Method for searching voltage dominant type cascading failure path in power electronic power system

Abstract

The application relates to a voltage leading type cascading failure path searching method of a power electronic power system, which can more completely reflect the evolution process of power system failure, identify the appearance sequence of voltage response events, serially determine the path of voltage leading type cascading failure in the power electronic power system, and further solve the path searching problem of the voltage leading type cascading failure in the system, thereby better providing technical support for the operation and maintenance of the power electronic power system; and the method is suitable for the path searching method of the voltage leading type cascading failure, and the path identification and determination of the voltage leading type cascading failure in the power electronic power system are more accurate, and the method has good technical implementation applicability and effectiveness.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power system protection and control technology, and particularly relates to a voltage dominant type cascading failure path search method for a power electronic power system. BACKGROUND

[0002] The development and application of new energy has become an important driving force for the transformation of the power industry. Due to the characteristics of high distribution concentration and reverse distribution with load center of new energy power generation, in order to transport the electric energy of new energy station to the load center, the direct current transmission system becomes one of the main means. The direct current transmission system includes conventional direct current transmission system and flexible direct current transmission system. New energy and direct current transmission both rely on power electronic devices to realize grid-connected operation, so the power electronic power system is developing. However, due to the high sensitivity and strong coupling of power electronic devices, the cascading failure in the power electronic power system is more complex and the risk is increasing.

[0003] In the traditional power system, the cascading failure of the power system is divided into overload dominant type, structure dominant type and coordination dominant type according to the cause of the cascading failure. The overload dominant type cascading failure is that the line is tripped due to fault or overload, which causes the redistribution of power flow and triggers the successive overload tripping of other lines. The structure dominant type cascading failure refers to the process that the key tie line or important node of the power grid is cut off or the device is tripped too early, which causes serious damage to the topology structure of the power grid. The coordination dominant type cascading failure is the process of fault aggravation caused by unreasonable setting value of secondary devices such as relay protection, low equipment reliability and hidden faults. From the perspective of methodology, the research on cascading failure is mainly divided into two categories. The first category analyzes the system safety from the overall characteristics, mainly including complex network theory and self-organized critical theory. The second category focuses on the reduction and path search of the cascading failure process, mainly including pattern search theory. The complex network theory evaluates the structural characteristics and potential vulnerability of the power system through the degree distribution, shortest path, average distance and clustering coefficient, and effectively identifies the key nodes and vulnerable lines. The self-organized critical theory considers that the system spontaneously reaches a critical state under certain conditions, and a small perturbation can trigger a large-scale cascading reaction, which mainly explains the mechanism of cascading failure from a macroscopic perspective. The pattern search theory establishes a fault propagation model based on the physical characteristics of the power system, simulates the transfer of power flow and device overload after the fault, and studies the development process of the large-scale blackout from the perspective of cascading failure sequence.

[0004] However, in power electronic power systems, new energy stations, multi-circuit DC transmission systems, and new energy and DC transmission systems are deeply coupled. Due to the vulnerability and controlled characteristics of power electronic devices, AC grid faults can cause control switching, commutation failure, or even locking or disconnection of power electronic devices. After the grid fault, the output characteristics of the power electronic devices change continuously, resulting in continuous changes in the grid voltage, which can cause changes in the output characteristics of the remaining power electronic devices, causing the fault to spread within the region or across regions in the electromagnetic transient time scale. In power electronic power systems, a cascading failure with obvious voltage dominant characteristics is presented. The cascading evolution of the fault is no longer caused by structural damage, power flow transfer, or incorrect action of protection and control, and the continuous change of the grid voltage becomes the cause of the cascading evolution of the fault. The generation mechanism and form of the voltage dominant cascading failure are different from those of the traditional cascading failure. The existing cascading failure research method focuses on describing the relationship between the change of grid power flow distribution and cascading failure under line overload and relay protection action, and cannot reflect the change of device operating state in the continuous change of voltage, and is difficult to apply to voltage dominant cascading failure.

[0005] In summary, how to consider the voltage coupling process between power electronic devices and between devices and the grid, fully analyze the evolution mechanism of voltage dominant cascading failure from the voltage angle, and propose a path search method for voltage dominant cascading failure has become a problem that needs to be solved by those skilled in the art. SUMMARY

[0006] In view of the above technical problems of the prior art, the present application provides a voltage dominant cascading failure path search method for power electronic power systems, which builds a voltage response event-grid multi-layer coupling network, calculates the correlation weight of the voltage response event, identifies the subsequent voltage response event, and solves the path search problem of the voltage dominant cascading failure in the system, thereby better providing technical support for the operation and maintenance of power electronic power systems.

[0007] In order to solve the above technical problems, the present application adopts the following technical solutions:

[0008] The voltage dominant cascading failure path search method for power electronic power systems comprises the following steps:

[0009] S1, collect the current working parameters of the power electronic devices in the power electronic power system, and determine the action state of each voltage response event in the current power system;

[0010] S2, determine the correlation process of each voltage response event in the power system and the corresponding correlation coefficient according to the action state of the voltage response event;

[0011] S3, constructing a voltage response event-power grid multi-layer coupling network according to the correlation process of the voltage response event and the correlation coefficient thereof; the voltage response event-power grid multi-layer coupling network is used to represent the correlation between the power grid nodes and the voltage response events and the weight values of the respective correlation;

[0012] S4, establishing a random walk model based on the meta-path of the voltage response event-power grid multi-layer coupling network;

[0013] S5, using a Markov chain Monte Carlo method to perform walk search and solve the random walk model based on the meta-path of the voltage response event-power grid multi-layer coupling network, and determine the sequence of occurrence of the voltage response events in the power system;

[0014] S6, determining the path of the voltage dominant cascading failure in the power electronic power system according to the sequence of occurrence of the voltage response events, and completing the failure path search.

[0015] Specifically, in step S1, the action state of the voltage response event includes starting, triggering, maintaining, and stopping, and the action state of any voltage response event i is determined in the following manner:

[0016]

[0017] In the formula, Sta i represents the action state of the voltage response event i; cU i and cT i are the voltage state variable and the time state variable of the voltage response event i, respectively;

[0018] wherein the voltage state variable cU i and the time state variable cT i of the voltage response event i are determined in the following manner:

[0019]

[0020] In the formula, U i is the grid-connected node voltage of the power electronic device corresponding to the voltage response event; C UThi represents the voltage range of the voltage response event i; T Thi represents the action time of the voltage response event i; t cUi represents the change time of the voltage state variable of the voltage response event i; t is the current time.

[0021] Specifically, in step S2, the correlation process of the voltage response event includes the correlation between the triggering or stopping of the voltage response event and the power exchange, the correlation between the power exchange and the voltage change, the correlation between the voltage change and the starting or maintaining of the voltage response event, and the correlation between the starting or maintaining of the voltage response event and the triggering or stopping.

[0022] The correlation coefficients corresponding to the voltage response event correlation process include a correlation coefficient of the voltage response event i triggering or stopping power exchange with the power grid node p, a correlation coefficient of power exchange of the power grid node p and voltage change of the power grid node q, a correlation coefficient of voltage change of the power grid node q and the voltage response event j starting or maintaining, and a correlation coefficient of the voltage response event j starting or maintaining and triggering or stopping.

[0023] Specifically, in step S2, the correlation coefficient of the voltage response event i triggering or stopping power exchange with the power grid node p is calculated in the following manner:

[0024]

[0025] The correlation coefficient of power exchange of the power grid node p and voltage change of the power grid node q is calculated in the following manner:

[0026]

[0027] The correlation coefficient of voltage change of the power grid node q and the voltage response event j starting or maintaining is calculated in the following manner:

[0028]

[0029] The correlation coefficient of the voltage response event j starting or maintaining and triggering or stopping is calculated in the following manner:

[0030]

[0031] wherein, E p represents a set of voltage response events of the power electronic device of the node p; α ip = 0 indicates that the voltage response event i is irrelevant to power of the power grid node p; α ip = 1 indicates that the voltage response event i is directly related to power of the power grid node p; ΔU q is a voltage change of the power grid node q; ΔP p and ΔQ p respectively represent active power exchange change and reactive power exchange change of the power electronic device and the power grid node p; k P and k Q respectively represent weights of active voltage sensitivity and reactive voltage sensitivity; c represents a correlation coefficient bias coefficient; τ j represents a remaining action time of the voltage response event j, which is a difference between an event action time and a current time.

[0032] Specifically, in step S3, the voltage response event-power grid multi-layer coupling network is established in the following manner:

[0033] CN = (N U E, CX W X );

[0034] Where N = {n1, n2, ..., n} k} represents the set of power grid nodes to which power electronic equipment is connected, and k is the number of power grid nodes to which the power electronic equipment is connected; E = {e1, e2, ..., e m-1 ,e m} represents the set of voltage response events, where m is the number of voltage response events; W X This represents the weight matrix indicating the relationships between multiple layers of coupled networks.

[0035] The voltage response event-grid multilayer coupling network includes four single-layer association networks: a grid node association network, a voltage response event association network, a directed association network between voltage response events and grid nodes, and a directed association network between grid nodes and voltage response events. X For voltage response events, there is a set of multi-layered correlations between the power grid and the network, and C X =C NN ∪C EE ∪C NE ∪C EN , where C NN For the power grid node association network L NN The set of relationships, C EE For voltage response event correlation network L EE The set of relationships, C EN L is a directed network of voltage response events and grid nodes. EN The set of relationships, C NE L is a directed correlation network of grid nodes and voltage response events. NE The sets of relationships are determined as follows:

[0036] Power grid node association network L NN The set of associations C NN Let N(p) → N(q) be the set of all N(p) → N(q) relationships, where N(p) → N(q) represents the relationship between grid node p and grid node q, and N(p) and N(q) represent grid node p and grid node q, respectively.

[0037] Voltage response event association network L EE The set of associations C EE Let E(i) → E(j) be the set of all E(i) → E(j) relationships, where E(i) → E(j) represents the relationship between voltage response event i and voltage response event j, and E(i) and E(j) represent voltage response event i and voltage response event j, respectively.

[0038] Voltage response events and directed association network of grid nodes LEN a set of association relations C EN is a set of all E(i)→N(p) association relations, where E(i)→N(p) represents a directed association relation between voltage response event i and grid node p;

[0039] a directed association network L between grid nodes and voltage response events NE a set of association relations C NE is a set of all N(q)→E(j) association relations, where N(q)→E(j) represents a directed association relation between grid node q and voltage response event j.

[0040] Specifically, the grid node association network L NN , the voltage response event association network L EE , the directed association network L EN between voltage response events and grid nodes, and the directed association network L NE between grid nodes and voltage response events are respectively represented as:

[0041] L NN = (N, C NN , W NN ) ;

[0042] L EE = (E, C EE , W EE ) ;

[0043] L EN = (E∪N, C EN , W EN ) ;

[0044] L NE = (N∪E, C NE , W NE ) ;

[0045] In the formula, W NN is a weight value matrix of association relations between grid nodes; W EE is a weight value matrix of unknown association relations of voltage response events; W EN is a weight value matrix of directed association relations between voltage response events and grid nodes; W NE is a weight value matrix of directed association relations between grid nodes and voltage response events; and the weight value matrix W X of the multi-layer coupled network association relation is composed of W NN , W EN , W NE , and W EE , and is represented as:

[0046]

[0047] Specifically, the weight value matrix W of the association relationship between the grid nodes NN , the weight value matrix W of the directed association relationship between the voltage response event and the grid node EN , the weight value matrix W of the directed association relationship between the grid node and the voltage response event NE , and the weight value matrix W of the unknown association relationship of the voltage response event EE The weight values in the above matrices are determined in the following manner, respectively:

[0048]

[0049] In the formula, W NN (p,q)∈W NN represents the weight value of the association relationship between the grid node p and the grid node q, β pq is the correlation coefficient of the power exchange of the grid node p and the voltage change of the grid node q; W EN (i,p)∈W EN represents the weight value of the directed association relationship between the voltage response event i and the grid node p, α ip is the correlation coefficient of the triggering or termination of the voltage response event i and the power exchange of the grid node p; W NE (q,j)∈W NE represents the weight value of the directed association relationship between the grid node q and the voltage response event j, θ qj is the correlation coefficient of the voltage change of the grid node q and the starting or maintaining of the voltage response event j; γ j is the correlation coefficient of the starting or maintaining and the triggering or termination of the voltage response event j; W EE (e i )∈W EE represents the weight value vector of the unknown association relationship between the voltage response event i and other voltage response events in the power system; W EE (e i ) The jth element w p (e i ,e j ) represents the weight value of the association relationship between the voltage response event j and the voltage response event i after the occurrence of any voltage response event i, which is an unknown quantity and needs to be determined by solving.

[0050] Specifically, in step S4, the random walk model of the voltage response event-grid multi-layer coupled network meta-path is established in the following manner:

[0051]

[0052] In the formula, CN is the voltage response event-grid multi-layer coupled network; MP is the random walk meta-path; v initis the initial node of the walk; W path (v init ) represents the walk path weight value matrix of the different walk paths starting from the initial node v init , e, n respectively represent the voltage response event node and the power grid node in the voltage response event-power grid multi-layer coupling network.

[0053] Specifically, in step S5, the walk search solving mode based on the random walk model of the voltage response event-power grid multi-layer coupling network meta path is as follows:

[0054] The voltage response event in the action state is the initial node of the particle random walk, and according to the random walk model of the voltage response event-power grid multi-layer coupling network meta path, the next node of each walk is randomly selected according to the meta path e→n→n→e by using the Markov chain Monte Carlo method, the corresponding walk path weight is calculated, and thus the current initial node v init and other different voltage response events as the walk path end v end walk path weight of the current initial node v init , and according to the order from large to small of the weight values of the correlation between the voltage response event node and other voltage response events, the order of occurrence of each voltage response event in the power system is determined.

[0055] Specifically, the step S5 is specifically as follows:

[0056] S501, according to the action state of each voltage response event in the current power system determined in step S1, the voltage response event in the action state Sta i is started, and the action time T Thi is the shortest, as the initial node v init of the particle random walk in the current walk path.

[0057] S502, according to the action state of each voltage response event in the current power system, the correlation process of each voltage response event in the current power system and the corresponding correlation coefficient are determined, and the correlation relationship weight value matrix of the voltage response event-power grid multi-layer coupling network is obtained;

[0058] S503, taking the current initial node v init as the starting point, according to the random walk model of the voltage response event-power grid multi-layer coupling network meta path, the next node of each walk is randomly selected according to the meta path MP=e→n→n→e by using the Markov chain Monte Carlo method, and the corresponding walk path weight is calculated:

[0059] w p (v init ,vend ) = w(v init |v n1 ) · w(v n1 |v n2 ) · w(v n2 |v end );

[0060] wherein w(v p , v init , v end ) is the walk path weight from the initial node v init to the voltage response event node as the walk path end v end , is an unknown to be solved, representing that from the initial node v init , the meta path e→n→n→e is walked and stopped at v end , the walk path is w(v init |v n1 ) · w(v n2 |v end ); v n1 , v n2 represent two power grid nodes passed in the walk path; w(v init |v n1 ) represents the walk path weight from the initial node v init to the first power grid node v n1 in the meta path, w(v n1 |v n2 ) represents the walk path weight from the first power grid node v n1 to the second power grid node v n2 in the meta path, w(v n2 |v end ) represents the walk path weight from the second power grid node v n2 to the walk path end v end , all of which are known quantities that can be determined by the correlation weight value matrix of the voltage response event-power grid multi-layer coupled network;

[0061] Thus, the walk path weight of the current initial node v init and other different voltage response events as the walk path end v end is obtained by traversal, as the weight value of the correlation between the voltage response event node of the current initial node v init and other voltage response events;

[0062] S504, judging whether the current initial node v initwhether the weight values of the association relationship between the voltage response event node and other voltage response events are all 0; if yes, the walk search is ended; otherwise, one walk path end point v end The corresponding voltage response event is determined as a subsequent voltage response event;

[0063] S505, taking the subsequent voltage response event as a new initial node, re-collecting the current working parameters of the power electronic equipment in the power electronic power system, re-determining the action state of each voltage response event in the current power system, and jumping back to step S502;

[0064] S506, the steps S502 to S505 are circularly executed until the walk search is ended, and the order of occurrence of each voltage response event in the power system is determined.

[0065] Compared with the prior art, the present application has the following beneficial effects:

[0066] 1. The fault analysis technology of the existing power electronic equipment only focuses on the change of the running state of the power electronic equipment caused by the voltage and the mutual influence of the running state of the power electronic equipment through the voltage. However, the method of the present application further considers the process of coupling between the power electronic equipment and the power grid through the voltage, constructs a voltage response event-power grid multi-layer coupling network, can more completely reflect the evolution process of the power system fault, identify the occurrence order of the voltage response event, determine the path of the voltage dominant cascading failure in the power electronic power system in series, and then solve the path search problem of the voltage dominant cascading failure in the system, so as to better provide technical support for the operation and maintenance of the power electronic power system.

[0067] 2. The prior art is only applicable to the overload dominant, structure dominant and coordination dominant cascading failures of the power system, and cannot be applied to the voltage dominant cascading failure of the power system caused by the large-scale application of power electronics. According to the difference between the voltage dominant cascading failure and the overload dominant, structure dominant and coordination dominant cascading failure in physical mechanism, the method of the present application proposes a cascading failure path search method suitable for the voltage dominant cascading failure, and the path identification and determination of the voltage dominant cascading failure in the power electronic power system are more accurate, and the method has good technical implementation applicability and effectiveness. BRIEF DESCRIPTION OF DRAWINGS

[0068] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below with reference to the drawings, in which:

[0069] Figure 1 Flow chart of the power system voltage dominant cascading failure path search method;

[0070] Figure 2 Structure diagram of power system with power electronic equipment access;

[0071] Figure 3 Voltage response event correlation prediction result diagram for line 14-16 fault in the embodiment of the application;

[0072] Figure 4 Power electronic equipment grid-connected node voltage for line 14-16 fault in the embodiment of the application. DETAILED DESCRIPTION

[0073] In order to make the objects, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some but not all of the embodiments of the application. The components of the embodiments of the application described and shown in the drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the application.

[0074] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some but not all of the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the application.

[0075] The application provides a voltage dominant cascading failure path search method for a power electronic power system, which comprises the following steps:

[0076] Based on the above technical design idea, the specific process of the voltage dominant cascading failure path search method for a power electronic power system of the application is shown in Figure 1 The specific process of the voltage dominant cascading failure path search method for a power electronic power system of the application is shown in

[0077] S1, collect the current working parameters of the power electronic equipment in the power electronic power system, and determine the action state of each voltage response event in the current power system;

[0078] S2, determining the correlation process of each voltage response event and the corresponding correlation coefficient in the power system according to the action state of the voltage response event;

[0079] S3, constructing a voltage response event-grid multilayer coupling network according to the correlation process of the voltage response event and the correlation coefficient; the voltage response event-grid multilayer coupling network is used to represent the correlation between the grid node and the voltage response event and the weight value of each correlation;

[0080] S4, establishing a random walk model based on the meta-path of the voltage response event-grid multilayer coupling network;

[0081] S5, using the Markov chain Monte Carlo method to perform walk search and solve the random walk model based on the meta-path of the voltage response event-grid multilayer coupling network, and determine the sequence of occurrence of each voltage response event in the power system;

[0082] S6, determining the path of the voltage dominant cascading failure in the power electronic power system according to the sequence of occurrence of the voltage response event, and completing the fault path search.

[0083] In specific implementation, the working parameters of the power electronic equipment in the power electronic power system collected in step S1 can include AC bus voltage, voltage duration and power exchange amount, etc. These working parameters are mainly used to determine the action state of each voltage response event in the current power system, and further determine the correlation process of each voltage response event in the power system and the corresponding correlation coefficient.

[0084] In specific implementation, in step S1, the action state of the voltage response event includes starting, triggering, maintaining, and stopping, and the action state of any voltage response event i is determined in the following manner:

[0085]

[0086] In the formula, Sta i represents the action state of the voltage response event i; cU i and cT i are the voltage state variable and the time state variable of the voltage response event i, respectively.

[0087] Wherein, the voltage state variable and the time state variable of the voltage response event i are determined in the following manner:

[0088]

[0089] In the formula, U i is the grid node voltage of the power electronic equipment corresponding to the voltage response event; C UThirepresents the voltage range of the voltage response event i; T Thi represents the action time of the voltage response event i; t cUi represents the change time of the voltage state variable of the voltage response event i; t is the current time.

[0090] According to formula (2), the value of the voltage state variable and the time state variable of the voltage response event is as follows: if the voltage meets the voltage range of the voltage response event, the voltage state variable of the voltage response event is 1, otherwise 0; if the duration of the voltage meeting the voltage range reaches the action time of the voltage response event, the time state variable of the voltage response event is 1, otherwise 0.

[0091] In specific implementation, in step S2, the association process of the voltage response event includes the association of the triggering or termination of the voltage response event with the power exchange of the grid node p, the association of the power exchange with the voltage change of the grid node q, the association of the voltage change with the start or maintenance of the voltage response event j, and the association of the start or maintenance of the voltage response event j with the triggering or termination.

[0092] Correspondingly, the correlation coefficient corresponding to the association process of the voltage response event includes the correlation coefficient of the triggering or termination of the voltage response event i and the power exchange of the grid node p, the correlation coefficient of the power exchange of the grid node p and the voltage change of the grid node q, the correlation coefficient of the voltage change of the grid node q and the start or maintenance of the voltage response event j, and the correlation coefficient of the start or maintenance of the voltage response event j and the triggering or termination.

[0093] In specific implementation, the correlation coefficient of the triggering or termination of the voltage response event i and the power exchange of the grid node p is calculated as follows:

[0094]

[0095] In the formula, E p represents the voltage response event set of the power electronic device of the grid node p; α ip = 0 indicates that the voltage response event i is not associated with the power of the grid node p; α ip = 1 indicates that the voltage response event i is directly associated with the power of the grid node p.

[0096] In specific implementation, the correlation coefficient of the power exchange of the grid node p and the voltage change of the grid node q is calculated as follows:

[0097]

[0098] In the formula, ΔU q is the voltage change of the grid node q; ΔP p , and ΔQ p respectively represent the active power exchange change and the reactive power exchange change of the power electronic device and the grid node p; kP and k Q respectively represent the weight of active voltage sensitivity and reactive voltage sensitivity; c represents the correlation coefficient bias coefficient.

[0099] In the specific implementation, the correlation coefficient of the voltage change of the grid node q and the start or maintenance of the voltage response event j is calculated in the following manner:

[0100]

[0101] In the specific implementation, the correlation coefficient of the voltage response event j start or maintenance and the trigger or termination is calculated in the following manner:

[0102]

[0103] wherein τ j represents the remaining action time of the voltage response event j, which is the difference between the event action time and the current time.

[0104] In the specific implementation, in step S3, the voltage response event-grid multi-layer coupled network is established in the following manner:

[0105] CN=(N∪E,C X ,W X ) (7)

[0106] wherein N={n1,n2,…,n k} represents the grid node set accessed by the power electronic device, k is the number of grid node sets accessed by the power electronic device; E={e1,e2,…,e m-1 ,e m} represents the voltage response event set, m is the number of voltage response events; W X represents the multi-layer coupled network correlation relationship weight value matrix.

[0107] The established voltage response event-grid multi-layer coupled network includes four single-layer correlation networks, i.e., the grid node correlation network, the voltage response event correlation network, the voltage response event and grid node directed correlation network, and the grid node and voltage response event directed correlation network, C X is the voltage response event-grid multi-layer correlation relationship set, and C X =C NN ∪C EE ∪C NE ∪C EN , wherein C NN is the correlation relationship set of the grid node correlation network L NN , C EE is the correlation relationship set of the voltage response event correlation network L EE , C ENThe association relationship set C of the grid node associated network L EN The association relationship set C of the grid node associated network L NE The association relationship set C of the grid node associated network L NE is determined in the following manner respectively:

[0108] The association relationship set C of the grid node associated network L NN The association relationship set C of the grid node associated network L NN is a set composed of all N(p)→N(q) association relationships, wherein N(p)→N(q) represents an association relationship between the grid node p and the grid node q, and N(p) and N(q) represent the grid node p and the grid node q respectively;

[0109] The association relationship set C of the grid node associated network L EE The association relationship set C of the grid node associated network L EE is a set composed of all E(i)→E(j) association relationships, wherein E(i)→E(j) represents an association relationship between the voltage response event i and the voltage response event j, and E(i) and E(j) represent the voltage response event i and the voltage response event j respectively;

[0110] The association relationship set C of the grid node associated network L EN The association relationship set C of the grid node associated network L EN is a set composed of all E(i)→N(p) association relationships, wherein E(i)→N(p) represents a directed association relationship between the voltage response event i and the grid node p;

[0111] The association relationship set C of the grid node associated network L NE The association relationship set C of the grid node associated network L NE is a set composed of all N(q)→E(i) association relationships, wherein N(q)→E(i) represents a directed association relationship between the grid node q and the voltage response event i.

[0112] Specifically, the grid node associated network L NN , the voltage response event associated network L EE , the voltage response event and grid node directed association network L EN , and the grid node and voltage response event directed association network L NE are represented as:

[0113] L NN = (N, C NN , W NN ) (8)

[0114] L EE = (E, C EE , W EE ) (9)

[0115] L EN =(E∪N,C EN ,W EN ) (10)

[0116] L NE =(N∪E,C NE ,W NE ) (11)

[0117] In the formula, W NN is the weight value matrix of the correlation between the grid nodes; W EE is the weight value matrix of the unknown correlation of the voltage response event; W EN is the weight value matrix of the directed correlation between the voltage response event and the grid node; and W NE is the weight value matrix of the directed correlation between the grid node and the voltage response event.

[0118] The weight value matrix W X of the correlation of the multi-layer coupled network is composed of W NN , W EN , W NE and W EE , and is expressed as:

[0119]

[0120] In a specific implementation, the weight values in the weight value matrix W NN of the correlation between the grid nodes, the weight value matrix W EN of the directed correlation between the voltage response event and the grid node, the weight value matrix W NE of the directed correlation between the grid node and the voltage response event, and the weight value matrix W EE of the unknown correlation of the voltage response event are respectively determined in the following manner:

[0121]

[0122] In the formula, W NN (p,q)∈W NN represents the weight value of the correlation between the grid node p and the grid node q, β pq is the correlation coefficient of the power exchange of the grid node p and the voltage change of the grid node q; W EN (i,p)∈W EN represents the weight value of the directed correlation between the voltage response event i and the grid node p, α ip is the correlation coefficient of the triggering or termination of the voltage response event i and the power exchange of the grid node p; W NE (q,j)∈W NEθ qj is the correlation coefficient of voltage variation of grid node q and voltage response event j starting or maintaining; γ j is the correlation coefficient of voltage response event j starting or maintaining and triggering or stopping; W EE (e i )∈W EE is the weight value vector of unknown correlation relationship between voltage response event i and other voltage response events in power system; W EE (e i ) is the jth element w p (e i ,e j ) represents the weight value of correlation relationship between voltage response event j and voltage response event i after the occurrence of any voltage response event i, which is an unknown quantity and needs to be determined by solving. The expression {A→B}∈C in the formula represents that the correlation from A to B belongs to the correlation set C.

[0123] In specific implementation, in step S4, the random walk model based on the voltage response event-grid multilayer coupling network element path is established in the following manner:

[0124]

[0125] In the formula, CN is the voltage response event-grid multilayer coupling network; MP is the random walk element path; v init is the initial node of the walk; W path (v init ) represents the weight value matrix of different walk paths of the element path MP starting from the initial node v init ; e and n respectively represent the voltage response event node and the grid node in the voltage response event-grid multilayer coupling network.

[0126] In specific implementation, in step S5, the manner of walk search solving based on the random walk model of the voltage response event-grid multilayer coupling network element path is as follows:

[0127] The voltage response event in the action state as the starting is the initial node of the particle random walk, according to the random walk model of the voltage response event-grid multilayer coupling network element path, the next level node of each walk is randomly selected according to the element path e→n→n→e by using the Markov chain Monte Carlo method, the corresponding walk path weight is calculated, and thus the current initial node v init and other different voltage response events as the walk path terminal v end are obtained, and the walk path weight is taken as the current initial node v initThe weight value of the association relationship between the voltage response event node and other voltage response events, and according to the order from large to small of the weight value of the association relationship, the order of occurrence of each voltage response event in the power system is determined.

[0128] Specifically, step S5 is specifically:

[0129] S501, according to the action state of each voltage response event in the current power system determined in step S1, the action state Sta i is started, and the action time T Thi The shortest one is the initial node v init of the random walk of the particle in the current walk path.

[0130] S502, according to the action state of each voltage response event in the current power system, the association process of each voltage response event in the power system and the corresponding correlation coefficient are determined, and the weight value matrix of the current association relationship of the voltage response event-power grid multi-layer coupling network is obtained.

[0131] S503, taking the current initial node v init as the starting point, according to the random walk model of the meta-path of the voltage response event-power grid multi-layer coupling network, according to the meta-path MP=e→n→n→e, the next level node of each walk is randomly selected by using the Markov chain Monte Carlo method, and the corresponding walk path weight is calculated:

[0132] w p (v init ,v end )=w(v init |v n1 )·w(v n1 |v n2 )·w(v n2 |v end ) (18)

[0133] In the formula, w p (v init ,v end ) is the walk path weight from the initial node v init to the voltage response event node v end as the end point of the walk path through the meta-path MP, which is an unknown quantity to be solved, which represents that from the initial node v init , walk along the meta-path e→n→n→e and stop at v end , the walk path weight of the walk path v init →v n1 →v n2 →v end ; v n1 , v n2represents two grid nodes passed in the walk path; w(v init |v n1 ) represents the walk path weight from the initial node v init to the first grid node v n1 in the meta path, w(v n1 |v n2 ) represents the walk path weight from the first grid node v n1 to the second grid node v n2 in the meta path, and w(v n2 |v end ) represents the walk path weight from the second grid node v n2 to the walk path end node v end in the meta path, which are all known quantities that can be determined by the correlation weight value matrix of the voltage response event-grid multilayer coupling network.

[0134] For example, if the voltage response event as the initial node is i, the voltage response event node as the walk path end node v end is j, and the two grid nodes passed in the walk path are grid node p and grid node q, respectively, then the weight value w p (e x ,e i ) of the correlation between the voltage response event j and the voltage response event i is w p (v init ,v end ), which is an unknown quantity to be solved; w(v init |v n1 ) is the weight value W EN (i,p) of the directed correlation between the voltage response event i and the grid node p, i.e., w(v init |v n1 ) = W EN (i,p); w(v n1 |v n2 ) is the weight value W NN (p,q) of the correlation between the grid node p and the grid node q, i.e., w(v n1 |v n2 ) = W NN (p,q); and w(v n2 |v end ) is the weight value W NE (q,j) of the directed correlation between the grid node q and the voltage response event j, i.e., w(v n2 |v end ) = W NE (q,j); thus, w(v init |v n1), w(v init |v n1 ), w(v n1 |v n2 ) are known quantities determined by the voltage response event-grid multilayer coupling network correlation weight value matrix calculation.

[0135] Thus, the initial node v init and other different voltage response events as the walk path endpoint v end walk path weight, as the voltage response event node of the current initial node v init and other voltage response events between the weight value of the correlation.

[0136] S504, determine whether the weight value of the correlation between the voltage response event node of the current initial node v init and other voltage response events is 0; if so, end the walk search; otherwise, determine the corresponding voltage response event of the walk path endpoint v end with the largest correlation weight value as a subsequent voltage response event.

[0137] S505, take the subsequent voltage response event as a new initial node, re-collect the current working parameters of the power electronic equipment in the power electronic power system, re-determine the action state of each voltage response event in the current power system, and jump back to step S502.

[0138] S506, steps S502 to S505 are executed in a loop until the walk search is ended, and the order of occurrence of each voltage response event in the power system is determined.

[0139] In the above process, each time a voltage response event is determined as a random walk initial node v init , the correlation process and the corresponding correlation coefficient of each voltage response event in the power system are determined based on the action state of each voltage response event corresponding to the initial node v init , and the correlation relationship weight value matrix of the voltage response event-grid multilayer coupling network is obtained. The initial node v end and other different voltage response events as the walk path endpoint v initThe algorithm first assigns weight values ​​to the relationships between voltage response event nodes. Then, based on these weight values, it determines the next voltage response event as the initial node for a new random walk search. Based on this new initial node, it re-determines the operational state of each voltage response event, the association process of each event, and its corresponding correlation coefficient. This results in a new weight value matrix for the current relationship between the voltage response event and the multi-layer coupled network of the power grid, which is then used in a new round of random walk model calculations. This process continues until the random walk search ends. In this way, each time the voltage response event is re-determined as the initial node, the impact of previously occurring voltage response events on the output characteristics and changes in the operating state of the current power electronic equipment are considered through recalculation. This leads to more accurate path search and identification for cascading faults with voltage-dominated characteristics in the power system.

[0140] Example:

[0141] To verify the effectiveness of the present invention, as follows Figure 2 The following analysis will be based on a schematic diagram of a power system structure with connected power electronic equipment. Figure 2 In the embodiment shown, the flexible DC transmission system (flexible DC transmission system) is connected to the receiving-end grid via nodes 35, 36, 37, 38, and 39, with transmission capacities of 1488.20MW, 1488.20MW, 957.80MW, 634.70MW, and 1514.80MW, respectively. The wind farm (WF) is equipped with direct-drive wind turbines and is connected to the AC grid via nodes 18 and 33, with grid-connected capacities of 250.00MW and 350.00MW, respectively. The conventional DC transmission system (conventional DC transmission system) is connected to the receiving-end grid via nodes 11, 15, 17, and 34, with transmission capacities of 2816.20MW, 2856.00MW, 1160.00MW, and 5996.40MW, respectively.

[0142] The voltage response events and their voltage ranges are shown in Table 1.

[0143] Table 1 Voltage Response Event Voltage Range Conditions

[0144]

[0145] Conventional DC transmission systems are blocked after three consecutive commutation failures within 200ms. The disconnection time for WF and flexible DC transmission systems is calculated based on the relay protection's operating time. For WF systems, disconnection occurs without delay when the voltage drops below 0.2 pu. When the WF voltage is between 0.2 and 0.9 pu, the operating time increases linearly with voltage from 625 to 2000 ms. For WF voltages between 1.1 and 1.2 pu, 1.2 and 1.25 pu, 1.25 and 1.3 pu, and above 1.3 pu, the operating times are 10 s, 1 s, 0.5 s, and 0 s, respectively. For flexible DC transmission systems, disconnection time is 150 ms when the voltage drops below 0.2 pu; the operating time for other voltages is the same as for WF. The response delays of voltage support control and high-voltage ride-through in WF and flexible DC transmission systems are determined by the controller performance. The dynamic reactive current rise time and exit time are typically required to be less than 60ms, with 25ms, 45ms, and 45ms selected respectively. Commutation failure usually occurs within half a power frequency cycle after the fault occurs; 5ms is selected as the action time for commutation failure. The power changes caused by voltage support control and high-voltage ride-through in WF and flexible DC transmission systems are calculated based on the changes in the control reference value; after grid disconnection or blocking occurs, the power becomes 0.

[0146] This embodiment uses a three-phase short-circuit fault occurring at 10ms as an example, with the fault cleared after 140ms. When line 14-16 experiences a fault, the cascading fault path search results are shown in Table 2. The voltage changes at the nodes connected to the power electronic equipment are sequentially affected by: conventional DC transmission system commutation failure, WF voltage support control, flexible DC transmission system voltage support control, conventional DC transmission system commutation failure and blocking, conventional DC transmission system commutation failure, WF voltage support control, and flexible DC transmission system voltage support control, consistent with the cascading fault path search results.

[0147] Figure 3 The diagram shows the predicted results of voltage response event correlation for faults in lines 14-16 in this embodiment. Figure 4 The grid-connected node voltage of the power electronic equipment for faults in lines 14-16 in this embodiment is shown. Figure 3 The association weights of preceding and subsequent voltage response events are shown. Each association process with the highest weight constitutes the cascading failure path shown in Table 2. Figure 4 As can be seen, the voltage changes triggered by the voltage response events in the cascading fault paths described in Table 2 are related to... Figure 4 The voltage change trends obtained from the simulation are consistent, which effectively demonstrates the accuracy of the cascading fault path.

[0148] Table 2 Search Results for Chain Fault Paths of Lines 14-16

[0149]

[0150] It can be seen that the power electronic power system voltage dominant type cascading failure path search method provided by the application, on the basis of paying attention to the voltage caused power electronic equipment operating state change and the power electronic equipment operating state through the voltage mutual influence, further combines the voltage coupling process between the power electronic equipment and the device and the grid, constructs a voltage response event-grid multi-layer coupling network, can more completely reflect the evolution process of the power system fault, identifies the appearance order of the voltage response event, determines the voltage dominant type cascading failure path in the power electronic power system in series, and solves the path search problem of the voltage dominant type cascading failure in the system, to better provide technical support for the operation and maintenance of the power electronic power system. At the same time, according to the difference between the voltage dominant type cascading failure and the overload dominant type, the structure dominant type and the coordination dominant type cascading failure physical mechanism, the application method proposes a cascading failure path search method suitable for the voltage dominant type cascading failure, and the path identification and determination of the voltage dominant type cascading failure in the power electronic power system is more accurate, and has good technical implementation applicability and effectiveness.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the application and not to limit the technical solutions, and those of ordinary skill in the art should understand that those who modify or equivalently replace the technical solutions of the application without departing from the purpose and scope of the technical solutions should be covered in the scope of the claims of the application.

Claims

1. A voltage-dominant cascading failure path search method for power electronicized power systems, characterized by, The method comprises the following steps: S1, collecting the current working parameters of power electronic devices in the power electronic power system, and determining the action state of each voltage response event in the current power system; the action state of the voltage response event includes starting, triggering, maintaining, and stopping, and the action state of any voltage response event i is determined in the following manner: ； wherein Sta i represents the action state of the voltage response event i; cU i and cT i are the voltage state variable and the time state variable, respectively, of the voltage response event i; wherein the voltage state variable cU of the voltage response event i i and the time state variable cT i is determined in particular as follows: ； ； In the formula, U i is the grid-connected node voltage of the power electronic device corresponding to the voltage response event; C UThi represents the voltage range of the voltage response event i; T Thi represents the action time of the voltage response event i; t cUi represents the change moment of the voltage state variable of the voltage response event i; t is the current moment; S2, determining the correlation process of each voltage response event in the power system and the corresponding correlation coefficient according to the action state of the voltage response event; wherein the correlation process of the voltage response event includes the correlation of voltage response event triggering or stopping and power exchange, the correlation of power exchange and voltage change, the correlation of voltage change and voltage response event starting or maintaining, and the correlation of voltage response event starting or maintaining and triggering or stopping; the correlation coefficient corresponding to the correlation process of the voltage response event includes the correlation coefficient of voltage response event i triggering or stopping and power exchange of grid node p, the correlation coefficient of power exchange of grid node p and voltage change of grid node q, the correlation coefficient of voltage change of grid node q and voltage response event j starting or maintaining, and the correlation coefficient of voltage response event j starting or maintaining and triggering or stopping; S3, constructing a voltage response event-grid multilayer coupling network according to the correlation process of the voltage response event and the correlation coefficient; the voltage response event-grid multilayer coupling network is used to represent the correlation relationship between the grid node and the voltage response event and the weight value of each correlation relationship; S4, establishing a random walk model based on the meta-path of the voltage response event-grid multilayer coupling network; the random walk model based on the meta-path of the voltage response event-grid multilayer coupling network is established in the following manner: ； where CN is the voltage response event-power grid multi-layer coupled network; MP is the random walk meta path; v init is the initial node of the walk; W path (v init ) represents the different walk path weight value matrix of the meta path MP starting from the initial node v init ; e, n represent the voltage response event node and the power grid node in the voltage response event-power grid multi-layer coupled network, respectively. S5, using the Markov chain Monte Carlo method to perform walk search and solve the random walk model based on the meta-path of the voltage response event-grid multilayer coupling network, and determining the sequence of occurrence of each voltage response event in the power system; S6, determining the path of voltage dominant cascading failure in the power electronic power system according to the sequence of occurrence of the voltage response event, and completing the fault path search.

2. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 1, characterized by, In step S2, the correlation coefficient of voltage response event i triggering or stopping and power exchange of grid node p is calculated in the following manner: ； The correlation coefficient of power exchange of grid node p and voltage change of grid node q is calculated in the following manner: ； The correlation coefficient of voltage change of grid node q and voltage response event j starting or maintaining is calculated in the following manner: ； The correlation coefficient of voltage response event j starting or maintaining and triggering or stopping is calculated in the following manner: ； wherein E p represents the set of voltage response events of the power electronics device of node p; a ip = 0 indicates that the voltage response event i is not associated with the grid node p power; a ip = 1 indicates that the voltage response event i is directly associated with the grid node p power; AU q is the voltage variation of the grid node q; AP p , AQ p are the active and reactive power exchange variation of the power electronics device and the grid node p, respectively; k P and k Q are the weights of the active voltage sensitivity and the reactive voltage sensitivity, respectively; c is the correlation coefficient bias factor; and t j is the remaining action time of the voltage response event j, which is the difference between the event action time and the current time.

3. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 1, characterized by, In step S3, the voltage response event-grid multilayer coupling network is established in the following manner: ； Wherein, N={n1, n2, …, n k} represents the set of power grid nodes accessed by power electronic devices, k is the number of sets of power grid nodes accessed by power electronic devices; E={e1, e2, …, e m-1 , m} represents a set of voltage response events, m is the number of voltage response events; W X represents a weight value matrix of the association relationship of the multi-layer coupled network; The voltage response event-power grid multilayer coupling network includes four single-layer association networks of a power grid node association network, a voltage response event association network, a voltage response event and power grid node directed association network, and a power grid node and voltage response event directed association network, C X is a voltage response event-power grid multilayer association relationship set, and C X =C NN ∪C EE ∪C NE ∪C EN , wherein C NN is an association relationship set of the power grid node association network L NN , C EE is an association relationship set of the voltage response event association network L EE , C EN is an association relationship set of the voltage response event and power grid node directed association network L EN , and C NE is an association relationship set of the power grid node and voltage response event directed association network L NE is determined in the following manner, respectively: Network L of grid node associations NN Set C of association relationships NN is a set of all N(p)→N(q) association relationships, where N(p)→N(q) represents an association relationship between grid node p and grid node q, and N(p) and N(q) represent grid node p and grid node q, respectively. Voltage response event correlation network L EE Correlation set C EE is a set of all E(i)→E(j) correlation relations, where E(i)→E(j) represents a correlation relation between voltage response event i and voltage response event j, and E(i) and E(j) represent voltage response event i and voltage response event j, respectively; Voltage response event and grid node directed association network L EN Set of association relations C EN is a set of all E(i)→N(p) association relations, where E(i)→N(p) represents a directed association relation between voltage response event i and grid node p; A set of associations C between grid nodes and voltage response events NE A set of associations C between grid nodes and voltage response events NE N(q)→E(j) for all N(q)→E(j) associations, where N(q)→E(j) denotes a directed association between grid node q and voltage response event j.

4. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 3, characterized by, The power grid node association network , voltage response event association network , voltage response event and power grid node directed association network , power grid node and voltage response event directed association network are respectively represented as: ； ； ； ； In the formula, W NN is the weight value matrix of the correlation between the nodes of the power grid; W EE is the weight value matrix of the unknown correlation of the voltage response event W EN is a weight value matrix of the voltage response events and the grid nodes having a directional correlation relationship; W NE is a weight value matrix of the directed association relationship between the grid nodes and the voltage response events; The multi-layer coupling network association relationship weight value matrix W X is composed of W NN , W EN , W NE , W EE , and is expressed as: 。 5. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 4, characterized by, a weight value matrix W of the association relationship between the grid nodes NN a weight value matrix W of the directed association relationship between the voltage response events and the grid nodes EN a weight value matrix W of the directed association relationship between the grid nodes and the voltage response events NE a weight value matrix W of the unknown association relationship of the voltage response events EE The weight values in the above weight value matrix W are determined in the following manner respectively: ； ； ； ； wherein, represents the weight value of the association relationship between the grid node p and the grid node q, is the correlation coefficient of the power exchange of the grid node p and the voltage change of the grid node q; represents the weight value of the directed association relationship between the voltage response event i and the grid node p, is the correlation coefficient of the triggering or suspension of the voltage response event i and the power exchange of the grid node p; represents the weight value of the directed association relationship between the grid node q and the voltage response event j, is the correlation coefficient of the voltage change of the grid node q and the starting or maintaining of the voltage response event j; is the correlation coefficient of the starting or maintaining and the triggering or suspension of the voltage response event j; represents the weight value vector of the unknown association relationship between the voltage response event i and other voltage response events in the power system; the jth element in represents the weight value of the association relationship between the voltage response event j and the voltage response event i after the occurrence of the voltage response event i, which is an unknown quantity and needs to be determined by solving.

6. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 1, characterized by, In step S5, the way of performing walk search and solving the random walk model based on the meta-path of the voltage response event-grid multilayer coupling network is as follows: The action state is the starting voltage response event of the random walk of particles, and the random walk model of the voltage response event-power grid multi-layer coupling network meta path is used according to the meta path e→n→n→e, the next level node of each walk is randomly selected by using the Markov chain Monte Carlo method, and the corresponding walk path weight is calculated, so that the current initial node v init The walk path weight of each different voltage response event as the end point v end of the walk path is obtained, which is the weight value of the correlation between the voltage response event node of the current initial node v init and other voltage response events, and the order of occurrence of each voltage response event in the power system is determined according to the order from large to small of the weight value of the correlation.

7. The power electronics-based power system voltage-dominance-type cascading failure path search method according to claim 6, characterized by, The step S5 is specifically as follows: S501, the action state of each voltage response event in the current power system determined according to step S1, to the action state Sta i is started, and the action time T Thi The shortest one is the initial node v of the random walk of the particle in the current walk path init ; S502, determining the correlation process of each voltage response event in the power system and the corresponding correlation coefficient according to the action state of each voltage response event in the current power system, and obtaining the correlation relationship weight value matrix of the voltage response event-grid multilayer coupling network at present; S503、with the current initial node v init As the starting point, according to the random walk model of the meta-path of the voltage response event-power grid multi-layer coupled network, the next level node of each walk is randomly selected by using the Markov chain Monte Carlo method according to the meta-path MP = e→n→n→e, and the corresponding walk path weight is calculated: ； wherein w p (v init ) is the walk path weight from the initial node v end to the voltage response event node as the walk path end node v init , is an unknown to be solved, which represents that from the initial node v end , the meta path e→n→n→e is walked and stopped at v init , and the walk path is w end →w init →w n1 →w n2 , and the walk path weight is w end ; v n1 and v n2 represent two power grid nodes passed in the walk path; represents the walk path weight from the initial node v init to the first power grid node v n1 in the meta path, represents the walk path weight from the first power grid node v n1 to the second power grid node v n2 in the meta path, represents the walk path weight from the second power grid node v n2 to the walk path end node v end in the meta path, and all are known quantities which can be determined by the correlation weight matrix of the voltage response event-power grid multi-layer coupled network. Thus, the current initial node v is obtained by traversing init The walk path weight of each different voltage response event as the end point v of the walk path end The weight value of the association relationship between the voltage response event node of the current initial node v init and other voltage response events; S504, judging whether the weight values of the correlation relationship between the voltage response event node as the current initial node v init and other voltage response events are all 0; If yes, it is determined that the walk search is ended; otherwise, one walk path end point v with the largest weight value of the association relationship is determined end The corresponding voltage response event is determined as a subsequent voltage response event S505, taking the next voltage response event as a new initial node, re-acquiring the current working parameters of the power electronic device in the power electronic power system, re-determining the action state of each voltage response event in the current power system, and jumping back to step S502; S506, cyclically executing steps S502 to S505 until the end of the wandering search, and determining the order of occurrence of each voltage response event in the power system.

Images