Method and system for evaluating new energy and energy storage planning implementation path of block power grid

Abstract

The invention discloses a block power grid new energy and energy storage planning implementation path evaluation method and system, and the method comprises the steps: obtaining multi-source heterogeneous data of a target region, and constructing a power grid digital twin pedestal through the analysis of the multi-source heterogeneous data; planning periods are divided based on the base, multi-stage capacity expansion planning and safety margin calculation are carried out through macroscopic-microscopic bidirectional coupling optimization, and a staged implementation path scheme set is obtained; introducing an adversarial game mechanism, constructing a virtual pressure test scene to evaluate path adaptability, and performing screening to form an optimized implementation path set; constructing an evolution topological graph by adopting dynamic network analysis based on the optimized implementation path set, and generating a path evolution association graph through node centrality analysis and trajectory mining; and finally, in combination with real-time power grid data, path performance deviation degree evaluation and root cause diagnosis are performed, rolling optimization and dynamic adjustment are realized, and an optimization scheme is generated and pushed. According to the invention, full-life-cycle dynamic evaluation and optimization of the planned path are realized.

Description

Technical Field

[0001] This invention relates to the fields of power system planning and energy technology, and in particular to an evaluation method and system for the implementation path of new energy and energy storage planning in block power grids. Background Technology

[0002] With the acceleration of the global energy transition, the proportion of new energy power generation, represented by wind power and photovoltaics, continues to increase, while the development of energy storage resources such as ocean energy and seawater desalination is also receiving increasing attention. In block grids with complex grid structures and dense loads, the large-scale integration of new energy sources and energy storage brings new challenges to grid planning. Traditional grid planning methods mainly target single energy types or static load conditions, making it difficult to adapt to the collaborative planning needs of multi-energy complementary systems. Existing planning methods often focus on optimizing economic indicators, lacking a comprehensive consideration of system operational safety, environmental benefits, and grid adaptability, and failing to effectively integrate the management model changes brought about by emerging technologies such as blockchain.

[0003] Currently, in the field of coordinated planning for new energy and energy storage, a systematic method for evaluating implementation paths has not yet been established. Existing technical solutions typically employ static cost-benefit analysis or simple scenario analysis, which cannot characterize the uncertainties faced by the planned path throughout its entire lifecycle. For systems like block grids, which have distinct regional characteristics and complex constraints, planning schemes need to be deeply coupled with the actual evolution of the grid infrastructure. However, traditional methods struggle to effectively address the impact of dynamic factors such as grid expansion timing and equipment commissioning plans on the planned path, leading to a disconnect between the planned scheme and the actual construction process.

[0004] Therefore, there is an urgent need for a planning and implementation path evaluation method that can comprehensively consider the characteristics of multi-energy complementarity, the timing of power grid evolution, and operational uncertainties, so as to provide scientific and reliable decision support for the planning of new energy and energy storage in block power grids. Summary of the Invention

[0005] This invention overcomes the shortcomings of the prior art and provides an evaluation method and system for the implementation path of new energy and energy storage planning in block power grids. Its important purpose is to provide scientific and reliable decision support for the planning of new energy and energy storage in block power grids.

[0006] To achieve the above objectives, the first aspect of this invention provides an evaluation method for the implementation path of new energy and energy storage planning in block power grids, comprising: Acquire data on grid load characteristics, distribution of new energy and energy storage resources, grid topology, and energy storage system parameters for the target area. Simultaneously, collect data on the expansion plan and capacity improvement roadmap of the regional power grid. Construct a digital twin base for the power grid through multi-source heterogeneous data analysis. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, resulting in a set of phased implementation path schemes that are coordinated with power grid evolution. An adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path in the whole life cycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set. Based on the optimized implementation path set, a dynamic network analysis method is used to construct the evolution topology of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining. The system acquires real-time power grid construction progress and operation status data of the target block, combines the path evolution correlation map to evaluate path performance deviation and diagnose root causes, and performs rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, generating and pushing out optimized plans for the planned implementation path.

[0007] In this solution, the acquisition of power grid load characteristic data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters for the target area, along with the collection of regional power grid expansion plans and capacity enhancement roadmaps, and the construction of a power grid digital twin base through multi-source heterogeneous data analysis, specifically includes: By introducing a big data network, the target area's power grid load characteristics data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters are obtained through big data retrieval. At the same time, the expansion plan and capacity improvement roadmap of the regional power grid are collected to obtain a multi-source heterogeneous dataset. Outliers and missing values ​​are identified and repaired in the multi-source heterogeneous dataset. The seven-parameter coordinate transformation method is used to unify the spatial reference systems of different data sources to the same coordinate system. The linear interpolation method is used to normalize various time series data to the same time granularity. A globally unique identifier is created for each power grid entity based on the preset coding rules to generate a standardized dataset. Based on the standardized dataset, the power grid topology features are extracted and imported into the Neo4j graph database. They are then converted into a set of nodes and branches using a CIM / XML parser based on the SAX parser. Substation buses are abstracted as vertices, and transmission lines and transformers are abstracted as edges with resistance, reactance, capacitance to ground, and susceptance parameters. A power grid topology attribute graph that supports shortest path, connected component, and power flow tracking calculations is then established. Based on the power grid topology attribute map, the ArcGIS Pro spatial analysis engine is integrated to perform geographic information fusion. By establishing spatial indexes and coordinate mapping relationships, a dynamic buffer area is created for each electrical node. The resource density and load distribution characteristics within each buffer are calculated using a partitioning statistical tool. These indicators are used as derived attributes and updated to the corresponding nodes through node ID association to generate a spatiotemporally enhanced power grid topology map. Deep semantic parsing is performed on unstructured planning texts in standardized datasets. A pre-trained language model based on the BERT architecture is used for deep semantic understanding. Semantic role labeling technology is used to identify the predicate-argument structure in the text. Conditional random fields are used for named entity recognition. Structured metadata quadruples containing latitude and longitude coordinates, technology type, capacity target and time node are extracted to obtain the power grid evolution event table. Using the spatiotemporally enhanced power grid topology map as input, feature vectors of all nodes in the map are extracted to construct a regional feature matrix. Spectral clustering algorithm is used for block division and potential level evaluation to generate a potential point-marked topology map. The system is then integrated with a standardized dataset, the spatiotemporally enhanced power grid topology map, and a power grid evolution event table to construct a digital twin base for the power grid.

[0008] In this scheme, the step of using the spatiotemporal enhanced power grid topology map as input, extracting feature vectors of all nodes in the map to construct a regional feature matrix, employing a spectral clustering algorithm for block division and potential level evaluation, and generating a potential point-marked topology map specifically includes: A spatiotemporal enhanced power grid topology map is obtained, and features are extracted from all graph nodes of the spatiotemporal enhanced power grid topology map, including resource density percentile value, load growth rate, topological centrality index and electrical betweenness. The extracted features are then Z-score standardized to form a multidimensional node feature matrix. A spectral clustering algorithm is introduced for block partitioning. The similarity matrix between nodes is calculated based on the multidimensional node feature matrix. An improved Gaussian kernel function is used to define the similarity between nodes, where the kernel parameter is determined based on the k-nearest neighbor distance. The degree matrix is ​​constructed using the calculated similarity matrix. The normalized Laplacian matrix is ​​calculated using the constructed degree matrix. The normalized Laplacian matrix is ​​then subjected to eigenvalue decomposition. The eigenvectors corresponding to the k smallest eigenvalues ​​are selected to form a dimensionality-reduced feature space. Subsequently, the K-means clustering algorithm is used for clustering. The optimal number of clusters is determined by the silhouette coefficient method. All network nodes are divided into k blocks with similar features, and the block division results are output. A multi-dimensional potential evaluation system is established, which includes resource endowment index, load concentration index, power grid support index and development coordination index. The weights of the indexes are determined by using the entropy weight CRITIC combination weighting method. The development coordination index is quantified by calculating the spatiotemporal matching degree between the existing line capacity utilization rate and the planned expansion project within the block. The comprehensive potential score of each block is obtained by using a multi-dimensional potential evaluation system, and a regional score distribution map is generated. The blocks are divided into three levels: high potential area, medium potential area and low potential area by combining the preset potential level threshold. The block potential evaluation results are mapped to the spatiotemporal enhanced power grid topology map and each node is assigned a potential level label of its block. Finally, a potential point labeled topology map is obtained.

[0009] In this scheme, based on the power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is used for multi-stage capacity expansion planning and power grid safety margin calculation, resulting in a set of phased implementation path schemes coordinated with power grid evolution, specifically including: Obtain the power grid digital twin base, extract the power grid evolution event table based on the power grid digital twin base, parse the power grid upgrade project information recorded in the event table, extract the planned commissioning time node, construction duration and logical dependency relationship between projects for each project, and construct a project dependency relationship network, where nodes represent power grid construction projects and directed edges represent the sequential dependency relationship between projects; Based on the project dependency network, the critical path method is used to calculate the earliest start time, latest finish time and total float of each node. The project sequence with zero total float is taken as the critical path. Jenks' natural break method is introduced. The time point sequence of the critical path projects is taken as input. The optimal segmentation scheme is determined by calculating the intra-class variance and inter-class variance ratio. Finally, the stage division scheme is output. A macro-micro bidirectional coupling optimization architecture is constructed, which includes a macro layer and a micro layer. In the macro layer, a linear programming model is established with the optimization objective of minimizing the present value of the total life cycle cost. The decision variables are the investment decisions and capacity variables of each stage and each candidate node. Constraints are set according to Kirchhoff's current law, transmission capacity constraints, and energy supply and demand balance constraints. The aforementioned stage division scheme is input into the linear programming model established in the macro layer, and the branch and bound algorithm is used to solve it. The optimal solution is gradually approximated through linear relaxation, branch operation and cutting plane method, and the macro capacity expansion planning scheme containing the investment time sequence and capacity scale of each stage is output. At the micro-operation layer, the capacity expansion planning scheme generated at the macro-level is dynamically coupled with the spatiotemporal enhanced grid topology map in the grid digital twin base. Based on the engineering construction sequence determined by critical path analysis, the completion time of each project on the critical path is modeled using the PERT-based three-point estimation method. Several engineering construction progress scenarios are generated through Beta distribution to obtain a comprehensive test scenario set that includes grid structure evolution and resource fluctuations. For each comprehensive test scenario, hourly operation simulation of the annual operation cycle is performed based on a preset time step. At each simulation time step, a node admittance matrix is ​​established based on the grid structure and power generation resources that are in operation at the current moment. The forward-backward power flow algorithm is used to calculate the system power flow distribution, record the deviation of the node voltage amplitude from the rated value, and the ratio of the active power flow to the thermal stability limit of each transmission line. The annual hourly simulation results are then output. By using the hourly simulation results throughout the year, the voltage qualification rate of each key node and the load rate of key lines are statistically analyzed to generate power grid safety margin assessment information and compare it with the preset safety standards. If it does not meet the preset safety standards, a penalty function is generated based on the deviation from the preset safety standards and fed back to the objective function of the macro level for re-solution, thereby achieving bidirectional coupling optimization. The optimization is repeatedly iterated through a two-way coupling optimization mechanism. The optimization is terminated when the solution set meets the convergence condition of the hypervolume index. Finally, a set of phased implementation path schemes that are coordinated with the actual evolution process of the power grid infrastructure is output.

[0010] In this solution, an adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path throughout its entire lifecycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set, specifically including: A scene generator agent containing a generator and a discriminator is constructed using a conditional generative adversarial network. The generator learns the joint distribution of potential risk factors and the reconstruction of virtual stress test scenarios through an encoder-decoder structure, while the discriminator distinguishes between real historical events and generated scenarios based on a spatiotemporal convolutional neural network. Using historical extreme weather events and equipment failure records as training samples, the scene generator agent that meets the expectations is obtained after training through minimax game. A phased implementation path scheme set is obtained. The scenario generator intelligent agent generates a virtual stress test scenario for each candidate implementation path in the phased implementation path scheme set. Each candidate path and the corresponding stress test scenario are input into the power grid digital twin base. The sequential Monte Carlo simulation method is used to perform full life cycle adaptive analysis, calculate the key performance indicators of each path under multiple stress scenarios, and generate a performance profile of each candidate implementation path. Based on key performance indicators, an adaptive evaluation system is constructed using the entropy weight method. According to the performance profile of each candidate implementation path, the adaptive evaluation system is used to weight and aggregate each performance indicator to obtain the adaptive score of each candidate implementation path under different pressure scenarios, thus obtaining path adaptive evaluation information. Based on the path adaptability evaluation information, candidate implementation paths with a value greater than the preset adaptability threshold are selected as high-quality implementation paths at the current adversarial game moment to generate a subset of high-quality paths. The performance characteristics of the high-quality paths are then fed back to the scene generator agent for generation strategy adjustment, thus completing one adversarial game optimization. Repeat the above adversarial game optimization steps. When the coefficient of variation of the path fitness score of the latest generation path set is less than the preset coefficient of variation threshold and the scene generator agent cannot generate a new scene that causes the path fitness score to drop by more than the preset drop, terminate the game process and output the optimized implementation path set.

[0011] In this scheme, the step of constructing a phased path evolution topology map based on the optimized implementation path set using dynamic network analysis methods, and generating a path evolution correlation graph through stage node centrality analysis and path evolution trajectory mining, specifically includes: Obtain an optimized implementation path set, and perform full-cycle feature extraction on each optimized implementation path based on the optimized implementation path set to generate a time-series path feature sequence; define Euclidean distance as a local similarity measure, and use the dynamic time warping algorithm to calculate the similarity between paths based on the time-series path feature sequence to finally obtain a path similarity matrix; A path evolution topology graph is constructed based on the path similarity matrix, where nodes represent different implementation paths, edges represent the similarity relationship between paths, and the weight of the edges is determined by the path similarity. A community detection algorithm is used to divide the path evolution topology graph into communities, and multi-level clustering is used to identify route clusters, resulting in an undirected weighted network graph that reveals the potential transformation relationship between paths. Based on the undirected weighted network graph, a multi-dimensional node centrality analysis is performed. The PageRank algorithm based on random walk is used to calculate the global influence of nodes through multiple rounds of iterative propagation. The betweenness centrality index based on the shortest path is used to identify key bridge nodes in the network. The connection quality importance of nodes is analyzed by combining the feature vector centrality, and finally the node centrality analysis results are obtained. Based on the undirected weighted network graph and node centrality analysis results, the path evolution trajectory is deeply mined. The identified technical route clusters are taken as hidden states, and the temporal transition relationships between paths are taken as observation sequences. The state transition probability matrix and observation probability matrix are calculated by the Baum-Welch algorithm, and the Viterbi algorithm is used to decode the optimal state sequence to identify the evolution mode, thereby obtaining path evolution trajectory analysis information. Based on the undirected weighted network graph, node centrality analysis results, and path evolution trajectory analysis information, a path evolution association graph is constructed. A force-oriented layout algorithm based on repulsion and attraction is used for visual spatial arrangement. The node size is set according to the node centrality analysis results, color coding is used to distinguish them according to the community division results, the thickness of the connecting edges is determined according to the similarity weight, and key bifurcation points, turning points, and typical evolution trajectories of the technical route are marked. Finally, the path evolution association graph is output.

[0012] In this solution, the process of acquiring real-time power grid construction progress and operation status data of the target area, combining this data with the path evolution correlation map to assess path performance deviation and diagnose root causes, and optimizing the planned implementation path based on the diagnostic results, generating an optimized planned implementation path scheme for submission, specifically includes: The real-time power grid construction progress and operation status data of the target area are obtained, the obtained real-time power grid construction progress and operation status data are preprocessed, and spatiotemporal alignment matching is performed with the expected status characteristics of the corresponding time nodes in the path evolution association map to form an implementation status feature vector reflecting the actual construction and operation status. The expected state sequence of each reference path is extracted by the path evolution association map. The instantaneous deviation in three dimensions—installed capacity, power grid structure parameters, and operation performance indicators—is calculated by combining the implementation state feature vector with the dynamic time warping algorithm. The cumulative deviation of each dimension is obtained by time integration and the overall deviation of the path is obtained by weighted fusion. The overall deviation of the path is compared with a preset threshold. If it is greater than the preset deviation threshold, the root cause diagnosis analysis is performed based on the isolated forest algorithm. The indicators of each dimension in the path feature sequence are used as input features. Multiple isolated trees are constructed and the path length of data points is calculated to identify the abnormal feature dimensions and obtain abnormal feature identification information. The node centrality features and anomaly features obtained by the path evolution association graph are input into the pre-trained Bayesian inference network. The root cause of the path deviation and its propagation path are determined by posterior probability inference, and root cause diagnosis information is generated. Based on root cause diagnosis information and path evolution trajectory information in the path evolution correlation map, a constrained model predictive control method is used for rolling optimization calculation. With the optimization objectives of minimizing path adjustment costs and maximizing path fit in subsequent stages, the capacity configuration scheme and project construction sequence of subsequent construction stages are replanned based on the current actual path state. Power grid safety operation constraints and technical feasibility constraints are set, and the optimal adjustment strategy is solved through a sequential quadratic programming algorithm. Path optimization suggestions are generated and pushed out to realize the optimization and dynamic adjustment of the planned implementation path.

[0013] A second aspect of the present invention provides an evaluation system for the implementation path of new energy and energy storage planning in a blockchain power grid. The system includes a memory, a processor, and a communication interface. The memory contains a program for evaluating the implementation path of new energy and energy storage planning in a blockchain power grid. When the processor executes the program for evaluating the implementation path of new energy and energy storage planning in a blockchain power grid, it performs the following steps: Acquire data on grid load characteristics, distribution of new energy and energy storage resources, grid topology, and energy storage system parameters for the target area. Simultaneously, collect data on the expansion plan and capacity improvement roadmap of the regional power grid. Construct a digital twin base for the power grid through multi-source heterogeneous data analysis. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, resulting in a set of phased implementation path schemes that are coordinated with power grid evolution. An adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path in the whole life cycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set. Based on the optimized implementation path set, a dynamic network analysis method is used to construct the evolution topology of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining. The system acquires real-time power grid construction progress and operation status data of the target block, combines the path evolution correlation map to evaluate path performance deviation and diagnose root causes, and performs rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, generating and pushing out optimized plans for the planned implementation path.

[0014] A third aspect of the present invention provides a computer-readable storage medium comprising an evaluation method program for the implementation path of new energy and energy storage planning in a block grid. When the evaluation method program for the implementation path of new energy and energy storage planning in a block grid is executed by a processor, it implements the steps of the evaluation method for the implementation path of new energy and energy storage planning in a block grid as described in any of the preceding claims. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments or examples of the present invention, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0016] Figure 1A flowchart of the first method for evaluating the implementation path of new energy and energy storage planning in a block grid, as provided in an embodiment of the present invention; Figure 2 A flowchart of the second method for evaluating the implementation path of new energy and energy storage planning in a block grid, as provided in an embodiment of the present invention; Figure 3 A block diagram of an evaluation system for the planning and implementation path of new energy and energy storage in a block power grid, provided as an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0018] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0019] Figure 1 A flowchart of the first method for evaluating the implementation path of new energy and energy storage planning in a block grid, as provided in an embodiment of the present invention; like Figure 1 As shown, the present invention provides a first method flowchart for evaluating the implementation path of new energy and energy storage planning in block power grids, including: S102: Acquire grid load characteristic data, new energy and energy storage resource distribution data, grid topology data and energy storage system parameters of the target area. At the same time, collect the expansion plan and capacity improvement roadmap of the regional grid. Construct a digital twin base for the grid through multi-source heterogeneous data analysis. S104. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, thereby obtaining a set of phased implementation path schemes that are coordinated with the power grid evolution. S106, Introduce an adversarial game mechanism, construct a virtual stress test scenario based on the phased implementation path scheme set, evaluate the adaptability of each path throughout the entire life cycle, obtain the optimal implementation path through adaptability screening, and form an optimized implementation path set; S108, Based on the optimized implementation path set, a dynamic network analysis method is used to construct an evolution topology map of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining; S110: Obtain real-time power grid construction progress and operation status data of the target block, combine the path evolution correlation map to evaluate the path performance deviation and diagnose the root cause, and perform rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, and generate and push the optimized plan implementation path scheme.

[0020] Furthermore, in a preferred embodiment of the present invention, the acquisition of power grid load characteristic data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters of the target area, while simultaneously collecting the regional power grid expansion plan and capacity improvement roadmap, and constructing a power grid digital twin base through multi-source heterogeneous data analysis, specifically includes: By introducing a big data network, the target area's power grid load characteristics data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters are obtained through big data retrieval. At the same time, the expansion plan and capacity improvement roadmap of the regional power grid are collected to obtain a multi-source heterogeneous dataset. Outliers and missing values ​​are identified and repaired in the multi-source heterogeneous dataset. The seven-parameter coordinate transformation method is used to unify the spatial reference systems of different data sources to the same coordinate system. The linear interpolation method is used to normalize various time series data to the same time granularity. A globally unique identifier is created for each power grid entity based on the preset coding rules to generate a standardized dataset. Based on the standardized dataset, the power grid topology features are extracted and imported into the Neo4j graph database. They are then converted into a set of nodes and branches using a CIM / XML parser based on the SAX parser. Substation buses are abstracted as vertices, and transmission lines and transformers are abstracted as edges with resistance, reactance, capacitance to ground, and susceptance parameters. A power grid topology attribute graph that supports shortest path, connected component, and power flow tracking calculations is then established. Based on the power grid topology attribute map, the ArcGIS Pro spatial analysis engine is integrated to perform geographic information fusion. By establishing spatial indexes and coordinate mapping relationships, a dynamic buffer area is created for each electrical node. The resource density and load distribution characteristics within each buffer are calculated using a partitioning statistical tool. These indicators are used as derived attributes and updated to the corresponding nodes through node ID association to generate a spatiotemporally enhanced power grid topology map. Deep semantic parsing is performed on unstructured planning texts in standardized datasets. A pre-trained language model based on the BERT architecture is used for deep semantic understanding. Semantic role labeling technology is used to identify the predicate-argument structure in the text. Conditional random fields are used for named entity recognition. Structured metadata quadruples containing latitude and longitude coordinates, technology type, capacity target and time node are extracted to obtain the power grid evolution event table. Using the spatiotemporally enhanced power grid topology map as input, feature vectors of all nodes in the map are extracted to construct a regional feature matrix. Spectral clustering algorithm is used for block division and potential level evaluation to generate a potential point-marked topology map. The system is then integrated with a standardized dataset, the spatiotemporally enhanced power grid topology map, and a power grid evolution event table to construct a digital twin base for the power grid.

[0021] It should be noted that by constructing a distributed big data retrieval network, real-time load characteristic curves, raster data of new energy and energy storage resource distribution, power grid topology connections, energy storage system parameters, and unstructured power grid expansion planning texts for the target area are simultaneously acquired from multiple heterogeneous data sources, including power dispatching systems, resource monitoring platforms, geographic information databases, and planning document repositories. This forms a raw, multi-source heterogeneous dataset containing spatiotemporal attributes, electrical parameters, and textual information. To address the issues of outliers, missing values, and inconsistent spatiotemporal benchmarks in the dataset, a data preprocessing process is implemented. An anomaly detection algorithm based on statistical distribution is used to identify and repair data anomalies. The KNN interpolation method is applied to complete missing records. A seven-parameter coordinate transformation model is used to unify spatial data from different sources to the national geodetic coordinate system. Linear interpolation technology is used to normalize various time series data to a uniform 15-minute granularity. Simultaneously, a globally unique identifier is created for each power grid entity according to the "regional code-voltage level-facility sequence" rule, ultimately generating a standardized dataset with a consistent spatiotemporal benchmark and a complete identification system.

[0022] Based on a standardized dataset, a CIM / XML parser based on a SAX parser is used to convert power grid topology data into a set of nodes and branches, which is then imported into the Neo4j graph database to construct a power grid topology attribute map. In this map, substation buses are abstracted as vertices, while transmission lines and transformers are abstracted as edges carrying detailed electrical parameters such as resistance, reactance, capacitance to ground, and susceptance to ground, forming a digital power grid skeleton that supports complex graph theory algorithms (such as shortest path and connected component analysis) and can perform power flow tracking calculations. To deeply integrate electrical characteristics with geographic resources and load distribution, this technology integrates the ArcGIS Pro spatial analysis engine. By establishing spatial indexes and coordinate mapping, a dynamic buffer is created for each electrical node, and the density of new energy resources and load distribution characteristics within each buffer are accurately calculated using partitioned statistical tools. These key spatial statistical indicators are then used as derived attributes and associated back with the corresponding electrical nodes, ultimately generating a spatiotemporally enhanced power grid topology map that tightly couples the electrical connections of the power grid with the resource and load distribution in geographic space.

[0023] For unstructured planning documents, a pre-trained language model based on the BERT architecture is used to perform deep semantic understanding of the planning text. Semantic role labeling technology is used to parse the predicate-argument structure in sentences, and then a conditional random field model is combined for named entity recognition. This allows for the accurate extraction of structured metadata quadruples containing latitude and longitude coordinates, technology type, capacity target, and time nodes from lengthy texts, forming a power grid evolution event table. This transforms ambiguous policy language into clear, structured planning instructions. Finally, using a spatiotemporally enhanced power grid topology map as input, multi-dimensional feature vectors of all nodes in the map are extracted to construct a regional feature matrix. Spectral clustering algorithms are used for block partitioning and potential level evaluation, generating a potential point-marked topology map. Integrating standardized datasets, spatiotemporally enhanced power grid topology maps, power grid evolution event tables, and potential point-marked topology maps, a digital twin foundation for the power grid is constructed, laying a solid data foundation for subsequent accurate planning and evaluation.

[0024] Furthermore, in a preferred embodiment of the present invention, based on the power grid digital twin base, the planning period is divided into multiple construction phases, and a macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid safety margin calculation, thereby obtaining a set of phased implementation path schemes coordinated with power grid evolution, specifically including: Obtain the power grid digital twin base, extract the power grid evolution event table based on the power grid digital twin base, parse the power grid upgrade project information recorded in the event table, extract the planned commissioning time node, construction duration and logical dependency relationship between projects for each project, and construct a project dependency relationship network, where nodes represent power grid construction projects and directed edges represent the sequential dependency relationship between projects; Based on the project dependency network, the critical path method is used to calculate the earliest start time, latest finish time and total float of each node. The project sequence with zero total float is taken as the critical path. Jenks' natural break method is introduced. The time point sequence of the critical path projects is taken as input. The optimal segmentation scheme is determined by calculating the intra-class variance and inter-class variance ratio. Finally, the stage division scheme is output. A macro-micro bidirectional coupling optimization architecture is constructed, which includes a macro layer and a micro layer. In the macro layer, a linear programming model is established with the optimization objective of minimizing the present value of the total life cycle cost. The decision variables are the investment decisions and capacity variables of each stage and each candidate node. Constraints are set according to Kirchhoff's current law, transmission capacity constraints, and energy supply and demand balance constraints. The aforementioned stage division scheme is input into the linear programming model established in the macro layer, and the branch and bound algorithm is used to solve it. The optimal solution is gradually approximated through linear relaxation, branch operation and cutting plane method, and the macro capacity expansion planning scheme containing the investment time sequence and capacity scale of each stage is output. At the micro-operation layer, the capacity expansion planning scheme generated at the macro-level is dynamically coupled with the spatiotemporal enhanced grid topology map in the grid digital twin base. Based on the engineering construction sequence determined by critical path analysis, the completion time of each project on the critical path is modeled using the PERT-based three-point estimation method. Several engineering construction progress scenarios are generated through Beta distribution to obtain a comprehensive test scenario set that includes grid structure evolution and resource fluctuations. For each comprehensive test scenario, hourly operation simulation of the annual operation cycle is performed based on a preset time step. At each simulation time step, a node admittance matrix is ​​established based on the grid structure and power generation resources that are in operation at the current moment. The forward-backward power flow algorithm is used to calculate the system power flow distribution, record the deviation of the node voltage amplitude from the rated value, and the ratio of the active power flow to the thermal stability limit of each transmission line. The annual hourly simulation results are then output. By using the hourly simulation results throughout the year, the voltage qualification rate of each key node and the load rate of key lines are statistically analyzed to generate power grid safety margin assessment information and compare it with the preset safety standards. If it does not meet the preset safety standards, a penalty function is generated based on the deviation from the preset safety standards and fed back to the objective function of the macro level for re-solution, thereby achieving bidirectional coupling optimization. The optimization is repeatedly iterated through a two-way coupling optimization mechanism. The optimization is terminated when the solution set meets the convergence condition of the hypervolume index. Finally, a set of phased implementation path schemes that are coordinated with the actual evolution process of the power grid infrastructure is output.

[0025] It should be noted that, Based on the power grid evolution event table in the power grid digital twin base, the planned commissioning time, construction period, and logical dependencies between all planned upgrade projects are analyzed, and a project dependency network is constructed accordingly. In this network, nodes represent specific engineering projects, and directed edges represent the necessary construction order between projects. The critical path method is used to analyze this network, calculating the earliest and latest time parameters of each node, identifying the critical path with zero total float, which is the core project sequence that determines the shortest construction period for the entire planning period. Subsequently, Jenks' natural breakpoint method is introduced to optimally segment the time points of this critical project sequence. By minimizing intra-class variance and maximizing inter-class variance, a reasonable construction phase division scheme within the planning period is determined, providing a time framework for subsequent phased investment decisions.

[0026] Based on the established phase division scheme, a macro- and micro-level bidirectional coupling optimization architecture is constructed. The macro-level layer, with the core objective of minimizing the present value of the entire lifecycle cost during the planning period, establishes a linear programming model. Its decision variables are the investment decisions and capacity allocation variables for each stage and each candidate geographical location, and are subject to physical constraints such as Kirchhoff's Current Law, transmission channel capacity limits, and real-time energy supply and demand balance. A branch-and-bound algorithm is used to solve this model, gradually approximating the optimal solution through linear relaxation, branch operations, and cutting plane generation, outputting a macro-level capacity expansion planning scheme that clarifies the types of facilities, capacity scale, and investment sequence to be constructed in each stage. To verify the engineering feasibility of the macro-level scheme, it is dynamically coupled with the spatiotemporal enhanced topology map in the power grid digital twin base, entering the micro-level operation layer. Considering the potential uncertainties in engineering construction on the critical path, the three-point estimation method of the Program Evaluation and Review Technique (PERT) is used to model its completion time, and a large number of possible engineering construction progress scenarios are generated through Beta distribution to form a test scenario set. For each test scenario, a year-long operational simulation is conducted with a preset time step. Within each time step, a node admittance matrix is ​​established based on the current operational power grid structure, and a forward-backward power flow algorithm is used to accurately calculate the system power flow distribution. Key safety indicators such as node voltage deviation and line load rate are recorded. Based on the extensive results generated from the micro-operational simulation, the voltage compliance rate of key nodes and the load rate of key lines are statistically evaluated, generating a quantitative power grid safety margin assessment report.

[0027] The assessment results are compared with preset safety standards. If any violations are found, a corresponding penalty function is constructed based on the severity of the violation, and this penalty term is fed back into the objective function of the macro-level optimization model. The macro-level model recalculates based on the feedback information and adjusts the capacity expansion scheme. Through this two-way coupling mechanism of "macro-optimization-micro-verification-feedback correction," multiple iterations are performed. When the set of solutions meets the convergence condition in terms of performance indicators (measured by the hypervolume index), the optimization process terminates. The final output is a set of phased implementation path schemes that not only meet the economic objectives but also ensure coordination with the actual evolution of the power grid at each construction stage and possess high engineering feasibility.

[0028] Furthermore, in a preferred embodiment of the present invention, the introduction of an adversarial game mechanism, constructing a virtual stress test scenario based on the phased implementation path scheme set, evaluating the adaptability of each path throughout its entire lifecycle, and obtaining the optimal implementation path through adaptability screening to form an optimized implementation path set, specifically includes: A scene generator agent containing a generator and a discriminator is constructed using a conditional generative adversarial network. The generator learns the joint distribution of potential risk factors and the reconstruction of virtual stress test scenarios through an encoder-decoder structure, while the discriminator distinguishes between real historical events and generated scenarios based on a spatiotemporal convolutional neural network. Using historical extreme weather events and equipment failure records as training samples, the scene generator agent that meets the expectations is obtained after training through minimax game. A phased implementation path scheme set is obtained. The scenario generator intelligent agent generates a virtual stress test scenario for each candidate implementation path in the phased implementation path scheme set. Each candidate path and the corresponding stress test scenario are input into the power grid digital twin base. The sequential Monte Carlo simulation method is used to perform full life cycle adaptive analysis, calculate the key performance indicators of each path under multiple stress scenarios, and generate a performance profile of each candidate implementation path. Based on key performance indicators, an adaptive evaluation system is constructed using the entropy weight method. According to the performance profile of each candidate implementation path, the adaptive evaluation system is used to weight and aggregate each performance indicator to obtain the adaptive score of each candidate implementation path under different pressure scenarios, thus obtaining path adaptive evaluation information. Based on the path adaptability evaluation information, candidate implementation paths with a value greater than the preset adaptability threshold are selected as high-quality implementation paths at the current adversarial game moment to generate a subset of high-quality paths. The performance characteristics of the high-quality paths are then fed back to the scene generator agent for generation strategy adjustment, thus completing one adversarial game optimization. Repeat the above adversarial game optimization steps. When the coefficient of variation of the path fitness score of the latest generation path set is less than the preset coefficient of variation threshold and the scene generator agent cannot generate a new scene that causes the path fitness score to drop by more than the preset drop, terminate the game process and output the optimized implementation path set.

[0029] It should be noted that an adversarial game mechanism is introduced to select highly robust planning paths through dynamic game among agents. A conditional generative adversarial network architecture is used to construct the scenario generator agent. The generator is based on an encoder-decoder structure. The encoder maps risk factors such as historical extreme weather events and equipment failure records to a latent space, learning their joint probability distribution. The decoder then reconstructs challenging virtual stress test scenarios based on this distribution. The discriminator uses a spatiotemporal convolutional neural network, leveraging its powerful spatiotemporal feature extraction capabilities to accurately distinguish between real historical events and generated scenarios. The two are trained adversarially through a minimax game, ultimately resulting in a mature agent capable of generating reasonable and severe test scenarios.

[0030] After obtaining a well-trained scenario generator, it is applied to each candidate path in the phased implementation path scheme set. The generator generates a series of virtual stress test scenarios for each path, targeting its technical characteristics and weaknesses. These scenarios may include adverse conditions such as extreme weather sequences, cascading equipment failures, and sudden changes in market demand. Each candidate path is combined with its corresponding stress test scenario and input into the power grid digital twin base. A sequential Monte Carlo simulation method is used for full lifecycle adaptive analysis. Through numerous repeated simulations, key performance indicators for each path under multiple stresses are calculated and statistically analyzed, such as system load failure probability, expected power supply shortage, and operating cost overrun rate. This generates a comprehensive performance profile for each candidate path, quantifying its performance under different adversities. Based on these key performance indicators, an adaptive evaluation system is constructed using the entropy weight method. The entropy weight method automatically assigns weights based on the dispersion of each indicator data, avoiding subjective assumptions. Subsequently, the performance profiles of each candidate path are weighted and aggregated using this evaluation system to calculate its comprehensive adaptive score under various stress scenarios. Finally, the adaptive evaluation information for all paths is obtained, revealing the risk resistance capabilities of each path.

[0031] Subsequently, an adversarial game optimization loop is initiated. Based on path fitness evaluation information, high-quality paths with fitness scores above a preset threshold are selected, forming a subset of high-quality paths. Then, the performance characteristics of these high-quality paths, especially their remaining relative weaknesses, are fed back to the scenario generator agent as feedback. The scenario generator adjusts its generation strategy accordingly, focusing on generating more challenging stress test scenarios targeting the weaknesses of these high-quality paths. This game step is repeated, with the scenario generator continuously attempting to "beat" the existing high-quality paths, while the path set evolves under stress testing. When the coefficient of variation of the fitness scores of the high-quality path subset stabilizes, and the scenario generator struggles to generate new scenarios that significantly reduce path fitness scores, the game is considered to have reached equilibrium, the optimization process terminates, and the final output is the optimized implementation path set.

[0032] Furthermore, in a preferred embodiment of the present invention, the step of constructing an evolutionary topology graph of phased paths using dynamic network analysis based on the optimized implementation path set, and generating a path evolution correlation graph through stage node centrality analysis and path evolution trajectory mining, specifically includes: Obtain an optimized implementation path set, and perform full-cycle feature extraction on each optimized implementation path based on the optimized implementation path set to generate a time-series path feature sequence; define Euclidean distance as a local similarity measure, and use the dynamic time warping algorithm to calculate the similarity between paths based on the time-series path feature sequence to finally obtain a path similarity matrix; A path evolution topology graph is constructed based on the path similarity matrix, where nodes represent different implementation paths, edges represent the similarity relationship between paths, and the weight of the edges is determined by the path similarity. A community detection algorithm is used to divide the path evolution topology graph into communities, and multi-level clustering is used to identify route clusters, resulting in an undirected weighted network graph that reveals the potential transformation relationship between paths. Based on the undirected weighted network graph, a multi-dimensional node centrality analysis is performed. The PageRank algorithm based on random walk is used to calculate the global influence of nodes through multiple rounds of iterative propagation. The betweenness centrality index based on the shortest path is used to identify key bridge nodes in the network. The connection quality importance of nodes is analyzed by combining the feature vector centrality, and finally the node centrality analysis results are obtained. Based on the undirected weighted network graph and node centrality analysis results, the path evolution trajectory is deeply mined. The identified technical route clusters are taken as hidden states, and the temporal transition relationships between paths are taken as observation sequences. The state transition probability matrix and observation probability matrix are calculated by the Baum-Welch algorithm, and the Viterbi algorithm is used to decode the optimal state sequence to identify the evolution mode, thereby obtaining path evolution trajectory analysis information. Based on the undirected weighted network graph, node centrality analysis results, and path evolution trajectory analysis information, a path evolution association graph is constructed. A force-oriented layout algorithm based on repulsion and attraction is used for visual spatial arrangement. The node size is set according to the node centrality analysis results, color coding is used to distinguish them according to the community division results, the thickness of the connecting edges is determined according to the similarity weight, and key bifurcation points, turning points, and typical evolution trajectories of the technical route are marked. Finally, the path evolution association graph is output.

[0033] It should be noted that, firstly, the optimized implementation path set is subjected to feature serialization processing to extract the core technical and economic indicators of each path at each stage within the entire planning cycle, forming a path feature sequence with temporal characteristics. Based on this, a dynamic time warping algorithm is used to calculate the morphological similarity between paths. This algorithm effectively overcomes the phase differences in the construction time sequence of different paths by constructing a cumulative distance matrix and finding the optimal curved path. Using Euclidean distance as a local similarity measure, a path similarity matrix that accurately reflects the overall evolutionary similarity between paths is finally obtained. A path evolution topology network graph is constructed based on the path similarity matrix, where nodes represent different implementation paths, edges represent the similarity relationships between paths, and the edge weights are determined by the path similarity metric. The Louvain community detection algorithm is used to partition the network into communities. Through a multi-level clustering process with modularity optimization, tightly connected technical route clusters are identified, forming an undirected weighted network graph that reveals the potential transformation relationships between paths. Based on this network structure, a multi-dimensional node centrality analysis is conducted. The PageRank algorithm based on the random walk model is used to calculate the global influence of nodes through multiple rounds of iterative propagation. The betweenness centrality is combined to identify key bridge nodes in the network, and the eigenvector centrality is used to evaluate the importance of node connectivity quality. Finally, a comprehensive node centrality evaluation system is established.

[0034] Furthermore, based on the undirected weighted network graph and node centrality analysis results, a Hidden Markov Model (HMM) is employed for in-depth mining of path evolution trajectories. The identified technical route clusters are used as the hidden states of the model, and the temporal transition relationships between paths are used as the observation sequences. The Baum-Welch algorithm is used to train the model parameters, obtaining the state transition probability matrix and the observation probability matrix. The Viterbi algorithm is used to decode the optimal state sequence, identify typical evolution patterns, and calculate characteristic parameters such as the first passage time between states, thereby obtaining complete path evolution trajectory analysis information. Finally, the undirected weighted network graph, node centrality analysis results, and path evolution trajectory analysis information are integrated to construct a multi-dimensional path evolution association graph. A force-directed layout algorithm based on repulsion and attraction is used for visual spatial arrangement. The display size is set according to the PageRank value of the nodes, color coding is used to distinguish them based on the community partitioning results, the thickness of the connecting edges is determined according to the similarity weight, and key bifurcation points, strategic turning points, and typical evolution trajectories of the technical routes are clearly marked in the graph. The final output is a path evolution correlation map that can intuitively show the correlation strength, evolution direction and conversion probability between paths.

[0035] Furthermore, in a preferred embodiment of the present invention, the step of acquiring real-time power grid construction progress and operation status data of the target area, combining the path evolution correlation map to perform path performance deviation assessment and root cause diagnosis, and optimizing the planned implementation path based on the diagnosis results, generating an optimized planned implementation path scheme for push, specifically includes: The real-time power grid construction progress and operation status data of the target area are obtained, the obtained real-time power grid construction progress and operation status data are preprocessed, and spatiotemporal alignment matching is performed with the expected status characteristics of the corresponding time nodes in the path evolution association map to form an implementation status feature vector reflecting the actual construction and operation status. The expected state sequence of each reference path is extracted by the path evolution association map. The instantaneous deviation in three dimensions—installed capacity, power grid structure parameters, and operation performance indicators—is calculated by combining the implementation state feature vector with the dynamic time warping algorithm. The cumulative deviation of each dimension is obtained by time integration and the overall deviation of the path is obtained by weighted fusion. The overall deviation of the path is compared with a preset threshold. If it is greater than the preset deviation threshold, the root cause diagnosis analysis is performed based on the isolated forest algorithm. The indicators of each dimension in the path feature sequence are used as input features. Multiple isolated trees are constructed and the path length of data points is calculated to identify the abnormal feature dimensions and obtain abnormal feature identification information. The node centrality features and anomaly features obtained by the path evolution association graph are input into the pre-trained Bayesian inference network. The root cause of the path deviation and its propagation path are determined by posterior probability inference, and root cause diagnosis information is generated. Based on root cause diagnosis information and path evolution trajectory information in the path evolution correlation map, a constrained model predictive control method is used for rolling optimization calculation. With the optimization objectives of minimizing path adjustment costs and maximizing path fit in subsequent stages, the capacity configuration scheme and project construction sequence of subsequent construction stages are replanned based on the current actual path state. Power grid safety operation constraints and technical feasibility constraints are set, and the optimal adjustment strategy is solved through a sequential quadratic programming algorithm. Path optimization suggestions are generated and pushed out to realize the optimization and dynamic adjustment of the planned implementation path.

[0036] It should be noted that construction progress and operational status data are acquired through a data acquisition system. After preprocessing, this data is spatiotemporally aligned and matched with the expected state characteristics of corresponding time nodes in the path evolution correlation map to form an implementation state feature vector reflecting the actual construction and operation status. Based on this, the expected state sequences of each reference path are extracted from the path evolution correlation map. Combining this with the implementation state feature vector, a dynamic time warping algorithm is used to calculate the instantaneous deviation in three key dimensions: installed capacity, power grid structure parameters, and operational performance indicators. The cumulative deviation in each dimension is then obtained through time integration. Finally, a weighted fusion based on weights determined by the entropy weight method is performed to obtain a quantified overall path deviation index.

[0037] When the overall deviation of the path exceeds a preset threshold, the root cause diagnosis and analysis process is initiated. First, multiple isolated trees are constructed based on the isolated forest algorithm. Each dimension of the path feature sequence is used as input features, and the path length of data points is calculated to identify abnormal feature dimensions, obtaining preliminary abnormal feature identification information. Subsequently, the node centrality features in the path evolution correlation graph and the abnormal feature identification information are combined and input into a pre-trained Bayesian network for deep inference. Posterior probability derivation is performed using methods such as Gibbs sampling to determine the root cause of the path deviation and its propagation path among technical route clusters, generating root cause diagnosis information containing primary causes, secondary factors, and influencing paths. Based on the root cause diagnosis information and the path evolution trajectory information in the path evolution correlation graph, a constrained model predictive control method is used for rolling optimization calculations. With the dual optimization objectives of minimizing path adjustment costs and maximizing path fit in subsequent stages, and considering the current actual path state, the capacity configuration scheme and project construction sequence for subsequent construction stages are replanned, taking into full account power grid safety constraints and technical feasibility constraints. The optimal adjustment strategy is solved by using a sequential quadratic programming algorithm, generating a path optimization suggestion scheme that includes a detailed implementation plan, resource allocation plan and time node arrangement. This scheme is then pushed to the relevant decision management system, thereby realizing the continuous optimization and dynamic adjustment of the planned implementation path.

[0038] Figure 2A flowchart of the second method for evaluating the implementation path of new energy and energy storage planning in a block grid, as provided in an embodiment of the present invention; like Figure 2 As shown, the present invention provides a second method flowchart for evaluating the implementation path of new energy and energy storage planning in block power grids, including: S202, Obtain the spatiotemporal enhanced power grid topology map, extract features from all graph nodes of the spatiotemporal enhanced power grid topology map, including resource density percentile value, load growth rate, topological centrality index and electrical betweenness, and form a multidimensional node feature matrix after Z-score normalization of the extracted features; S204, introduce spectral clustering algorithm for block division, calculate the similarity matrix between nodes based on the multidimensional node feature matrix, use an improved Gaussian kernel function to define the similarity between nodes, where the kernel parameter is determined according to the k nearest neighbor distance, and construct the degree matrix through the calculated similarity matrix; S206, calculate the normalized Laplacian matrix using the constructed degree matrix, perform eigenvalue decomposition on the normalized Laplacian matrix, select the eigenvectors corresponding to the first k smallest eigenvalues ​​to form a dimension-reduced feature space, then use the K-means clustering algorithm to perform clustering, determine the optimal number of clusters using the silhouette coefficient method, divide all network nodes into k blocks with similar features, and output the block division results; S208. Establish a multi-dimensional potential evaluation system, which includes resource endowment index, load concentration index, power grid support index and development coordination index. The weight of the index is determined by using the entropy weight CRITIC combination weighting method. The development coordination index is quantified by calculating the spatiotemporal matching degree between the existing line capacity utilization rate and the planned expansion project in the block. S210 utilizes a multi-dimensional potential evaluation system to obtain the comprehensive potential score of each block, generates a regional score distribution map, and divides the blocks into three levels: high potential area, medium potential area, and low potential area based on preset potential level thresholds. The block potential evaluation results are mapped to a spatiotemporal enhanced power grid topology map, and each node is assigned a potential level label of its block, ultimately obtaining a potential point-marked topology map.

[0039] It should be noted that the multi-dimensional feature extraction of all nodes in the power grid diagram covers key parameters such as resource density percentile, load growth rate, topological centrality index, and electrical betweenness. These features quantify the development value of nodes from different perspectives, including resource endowment, energy demand, network structure, and electrical function. After eliminating the influence of dimensions using the Z-score standardization method, a multi-dimensional node feature matrix is ​​formed. A spectral clustering algorithm is introduced for intelligent partitioning of power grid blocks. First, a Gaussian kernel function is used to calculate the similarity matrix between nodes, where the kernel parameter is adaptively determined based on the k-nearest neighbor distance. A degree matrix is ​​constructed using the similarity matrix, and a normalized Laplace matrix is ​​further calculated. After eigenvalue decomposition of this matrix, the eigenvectors corresponding to the k smallest eigenvalues ​​are selected to form a dimensionality-reduced feature space, mapping the original high-dimensional features to a low-dimensional spectral space. Subsequently, the K-means algorithm is applied to perform cluster analysis in this space, while the silhouette coefficient method is used to evaluate the clustering quality and determine the optimal number of clusters. Finally, all nodes in the network are divided into several blocks with similar technical characteristics.

[0040] Based on the block division, a multi-dimensional potential evaluation system is established, encompassing resource endowment, load concentration, grid support, and development coordination. Development coordination assesses infrastructure synergy by quantifying the spatiotemporal matching between existing line capacity utilization and planned expansion projects within a block. The CRITIC weighting method is used to objectively determine the weights of each indicator, considering both the dispersion of indicator data and the potential conflicts between indicators to ensure the rationality and scientific nature of weight allocation. This evaluation system is used to calculate the comprehensive potential score for each block, generating a visualized regional score distribution map. Finally, based on preset potential level thresholds, each block is divided into high, medium, and low potential levels, and the evaluation results are fed back into the spatiotemporally enhanced grid topology map, assigning a corresponding potential level label to each node. The resulting potential point-marked topology map not only retains the original grid topology and spatiotemporal characteristics but also overlays scientific block division and potential evaluation information, providing an intuitive and quantitative spatial reference for grid planning decisions and effectively supporting the orderly advancement of new energy and energy storage development and the optimal allocation of resources.

[0041] Figure 3 An evaluation system 3 for the implementation path of new energy and energy storage planning in a block grid is provided in one embodiment of the present invention. The system includes: a memory 301, a processor 302, and a communication interface 303. The memory 301 contains a program for evaluating the implementation path of new energy and energy storage planning in a block grid. When the program for evaluating the implementation path of new energy and energy storage planning in a block grid is executed by the processor 302, it performs the following steps: Acquire data on grid load characteristics, distribution of new energy and energy storage resources, grid topology, and energy storage system parameters for the target area. Simultaneously, collect data on the expansion plan and capacity improvement roadmap of the regional power grid. Construct a digital twin base for the power grid through multi-source heterogeneous data analysis. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, resulting in a set of phased implementation path schemes that are coordinated with power grid evolution. An adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path in the whole life cycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set. Based on the optimized implementation path set, a dynamic network analysis method is used to construct the evolution topology of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining. The system acquires real-time power grid construction progress and operation status data of the target block, combines the path evolution correlation map to evaluate path performance deviation and diagnose root causes, and performs rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, generating and pushing out optimized plans for the planned implementation path.

[0042] In another aspect, the present invention provides a computer-readable storage medium including a method program for evaluating the implementation path of new energy and energy storage planning for block grids. When the method program is executed by a processor, it implements the steps of the method for evaluating the implementation path of new energy and energy storage planning for block grids as described in any of the preceding claims.

[0043] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0044] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0045] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0046] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0047] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0048] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An evaluation method for the implementation path of new energy and energy storage planning in block power grids, characterized in that, include: Acquire data on grid load characteristics, distribution of new energy and energy storage resources, grid topology, and energy storage system parameters for the target area. Simultaneously, collect data on the expansion plan and capacity improvement roadmap of the regional power grid. Construct a digital twin base for the power grid through multi-source heterogeneous data analysis. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, resulting in a set of phased implementation path schemes that are coordinated with power grid evolution. An adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path in the whole life cycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set. Based on the optimized implementation path set, a dynamic network analysis method is used to construct the evolution topology of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining. The system acquires real-time power grid construction progress and operation status data of the target block, combines the path evolution correlation map to evaluate path performance deviation and diagnose root causes, and performs rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, generating and pushing out optimized plans for the planned implementation path.

2. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 1, characterized in that, The process involves acquiring power grid load characteristic data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters for the target area. Simultaneously, it collects regional power grid expansion plans and capacity enhancement roadmaps. A digital twin foundation for the power grid is constructed through multi-source heterogeneous data analysis. Specifically, this includes: By introducing a big data network, the target area's power grid load characteristics data, new energy and energy storage resource distribution data, power grid topology data, and energy storage system parameters are obtained through big data retrieval. At the same time, the expansion plan and capacity improvement roadmap of the regional power grid are collected to obtain a multi-source heterogeneous dataset. Outliers and missing values ​​are identified and repaired in the multi-source heterogeneous dataset. The seven-parameter coordinate transformation method is used to unify the spatial reference systems of different data sources to the same coordinate system. The linear interpolation method is used to normalize various time series data to the same time granularity. A globally unique identifier is created for each power grid entity based on the preset coding rules to generate a standardized dataset. Based on the standardized dataset, the power grid topology features are extracted and imported into the Neo4j graph database. They are then converted into a set of nodes and branches using a CIM / XML parser based on the SAX parser. Substation buses are abstracted as vertices, and transmission lines and transformers are abstracted as edges with resistance, reactance, capacitance to ground, and susceptance parameters. A power grid topology attribute graph that supports shortest path, connected component, and power flow tracking calculations is then established. Based on the power grid topology attribute map, the ArcGIS Pro spatial analysis engine is integrated to perform geographic information fusion. By establishing spatial indexes and coordinate mapping relationships, a dynamic buffer area is created for each electrical node. The resource density and load distribution characteristics within each buffer are calculated using a partitioning statistical tool. These indicators are used as derived attributes and updated to the corresponding nodes through node ID association to generate a spatiotemporally enhanced power grid topology map. Deep semantic parsing is performed on unstructured planning texts in standardized datasets. A pre-trained language model based on the BERT architecture is used for deep semantic understanding. Semantic role labeling technology is used to identify the predicate-argument structure in the text. Conditional random fields are used for named entity recognition. Structured metadata quadruples containing latitude and longitude coordinates, technology type, capacity target and time node are extracted to obtain the power grid evolution event table. Using the spatiotemporally enhanced power grid topology map as input, feature vectors of all nodes in the map are extracted to construct a regional feature matrix. Spectral clustering algorithm is used for block division and potential level evaluation to generate a potential point-marked topology map. The system is then integrated with a standardized dataset, the spatiotemporally enhanced power grid topology map, and a power grid evolution event table to construct a digital twin base for the power grid.

3. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 2, characterized in that, The process of using the spatiotemporal enhanced power grid topology map as input, extracting feature vectors from all nodes in the map to construct a regional feature matrix, employing a spectral clustering algorithm for block division and potential level evaluation, and generating a potential point-marked topology map specifically includes: A spatiotemporal enhanced power grid topology map is obtained, and features are extracted from all graph nodes of the spatiotemporal enhanced power grid topology map, including resource density percentile value, load growth rate, topological centrality index and electrical betweenness. The extracted features are then Z-score standardized to form a multidimensional node feature matrix. A spectral clustering algorithm is introduced for block partitioning. The similarity matrix between nodes is calculated based on the multidimensional node feature matrix. An improved Gaussian kernel function is used to define the similarity between nodes, where the kernel parameter is determined based on the k-nearest neighbor distance. The degree matrix is ​​constructed using the calculated similarity matrix. The normalized Laplacian matrix is ​​calculated using the constructed degree matrix. The normalized Laplacian matrix is ​​then subjected to eigenvalue decomposition. The eigenvectors corresponding to the k smallest eigenvalues ​​are selected to form a dimensionality-reduced feature space. Subsequently, the K-means clustering algorithm is used for clustering. The optimal number of clusters is determined by the silhouette coefficient method. All network nodes are divided into k blocks with similar features, and the block division results are output. A multi-dimensional potential evaluation system is established, which includes resource endowment index, load concentration index, power grid support index and development coordination index. The weights of the indexes are determined by using the entropy weight CRITIC combination weighting method. The development coordination index is quantified by calculating the spatiotemporal matching degree between the existing line capacity utilization rate and the planned expansion project within the block. The comprehensive potential score of each block is obtained by using a multi-dimensional potential evaluation system, and a regional score distribution map is generated. The blocks are divided into three levels: high potential area, medium potential area and low potential area by combining the preset potential level threshold. The block potential evaluation results are mapped to the spatiotemporal enhanced power grid topology map and each node is assigned a potential level label of its block. Finally, a potential point labeled topology map is obtained.

4. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 1, characterized in that, Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is used for multi-stage capacity expansion planning and power grid safety margin calculation, resulting in a set of phased implementation path schemes coordinated with power grid evolution, specifically including: Obtain the power grid digital twin base, extract the power grid evolution event table based on the power grid digital twin base, parse the power grid upgrade project information recorded in the event table, extract the planned commissioning time node, construction duration and logical dependency relationship between projects for each project, and construct a project dependency relationship network, where nodes represent power grid construction projects and directed edges represent the sequential dependency relationship between projects; Based on the project dependency network, the critical path method is used to calculate the earliest start time, latest finish time and total float of each node. The project sequence with zero total float is taken as the critical path. Jenks' natural break method is introduced. The time point sequence of the critical path projects is taken as input. The optimal segmentation scheme is determined by calculating the intra-class variance and inter-class variance ratio. Finally, the stage division scheme is output. A macro-micro bidirectional coupling optimization architecture is constructed, which includes a macro layer and a micro layer. In the macro layer, a linear programming model is established with the optimization objective of minimizing the present value of the total life cycle cost. The decision variables are the investment decisions and capacity variables of each stage and each candidate node. Constraints are set according to Kirchhoff's current law, transmission capacity constraints, and energy supply and demand balance constraints. The aforementioned stage division scheme is input into the linear programming model established in the macro layer, and the branch and bound algorithm is used to solve it. The optimal solution is gradually approximated through linear relaxation, branch operation and cutting plane method, and the macro capacity expansion planning scheme containing the investment time sequence and capacity scale of each stage is output. At the micro-operation layer, the capacity expansion planning scheme generated at the macro-level is dynamically coupled with the spatiotemporal enhanced grid topology map in the grid digital twin base. Based on the engineering construction sequence determined by critical path analysis, the completion time of each project on the critical path is modeled using the PERT-based three-point estimation method. Several engineering construction progress scenarios are generated through Beta distribution to obtain a comprehensive test scenario set that includes grid structure evolution and resource fluctuations. For each comprehensive test scenario, hourly operation simulation of the annual operation cycle is performed based on a preset time step. At each simulation time step, a node admittance matrix is ​​established based on the grid structure and power generation resources that are in operation at the current moment. The forward-backward power flow algorithm is used to calculate the system power flow distribution, record the deviation of the node voltage amplitude from the rated value, and the ratio of the active power flow to the thermal stability limit of each transmission line. The annual hourly simulation results are then output. By using the hourly simulation results throughout the year, the voltage qualification rate of each key node and the load rate of key lines are statistically analyzed to generate power grid safety margin assessment information and compare it with the preset safety standards. If it does not meet the preset safety standards, a penalty function is generated based on the deviation from the preset safety standards and fed back to the objective function of the macro level for re-solution, thereby achieving bidirectional coupling optimization. The optimization is repeatedly iterated through a two-way coupling optimization mechanism. The optimization is terminated when the solution set meets the convergence condition of the hypervolume index. Finally, a set of phased implementation path schemes that are coordinated with the actual evolution process of the power grid infrastructure is output.

5. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 1, characterized in that, The introduction of an adversarial game mechanism involves constructing a virtual stress test scenario based on the phased implementation path scheme set, evaluating the adaptability of each path throughout its lifecycle, and obtaining the optimal implementation path through adaptability screening to form an optimized implementation path set, specifically including: A scene generator agent containing a generator and a discriminator is constructed using a conditional generative adversarial network. The generator learns the joint distribution of potential risk factors and the reconstruction of virtual stress test scenarios through an encoder-decoder structure, while the discriminator distinguishes between real historical events and generated scenarios based on a spatiotemporal convolutional neural network. Using historical extreme weather events and equipment failure records as training samples, the scene generator agent that meets the expectations is obtained after training through minimax game. A phased implementation path scheme set is obtained. The scenario generator intelligent agent generates a virtual stress test scenario for each candidate implementation path in the phased implementation path scheme set. Each candidate path and the corresponding stress test scenario are input into the power grid digital twin base. The sequential Monte Carlo simulation method is used to perform full life cycle adaptive analysis, calculate the key performance indicators of each path under multiple stress scenarios, and generate a performance profile of each candidate implementation path. Based on key performance indicators, an adaptive evaluation system is constructed using the entropy weight method. According to the performance profile of each candidate implementation path, the adaptive evaluation system is used to weight and aggregate each performance indicator to obtain the adaptive score of each candidate implementation path under different pressure scenarios, thus obtaining path adaptive evaluation information. Based on the path adaptability evaluation information, candidate implementation paths with a value greater than the preset adaptability threshold are selected as high-quality implementation paths at the current adversarial game moment to generate a subset of high-quality paths. The performance characteristics of the high-quality paths are then fed back to the scene generator agent for generation strategy adjustment, thus completing one adversarial game optimization. Repeat the above adversarial game optimization steps. When the coefficient of variation of the path fitness score of the latest generation path set is less than the preset coefficient of variation threshold and the scene generator agent cannot generate a new scene that causes the path fitness score to drop by more than the preset drop, terminate the game process and output the optimized implementation path set.

6. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 1, characterized in that, The step of constructing a phased path evolution topology map based on the optimized implementation path set using dynamic network analysis methods, and generating a path evolution correlation graph through stage node centrality analysis and path evolution trajectory mining, specifically includes: Obtain an optimized implementation path set, and perform full-cycle feature extraction on each optimized implementation path based on the optimized implementation path set to generate a time-series path feature sequence; define Euclidean distance as a local similarity measure, and use the dynamic time warping algorithm to calculate the similarity between paths based on the time-series path feature sequence to finally obtain a path similarity matrix; A path evolution topology graph is constructed based on the path similarity matrix, where nodes represent different implementation paths, edges represent the similarity relationship between paths, and the weight of the edges is determined by the path similarity. A community detection algorithm is used to divide the path evolution topology graph into communities, and multi-level clustering is used to identify route clusters, resulting in an undirected weighted network graph that reveals the potential transformation relationship between paths. Based on the undirected weighted network graph, a multi-dimensional node centrality analysis is performed. The PageRank algorithm based on random walk is used to calculate the global influence of nodes through multiple rounds of iterative propagation. The betweenness centrality index based on the shortest path is used to identify key bridge nodes in the network. The connection quality importance of nodes is analyzed by combining the feature vector centrality, and finally the node centrality analysis results are obtained. Based on the undirected weighted network graph and node centrality analysis results, the path evolution trajectory is deeply mined. The identified technical route clusters are taken as hidden states, and the temporal transition relationships between paths are taken as observation sequences. The state transition probability matrix and observation probability matrix are calculated by the Baum-Welch algorithm, and the Viterbi algorithm is used to decode the optimal state sequence to identify the evolution mode, thereby obtaining path evolution trajectory analysis information. Based on the undirected weighted network graph, node centrality analysis results, and path evolution trajectory analysis information, a path evolution association graph is constructed. A force-oriented layout algorithm based on repulsion and attraction is used for visual spatial arrangement. The node size is set according to the node centrality analysis results, color coding is used to distinguish them according to the community division results, the thickness of the connecting edges is determined according to the similarity weight, and key bifurcation points, turning points, and typical evolution trajectories of the technical route are marked. Finally, the path evolution association graph is output.

7. The evaluation method for the implementation path of new energy and energy storage planning in a block power grid according to claim 1, characterized in that, The process involves acquiring real-time power grid construction progress and operational status data for the target area, combining this data with the path evolution correlation map to assess path performance deviation and diagnose root causes, optimizing the planned implementation path based on the diagnostic results, and generating and pushing out an optimized planned implementation path scheme. Specifically, this includes: The real-time power grid construction progress and operation status data of the target area are obtained, the obtained real-time power grid construction progress and operation status data are preprocessed, and spatiotemporal alignment matching is performed with the expected status characteristics of the corresponding time nodes in the path evolution association map to form an implementation status feature vector reflecting the actual construction and operation status. The expected state sequence of each reference path is extracted by the path evolution association map. The instantaneous deviation in three dimensions—installed capacity, power grid structure parameters, and operation performance indicators—is calculated by combining the implementation state feature vector with the dynamic time warping algorithm. The cumulative deviation of each dimension is obtained by time integration and the overall deviation of the path is obtained by weighted fusion. The overall deviation of the path is compared with a preset threshold. If it is greater than the preset deviation threshold, the root cause diagnosis analysis is performed based on the isolated forest algorithm. The indicators of each dimension in the path feature sequence are used as input features. Multiple isolated trees are constructed and the path length of data points is calculated to identify the abnormal feature dimensions and obtain abnormal feature identification information. The node centrality features and anomaly features obtained by the path evolution association graph are input into the pre-trained Bayesian inference network. The root cause of the path deviation and its propagation path are determined by posterior probability inference, and root cause diagnosis information is generated. Based on root cause diagnosis information and path evolution trajectory information in the path evolution correlation map, a constrained model predictive control method is used for rolling optimization calculation. With the optimization objectives of minimizing path adjustment costs and maximizing path fit in subsequent stages, the capacity configuration scheme and project construction sequence of subsequent construction stages are replanned based on the current actual path state. Power grid safety operation constraints and technical feasibility constraints are set, and the optimal adjustment strategy is solved through a sequential quadratic programming algorithm. Path optimization suggestions are generated and pushed out to realize the optimization and dynamic adjustment of the planned implementation path.

8. An evaluation system for the planning and implementation path of new energy and energy storage in block grids, characterized in that, The system includes: a memory, a processor, and a communication interface. The memory contains an evaluation method program for the implementation path of new energy and energy storage planning in the blockchain power grid. When the processor executes the evaluation method program for the implementation path of new energy and energy storage planning in the blockchain power grid, it performs the following steps: Acquire data on grid load characteristics, distribution of new energy and energy storage resources, grid topology, and energy storage system parameters for the target area. Simultaneously, collect data on the expansion plan and capacity improvement roadmap of the regional power grid. Construct a digital twin base for the power grid through multi-source heterogeneous data analysis. Based on the aforementioned power grid digital twin base, the planning period is divided into multiple construction phases. A macro-micro bidirectional coupling optimization architecture is adopted to perform multi-stage capacity expansion planning and power grid security margin calculation, resulting in a set of phased implementation path schemes that are coordinated with power grid evolution. An adversarial game mechanism is introduced to construct a virtual stress test scenario based on the phased implementation path scheme set, and the adaptability of each path in the whole life cycle is evaluated. The optimal implementation path is obtained through adaptability screening, forming an optimized implementation path set. Based on the optimized implementation path set, a dynamic network analysis method is used to construct the evolution topology of the phased paths, and a path evolution correlation map is generated through stage node centrality analysis and path evolution trajectory mining. The system acquires real-time power grid construction progress and operation status data of the target block, combines the path evolution correlation map to evaluate path performance deviation and diagnose root causes, and performs rolling optimization and dynamic adjustment of the planned implementation path based on the diagnosis results, generating and pushing out optimized plans for the planned implementation path.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes an evaluation method program for the implementation path of new energy and energy storage planning in a block grid. When the evaluation method program for the implementation path of new energy and energy storage planning in a block grid is executed by a processor, it implements the steps of the evaluation method for the implementation path of new energy and energy storage planning in a block grid as described in any one of claims 1 to 7.

Images