Novel power system stability pre-decision dynamic disturbance set construction method and system

Abstract

The invention discloses a novel power system stability pre-decision dynamic disturbance set construction method and system, and the method comprises the steps: carrying out the visualization of the connection conditions of internal nodes and branches of a power grid, and obtaining a power grid topology adjacency matrix; screening the power grid topology adjacency matrix according to a preset resource equipment table and a preset fault equipment type to obtain an anticipated equipment set; according to the load rate and power flow data of each power supply line in the anticipated equipment set, a preset branch weight and a preset entropy model, constructing a double-entropy index Sudoku; and according to a preset power grid fault behavior and power grid historical operation data, dynamically updating node information in the double-entropy index Sudoku to obtain a dynamic disturbance set. According to the method, the dynamic disturbance set capable of comprehensively covering the fault type of the power system is constructed, and reliable data support is provided for risk assessment of the power system.

Description

Technical Field

[0001] The present application belongs to the technical field of power grid risk assessment, and specifically relates to a novel method and system for constructing a dynamic disturbance set for pre-decision making on power system stability. Background Art

[0002] Studying the safety and stability risks of power systems requires checking the system response by applying disturbance types of different natures to the system based on a certain operating state of the system. When faced with large-scale systems and diverse disturbances, it is difficult to analyze all possible scenarios one by one due to the constraints of computing resources and computing efficiency. Therefore, it is necessary to construct the corresponding expected disturbance set for specific problems to analyze and test the system operation. This disturbance set can provide scenario conditions for the formulation of safety and stability pre-decision-making for the power system. In terms of power grid security analysis, the methods for generating disturbance sets are mostly based on experience and rely on the structural connection characteristics of the power grid. Therefore, when reflecting the problems existing in the operation of the power grid, it is necessary to exhaustively enumerate massive fault items to increase the fault types covered by the disturbance set, thereby increasing the computational burden. In this regard, how to screen the disturbance set of new power systems and construct a dynamic disturbance set that can efficiently evaluate the reliability of new power systems is a major problem that needs to be solved urgently.

[0003] The existing power grid security assessment uses static disturbance sets. When the system operating state changes, the disturbance set cannot be changed. The historical calculation results of the disturbance set are not used, and the fault types covered by the disturbance set are not comprehensive. The system cannot perform online reliability assessment based on the disturbance set. Summary of the invention

[0004] This application proposes a novel method and system for constructing a dynamic disturbance set for power system stability pre-decision, constructing a dynamic disturbance set that can comprehensively cover the types of power system faults, and providing reliable data support for risk assessment of the power system.

[0005] The first aspect of the present application provides a novel method for constructing a dynamic disturbance set for power system stability pre-decision making, the method comprising:

[0006] Visualize the internal nodes and branch connections of the power grid to obtain the power grid topology adjacency matrix;

[0007] According to the preset resource equipment table and the preset fault equipment type, the grid topology adjacency matrix is ​​screened to obtain the expected equipment set;

[0008] Constructing a double entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model;

[0009] According to the preset power grid fault behavior and the historical power grid operation data, the node information in the double entropy index nine-square grid is dynamically updated to obtain a dynamic disturbance set.

[0010] The above scheme first visualizes the physical structure of the power grid, constructs a power grid topology adjacency matrix that describes the connection between nodes and branches inside the power grid, then connects the corresponding resource information for the nodes in the power grid topology adjacency matrix, and screens the nodes by the type of faulty equipment, so as to screen the effective equipment from the power grid structure and obtain the expected equipment set; then, the power supply lines of the expected equipment are weighted according to the preset branch weights, and then, combined with the entropy model, a double entropy index nine-square grid that can describe the probability of disturbance and anti-disturbance capability of each power supply line is constructed, which can completely cover the types of power system faults. Finally, when a fault occurs in the power grid, the node information in the double entropy index nine-square grid is updated with the historical operation data of the power grid as prior knowledge, so as to dynamically update the disturbance set according to the changes in the system operation status, and provide reliable data support for the risk assessment of the power system.

[0011] In a possible implementation method of the first aspect, the power grid topology adjacency matrix is ​​screened according to a preset resource device table and a preset fault device type to obtain an expected device set, specifically:

[0012] Establishing corresponding resource codes for the diagonal nodes in the power grid topology adjacency matrix and the branches corresponding to the nodes, and generating the resource index table according to the resource codes;

[0013] According to the resource index table, the nodes of the power grid topology adjacency matrix are arranged in a preset order from both row and column directions to obtain an original adjacency matrix;

[0014] Dividing the resource index table according to two types of line connection branches and transformer connection branches to obtain a first branch type set and a second branch type set;

[0015] According to the resource coding, the first branch type set and the second branch type set are substituted into the original adjacency matrix to obtain the expected device set.

[0016] The above scheme first uniformly encodes the nodes in the grid topology adjacency matrix to ensure that the device information corresponding to the node can be accessed through a unique code. Then the nodes of the grid topology adjacency matrix are arranged through the resource index table. The original adjacency matrix obtained after sorting can represent the relationship between nodes, and the resource index table is divided into two types: line connection branches and transformer connection branches, to obtain two subsets. The original adjacency matrix is ​​processed through these two subsets to complete the screening of the original adjacency matrix in the grid structure, and further screen out the nodes that will cause disturbances.

[0017] In a possible implementation method of the first aspect, the envisioned device set is specifically:

[0018] The expression of the expected device set is specifically:

[0019]

[0020] Where M(V) is the expected equipment set, <l i |t j > is the combination of the first branch type set and the second branch type, v n The node code for the grid node.

[0021] In a possible implementation method of the first aspect, a double entropy index nine-square grid is constructed according to the load rate and flow data of each power supply line in the envisioned device set, a preset branch weight, and a preset entropy model, specifically:

[0022] Constructing the entropy model according to all operating states of the power grid and the probability of occurrence of each operating state;

[0023] Obtaining power data of each power supply line in the envisioned equipment set according to the power flow data of each power supply line;

[0024] According to the load rate of each power supply line in the expected equipment concentration, the power data of the power supply line and the entropy model, a power grid power flow entropy index is constructed when a disturbance shock occurs to the power grid;

[0025] According to the branch weight, the power data of the power supply line and the entropy model, construct a branch power flow disconnection entropy index when a disturbance shock occurs to the power grid;

[0026] The double entropy index nine-square grid is constructed according to the power grid power flow entropy index and the branch power flow breaking entropy index.

[0027] In a possible implementation method of the first aspect, a power flow entropy index of a power grid when a disturbance shock occurs to the power grid is constructed according to the load rate of each power supply line in the expected equipment set, the power data of the power supply line, and the entropy model, specifically:

[0028] According to a preset load rate interval and a load rate of each power supply line in the power grid, the power supply lines are divided to obtain a set of power supply lines corresponding to each load rate interval;

[0029] Obtaining a branch probability in each load rate interval according to the power supply line set and the load rate of the power supply line;

[0030] According to the branch probability of each load rate interval and the entropy model, a power grid power flow entropy index is constructed when a disturbance shock occurs to the power grid.

[0031] The above scheme first divides the power supply lines into a set of power supply lines corresponding to different load rates according to the load rate, and then obtains the branch probability to describe the proportion of the power supply lines included in each load rate interval in the power grid, so as to obtain the power grid flow entropy index by combining the entropy model later, and provide data support for calculating the probability of disturbance occurring in each power supply line.

[0032] In a possible implementation method of the first aspect, a branch power flow disconnection entropy index when a disturbance shock occurs to the power grid is constructed according to the branch weight, the power data of the power supply line and the entropy model, specifically:

[0033] Obtaining a power grid structure-operation characteristic index according to the branch weight, the power data of the power supply line and the power grid topology adjacency matrix;

[0034] By extracting the maximum and minimum phase angles of all nodes in the envisioned device set, a longest power transmission distance index is constructed;

[0035] The total active power transmission ratio is obtained by combining the grid structure-operation characteristic index and the longest power transmission distance index;

[0036] According to the power flow changes and power transmission distance changes before and after the disturbance shock of the power grid, the power flow relative transfer rate index is obtained based on the total active power transmission ratio;

[0037] According to the entropy model and the power flow relative transfer rate index, a branch power flow breaking entropy index is obtained.

[0038] The above scheme first assigns weights to each power supply line through branch weights, and combines the branch weights and the size of the active power transmitted on each power supply line to obtain the power grid structure-operation characteristic index; based on the power grid structure-operation characteristic index and the longest power transmission distance index, a total active power transmission ratio is constructed, and the total active power transmission ratio takes into account the power flow changes and power transmission distance changes before and after the power grid is disturbed, to obtain the branch power flow breaking entropy index that can reflect the anti-disturbance ability of the power supply line.

[0039] In a possible implementation method of the first aspect, the dual entropy index nine-square grid is constructed according to the power grid power flow entropy index and the branch power flow disconnection entropy index, specifically:

[0040] Sorting the power flow entropy indexes of the power grid to obtain a maximum power flow entropy value and a minimum power flow entropy value;

[0041] The branch power flow breaking entropy indexes are sorted by size to obtain a maximum branch power flow breaking entropy value and a minimum branch power flow breaking entropy value;

[0042] The coordinate line is drawn with the minimum power flow entropy value and the minimum branch power flow breaking entropy value as the origin, the maximum power flow entropy value as the end point of the vertical axis and the maximum branch power flow breaking entropy value as the end point of the horizontal axis;

[0043] The vertical axis range and the horizontal axis range are divided into three equal parts to obtain the nine-square grid of the dual entropy index;

[0044] According to a preset entropy value range, the dual entropy index nine-square grid is divided into a first threshold number of areas.

[0045] The above scheme uses the power grid flow entropy index as the vertical axis direction and the branch flow disconnection entropy index as the horizontal axis direction, and divides the vertical axis range and the horizontal axis range into three equal parts to construct a double entropy index nine-square grid. The power grid flow entropy index corresponding to the power supply line and the size of the branch flow disconnection entropy index are realized, and the probability of disturbance and anti-disturbance capability of each power supply line are measured, providing reliable data support for risk assessment of the power system.

[0046] In a possible implementation method of the first aspect, according to preset power grid fault behavior and power grid historical operation data, the node information in the dual entropy indicator nine-square grid is dynamically updated to obtain a dynamic disturbance set, specifically:

[0047] Based on the expected device set, the historical power flow data of each node in the power grid is obtained to construct the deep learning features of the original topology graph;

[0048] According to the preset power grid fault behavior, respectively obtain a first topology graph deep learning feature after the power grid fault behavior occurs, and a second topology graph deep learning feature after the faulty device corresponding to the fault behavior is removed;

[0049] The original topology map deep learning features, the first topology map deep learning features and the second topology map deep learning features are sequentially subjected to feature splicing, feature extraction and feature supplementation to obtain system map features after a fault behavior occurs in the power grid;

[0050] The node information in the expected device set is updated according to the characteristics of the system diagram, so as to update the node information in the double entropy index nine-square grid and obtain a dynamic disturbance set.

[0051] The above scheme uses the historical flow data of each node in the power grid as prior knowledge to construct the deep learning features of the original topology map; based on the deep learning features of the original topology map, the grid topology changes after the node failure are predicted, and the node information of the expected equipment set is updated to complete the update of the status of each node in the power grid. The double entropy indicator grid is updated with the updated node information to timely detect the faulty nodes and perform maintenance.

[0052] In a possible implementation method of the first aspect, the original topology map deep learning features, the first topology map deep learning features, and the second topology map deep learning features are sequentially subjected to feature splicing, feature extraction, and feature supplementation to obtain system map features after a fault behavior of the power grid occurs, specifically:

[0053] Perform feature splicing on the original topology map deep learning feature, the first topology map deep learning feature and the second topology map deep learning feature to obtain a system topology dynamic adjustment feature;

[0054] Differentiating and extracting nodes in the system topology dynamic adjustment feature according to a preset power grid fault behavior to obtain local topology difference information;

[0055] Introducing a preset learning coefficient to supplement the local topology difference information to obtain global topology difference information;

[0056] The system topology dynamic adjustment feature, the local topology difference information and the global topology difference information are spliced ​​to obtain the system graph feature.

[0057] The second aspect of the present application provides a novel power system stability pre-decision dynamic disturbance set construction system, the system comprising: a power grid topology visualization module, a pre-concept equipment set construction module, a dual entropy index nine-square grid construction module and a dynamic disturbance set generation module;

[0058] The power grid topology visualization module is used to visualize the internal nodes and branch connections of the power grid to obtain a power grid topology adjacency matrix;

[0059] The expected device set construction module is used to screen the power grid topology adjacency matrix according to the preset resource device table and the preset fault device type to obtain the expected device set;

[0060] The dual entropy index nine-square grid construction module is used to construct a dual entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model;

[0061] The dynamic disturbance set generation module is used to dynamically update the node information in the double entropy index nine-square grid according to the preset power grid fault behavior and the power grid historical operation data to obtain the dynamic disturbance set. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the implementation manner will be briefly introduced below. Obviously, the drawings described below are only some implementation manners of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0063] Figure 1 It is a specific flow chart of a novel method for constructing a dynamic disturbance set for power system stability pre-decision-making provided by a certain embodiment of the present application;

[0064] Figure 2 It is a schematic diagram of a double entropy index nine-square grid of a novel method for constructing a dynamic disturbance set for power system stability pre-decision-making provided by a certain embodiment of the present application;

[0065] Figure 3 A structural diagram of a new type of power system stability pre-decision dynamic disturbance set construction system provided by a certain embodiment of the present application. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0067] It should be understood that the step numbers used in this article are only for the convenience of description and are not intended to limit the order in which the steps are executed.

[0068] First embodiment

[0069] Nowadays, safety and stability pre-decision is often used to protect the power system from fault disturbance or malicious attack in order to maintain safe operation. The full name of safety and stability pre-decision is safety and stability control pre-decision. The corresponding disturbance set provides scenario conditions for the formulation of safety and stability pre-decision of the power system, so as to analyze and test the operation of the power system. However, the existing methods of constructing disturbance sets are mostly based on experience and rely on the structural connection characteristics of the power grid. Therefore, in order to reflect the problems existing in the operation of the power grid, it is necessary to exhaustively enumerate a large number of fault items to increase the types of faults covered by the fault set, thereby increasing the calculation burden. Moreover, the disturbance set obtained by the exhaustive method is difficult to find the dominant hazard fault items, which reduces the fault coverage and cannot provide reliable data support for the risk assessment of the power system.

[0070] like Figure 1 As shown, Figure 1 A specific flow chart of a novel method for constructing a dynamic disturbance set for pre-decision-making of a power system stability is provided for a certain embodiment of the present application. The novel method for constructing a dynamic disturbance set for pre-decision-making of a power system stability in this embodiment includes steps S1 to S4, which are described in detail as follows:

[0071] Step S1, visualizing the internal nodes and branch connections of the power grid to obtain a power grid topology adjacency matrix.

[0072] In the embodiment of the present application, a power grid topology adjacency matrix is ​​first constructed based on graph theory and the physical structure of the power grid to visualize the internal nodes and branch connections of the power grid. The power grid in the embodiment of the present application is based on a new type of power system.

[0073] Among them, the physical structure of the power grid is represented by graph G, and the specific expression is:

[0074] G = (V, E);

[0075] In the formula, V is the node set of graph G, E is the edge set of graph G, including node pairs (i, j), i and j are node numbers in set V. The edges of graph G can be formed by combining node numbers in pairs.

[0076] The busbars of the power grid are numbered to form the node set V of the graph G. The specific expression is:

[0077] V(G)=(v1,v2,…,v n )v∈[1 2...n];

[0078] In the formula, v1 to v n Number the node.

[0079] After determining the node set, the power grid topology adjacency matrix A(G) is constructed according to the node set and the edge set. The specific expression is:

[0080]

[0081] In the formula, a ij is the connection characteristic of the node; when nodes i and j are adjacent, a in the matrix ij =1; when nodes i and j are not adjacent or are represented as the same node, a ij = 0. The directions between nodes can be distinguished by positive and negative signs, and the positive direction can be defined by yourself.

[0082] Step S2, screening the grid topology adjacency matrix according to the preset resource equipment table and the preset fault equipment type to obtain the expected equipment set.

[0083] In an embodiment of the present application, resource coding is constructed for the diagonal nodes of the power grid topology adjacency matrix and the branches corresponding to the nodes, that is, the device information corresponding to these nodes is uniformly encoded to ensure that the device resources in the power grid can be accessed through a unique code.

[0084] Then these resource codes are integrated to construct the resource index table I, the specific expression is:

[0085] I = {1, 2, ..., N};

[0086] Where N is the resource code.

[0087] The resource index table mainly records the sufficient and necessary node information participating in the power grid calculation, and can exist in the form of database, XML, etc. Each resource code corresponds to one type of sufficient and necessary node information participating in the power grid calculation.

[0088] Then, based on the node set V, the nodes of the power grid topology adjacency matrix are arranged in a preset order from both the row and column directions to obtain the original adjacency matrix. The original adjacency matrix O(G) can be used to represent the association relationship between nodes, and its element a ij It is only stored in the upper or lower triangular matrix. The specific expression is:

[0089]

[0090] For example, in the embodiment of the present application, the node sorting is to sort the nodes from node number 1 to the right in the horizontal direction, and to sort the nodes from node number 1 to the bottom in the vertical direction. For example, 12 Represents the element at row 1 and column 2 in the original adjacency matrix.

[0091] Then, the device corresponding to the resource code in the resource index table is associated with the node in the original adjacency matrix to form an original device screening matrix. The corresponding device information can be accessed through the resource index table through the node sequence number in the original device screening matrix.

[0092] In order to complete the disturbance analysis from the perspective of the power grid structure through the original equipment screening matrix, the resource index table is first classified. The resource index table is divided into two types: line connection branches and transformer connection branches, and the first branch type set L(G) and the second branch type set T(G) are obtained. The specific expression is:

[0093] L(G)={l1 l2...l i};

[0094] T(G)={t1 t2...t j};

[0095] In the formula, l i is the power supply line number, t j is the transformer number. L(G) and T(G) are both subsets of the resource index table.

[0096] Then inject the above node set V into the diagonal elements of the original device screening matrix, and substitute L(G) and T(G) to obtain the expected device set. The specific expression is:

[0097]

[0098] In the formula, <l i |t j > is a combination of elements in the first branch type set and the second branch type set according to the actual power grid topology.

[0099] Step S3, constructing a double entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model.

[0100] In the embodiment of the present application, the entropy model is first constructed based on all the operating states of the power grid and the probability of occurrence of each operating state. Entropy is a measure of the orderly and disordered state of the power grid. Due to the strong intermittent nature of the output of renewable energy in the new power system, the new power system has multiple possible operating states. A single operating state is represented by X. i Indicates that the probability of each running state occurring is P(X i ), the value range is 0 to 1, then the specific expression of the entropy model is:

[0101]

[0102] Where m is the total number of operating states of the power grid, C is the model adjustment constant, H is the entropy value, and i is the i-th operating state.

[0103] The entropy model is used to measure the uncertainty of each operating state, that is, the risk of each operating state; if the order of the power grid is higher, the entropy value is smaller, on the contrary, the lower the order of the power grid, the larger the entropy value. When the power grid is impacted by a disturbance, the operation develops from order to disorder. If the anti-disturbance ability of the power grid is strong, it will quickly recover to an ordered state, that is, a new normal operating state, after being disturbed; if the operation of the power grid cannot be controlled to develop towards a disordered trend through protective actions, safety automatic device protection decoupling, or manual intervention, that is, the entropy value continues to increase, then when the entropy value reaches a critical point, the power grid will collapse. Optionally, the embodiment of the present application determines the probability of each operating state of the power grid through the output data of the power grid.

[0104] Then, according to the power flow data of the power grid, the active power and reactive power of each power supply line of the expected equipment concentration are calculated to obtain the power data of the power supply line. The specific calculation formula is:

[0105]

[0106] Where P i is the active power injected into node i on the power supply line, Q i is the reactive power injected into node i on the power supply line, V i is the voltage amplitude of node i on the power supply line, δ i is the voltage phase angle of node i on the power supply line, Y ij , B ij is the position admittance corresponding to nodes i and j on the power supply line, and N is the total number of nodes on the power supply line.

[0107] Then, the load factor ε of each power supply line is calculated using the power data of the power supply line. N , the specific calculation formula is:

[0108]

[0109] Where P N is the rated power of the power supply line, P0 is the input power of the power supply line under no-load condition, P s It is the actual operating power of the power supply line.

[0110] The power consumed by the power supply line under rated load operation is ∑P N , which can be written as:

[0111]

[0112] Where ηN It is the power supply efficiency of the power supply line at rated power.

[0113] According to the load rate of each power supply line and the preset load rate range, the power supply lines are divided into multiple power supply line sets, each power supply line set corresponds to a load rate range, so that power supply lines with similar load rates are treated as a set for subsequent calculations, thereby improving calculation efficiency.

[0114] Then, the branch probability p(k) in different load rate intervals is further calculated by combining the power supply line set and the load rate of each power supply line. The specific calculation formula is:

[0115]

[0116] In the formula, l k is the number of power supply lines in the power supply line set of the kth load rate interval, and K is the total number of load rate intervals. The branch probability can measure the proportion of the power supply lines included in each load rate interval to the total power supply lines in the power grid.

[0117] Then, according to the calculated branch probability of the load rate interval, by combining the entropy model, the power grid flow entropy index when the power grid is disturbed is constructed. The specific expression is:

[0118] H k = -Cp(k)lnp(k);

[0119]

[0120] In the formula, H k is the power flow entropy value of the kth load rate interval, and H is the power flow entropy value of the entire power grid.

[0121] The probability of a fault occurring in each power supply line can be calculated by using the power grid power flow entropy index.

[0122] For constructing the branch flow disconnection entropy index when the power grid is disturbed, firstly, the corresponding power supply line is weighted according to the preset branch weight. Optionally, in the embodiment of the present application, the reactance value of each power supply line is used as the branch weight, and the specific expression is as follows:

[0123] W={w1,w2,…,w k ,…,w m};

[0124] In the formula, w is the branch weight, and its subscript represents the number of the corresponding power supply line. The power supply line can be weighted by the number. W is the branch weight set, and m is the total number of power supply lines.

[0125] Then, according to the weighted power supply lines, the active power of the power supply lines and the grid topology adjacency matrix, the grid structure-operation characteristic index D is obtained. k , the specific expression is:

[0126] D k =|w k p k |(k=1,2,…,m);

[0127] In the formula, p k w k The amount of active power transmitted on the corresponding power supply line.

[0128] Then, the maximum and minimum phase angles of all nodes in the power grid are extracted through the expected device set to construct the longest power transmission distance indicator D max , the specific expression is as follows:

[0129] θ min =min(θ1,θ2,…,θ n );

[0130] θ max =max(θ1,θ2,…,θ n );

[0131] D max =|θ max -θ min |;

[0132] In the formula, θ min is the minimum phase angle of all nodes, θ max is the maximum value of the phase angle of all nodes, and n is the total number of connected nodes.

[0133] D k and D max Combined, we can get the branch transmission rate δ k , the specific expression is as follows:

[0134]

[0135] Among them, the size of the branch transmission rate can reflect the transmission distance of the active power of the power grid from the source end to the receiving end and the smoothness of the active power transmission process.

[0136] Furthermore, the total active power transmission ratio δ is constructed by the branch transmission rate. The specific expression is as follows:

[0137]

[0138] Because the branch power flow breaking entropy index is constructed when the power grid is disturbed, it is necessary to consider the power flow changes and power transmission distance changes of the power supply line before and after the power grid disturbance. Take the two time states before and after the disturbance, that is, t0 and t1, to establish the power flow relative transfer rate index. The specific expression is as follows:

[0139] Δδ=δ t1 -δ t0 ;

[0140] In the formula, δ t1 is the total active power transmission ratio at time t1, and is δ t0 is the total active power transmission ratio at time t0, δ t1 >δ t0 .

[0141] By combining the relative power transfer rate index and the entropy model, the branch power flow disconnection entropy index R when the power grid is disturbed can be obtained. The specific expression is as follows:

[0142] R = -CΔδlnΔδ;

[0143] The branch power flow breaking entropy index is used to measure the risk of the power grid approaching a critical state after the branch is disconnected, that is, to reflect the risk that the power grid cannot operate stably after the branch is disconnected.

[0144] After obtaining the power flow entropy index and the branch power flow breaking entropy index, firstly calculate the power flow entropy index {H k} sort by size to obtain the maximum power flow entropy value H max and the minimum power flow entropy value H min Then, with the minimum power flow entropy value H min As the origin of the coordinate system, draw the vertical axis with the upward direction as the positive direction, and the maximum tidal current entropy value H max is the end point of the vertical axis. max To the minimum power flow entropy value H min The distance is the vertical axis range, and the vertical axis range is divided into three equal parts to obtain two vertical axis equal points. The expression of each vertical axis equal point is:

[0145] H settmg.1 =H min +Δh;

[0146] H settmg.2 =H min +2Δh;

[0147]

[0148] In the formula, H setting.1 , H setting.2 is the value of the vertical axis dividing point on the vertical axis, and Δh is the size of the vertical axis range after being divided into three equal parts.

[0149] Then the branch power flow breaking entropy index {R k} sort by size to obtain the maximum branch flow breaking entropy value R max and the minimum branch power flow breaking entropy value R min Then the minimum branch flow breaking entropy value R min As the coordinate origin, the coordinate origin and the minimum power flow entropy value H min Overlap, draw the horizontal axis with the right as the positive direction, and use the maximum branch flow breaking entropy value R max is the end point of the horizontal axis. The maximum branch flow breaking entropy value R max To the minimum branch power flow breaking entropy value R min The distance is the horizontal axis range, and the horizontal axis range is divided into three equal parts to obtain two horizontal axis equal points. The expression of each horizontal axis equal point is:

[0150] R setting.1 =R min +Δr;

[0151] R setting.2 =R min +2Δr;

[0152]

[0153] In the formula, R setting.1 , R setting.2 is the value of the horizontal axis at the point where the horizontal axis is equally divided, and Δr is the size after the vertical and horizontal axis range is divided into three equal parts.

[0154] Coordinate lines are drawn according to the equal points of the horizontal axis and the vertical axis to obtain a double entropy index nine-square grid, and then the double entropy index nine-square grid is divided into three areas, namely area A, area B and area C, according to a preset entropy value range.

[0155] The coordinate range of the area A is:

[0156]

[0157] The coordinate range of the area B is:

[0158] (R min ,H max )→(R setting.1 ,H max )→(R setting.1 ,H setting.2 )→(R min ,H sup2 )→(R min ,H max )

[0159] ∪(R setting.1,H setting.2 )→(R setting.2 ,H setting.2 )→(R setting.2 ,H setting.1 )→(R setting.1 ,H setting.1 )→(R setting.1 ,H setting.2 )

[0160] ∪(R setting.2 ,H setting.1 )→(R max ,H setting.1 )→(R max ,H min )→(R setting.2 ,H min )→(R setting.2 ,H setting.1 )

[0161] The coordinate range of the area C is:

[0162]

[0163] In order to better display the double entropy index nine-square grid after regional division, Figure 2 A schematic diagram of the double entropy index nine-square grid is provided. There are three areas in the figure, the darker one is area A, the lighter one is area C, and the white one is area B. In the preset entropy value range, the larger the entropy value, the worse the anti-disturbance ability of the range, so area A represents the equipment branch with a higher probability of disturbance in the power grid; area B represents the equipment branch with a moderate probability of disturbance in the power grid; area C represents the equipment branch with a lower probability of disturbance in the power grid, and the equipment branch has a strong anti-disturbance ability.

[0164] Optionally, in order to reduce the amount of calculation, since the device branch represented by area C has strong anti-interference ability and is not prone to fault disturbance, the device branch represented by area C can be ignored from the expected device set when assessing the power grid risk.

[0165] Through the dual entropy indicator grid, it is possible to identify which equipment branches are prone to fault interference in the power grid and need to be focused on during risk assessment, thus achieving a one-time screening of the dynamic disturbance set in terms of the equipment's anti-interference ability.

[0166] Step S4, dynamically updating the node information in the dual entropy index nine-square grid according to the preset power grid fault behavior and the historical power grid operation data, to obtain a dynamic disturbance set.

[0167] In the embodiment of the present application, based on the expected device set constructed in step S1, the historical flow data of each node in the power grid is sorted to construct the deep learning features of the original topology graph, and the specific expression is:

[0168]

[0169] In the formula, V 0-,i ,θ 0-,i are the active power and reactive power of the load at node i in the power grid, the active power and reactive power of the distributed generation, the voltage amplitude and voltage phase angle at node i; N is the total number of nodes in the power grid, and 0- represents the moment before the node fault disturbance occurs.

[0170] The preset grid fault behavior is implemented on the grid simulation, the changed grid topology is recorded, and the equipment that has experienced fault disturbance, needs to be shut down or repaired, and the corresponding fault location and fault clearing location are obtained. The nodes of the expected equipment set are marked according to the fault location and fault clearing location to complete the node status update, and then the instantaneous power flow after the grid fault is estimated. The deep learning features of the first topology after the fault are constructed based on the updated expected equipment set. The specific expression is:

[0171]

[0172] Where, 0 + is the time before the node failure disturbance occurs.

[0173] Then clear the faulty device and update the expected device set again to obtain the node corresponding to the cleared faulty device, and replace the power of the node with the difference of the steady-state power before and after the fault. The specific calculation formula is:

[0174]

[0175] Where P (imp) and Q (imp) are the differences of steady-state active power and reactive power before and after the fault, Y c+ is the admittance matrix of the power grid after the fault is cleared, is the node voltage vector before the fault.

[0176] According to the node power after the fault is cleared and the expected device set after the second update, the deep learning feature X of the second topology graph can be obtained: c+ , the specific expression is:

[0177]

[0178] The system topology dynamic adjustment feature X can be obtained by combining the original topology map deep learning feature, the first topology map deep learning feature and the second topology map deep learning feature. D , the specific expression is:

[0179] X D =X 0- ||X 0+ ||||X c+ ;

[0180] In the formula, || means that the vectors in the feature are concatenated row by row.

[0181] In the second step, the node features corresponding to the system topology dynamic adjustment features are extracted according to the preset power grid fault behavior. Specifically, the system topology dynamic adjustment features and the preset graph module function are used to integrate the local topology difference information of the nodes before and after the topology change. L , the specific expression is:

[0182]

[0183] In the formula, A 0- , A c+ are respectively the grid topology adjacency matrices before and after the fault according to the branch weight assignment, and f GNN is a preset graph module function that can convert X D The information is mapped into the power grid topology adjacency matrix, and the specific operation process is similar to associating the resource index table with the nodes in the original adjacency matrix.

[0184] The third step is to introduce the preset learning coefficient to supplement the local topological difference information and obtain the global topological difference information. The specific expression is:

[0185]

[0186] Where W Q ,W K ,W V is the learning coefficient, f Norm It is a SoftMin function, which can assign a higher probability to a smaller value when providing a probability distribution. In the embodiment of the present application, it is used to extract the maximum difference information of the nodes of the power grid topology change before and after the fault occurs.

[0187] Finally, the system topology dynamic adjustment features, local topology difference information and global topology difference information are spliced ​​to obtain the system graph feature X H , the specific expression is:

[0188] X H =X D ||XL ||X G ;

[0189] Furthermore, the system diagram feature is actually the system diagram feature after a power grid failure, which contains information about all nodes in the power grid. Through the system diagram feature, the changes in the power grid node information after the preset fault behavior occurs can be obtained, and the resource index table and the expected equipment set are updated accordingly according to the changes in the node information, and the nodes corresponding to the equipment that will be shut down and to be repaired are deleted to complete the dynamic update of the information of each node in the power grid. Then, according to the updated expected equipment set, the node information in the double entropy index nine-square grid is dynamically updated, and the double entropy index nine-square grid is reconstructed through the effective power supply lines after the fault, and finally a dynamic disturbance set that can fully cover the fault types of the power system is obtained.

[0190] Implementing the embodiments of the present application has the following beneficial effects:

[0191] The embodiment of the present application first visualizes the physical structure of the power grid, constructs a power grid topology adjacency matrix that describes the connection between nodes and branches inside the power grid, then connects the corresponding resource information for the nodes in the power grid topology adjacency matrix, and screens the nodes by the type of faulty equipment, thereby screening the effective equipment from the power grid structure to obtain the expected equipment set; then, the power supply lines of the expected equipment concentration are weighted according to the preset branch weights, and then combined with the entropy model, a double entropy index nine-square grid that can describe the disturbance probability and anti-disturbance capability of each power supply line is constructed, which can not only completely cover the types of power system faults, but also screen out the equipment nodes with weak anti-disturbance capabilities for key attention during risk assessment. Finally, when a power grid fails, the node information in the double entropy index nine-square grid is updated with the historical operation data of the power grid as prior knowledge, so as to dynamically update the disturbance set according to the changes in the system operation status, delete the equipment that is out of repair to reduce the amount of calculation, and provide reliable data support for the risk assessment of the power system.

[0192] Second embodiment

[0193] Furthermore, in order to implement the novel power system stability pre-decision dynamic disturbance set construction system corresponding to the above method embodiment to achieve the corresponding functions and technical effects, Figure 3 A structural diagram of a novel power system stability pre-decision dynamic disturbance set construction system is provided. For ease of explanation, only the parts related to this embodiment are shown. The novel power system stability pre-decision dynamic disturbance set construction system provided by the embodiment of the present application includes:

[0194] The power grid topology visualization module 201 is used to visualize the internal nodes and branch connections of the power grid to obtain a power grid topology adjacency matrix.

[0195] In the embodiment of the present application, a power grid topology adjacency matrix is ​​first constructed based on graph theory and the physical structure of the power grid to visualize the internal nodes and branch connections of the power grid. The power grid in the embodiment of the present application is based on a new type of power system.

[0196] Among them, the physical structure of the power grid is represented by graph G, and the specific expression is:

[0197] G = (V, E);

[0198] Where V is the node set of graph G, and E is the edge set of graph G, which contains node pairs (i, j). i and j are node numbers in set V. The edges of graph G can be formed by combining node numbers in pairs.

[0199] The busbars of the power grid are numbered to form the node set V of the graph G. The specific expression is:

[0200] V(G)=(v1,v2,…,v n )v∈[1 2...n];

[0201] In the formula, v1 to v n Number the node.

[0202] After determining the node set, the power grid topology adjacency matrix A(G) is constructed according to the node set and the edge set. The specific expression is:

[0203]

[0204] In the formula, a ij is the connection characteristic of the node; when nodes i and j are adjacent, a in the matrix ij =1; when nodes i and j are not adjacent or are represented as the same node, a ij = 0. The directions between nodes can be distinguished by positive and negative signs, and the positive direction can be defined by oneself.

[0205] The expected device set building module 202 is used to screen the power grid topology adjacency matrix according to the preset resource device table and the preset fault device type to obtain the expected device set.

[0206] In an embodiment of the present application, corresponding resource codes are established for the diagonal nodes in the power grid topology adjacency matrix and the branches corresponding to the nodes, and the resource index table is generated according to the resource codes; according to the resource index table, the nodes of the power grid topology adjacency matrix are arranged in a preset order from both row and column directions to obtain an original adjacency matrix; the resource index table is divided into two types, namely, line connection branches and transformer connection branches, to obtain a first branch type set and a second branch type set; according to the resource code, the first branch type set and the second branch type set are substituted into the original adjacency matrix to obtain an expected device set.

[0207] The dual entropy index nine-square grid construction module 203 is used to construct a dual entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model.

[0208] In an embodiment of the present application, the entropy model is constructed based on all operating states of the power grid and the probability of occurrence of each operating state; the power data of the power supply line is obtained based on the flow data of each power supply line in the expected equipment set; the power grid flow entropy index when the power grid is subjected to a disturbance shock is constructed based on the load rate of each power supply line in the expected equipment set, the power data of the power supply line and the entropy model; the branch flow switching entropy index when the power grid is subjected to a disturbance shock is constructed based on the branch weight, the power data of the power supply line and the entropy model; the dual entropy index nine-square grid is constructed based on the power grid flow entropy index and the branch flow switching entropy index.

[0209] The dynamic disturbance set generation module 204 is used to dynamically update the node information in the double entropy index nine-square grid according to the preset power grid fault behavior and the power grid historical operation data to obtain the dynamic disturbance set.

[0210] In an embodiment of the present application, based on the expected equipment set, the historical flow data of each node in the power grid is obtained to construct the original topology map deep learning features; according to the preset power grid fault behavior, the first topology map deep learning features after the power grid occurs the fault behavior, and the second topology map deep learning features after the fault equipment corresponding to the fault behavior is cleared are obtained respectively; the original topology map deep learning features, the first topology map deep learning features and the second topology map deep learning features are sequentially subjected to feature splicing, feature extraction and feature supplementation to obtain the system map features after the power grid fails; according to the system map features, the node information in the expected equipment set is updated to realize the updating of the node information in the dual entropy indicator nine-square grid to obtain the dynamic disturbance set.

[0211] In some embodiments, the envisioned device set building module 202 further includes:

[0212] Resource coding is constructed for the diagonal nodes of the power grid topology adjacency matrix and the branches corresponding to the nodes, that is, the device information corresponding to these nodes is uniformly coded to ensure that the device resources in the power grid can be accessed through a unique code.

[0213] Then these resource codes are integrated to construct the resource index table I, the specific expression is:

[0214] I = {1, 2, ..., N};

[0215] Where N is the resource code.

[0216] The resource index table mainly records the sufficient and necessary node information participating in the power grid calculation, and can exist in the form of database, XML, etc. Each resource code corresponds to one type of sufficient and necessary node information participating in the power grid calculation.

[0217] Then, based on the node set V, the nodes of the power grid topology adjacency matrix are arranged in a preset order from both the row and column directions to obtain the original adjacency matrix. The original adjacency matrix O(G) can be used to represent the association relationship between nodes, and its element a ij It is only stored in the upper or lower triangular matrix. The specific expression is:

[0218]

[0219] For example, in the embodiment of the present application, the node sorting is to sort the nodes from node number 1 to the right in the horizontal direction, and to sort the nodes from node number 1 to the bottom in the vertical direction. For example, 12 Represents the element at row 1 and column 2 in the original adjacency matrix.

[0220] Then, the device corresponding to the resource code in the resource index table is associated with the node in the original adjacency matrix to form an original device screening matrix. The corresponding device information can be accessed through the resource index table through the node sequence number in the original device screening matrix.

[0221] In order to complete the disturbance analysis from the perspective of the power grid structure through the original equipment screening matrix, the resource index table is first classified. The resource index table is divided into two types: line connection branches and transformer connection branches, and the first branch type set L(G) and the second branch type set T(G) are obtained. The specific expression is:

[0222] L(G)={l1 l2...l i};

[0223] T(G)={t1 t2...t j};

[0224] In the formula, l i is the power supply line number, t j is the transformer number. L(G) and T(G) are both subsets of the resource index table.

[0225] Then inject the above node set V into the diagonal elements of the original device screening matrix, and substitute L(G) and T(G) to obtain the expected device set. The specific expression is:

[0226]

[0227] In the formula, <l i |t j > is a combination of elements in the first branch type set and the second branch type set according to the actual power grid topology.

[0228] In some embodiments, the dual entropy index nine-square grid construction module 203 further includes:

[0229] First, the entropy model is constructed based on all the operating states of the power grid and the probability of occurrence of each operating state. Entropy is a measure of the orderly and disordered state of the power grid. Due to the strong intermittent nature of the output of renewable energy in the new power system, the new power system has multiple possible operating states. A single operating state is represented by X. i Indicates that the probability of each running state occurring is P(X i ), the value range is 0 to 1, then the specific expression of the entropy model is:

[0230]

[0231] Where m is the total number of operating states of the power grid, C is the model adjustment constant, H is the entropy value, and i is the i-th operating state.

[0232] The entropy model is used to measure the uncertainty of each operating state, that is, the risk of each operating state. Optionally, the embodiment of the present application determines the probability of each operating state of the power grid through the output data of the power grid.

[0233] Then, according to the power flow data of the power grid, the active power and reactive power of each power supply line of the expected equipment concentration are calculated to obtain the power data of the power supply line. The specific calculation formula is:

[0234]

[0235] Where P i is the active power injected into node i on the power supply line, Q i is the reactive power injected into node i on the power supply line, V iis the voltage amplitude of node i on the power supply line, δ i is the voltage phase angle of node i on the power supply line, Y ij , B ij is the position admittance corresponding to nodes i and j on the power supply line, and N is the total number of nodes on the power supply line.

[0236] Then, the load factor ε of each power supply line is calculated using the power data of the power supply line. N , the specific calculation formula is:

[0237]

[0238] Where P N is the rated power of the power supply line, P0 is the input power of the power supply line under no-load condition, P s It is the actual operating power of the power supply line.

[0239] The power consumed by the power supply line under rated load operation is ΣP N , which can be written as:

[0240]

[0241] Where η N It is the power supply efficiency of the power supply line at rated power.

[0242] According to the load rate of each power supply line and the preset load rate range, the power supply lines are divided into multiple power supply line sets, each power supply line set corresponds to a load rate range, so that power supply lines with similar load rates are treated as a set for subsequent calculations, thereby improving calculation efficiency.

[0243] Then, the branch probability p(k) in different load rate intervals is further calculated by combining the power supply line set and the load rate of each power supply line. The specific calculation formula is:

[0244]

[0245] In the formula, l k is the number of power supply lines in the power supply line set of the kth load rate interval, and K is the total number of load rate intervals. The branch probability can measure the proportion of the power supply lines included in each load rate interval to the total power supply lines in the power grid.

[0246] Then, according to the calculated branch probability of the load rate interval, by combining the entropy model, the power grid flow entropy index when the power grid is disturbed is constructed. The specific expression is:

[0247] H k = -Cp(k)lnp(k);

[0248]

[0249] In the formula, H k is the power flow entropy value of the kth load rate interval, and H is the power flow entropy value of the entire power grid.

[0250] The power grid power flow entropy index can be used to calculate the probability of disturbances occurring in each power supply line, that is, to reflect the probability of failures occurring in power grid branches.

[0251] For constructing the branch flow disconnection entropy index when the power grid is disturbed, firstly, the corresponding power supply line is weighted according to the preset branch weight. Optionally, in the embodiment of the present application, the reactance value of each power supply line is used as the branch weight, and the specific expression is as follows:

[0252] W={w1,w2,…,w k ,…,w m};

[0253] In the formula, w is the branch weight, and its subscript represents the number of the corresponding power supply line. The power supply line can be weighted by the number. W is the branch weight set, and m is the total number of power supply lines.

[0254] Then, according to the weighted power supply lines, the active power of the power supply lines and the grid topology adjacency matrix, the grid structure-operation characteristic index D is obtained. k , the specific expression is:

[0255] D k =|w k p k |(k=1,2,…,m);

[0256] In the formula, p k w k The amount of active power transmitted on the corresponding power supply line.

[0257] Then, the maximum and minimum phase angles of all nodes in the power grid are extracted through the expected device set to construct the longest power transmission distance indicator D max , the specific expression is as follows:

[0258] θ min =min(θ1,θ2,…,θ n );

[0259] θ max =max(θ1,θ2,…,θ n );

[0260] D max =|θ max -θ min |;

[0261] In the formula, θ min is the minimum phase angle of all nodes, θ max is the maximum value of the phase angle of all nodes, and n is the total number of connected nodes.

[0262] D k and D max Combined, we can get the branch transmission rate δ k , the specific expression is as follows:

[0263]

[0264] Among them, the size of the branch transmission rate can reflect the transmission distance of the active power of the power grid from the source end to the receiving end and the smoothness of the active power transmission process.

[0265] Furthermore, the total active power transmission ratio δ is constructed by the branch transmission rate. The specific expression is as follows:

[0266]

[0267] Because the branch power flow breaking entropy index is constructed when the power grid is disturbed, it is necessary to consider the power flow changes and power transmission distance changes of the power supply line before and after the power grid disturbance. Take the two time states before and after the disturbance, that is, t0 and t1, to establish the power flow relative transfer rate index. The specific expression is as follows:

[0268] Δδ=δ t1 -δ t0 ;

[0269] In the formula, δ t1 is the total active power transmission ratio at time t1, and is δ t0 is the total active power transmission ratio at time t0, δ t1 >δ t0 .

[0270] By combining the relative power transfer rate index and the entropy model, the branch power flow disconnection entropy index R when the power grid is disturbed can be obtained. The specific expression is as follows:

[0271] R = -CΔδlnΔδ;

[0272] After obtaining the power flow entropy index and the branch power flow breaking entropy index, firstly calculate the power flow entropy index {H k} sort by size to obtain the maximum power flow entropy value H max and the minimum power flow entropy value H min Then, with the minimum power flow entropy value H min As the origin of the coordinate system, draw the vertical axis with the upward direction as the positive direction, and the maximum tidal current entropy value H maxis the end point of the vertical axis. max To the minimum power flow entropy value H min The distance is the vertical axis range, and the vertical axis range is divided into three equal parts to obtain two vertical axis equal points. The expression of each vertical axis equal point is:

[0273] H settmg.1 =H min +Δh;

[0274] H settmg.2 =H min +2Δh;

[0275]

[0276] In the formula, H setting.1 , H setting.2 is the value of the vertical axis dividing point on the vertical axis, and Δh is the size of the vertical axis range after being divided into three equal parts.

[0277] Then the branch power flow breaking entropy index {R k} sort by size to obtain the maximum branch flow breaking entropy value R max and the minimum branch power flow breaking entropy value R min Then the minimum branch flow breaking entropy value R min As the coordinate origin, the coordinate origin and the minimum power flow entropy value H min Overlap, draw the horizontal axis with the right as the positive direction, and use the maximum branch flow breaking entropy value R max is the end point of the horizontal axis. The maximum branch flow breaking entropy value R max To the minimum branch power flow breaking entropy value R min The distance is the horizontal axis range, and the horizontal axis range is divided into three equal parts to obtain two horizontal axis equal points. The expression of each horizontal axis equal point is:

[0278] R setting.1 =R min +Δr;

[0279] R setting.2 =R min +2Δr;

[0280]

[0281] In the formula, R setting.1 , R setting.2 is the value of the horizontal axis at the point where the horizontal axis is equally divided, and Δr is the size after the vertical and horizontal axis range is divided into three equal parts.

[0282] Coordinate lines are drawn according to the equal points of the horizontal axis and the equal points of the vertical axis to obtain a double entropy index nine-square grid, and then the double entropy index nine-square grid is divided into three areas, namely area A, area B and area C, according to a preset entropy value range.

[0283] The coordinate range of the area A is:

[0284]

[0285] The coordinate range of the area B is:

[0286] (R min ,H max )→(R setting.1 ,H max )→(R setting.1 ,H setting.2 )→(R min ,H sup2 )→(R min ,H max )

[0287] ∪(R setting.1 ,H setting.2 )→(R setting.2 ,H setting.2 )→(R setting.2 ,H setting.1 )→(R setting.1 ,H setting.1 )→(R setting.1 ,H setting.2 )

[0288] ∪(R setting.2 ,H setting.1 )→(R max ,H setting.1 )→(R max ,H min )→(R setting.2 ,H min )→(R setting.2 ,H setting.1 )

[0289] The coordinate range of the area C is:

[0290]

[0291] In order to better display the double entropy index nine-square grid after regional division, Figure 2 A schematic diagram of the double entropy index nine-square grid is provided. There are three areas in the figure, the darker one is area A, the lighter one is area C, and the white one is area B. In the preset entropy value range, the larger the entropy value, the worse the anti-disturbance ability of the range, so area A represents the equipment branch with a higher probability of disturbance in the power grid; area B represents the equipment branch with a moderate probability of disturbance in the power grid; area C represents the equipment branch with a lower probability of disturbance in the power grid, and the equipment branch has a strong anti-disturbance ability.

[0292] Optionally, in order to reduce the amount of calculation, since the device branch represented by area C has strong anti-interference ability and is not prone to fault disturbance, the device branch represented by area C can be ignored from the expected device set when evaluating the power grid risk.

[0293] Through the dual entropy indicator grid, it is possible to identify which equipment branches are prone to fault interference in the power grid and need to be focused on during risk assessment, thereby achieving a one-time screening of the dynamic disturbance set constructed in terms of the equipment's anti-interference ability.

[0294] In some embodiments, the dynamic perturbation set generation module 204 further includes:

[0295] Based on the expected device set constructed in step S1, the historical flow data of each node in the power grid is sorted to construct the deep learning features of the original topology graph. The specific expression is:

[0296]

[0297] In the formula, V 0-,i ,θ 0-,i are the active power and reactive power of the load at node i in the power grid, the active power and reactive power of the distributed generation, the voltage amplitude and voltage phase angle at node i; N is the total number of nodes in the power grid, and 0- represents the moment before the node fault disturbance occurs.

[0298] The preset grid fault behavior is implemented on the grid simulation, the changed grid topology is recorded, and the equipment that has experienced fault disturbance, needs to be shut down or repaired, and the corresponding fault location and fault clearing location are obtained. The nodes of the expected equipment set are marked according to the fault location and fault clearing location to complete the node status update, and then the instantaneous power flow after the grid fault is estimated. The deep learning features of the first topology after the fault are constructed based on the updated expected equipment set. The specific expression is:

[0299]

[0300] Where, 0 + is the time before the node failure disturbance occurs.

[0301] Then clear the faulty device and update the expected device set again to obtain the node corresponding to the cleared faulty device, and replace the power of the node with the difference of the steady-state power before and after the fault. The specific calculation formula is:

[0302]

[0303] Where P (imp) and Q (imp) are the differences of steady-state active power and reactive power before and after the fault, Y c+is the admittance matrix of the power grid after the fault is cleared, is the node voltage vector before the fault.

[0304] According to the node power after the fault is cleared and the expected device set after the second update, the deep learning feature X of the second topology graph can be obtained: c+ , the specific expression is:

[0305]

[0306] The system topology dynamic adjustment feature X can be obtained by combining the original topology map deep learning feature, the first topology map deep learning feature and the second topology map deep learning feature. D , the specific expression is:

[0307] X D =X 0- ||X 0+ ||X c+ ;

[0308] In the formula, || means that the vectors in the feature are concatenated row by row.

[0309] In the second step, the node features corresponding to the system topology dynamic adjustment features are extracted according to the preset power grid fault behavior. Specifically, the system topology dynamic adjustment features and the preset graph module function are used to integrate the local topology difference information of the nodes before and after the topology change. L , the specific expression is:

[0310]

[0311] In the formula, A 0- , A c+ are respectively the grid topology adjacency matrices before and after the fault according to the branch weight assignment, and f GNN is a preset graph module function that can convert X D The information is mapped into the power grid topology adjacency matrix, and the specific operation process is similar to associating the resource index table with the nodes in the original adjacency matrix.

[0312] The third step is to introduce the preset learning coefficient to supplement the local topological difference information and obtain the global topological difference information. The specific expression is:

[0313]

[0314] Where W Q ,W K ,W V is the learning coefficient, f NormIt is a SoftMin function, which can assign a higher probability to a smaller value when providing a probability distribution. In the embodiment of the present application, it is used to extract the maximum difference information of the node of the power grid topology change before and after the fault occurs.

[0315] Finally, the system topology dynamic adjustment features, local topology difference information and global topology difference information are spliced ​​to obtain the system graph feature X H , the specific expression is:

[0316] X H =X D ||X L ||X G ;

[0317] Furthermore, the system diagram feature is actually the system diagram feature after a power grid failure, which contains information about all nodes in the power grid. Through the system diagram feature, the changes in the power grid node information after the preset fault behavior occurs can be obtained, and the resource index table and the expected equipment set are updated accordingly according to the changes in the node information, and the nodes corresponding to the equipment that will be shut down and to be repaired are deleted to complete the dynamic update of the information of each node in the power grid. Then, according to the updated expected equipment set, the node information in the double entropy index nine-square grid is dynamically updated, and the double entropy index nine-square grid is reconstructed through the effective power supply lines after the fault, and finally a dynamic disturbance set that can fully cover the fault types of the power system is obtained.

[0318] Implementing the embodiments of the present application has the following beneficial effects:

[0319] The embodiment of the present application first visualizes the physical structure of the power grid, constructs a power grid topology adjacency matrix that describes the connection between nodes and branches inside the power grid, then connects the corresponding resource information for the nodes in the power grid topology adjacency matrix, and screens the nodes by the type of faulty equipment, thereby screening the effective equipment from the power grid structure to obtain the expected equipment set; then, weights are assigned to the power supply lines of the expected equipment concentration according to the preset branch weights, and then, combined with the entropy model, a double entropy index nine-square grid that can describe the probability of disturbance occurrence and anti-disturbance capability of each power supply line is constructed, and the nine-square grid can completely cover the types of power system faults. Finally, when a power grid fault occurs, the node information in the double entropy index nine-square grid is updated using the historical operation data of the power grid as prior knowledge, so as to dynamically update the disturbance set according to the changes in the system operation status, and provide reliable data support for the risk assessment of the power system.

[0320] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A novel method for constructing a dynamic disturbance set for power system stability pre-decision making, characterized in that: include: Visualize the internal nodes and branch connections of the power grid to obtain the power grid topology adjacency matrix; According to the preset resource equipment table and the preset fault equipment type, the grid topology adjacency matrix is ​​screened to obtain the expected equipment set; Constructing a double entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model; According to the preset power grid fault behavior and the historical power grid operation data, the node information in the double entropy index nine-square grid is dynamically updated to obtain a dynamic disturbance set.

2. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 1 is characterized in that: According to the preset resource equipment table and the preset fault equipment type, the grid topology adjacency matrix is ​​screened to obtain the expected equipment set, which is specifically: Establishing corresponding resource codes for the diagonal nodes in the power grid topology adjacency matrix and the branches corresponding to the nodes, and generating the resource index table according to the resource codes; According to the resource index table, the nodes of the power grid topology adjacency matrix are arranged in a preset order from both row and column directions to obtain an original adjacency matrix; Dividing the resource index table according to two types of line connection branches and transformer connection branches to obtain a first branch type set and a second branch type set; According to the resource coding, the first branch type set and the second branch type set are substituted into the original adjacency matrix to obtain the expected device set.

3. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 2 is characterized in that: The envisioned device set is specifically: The expression of the expected device set is specifically: Where M(V) is the expected equipment set, <l i |t j > is the combination of the first branch type set and the second branch type, v n The node code for the grid node.

4. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 1 is characterized in that: According to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model, a double entropy index nine-square grid is constructed, specifically: Constructing the entropy model according to all operating states of the power grid and the probability of occurrence of each operating state; Obtaining power data of each power supply line in the envisioned equipment set according to the power flow data of each power supply line; According to the load rate of each power supply line in the expected equipment concentration, the power data of the power supply line and the entropy model, a power grid power flow entropy index is constructed when a disturbance shock occurs to the power grid; According to the branch weight, the power data of the power supply line and the entropy model, construct a branch power flow disconnection entropy index when a disturbance shock occurs to the power grid; The double entropy index nine-square grid is constructed according to the power grid power flow entropy index and the branch power flow breaking entropy index.

5. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 4 is characterized in that: The power grid power flow entropy index when a disturbance shock occurs to the power grid is constructed according to the load rate of each power supply line in the expected equipment set, the power data of the power supply line and the entropy model, specifically: According to a preset load rate interval and a load rate of each power supply line in the power grid, the power supply lines are divided to obtain a set of power supply lines corresponding to each load rate interval; Obtaining a branch probability in each load rate interval according to the power supply line set and the load rate of the power supply line; According to the branch probability of each load rate interval and the entropy model, a power grid power flow entropy index is constructed when a disturbance shock occurs to the power grid.

6. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 4 is characterized in that: The branch power flow disconnection entropy index when a disturbance shock occurs in the power grid is constructed according to the branch weight, the power data of the power supply line and the entropy model, specifically: Obtaining a power grid structure-operation characteristic index according to the branch weight, the power data of the power supply line and the power grid topology adjacency matrix; By extracting the maximum and minimum phase angles of all nodes in the envisioned device set, a longest power transmission distance index is constructed; The total active power transmission ratio is obtained by combining the grid structure-operation characteristic index and the longest power transmission distance index; According to the power flow changes and power transmission distance changes before and after the disturbance shock of the power grid, the power flow relative transfer rate index is obtained based on the total active power transmission ratio; According to the entropy model and the power flow relative transfer rate index, a branch power flow breaking entropy index is obtained.

7. The novel method for constructing a dynamic disturbance set for power system stability pre-decision according to claim 4 is characterized in that: The double entropy index nine-square grid is constructed according to the power grid power flow entropy index and the branch power flow disconnection entropy index, specifically: Sorting the power flow entropy indexes of the power grid to obtain a maximum power flow entropy value and a minimum power flow entropy value; The branch power flow breaking entropy indexes are sorted by size to obtain a maximum branch power flow breaking entropy value and a minimum branch power flow breaking entropy value; The coordinate line is drawn with the minimum power flow entropy value and the minimum branch power flow breaking entropy value as the origin, the maximum power flow entropy value as the end point of the vertical axis and the maximum branch power flow breaking entropy value as the end point of the horizontal axis; The vertical axis range and the horizontal axis range are divided into three equal parts to obtain the nine-square grid of the dual entropy index; According to a preset entropy value range, the dual entropy index nine-square grid is divided into a first threshold number of areas.

8. The novel method for constructing a dynamic disturbance set for power system stability pre-decision making according to claim 1 is characterized in that: According to the preset power grid fault behavior and the historical power grid operation data, the node information in the double entropy index nine-square grid is dynamically updated to obtain a dynamic disturbance set, which is specifically: Based on the expected device set, the historical power flow data of each node in the power grid is obtained to construct the deep learning features of the original topology graph; According to the preset power grid fault behavior, respectively obtain a first topology graph deep learning feature after the power grid fault behavior occurs, and a second topology graph deep learning feature after the faulty device corresponding to the fault behavior is removed; The original topology map deep learning features, the first topology map deep learning features and the second topology map deep learning features are sequentially subjected to feature splicing, feature extraction and feature supplementation to obtain system map features after a fault behavior occurs in the power grid; The node information in the expected device set is updated according to the characteristics of the system diagram, so as to update the node information in the double entropy index nine-square grid and obtain a dynamic disturbance set.

9. The novel method for constructing a dynamic disturbance set for power system stability pre-decision making according to claim 8 is characterized in that: The original topology map deep learning features, the first topology map deep learning features and the second topology map deep learning features are sequentially subjected to feature splicing, feature extraction and feature supplementation to obtain the system map features after the fault behavior of the power grid occurs, specifically: Perform feature splicing on the original topology map deep learning feature, the first topology map deep learning feature and the second topology map deep learning feature to obtain a system topology dynamic adjustment feature; Differentiate and extract nodes in the system topology dynamic adjustment feature according to a preset power grid fault behavior to obtain local topology difference information; Introducing a preset learning coefficient to supplement the local topological difference information to obtain global topological difference information; The system topology dynamic adjustment feature, the local topology difference information and the global topology difference information are spliced ​​to obtain the system graph feature.

10. A new type of power system stability pre-decision dynamic disturbance set construction system, characterized by: include: Grid topology visualization module, anticipated equipment set construction module, dual entropy index nine-square grid construction module and dynamic disturbance set generation module; The power grid topology visualization module is used to visualize the internal nodes and branch connections of the power grid to obtain a power grid topology adjacency matrix; The expected device set construction module is used to screen the power grid topology adjacency matrix according to the preset resource device table and the preset fault device type to obtain the expected device set; The dual entropy index nine-square grid construction module is used to construct a dual entropy index nine-square grid according to the load rate and flow data of each power supply line in the expected equipment set, the preset branch weight and the preset entropy model; The dynamic disturbance set generation module is used to dynamically update the node information in the double entropy index nine-square grid according to the preset power grid fault behavior and the power grid historical operation data to obtain the dynamic disturbance set.

Images