CN-121997509-A - Simulation method and system for grid-connected operation of new energy station

CN121997509ACN 121997509 ACN121997509 ACN 121997509ACN-121997509-A

Abstract

The invention provides a simulation method and a system for grid-connected operation of a new energy station, which relate to the technical field of new energy power system simulation and comprise the following steps: by acquiring station operation configuration and power grid access information, extracting power generation unit characteristics, calculating electrical similarity, carrying out clustering grouping and equivalent parameter construction, identifying node cluster boundaries to construct partition topology, executing asynchronous parallel simulation, and carrying out time alignment processing on boundary interaction data. The invention can effectively reduce the simulation calculation complexity, improve the grid-connected simulation efficiency of the large-scale new energy station and ensure the calculation precision and the real-time performance.

Inventors

ZHENG JIANHU
WU DALI
WANG JUNFENG
Zong pu

Assignees

陕西华电新能源发电有限公司

Dates

Publication Date: 20260508
Application Date: 20260127

Claims (10)

1. The simulation method for the grid-connected operation of the new energy station is characterized by comprising the following steps of: Acquiring operation configuration information of a new energy station and access point information of a power grid side, extracting operation state characteristics of power generation units based on the operation configuration information, constructing a time sequence characteristic matrix, extracting dominant characteristic components corresponding to the time sequence characteristic matrix, constructing a characteristic vector set, and calculating an electrical similarity index between the power generation units; Grouping the power generation units through a condensation hierarchical clustering algorithm according to the electrical similarity index to obtain a plurality of power generation groups, calculating a self-adaptive weighting coefficient according to the electrical similarity index, weighting and polymerizing the electrical impedance characteristics of each power generation unit in the power generation groups to obtain equivalent electrical parameters, constructing equivalent nodes based on the equivalent electrical parameters, and determining a simplified topological structure; Determining boundary constraint conditions of a power grid side according to the access point information, applying the boundary constraint conditions to the simplified topological structure, determining a Laplacian matrix corresponding to the simplified topological structure, constructing a spectrum embedding space, mapping the equivalent nodes to the spectrum embedding space, identifying node cluster boundaries, and dividing the simplified topological structure into a plurality of sub-network areas to obtain a partition topological structure; And configuring independent simulation step sizes for the sub-network areas according to the partition topological structure, executing asynchronous parallel simulation, extracting boundary node data at preset synchronous time, carrying out time alignment processing on boundary interaction data by combining an extrapolation prediction algorithm, and transmitting the processed boundary interaction data to adjacent sub-network areas to obtain a global simulation result.
2. The method of claim 1, wherein obtaining operation configuration information of a new energy station and access point information on a power grid side, extracting operation state characteristics of power generation units based on the operation configuration information and constructing a time sequence characteristic matrix, extracting dominant characteristic components corresponding to the time sequence characteristic matrix to construct a characteristic vector set, and calculating an electrical similarity index between the power generation units comprises: collecting power output sequences and control response sequences of each power generation unit under a plurality of operation conditions, performing time stamp alignment and abnormal data point elimination to obtain an effective power sequence and an effective control sequence, and performing splicing construction according to a time dimension to obtain a time sequence feature matrix; Performing standardization processing on the time sequence feature matrix to obtain a standardized time sequence matrix, constructing a covariance matrix based on the standardized time sequence matrix, performing feature value decomposition operation, performing descending order according to the size of the feature values, screening feature vectors corresponding to a plurality of previous feature values according to the accumulated contribution rate to serve as principal component directions, and performing projective transformation on the standardized time sequence matrix to the principal component directions to obtain dominant feature components; Extracting the dominant characteristic components corresponding to each power generation unit, organizing according to the serial numbers of the power generation units to construct a characteristic vector set, executing covariance calculation on the characteristic vectors of any two power generation units in the characteristic vector set, constructing a covariance matrix, executing matrix inversion operation on the covariance matrix to obtain an inverse covariance matrix, calculating difference vectors among the characteristic vectors of different power generation units, executing weighted distance calculation with the corresponding inverse covariance matrix to obtain a mahalanobis distance, and executing normalization mapping on the mahalanobis distance to obtain an electric similarity index.
3. The method of claim 1, wherein grouping the power generation units by a hierarchical clustering algorithm to obtain a plurality of power generation groups according to the electrical similarity index, calculating an adaptive weighting coefficient according to the electrical similarity index, and performing weighted aggregation on electrical impedance characteristics of each power generation unit in the power generation groups to obtain equivalent electrical parameters, and constructing equivalent nodes based on the equivalent electrical parameters and determining a simplified topology structure comprises: Constructing a similarity matrix based on the electrical similarity index, taking the power generation units as independent clusters, calculating a power generation unit pair with the maximum similarity in the similarity matrix, performing cluster merging operation to obtain a merged cluster, updating the similarity matrix, combining with a contour coefficient evaluation algorithm to calculate to obtain cluster quality, and repeatedly performing cluster merging operation until the cluster quality converges to obtain a plurality of power generation groups; Calculating the electrical similarity index of each power generation unit in the power generation group and the power generation group center, performing reverse mapping to obtain an adaptive weighting coefficient, extracting the resistance parameter and the reactance parameter of each power generation unit in the power generation group from the operation configuration information, and constructing an electrical impedance feature vector; constructing an adjacent matrix based on the electrical connection relation between each power generation unit in the power generation group, performing symmetrical transformation on the adjacent matrix through a graph convolution propagation algorithm to obtain a symmetrical adjacent matrix, performing multi-hop propagation operation by combining the electrical impedance feature vector to obtain a multi-order neighborhood feature, performing layer-by-layer aggregation on the multi-order neighborhood feature to obtain a neighborhood aggregation feature, and performing weighted summation and nonlinear activation transformation operation by combining the adaptive weighting coefficient to obtain an equivalent electrical parameter; And constructing the equivalent node for each power generation group according to the equivalent electrical parameters, extracting a connection topological relation between the power generation unit and the confluence unit in the operation configuration information, and constructing the simplified topological structure based on the connection topological relation between the equivalent node and the confluence unit.
4. The method of claim 3, wherein performing a symmetric transformation on the adjacency matrix by a graph convolution propagation algorithm to obtain a symmetric adjacency matrix and performing a multi-hop propagation operation in conjunction with the electrical impedance feature vector to obtain a multi-order neighborhood feature comprises: Extracting connection relation information of nodes of each power generation unit in the adjacent matrix, counting node degree construction degree matrix, performing inverse transformation on the degree matrix to obtain a degree inverse matrix, performing bilateral symmetry transformation on the adjacent matrix to obtain a symmetrical adjacent matrix, constructing a topology perception mask matrix based on the symmetrical adjacent matrix, taking the electrical impedance characteristic vector as zero-order characteristic input, extracting capacity parameters and impedance parameters of each power generation unit from the operation configuration information, and constructing node attribute embedded vectors; performing neighborhood information aggregation on the current layer feature vector and the symmetrical adjacency matrix through a graph volume integration algorithm to obtain neighborhood propagation features, fusing the neighborhood propagation features and the node attribute embedded vectors through a feature fusion algorithm to obtain enhancement propagation features, performing leachable weight transformation on the enhancement propagation features, performing selective feature filtering by combining the topology perception mask matrix to obtain current layer output features, and solving based on the current layer output features and the current layer input features to obtain residual enhancement features; Judging whether the current propagation layer number reaches a preset maximum propagation layer number, if not, taking the residual enhancement feature as a next layer input feature and repeatedly executing neighborhood propagation operation, if so, extracting the residual enhancement feature output by each propagation layer, calculating the level importance weight based on the propagation depth, and aggregating the residual enhancement feature of each propagation layer and the corresponding level importance weight to obtain the multi-order neighborhood feature.
5. The method of claim 1, wherein determining a boundary constraint condition on a grid side according to the access point information and applying the boundary constraint condition to the simplified topology, determining a laplace matrix corresponding to the simplified topology and constructing a spectrum embedded space, mapping the equivalent node to the spectrum embedded space to identify a node cluster boundary and dividing the simplified topology into a plurality of sub-network areas to obtain a partitioned topology comprises: extracting an access point position identifier and grid side access capacity limit information from the access point information, constructing a power constraint set based on the grid side access capacity limit information, binding the power constraint set to equivalent nodes corresponding to the access point position identifier in the simplified topological structure, and constructing a boundary node marking matrix based on the equivalent nodes bound with the power constraint set; Extracting connection relations among equivalent nodes in the simplified topological structure, constructing a connection matrix, calculating connection numbers of all equivalent nodes based on the connection matrix, constructing a node degree vector, converting the node degree vector into a diagonal matrix form to obtain a degree diagonal matrix, and performing matrix difference operation on the degree diagonal matrix and the connection matrix to obtain the Laplace matrix; Performing eigenvalue decomposition on the Laplace matrix to obtain an eigenvector matrix, and constructing the spectrum embedding space based on the eigenvector matrix; projecting each equivalent node in the simplified topological structure to the spectrum embedding space to obtain spectrum coordinate representation, constructing a spectrum similarity matrix by combining a spectrum clustering algorithm, carrying out constraint adjustment on the spectrum similarity matrix based on the boundary node marking matrix, identifying the boundary position of a spectrum coordinate dense region and a spectrum coordinate sparse region, determining a node cluster boundary based on the boundary position, dividing the equivalent nodes in the simplified topological structure into different sub-network regions, constructing a sub-network region identification mapping table, and decomposing the simplified topological structure based on the sub-network region identification mapping table to obtain a partitioned topological structure.
6. The method of claim 1, wherein configuring independent simulation step sizes for the sub-network areas according to the partition topology structure and performing asynchronous parallel simulation to extract boundary node data at a preset synchronous moment, performing time alignment processing on boundary interaction data in combination with an extrapolation prediction algorithm, and then transmitting the boundary interaction data to adjacent sub-network areas to obtain global simulation results comprises: Extracting topology complexity parameters and node quantity parameters of each sub-network region from the partition topology structure, calculating calculation load quantity of each sub-network region based on the topology complexity parameters and the node quantity parameters, distributing independent simulation step sizes for each sub-network region through a load balancing optimization algorithm based on the calculation load quantity and coupling strengths among different sub-network regions, and constructing a step size configuration table based on the independent simulation step sizes; Identifying boundary connection relations among all sub-network areas from the partition topological structure, extracting boundary node identifiers, and constructing a boundary node list based on the boundary node identifiers; And driving each sub-network region to execute asynchronous parallel simulation according to the corresponding independent simulation step length according to the step length configuration table, extracting corresponding boundary node data from the boundary node list as boundary interaction data and recording corresponding actual extraction time when the simulation time of each sub-network region reaches the preset synchronization time, carrying out time alignment processing on the boundary interaction data through an extrapolation prediction algorithm based on the time deviation between the actual extraction time and the preset synchronization time to obtain synchronous boundary data, transmitting the synchronous boundary data to an adjacent sub-network region as boundary condition according to the boundary connection relation, repeatedly executing asynchronous parallel simulation until the simulation end time, and summarizing simulation output data of each sub-network region to obtain the global simulation result.
7. The method of claim 6, wherein extracting a topology complexity parameter and a number of nodes parameter for each sub-network region from the partitioned topology, calculating a calculated load amount for each sub-network region based on the topology complexity parameter and the number of nodes parameter, and assigning an independent simulation step size for each sub-network region by a load balancing optimization algorithm based on the calculated load amount and a coupling strength between different sub-network regions comprises: Traversing each sub-network area in the partition topological structure, counting the total number of nodes in each sub-network area to obtain a node quantity parameter, calculating the complexity of the connection relation between the nodes in each sub-network area, combining the topological hierarchy depth in the sub-network area to obtain the topological complexity parameter, and calculating the calculation load quantity of each sub-network area based on the node quantity parameter and the topological complexity parameter; Extracting the number of boundary node connections between different sub-network areas in the partition topological structure, counting the boundary node power exchange frequency between each sub-network area and the adjacent sub-network area, calculating the coupling strength between different sub-network areas based on the number of boundary node connections and the boundary node power exchange frequency, and constructing a coupling strength matrix; And constructing an optimization function aiming at minimizing the total simulation duration through a load balancing optimization algorithm based on the calculated load quantity and the coupling strength matrix, extracting electrical characteristic parameters of each sub-network region from the partition topological structure, constructing a simulation numerical stability criterion based on the electrical characteristic parameters and the preset synchronization moment, adding the simulation numerical stability criterion as a simulation stability constraint condition to the optimization function, and solving to obtain independent simulation step sizes of each sub-network region.
8. A simulation system for grid-connected operation of a new energy station, for implementing the method of any one of the preceding claims 1-7, comprising: The characteristic clustering module is used for acquiring operation configuration information of the new energy station and access point information of the power grid side, extracting operation state characteristics of the power generation units based on the operation configuration information, constructing a time sequence characteristic matrix, extracting dominant characteristic components corresponding to the time sequence characteristic matrix, constructing a characteristic vector set and calculating an electrical similarity index between the power generation units; The topology simplification module is used for grouping the power generation units through a condensation hierarchical clustering algorithm according to the electrical similarity index to obtain a plurality of power generation groups, calculating a self-adaptive weighting coefficient according to the electrical similarity index, carrying out weighted aggregation on the electrical impedance characteristics of each power generation unit in the power generation groups to obtain equivalent electrical parameters, constructing equivalent nodes based on the equivalent electrical parameters, and determining a simplified topology structure; The spectrum domain partitioning module is used for determining boundary constraint conditions of a power grid side according to the access point information and applying the boundary constraint conditions to the simplified topological structure, determining a Laplacian matrix corresponding to the simplified topological structure and constructing a spectrum embedding space, mapping the equivalent nodes to the spectrum embedding space, identifying node cluster boundaries and dividing the simplified topological structure into a plurality of sub-network areas to obtain a partitioned topological structure; And the asynchronous simulation module is used for configuring independent simulation step length for the sub-network area according to the partition topological structure, executing asynchronous parallel simulation, extracting boundary node data at a preset synchronous moment, carrying out time alignment processing on boundary interaction data by combining an extrapolation prediction algorithm, and transmitting the boundary node data to an adjacent sub-network area to obtain a global simulation result.
9. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 7.
10. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 7.

Description

Simulation method and system for grid-connected operation of new energy station Technical Field The invention relates to the technical field of new energy power system simulation, in particular to a simulation method and a simulation system for grid-connected operation of a new energy station. Background With the rapid development of renewable energy sources, the number of new energy stations is increasing, and the grid-connected operation of the new energy stations has an important influence on the stability of an electric power system. The new energy station consists of a large number of distributed power generation units, has the characteristics of intermittence, volatility and uncertainty, and has complex and changeable influence on a power grid. In order to ensure safe and stable operation of the power system, simulation analysis is required to be carried out on grid-connected operation of the new energy station, so that influence of the new energy station on a power grid is predicted, and a corresponding regulation strategy is formulated. The traditional new energy station grid-connected simulation method mainly adopts a full-model direct simulation mode, and all power generation units and control systems thereof are incorporated into a simulation model, so that the simulation system is huge in scale and high in calculation complexity, and the full-model direct simulation method is difficult to meet the requirement of real-time simulation along with the continuous expansion of the installed scale of new energy. At present, the simulation technology for grid-connected operation of the new energy station still has the problems that the large-scale new energy station cannot be effectively and reasonably simplified, the calculation load is heavy, the simulation efficiency is low, the self-adaptive clustering method based on physical characteristics is lacking, good balance between precision and efficiency is difficult to achieve, targeted optimization cannot be carried out according to the dynamic characteristics of different sub-network areas, the simulation speed and precision are low, and the like. Disclosure of Invention The embodiment of the invention provides a simulation method and a simulation system for grid-connected operation of a new energy station, which at least can solve part of problems in the prior art. In a first aspect of the embodiment of the present invention, a simulation method for grid-connected operation of a new energy station is provided, including: Acquiring operation configuration information of a new energy station and access point information of a power grid side, extracting operation state characteristics of power generation units based on the operation configuration information, constructing a time sequence characteristic matrix, extracting dominant characteristic components corresponding to the time sequence characteristic matrix, constructing a characteristic vector set, and calculating an electrical similarity index between the power generation units; Grouping the power generation units through a condensation hierarchical clustering algorithm according to the electrical similarity index to obtain a plurality of power generation groups, calculating a self-adaptive weighting coefficient according to the electrical similarity index, weighting and polymerizing the electrical impedance characteristics of each power generation unit in the power generation groups to obtain equivalent electrical parameters, constructing equivalent nodes based on the equivalent electrical parameters, and determining a simplified topological structure; Determining boundary constraint conditions of a power grid side according to the access point information, applying the boundary constraint conditions to the simplified topological structure, determining a Laplacian matrix corresponding to the simplified topological structure, constructing a spectrum embedding space, mapping the equivalent nodes to the spectrum embedding space, identifying node cluster boundaries, and dividing the simplified topological structure into a plurality of sub-network areas to obtain a partition topological structure; And configuring independent simulation step sizes for the sub-network areas according to the partition topological structure, executing asynchronous parallel simulation, extracting boundary node data at preset synchronous time, carrying out time alignment processing on boundary interaction data by combining an extrapolation prediction algorithm, and transmitting the processed boundary interaction data to adjacent sub-network areas to obtain a global simulation result. In an alternative embodiment of the present invention, Acquiring operation configuration information of a new energy station and access point information of a power grid side, extracting operation state characteristics of power generation units based on the operation configuration information and constructing a time sequence characteristic matr