate system.

[0037] In one embodiment, the first local quantity measurement y of the i-th high-voltage direct current transmission line can be obtained. i First time series data at m time points In one example, the control quantity u is a vector composed of the DC line reference active power regulation of each LCC-HVDC. When obtaining Y... i At that time, u can be set to 0.

[0038] Furthermore, it can be based on Construct the time-delay coordinates (corresponding to the first time series matrix under the time-delay coordinates) as shown in formula (1):

[0039]

[0040] Where 'a' represents the dimension of the time-delay coordinate, and 'e' represents the number of delayed timestamps. Thus, the quantity measured in the time-delay coordinate (corresponding to the first time-series matrix in the time-delay coordinate) X is obtained. i .

[0041] In one example, the first time series matrix X can be defined in time-delay coordinates. i As the first state variable.

[0042] It should be noted that the relationship between the preset number of time points m, the dimension a of the time lag coordinate, and the number of lag timestamps e can be expressed as m≥a+e.

[0043] Furthermore, a base function library can be built based on the first state variables.

[0044] It should be noted that, for a given dynamic system, the selection of basis functions based on the known physical laws of the system requires judgment from personnel with relevant technical backgrounds; and, in order for the Koopman eigenfunction to be well approximated by the basis function library, the basis function library should include as many functions as possible, i.e., L should be as large as possible, where L represents the number of basis functions in the basis function library. Understandably, when L is too large, due to the eigenfunction pair... In the identification step, each update of the Koopman feature value uses the least squares algorithm (computational complexity O(n log n)). A large L will increase the computational cost of feature pair identification. Additionally, the set of state variables that act as the control object in the state vector (corresponding to the first state variable) x is denoted as... Since Ξ is a sparse matrix, it is identified according to KRONIC. It may not include Each component in the equation presents challenges for the design of subsequent linear quadratic regulators.

[0045] However, in this embodiment, a basis function library is constructed based on the first state variable. This library can be combined with nonlinear terms in the power system dynamic equations to construct basis functions. Then, a basis function ensemble learning method is introduced. This ensures that the Koopman feature function is accurately approximated by the basis function library while reducing the computational cost during the identification process and increasing the accuracy of the KRONIC-identified features. Includes The possibilities of all objects in the array.

[0046] In step 120, the basis function library is sampled to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state variable.

[0047] In one embodiment, the basis function library Θ can be sampled to form different subsets as a sub-function library Θ. d (d = 1, 2, ..., D). Each sub-function library contains a first state variable x. It should be noted that the first state variable x is also the first time series matrix X in time-delay coordinates. i .

[0048] In step 130, ensemble learning is performed based on multiple sub-function libraries to obtain the target Koopman feature set and the target Koopman feature function set.

[0049] In one embodiment, the target Koopman feature set and the target Koopman feature function set can be obtained through basis function ensemble learning.

[0050] In step 140, under a preset fault, the reference active power adjustment of the DC line in the multi-feed DC system is set, wherein the reference active power adjustment of the DC line is determined based on the transmission power of the high-voltage DC transmission line.

[0051] In step 150, the control matrix is ​​determined based on the DC line reference active power adjustment.

[0052] In one embodiment, for the same multi-infeed DC system, by setting the same preset fault as in step 110, the reference active power regulation amount u of the DC line of the multi-infeed DC system (which also corresponds to the control amount described above) is set. In one example, the DC line reference active power regulation amount u can be set to the transmission power p of the high-voltage DC transmission line. i dc The random control input is between the rated values ​​[-20%, +10%]. Further, the control matrix is ​​determined based on the DC line reference active power regulation.

[0053] In this embodiment, determining the control matrix by using the DC line reference active power adjustment can overcome the current problem that it cannot be applied to an unknown control matrix B.

[0054] In step 160, based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, the target active power regulation of the multi-infeed DC system is determined to stabilize the system frequency of the multi-infeed DC system.

[0055] In one embodiment, the target active power regulation of the multi-infeed DC system can be determined based on the control matrix B, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, so as to stabilize the system frequency of the multi-infeed DC system. This enables emergency frequency support for the multi-infeed DC system and avoids economic losses caused by direct generator tripping.

[0056] In one embodiment, the optimal control problem solved by the linear quadratic regulator of the emergency control of the DC additional frequency of the i-th high-voltage DC transmission line (corresponding to determining the target active power regulation of the multi-feed DC system) can be expressed by formula (2):

[0057]

[0058]

[0059]

[0060] Where Q is a positive definite matrix, representing the weight matrix of the state variable deviation; R is a positive number, representing the weight of the control variable; X iThis is the first time series matrix in time-delay coordinates; Let the variable be X i The Koopman characteristic equation; u i B represents the reference active power adjustment for the i-th DC line; B represents the control matrix. Let the variable be X i Reference values ​​for the Koopman characteristic equation.

[0061] Furthermore, the optimal control of this problem (The corresponding target active power adjustment) can be transformed into solving the state-dependent ricatti (SDRE) equations shown in formulas (3) to (4):

[0062]

[0063] in, Let P be the reference power adjustment of LCC-HVDC i under system fault conditions, where P satisfies formula (3):

[0064]

[0065] Where P represents a positive definite matrix.

[0066] In summary, the aforementioned quadratic regulator calculates the control quantity (corresponding to the target active power adjustment quantity) within each control cycle based on the measured state quantity at the start of the control cycle, as shown in equations (3)-(4). This control quantity remains unchanged within the control cycle. Through this embodiment, system frequency stability of the multi-infeed DC system can be achieved, realizing emergency frequency support for the multi-infeed DC system.

[0067] The present invention provides an emergency frequency control method for multi-infeed DC systems. It constructs a basis function library based on the first state variables of the multi-infeed DC system under a preset fault condition, and then uses this library to obtain multiple sub-function libraries. Based on these sub-function libraries, it performs ensemble learning to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set. This ensures the accuracy of the approximate representation of the Koopman eigenfunction by the basis function library, reducing computational costs during the identification process. Furthermore, based on the reference active power adjustment of the DC line, a control matrix is ​​determined. Then, based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, the target active power adjustment of the multi-infeed DC system is determined to stabilize the system frequency, thus achieving emergency frequency support for the multi-infeed DC system.

[0068] Figure 2 This is a schematic diagram of the process of constructing a basis function library based on the first state variable provided by the present invention.

[0069] The following will combine Figure 2 The process of constructing a basis function library based on the first state variable is explained.

[0070] In an exemplary embodiment of the present invention, combined with Figure 2 As can be seen, constructing the basis function library based on the first state variable can include steps 210 and 220, which will be described in detail below.

[0071] In step 210, based on the first state variable, the trigonometric function and polynomial function of the first state variable are determined respectively.

[0072] In step 220, a basis function library is constructed based on the first state variables, trigonometric functions, and polynomial functions.

[0073] In one embodiment, information about the first state variable X can be obtained. i Trigonometric functions and polynomial functions. In application, combined with the known nonlinear terms in the dynamic equations of the power system, sinX can be... i cosX i and A basis function library is introduced. Since the current method cannot fully characterize the nonlinearity of the system, and considering that any function can be Taylor expanded, in this embodiment, the polynomial function X of the state vector is used. i The power of k (k≥2) is introduced into the basis function library, where k represents the order of the largest polynomial in the basis function library.

[0074] In yet another example, the aforementioned basis function library Θ can be sampled to form different subsets as a sub-function library Θ. d (d = 1, 2, ..., D). In this embodiment, each sub-function library contains a state variable x. Furthermore, four nonlinear terms (i.e., sinX) i cosX i , and X i The power of k (k≥2) appears in each sub-function library with a 50% probability, thus allowing the construction of a total D=2. 4 =16 different sub-function libraries.

[0075] Figure 3 This is a flowchart illustrating the process of obtaining the target Koopman feature set and the target Koopman feature function set through ensemble learning based on multiple sub-function libraries, as provided by this invention.

[0076] The following will combine Figure 3 The process of obtaining the target Koopman feature set and the target Koopman feature function set by ensemble learning based on multiple sub-function libraries is explained.

[0077] In an exemplary embodiment of the present invention, combined with Figure 3 As can be seen, the ensemble learning based on multiple sub-function libraries to obtain the target Koopman feature set and the target Koopman feature function set may include steps 310 to 340, which will be described in detail below.

[0078] In step 310, based on the sub-function library, the set of Koopman feature values ​​and the set of Koopman feature functions corresponding to each sub-function library are obtained through the feature pair identification method.

[0079] In one embodiment, continuing with the aforementioned embodiment as an example, for the 16 sub-function libraries obtained above, the Koopman feature pair identification method in KRONIC is used to form the Koopman feature value set Λ and the Koopman feature function set for each sub-function library. It should be noted that the feature pair identification method is not limited to the Koopman feature pair identification method mentioned above, and the feature pair identification method can also be adjusted according to the actual situation.

[0080] In step 320, a candidate set of Koopman features is obtained based on each set of Koopman features.

[0081] In step 330, a candidate set of Koopman feature functions is obtained based on each set of Koopman feature functions.

[0082] In one embodiment, the union of the Koopman eigenvalue sets under each sub-function library can be taken to obtain the candidate set of Koopman eigenvalues.

[0083] In another embodiment, the candidate set of Koopman feature functions can be obtained by taking the union of the sets of Koopman feature functions under each sub-function library.

[0084] In step 340, the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions are filtered to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions.

[0085] Figure 4 This is a flowchart illustrating the process of filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions, as provided by the present invention.

[0086] The following will combine Figure 4The process of filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions is explained.

[0087] In an exemplary embodiment of the present invention, combined with Figure 4 As can be seen, filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions can include steps 410 to 430, which will be described in detail below.

[0088] In step 410, feature pairs under the first state variable are determined based on the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions.

[0089] In step 420, the prediction error is determined based on the feature pairs under the first state variables.

[0090] In step 430, target feature pairs with prediction errors less than the error threshold are selected, and based on the target feature pairs, the candidate set of Koopman feature values ​​and the candidate set of Koopman feature functions are selected respectively to obtain the target set of Koopman feature values ​​and the target set of Koopman feature functions.

[0091] In one embodiment, the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions can be filtered according to the prediction error of each eigenvalue and the corresponding eigenfunction to finally obtain the set of Koopman eigenvalues ​​(corresponding to the target set of Koopman eigenvalues) and the set of Koopman eigenfunctions (corresponding to the target set of Koopman eigenfunctions).

[0092] In one example, in the first state variable X i When performing the evaluation, the s-th feature pair The prediction error is defined by the following formula (5), where m represents m time points; λ s Represents the Koopman eigenvalues. This represents the Koopman feature function.

[0093]

[0094] Where a represents the dimension of the time delay coordinate; t represents the time point (t=1,2,...,a); This represents the first state variable at time t.

[0095] Furthermore, based on the prediction error E sThe identified feature pairs are sorted, and all feature pairs with errors below a certain error threshold are selected, ultimately yielding the Koopman feature set (corresponding to the target Koopman feature set) and the Koopman feature function set (corresponding to the target Koopman feature function set).

[0096] In this embodiment, the sub-function library Θ is recorded. d The number of neutron functions is |Θ d At this point, the computational complexity of each sub-function library obtaining the Koopman feature set and the Koopman feature function set is... The computational complexity of each algorithm is no greater than that of the original algorithm. Furthermore, the calculations of each sub-function library can be performed in parallel, which can reduce computational costs and thus improve control efficiency through this embodiment.

[0097] Figure 5 This is a flowchart illustrating the process of determining the control matrix based on the reference active power adjustment of a DC line, as provided by the present invention.

[0098] The following will combine Figure 5 The process of determining the control matrix based on the reference active power regulation of the DC line is explained.

[0099] In an exemplary embodiment of the present invention, combined with Figure 5 As can be seen, determining the control matrix based on the DC line reference active power adjustment can include steps 510 to 530, which will be described in detail below.

[0100] In step 510, a second state quantity corresponding to the DC line reference active power adjustment is determined based on the DC line reference active power adjustment.

[0101] In step 520, based on the second state quantity, the timing data of the second state quantity at a preset number of time points are determined.

[0102] In one embodiment, the second state quantity can be used to characterize a second local quantity measurement corresponding to the DC line reference active power regulation in a time-delay coordinate system. In one example, the second local quantity measurement can be obtained based on the DC line reference active power regulation u. Where, ω i,u This indicates the frequency deviation relative to the rated frequency (50Hz or 60Hz) measured under u. This represents the transmission power of the i-th high-voltage direct current transmission line obtained under u.

[0103] In yet another exemplary embodiment of the present invention, the second state quantity can be determined in the following manner:

[0104] The second local quantity measurement is determined based on the DC line reference active power regulation.

[0105] Based on the second local quantity measurement, determine the second time series data of the second local quantity measurement at a preset number of time points;

[0106] Based on the second time series data, a second time series matrix is ​​constructed in time-delay coordinates;

[0107] The second state variable is determined based on the second time series matrix in the time-delay coordinate system.

[0108] In one embodiment, the second local quantity measurement can be determined based on the DC line reference active power regulation, thus obtaining y. i,u Furthermore, the second local quantity measurement y of the i-th high-voltage direct current transmission line can be obtained. i,u The second time-series data at time points *a*. Furthermore, based on the second time-series data, and in the same manner as the first time-series matrix constructed in time-delay coordinates, a second time-series matrix *X* is constructed in time-delay coordinates. iu The second time-series matrix X in the time-delay coordinate system will be... iu As the second state quantity.

[0109] Furthermore, based on the second state variable, the time-series data of the second state variable at a preset number of time points are determined. In one example, this can be based on the second state variable X. iu Obtain time series data X at a time points. i,u =[x1 iu x2 iu , ...x a iu ].

[0110] In step 530, the control matrix is ​​determined based on the time series data.

[0111] In an exemplary embodiment of the present invention, the control matrix can be determined based on time-series data in the following manner:

[0112] The control matrix is ​​determined based on time series data, the target Koopman eigenvalue set, and the target Koopman eigenfunction set.

[0113] In one embodiment, continuing with the previously described embodiments, the data can also be based on timing data X. i,u Calculate time series data X i,u The derivative at a time points

[0114] For the state variables at time t (t = 1, 2, ..., a) (Corresponding to the second state variable), defined by formula (6):

[0115]

[0116] Where, θ h (.) represents a preset function, where h∈[1,p].

[0117] Furthermore, the control matrix B can be calculated according to formula (7):

[0118]

[0119] in, b n This represents the nth row of B. Let Ξ denote the Kronecker product. Ξ is an L×S complex matrix, where L represents the number of basis functions in the basis function library, and S represents the number of eigenfunctions; in, Represents the set of Koopman feature functions of the target; u1……u a Λ represents the DC line reference active power adjustment at different time points (from 1 to a); Λ represents the target Koopman characteristic function set.

[0120] This invention solves the problem that only some state variables can be observed under emergency frequency control by constructing local quantity measurements of high-voltage direct current transmission lines in time-delay coordinates, thereby realizing the adjustment of the reference active power of high-voltage direct current transmission lines.

[0121] The frequency emergency control method for multi-infeed DC systems provided by this invention is a Koopman feature pair construction method for frequency stability problems in multi-infeed DC systems, enabling the MIDC system to be linearized in the coordinate system of the feature function. The key features of this feature pair construction method are: by sampling a basis function library to form multiple sub-basis function libraries, it achieves integrated learning of the Koopman eigenvalue set and eigenvector set, ensuring the accuracy of the approximate representation of the Koopman feature function by the basis function library while reducing the computational cost in the identification process, and increasing the accuracy of the KRONIC identification. This includes the possibility of all controlled objects. Furthermore, based on the Koopman feature pairs for frequency stability problems in multi-infeed DC systems, an online control method for DC power reference values ​​in DC emergency supplementary control is proposed. This method is a data-driven approach: it learns Koopman feature pairs through offline simulation and innovatively utilizes these pairs to mine unknown control matrices from the data. Moreover, this control method is adaptable to local measurement problems in emergency control.

[0122] As described above, the frequency emergency control method for multi-infeed DC systems provided by this invention constructs a basis function library based on the first state variables of the multi-infeed DC system under a preset fault. Multiple sub-function libraries are obtained by employing these basis function libraries, and then integrated learning is performed based on these sub-function libraries to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set. This ensures that the Koopman eigenfunction is approximately represented by the basis function library with high accuracy, reducing computational costs during the identification process. Furthermore, a control matrix is ​​determined based on the reference active power adjustment of the DC line. Based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, the target active power adjustment of the multi-infeed DC system is determined to stabilize the system frequency, thus achieving emergency frequency support for the multi-infeed DC system.

[0123] Based on the same concept, the present invention also provides an emergency frequency control device for a multi-feed DC system.

[0124] The following describes the emergency frequency control device for a multi-infeed DC system provided by the present invention. The emergency frequency control device for a multi-infeed DC system described below can be referred to in correspondence with the emergency frequency control method for a multi-infeed DC system described above.

[0125] Figure 6 This is a schematic diagram of the structure of the emergency frequency control device for a multi-feed DC system provided by the present invention.

[0126] In an exemplary embodiment of the present invention, the frequency emergency control device for a multi-infeed DC system can be applied to a multi-infeed DC system. The multi-infeed DC system may include an AC system and multiple high-voltage DC transmission lines. (In conjunction with...) Figure 6 As can be seen, the frequency emergency control device for a multi-infeed DC system may include the first module 610 to the sixth module 660, and each module will be described below.

[0127] The first module 610 can be configured to acquire the first state quantity of the multi-infeed DC system under a preset fault, and construct a basis function library based on the first state quantity, wherein the first state quantity is used to characterize the first local quantity measurement in the time-delay coordinate.

[0128] The second module 620 can be configured to sample the base function library to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state variable;

[0129] The third module 630 can be configured to perform ensemble learning based on multiple sub-function libraries to obtain the target Koopman feature set and the target Koopman feature function set.

[0130] The fourth module 640 can be configured to set the DC line reference active power adjustment amount of the multi-infeed DC system under preset fault conditions, wherein the DC line reference active power adjustment amount is determined according to the transmission power of the high voltage DC transmission line.

[0131] The fifth module 650 can be configured to determine the control matrix based on the DC line reference active power regulation.

[0132] The sixth module 660 can be configured to determine the target active power regulation of a multi-infeed DC system based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, so as to stabilize the system frequency of the multi-infeed DC system.

[0133] In an exemplary embodiment of the present invention, the first module 610 may determine the first state quantity in the following manner:

[0134] Based on the first local quantity measurement, determine the first time series data of the first local quantity measurement at a preset number of time points;

[0135] Based on the first time series data, a first time series matrix is ​​constructed in time-delay coordinates;

[0136] The first state variable is determined based on the first time series matrix in the time-delay coordinate system.

[0137] In an exemplary embodiment of the present invention, the first module 610 may construct a basis function library based on the first state variable in the following manner:

[0138] Based on the first state variable, determine the trigonometric function and polynomial function of the first state variable respectively;

[0139] A basis function library is constructed based on the first state variable, trigonometric functions, and the polynomial functions.

[0140] In an exemplary embodiment of the present invention, the third module 630 can perform ensemble learning based on multiple sub-function libraries to obtain the target Koopman feature value set and the target Koopman feature function set in the following manner:

[0141] Based on the sub-function library, the Koopman feature value set and Koopman feature function set corresponding to each sub-function library are obtained through the feature pair identification method;

[0142] Based on each set of Koopman eigenvalues, a candidate set of Koopman eigenvalues ​​is obtained;

[0143] Based on each set of Koopman feature functions, a candidate set of Koopman feature functions is obtained;

[0144] The candidate sets of Koopman eigenvalues ​​and Koopman eigenfunctions are filtered separately to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions.

[0145] In an exemplary embodiment of the present invention, the third module 630 can perform the following methods to filter the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions respectively, to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions:

[0146] Based on the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions, feature pairs are determined under the first state variable;

[0147] Based on the feature pairs under the first state quantity, the prediction error is determined;

[0148] Target feature pairs with prediction errors less than the error threshold are selected, and based on the target feature pairs, the candidate sets of Koopman feature values ​​and Koopman feature functions are selected respectively to obtain the target Koopman feature set and the target Koopman feature function set.

[0149] In an exemplary embodiment of the present invention, the fifth module 650 may determine the control matrix based on the DC line reference active power adjustment in the following manner:

[0150] Based on the DC line reference active power adjustment, a second state quantity corresponding to the DC line reference active power adjustment is determined, wherein the second state quantity is used to characterize the second local quantity measurement corresponding to the DC line reference active power adjustment in the time delay coordinate.

[0151] Based on the second state quantity, determine the time series data of the second state quantity at a preset number of time points;

[0152] The control matrix is ​​determined based on time series data.

[0153] In an exemplary embodiment of the present invention, the fifth module 650 may determine the second state quantity corresponding to the DC line reference active power adjustment amount based on the DC line reference active power adjustment amount in the following manner:

[0154] The second local quantity measurement is determined based on the DC line reference active power regulation.

[0155] Based on the second local quantity measurement, determine the second time series data of the second local quantity measurement at a preset number of time points;

[0156] Based on the second time series data, a second time series matrix is ​​constructed in time-delay coordinates;

[0157] The second state variable is determined based on the second time series matrix in the time-delay coordinate system.

[0158] In an exemplary embodiment of the present invention, the fifth module 650 can determine the control matrix from the time-series data in the following manner:

[0159] The control matrix is ​​determined based on time series data, the target Koopman eigenvalue set, and the target Koopman eigenfunction set.

[0160] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute a frequency emergency control method for a multi-infeed DC system. The multi-infeed DC system includes an AC system and multiple high-voltage DC transmission lines. The method includes: acquiring a first state quantity of the multi-infeed DC system under a preset fault, and constructing a basis function library based on the first state quantity, wherein the first state quantity is used to characterize a first local quantity measurement in a time-delay coordinate; sampling the basis function library to obtain multiple sub-function libraries, wherein each sub-function library includes at least the first state quantity; and performing ensemble learning based on the multiple sub-function libraries to obtain a target... A set of Koopman eigenvalues ​​and a set of target Koopman eigenfunctions are defined. Under the preset fault condition, a reference active power adjustment for the DC lines of the multi-infeed DC system is set, wherein the reference active power adjustment is determined based on the transmission power of the high-voltage DC transmission lines. A control matrix is ​​determined based on the reference active power adjustment. Based on the control matrix, the set of target Koopman eigenvalues, and the set of target Koopman eigenfunctions, a target active power adjustment for the multi-infeed DC system is determined to stabilize the system frequency of the multi-infeed DC system.

[0161] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0162] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the frequency emergency control method for a multi-infeed DC system provided by the above methods. The multi-infeed DC system includes an AC system and multiple high-voltage DC transmission lines. The method includes: acquiring a first state quantity of the multi-infeed DC system under a preset fault, and constructing a basis function library based on the first state quantity, wherein the first state quantity is used to characterize a first local quantity measurement in a time-delay coordinate; sampling the basis function library to obtain multiple sub-function libraries, wherein the sub-function libraries contain... The process includes at least the first state variable; ensemble learning based on multiple sub-function libraries to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set; under the preset fault, setting a reference active power adjustment for the DC lines of the multi-infeed DC system, wherein the reference active power adjustment is determined based on the transmission power of the high-voltage DC transmission line; determining a control matrix based on the reference active power adjustment; and determining a target active power adjustment for the multi-infeed DC system based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, so as to stabilize the system frequency of the multi-infeed DC system.

[0163] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the frequency emergency control method for a multi-infeed DC system provided by the methods described above. The multi-infeed DC system includes an AC system and multiple high-voltage DC transmission lines. The method includes: acquiring a first state quantity of the multi-infeed DC system under a preset fault, and constructing a basis function library based on the first state quantity, wherein the first state quantity is used to characterize a first local quantity measurement in a time-delay coordinate; sampling the basis function library to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state quantity; and based on the multi-infeed DC system... The sub-function libraries are integrated for learning to obtain a target Koopman eigenvalue set and a target Koopman eigenfunction set. Under the preset fault, a reference active power adjustment amount for the DC lines of the multi-infeed DC system is set, wherein the reference active power adjustment amount for the DC lines is determined based on the transmission power of the high-voltage DC transmission lines. Based on the reference active power adjustment amount for the DC lines, a control matrix is ​​determined. Based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, a target active power adjustment amount for the multi-infeed DC system is determined to stabilize the system frequency of the multi-infeed DC system.

[0164] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0165] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0166] It is further understood that although the operations are described in a specific order in the accompanying drawings in the embodiments of the present invention, this should not be construed as requiring these operations to be performed in the specific order or serial order shown, or requiring all the operations shown to obtain the desired result. In certain environments, multitasking and parallel processing may be advantageous.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for emergency frequency control of a multi-infeed DC system, characterized in that, The method, applied to a multi-infeed DC system comprising an AC system and multiple high-voltage DC transmission lines, includes: Obtain the first state quantity of the multi-infeed DC system under a preset fault, and construct a basis function library based on the first state quantity, wherein the first state quantity is used to characterize the first local quantity measurement in the time-delay coordinate. The basis function library is sampled to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state variable; Based on ensemble learning of multiple sub-function libraries, a target Koopman feature value set and a target Koopman feature function set are obtained; Under the preset fault condition, the DC line reference active power adjustment amount of the multi-infeed DC system is set, wherein the DC line reference active power adjustment amount is determined according to the transmission power of the high voltage DC transmission line. A control matrix is ​​determined based on the DC line reference active power adjustment, wherein determining the control matrix based on the DC line reference active power adjustment includes: Based on the DC line reference active power adjustment, a second state quantity corresponding to the DC line reference active power adjustment is determined, wherein the second state quantity is used to characterize the second local quantity measurement corresponding to the DC line reference active power adjustment in the time delay coordinate. Based on the second state quantity, determine the time series data of the second state quantity at a preset number of time points; The control matrix is ​​determined based on the time series data; Based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, the target active power regulation of the multi-infeed DC system is determined to stabilize the system frequency of the multi-infeed DC system.

2. The emergency frequency control method for a multi-infeed DC system according to claim 1, characterized in that, The first state variable is determined in the following way: Based on the first local quantity measurement, determine the first time series data of the first local quantity measurement at a preset number of time points; Based on the first time series data, a first time series matrix is ​​constructed under the time delay coordinates; The first state variable is determined based on the first time series matrix in the time-delay coordinates.

3. The emergency frequency control method for a multi-infeed DC system according to claim 1, characterized in that, The construction of the basis function library based on the first state variable specifically includes: Based on the first state variable, determine the trigonometric function and polynomial function of the first state variable respectively; A basis function library is constructed based on the first state variable, the trigonometric function, and the polynomial function.

4. The emergency frequency control method for a multi-infeed DC system according to any one of claims 1 to 3, characterized in that, The ensemble learning based on multiple sub-function libraries to obtain the target Koopman feature value set and the target Koopman feature function set specifically includes: Based on the sub-function library, the Koopman feature value set and Koopman feature function set corresponding to each sub-function library are obtained by the feature pair identification method; Based on each of the aforementioned Koopman feature sets, a candidate set of Koopman features is obtained; Based on each of the aforementioned sets of Koopman feature functions, a candidate set of Koopman feature functions is obtained; The candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions are filtered respectively to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions.

5. The emergency frequency control method for a multi-infeed DC system according to claim 4, characterized in that, The step of filtering the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions to obtain the target set of Koopman eigenvalues ​​and the target set of Koopman eigenfunctions specifically includes: Based on the candidate set of Koopman eigenvalues ​​and the candidate set of Koopman eigenfunctions, feature pairs are determined under the first state variable; Based on the feature pairs under the first state quantity, the prediction error is determined; Target feature pairs with prediction errors less than an error threshold are selected, and based on the target feature pairs, the candidate set of Koopman feature values ​​and the candidate set of Koopman feature functions are selected respectively to obtain the target set of Koopman feature values ​​and the target set of Koopman feature functions.

6. The emergency frequency control method for a multi-infeed DC system according to claim 1, characterized in that, The step of determining the second state quantity corresponding to the DC line reference active power adjustment based on the DC line reference active power adjustment specifically includes: Based on the DC line reference active power adjustment, determine the second local quantity measurement; Based on the second local quantity measurement, determine the second time-series data of the second local quantity measurement at a preset number of time points; Based on the second time series data, a second time series matrix is ​​constructed under the time delay coordinates; The second state quantity is determined based on the second time series matrix in the time delay coordinates.

7. The emergency frequency control method for a multi-infeed DC system according to claim 1, characterized in that, Determining the control matrix based on the time-series data specifically includes: The control matrix is ​​determined based on the time series data, the target Koopman eigenvalue set, and the target Koopman eigenfunction set.

8. A frequency emergency control device for a multi-feed DC system, characterized in that, An apparatus for use in multi-infeed DC systems, the multi-infeed DC system comprising an AC system and multiple high-voltage DC transmission lines, the apparatus being used to implement the frequency emergency control method for multi-infeed DC systems according to any one of claims 1 to 7, the apparatus comprising: The first module is used to obtain the first state quantity of the multi-infeed DC system under a preset fault, and to construct a basis function library based on the first state quantity, wherein the first state quantity is used to characterize the first local quantity measurement in the time-delay coordinate. The second module is used to sample the base function library to obtain multiple sub-function libraries, wherein the sub-function libraries include at least the first state variable; The third module is used to perform integrated learning based on multiple sub-function libraries to obtain the target Koopman feature value set and the target Koopman feature function set. The fourth module is used to set the DC line reference active power adjustment amount of the multi-infeed DC system under the preset fault, wherein the DC line reference active power adjustment amount is determined according to the transmission power of the high voltage DC transmission line. The fifth module is used to determine the control matrix based on the reference active power adjustment of the DC line; The sixth module is used to determine the target active power regulation of the multi-infeed DC system based on the control matrix, the target Koopman eigenvalue set, and the target Koopman eigenfunction set, so as to stabilize the system frequency of the multi-infeed DC system.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the emergency frequency control method for a multi-feed DC system as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the emergency frequency control method for a multi-feed DC system as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the emergency frequency control method for a multi-feed DC system as described in any one of claims 1 to 7.

Images