Power transmission congestion management partition pricing method and system considering dynamic topology identification

Abstract

The invention discloses a power transmission congestion management partition pricing method and system considering dynamic topology recognition, and belongs to the field of power market operation and intelligent power grid control, and the method comprises the steps: obtaining real-time operation data and high-frequency synchronous measurement data, and recognizing a current operation mode fingerprint of a power grid through an unsupervised machine learning model; according to the method, the blocking section is analyzed based on the mode fingerprint, the evaluation result is generated in combination with the physical elastic margin, then the multi-target optimization algorithm with optimization of partition price distortion and partition internal physical elastic margin as targets is constructed, the dynamic partition set is solved and generated, and the optimal power flow is operated on the basis to calculate and generate the partition marginal electricity price. According to the method, a data-driven operation mode recognition and elastic safety coupling optimization mechanism is adopted, the market partition can be adaptively adjusted in real time, the economic value of physical limitation is accurately reflected, and the efficiency of power transmission congestion management and the robustness of power grid operation are improved.

Description

Technical Field

[0001] This application relates to the fields of power market operation and smart grid control, and in particular to a transmission congestion management zone pricing method and system that takes into account dynamic topology identification. Background Technology

[0002] Transmission congestion management is a crucial aspect of ensuring the safe and stable operation of the power system and the efficient functioning of the electricity market. In regional electricity markets, zonal pricing is a widely adopted congestion management mechanism. It divides the entire power grid into several pricing zones, with a uniform electricity price within each zone, while price differences between zones reflect the cost of transmission congestion. This mechanism sends price signals to market participants, guiding the optimal adjustment of resources on both the generation and load sides, thereby economically and effectively alleviating transmission bottlenecks and ensuring a reliable supply of electricity.

[0003] In existing technologies, regional pricing schemes in electricity markets typically employ a static regional approach. These regions are pre-defined and fixed based on long-term power grid planning studies, historical power flow analysis, or administrative and geographical boundaries. During market operation, dispatching agencies perform market clearing and safety checks based on these fixed regions, determining the marginal price for each region through optimal power flow calculations. When transmission congestion occurs between regions, the prices in different regions will differ to manage congestion.

[0004] However, the aforementioned existing technical solutions have significant limitations. The operating state of a power system is highly dynamic. The start-up and shutdown of generator units, maintenance or faults in transmission lines, random fluctuations in renewable energy output, and real-time changes in load all lead to real-time shifts in power flow distribution and congestion sections of the grid. Static zoning cannot adapt to these dynamic changes. Fixed zoning boundaries may not match the actual physical congestion sections under certain operating conditions, resulting in distorted price signals. This not only fails to effectively alleviate congestion but may also lead to unfair market settlements and reduce market efficiency. Furthermore, static zoning often focuses on economic efficiency or historical congestion patterns, insufficiently considering the dynamic safety margin of the grid under different operating modes. This may create areas with poor safety stability under specific operating conditions, posing a threat to the safe operation of the power grid. Summary of the Invention

[0005] This application provides a transmission congestion management zone pricing method and system that takes into account dynamic topology identification. It uses unsupervised machine learning to identify the real-time operation mode of the power grid and combines it with a multi-objective optimization algorithm to generate dynamic zones that take into account both economy and safety. It can generate a congestion management pricing scheme that adapts to the power grid state, thereby improving the efficiency of the electricity market and the reliability of system operation.

[0006] To achieve the above objectives, this application adopts the following technical solution: Firstly, a method for regional pricing of transmission congestion management considering dynamic topology identification is provided, comprising: acquiring physical parameters, topology state parameters, and market information parameters of the power grid, as well as high-frequency synchronous measurement data, to obtain real-time operation data of the power grid, and processing the real-time operation data to generate fused data; inputting the fused data into a preset unsupervised machine learning model that classifies the power grid operation status based on data similarity, identifying and outputting the current operation mode fingerprint of the power grid; analyzing key congestion sections in the power grid based on the current operation mode fingerprint, and calculating the physical elasticity margin of each region in combination with high-frequency synchronous measurement data to generate congestion and elasticity assessment results; constructing a multi-objective optimization algorithm based on the congestion and elasticity assessment results, aiming to optimize regional price distortion and the physical elasticity margin within the region, solving and generating a dynamic regional set; running optimal power flow calculation based on the dynamic regional set to generate the regional marginal electricity price of each region; and publishing the regional marginal electricity price to the electricity market to guide the trading behavior of market participants.

[0007] Based on the above technical solution, in the transmission congestion management zoning pricing method that takes into account dynamic topology identification provided in this application, unsupervised machine learning is used to identify the real-time operation mode of the power grid, and a multi-objective optimization algorithm is combined to generate dynamic zoning that takes into account both economy and safety. This can generate a congestion management pricing scheme that adapts to the power grid state, thereby improving the efficiency of the electricity market and the reliability of system operation.

[0008] In conjunction with the first aspect above, in one possible implementation, identifying and outputting the current operating mode fingerprint of the power grid includes: extracting features from the fused data to generate a feature vector characterizing the operating state of the power grid; inputting the feature vector into an unsupervised machine learning model for cluster analysis to obtain clustering results; and outputting the clustering results as the current operating mode fingerprint.

[0009] In conjunction with the first aspect above, in one possible implementation, the calculation of the physical resilience margin of each region using high-frequency synchronous measurement data includes: determining the set of critical transmission branches associated with the current operating mode fingerprint, performing power flow limit verification on the set of critical transmission branches, and identifying critical congestion sections in the power grid by combining the power flow distribution factor; calculating a transient stability index characterizing power angle stability and a voltage stability margin index characterizing voltage stability based on the high-frequency synchronous measurement data; performing weighted or coupled calculations on the transient stability index and the voltage stability margin index to obtain a comprehensive regional resilience index; and quantifying the critical congestion sections and the comprehensive regional resilience index into a congestion and resilience assessment result.

[0010] In conjunction with the first aspect above, in one possible implementation, the construction of a multi-objective optimization algorithm aimed at optimizing partition price distortion and internal physical elasticity margin of partitions includes: constructing an optimization objective function containing a first objective function for optimizing partition price distortion and a second objective function for optimizing internal physical elasticity margin of partitions; setting node connectivity constraints and blocking constraints as solution constraints for the multi-objective optimization algorithm; and solving the first and second optimization objective functions under the solution constraints to generate a dynamic partition set.

[0011] In conjunction with the first aspect above, in one possible implementation, the construction of the optimization objective function, which includes a first objective function for optimizing partitioned price distortion and a second objective function for optimizing the physical elasticity margin within the partition, includes: obtaining the node marginal electricity price and the uniform electricity price within the partition for each node, and calculating the value of the first objective function based on the deviation between the node marginal electricity price and the uniform electricity price within the partition; obtaining the physical elasticity margin, and calculating the value of the second objective function based on the physical elasticity margin.

[0012] In conjunction with the first aspect above, in one possible implementation, the optimal power flow calculation includes: obtaining generator quotations and load demands based on the dynamic partition set, generating generation cost parameters and load demand parameters; solving the optimal power flow problem under grid operation constraints with the goal of optimizing total generation cost, and obtaining the optimal power flow solution result; and determining the partition marginal electricity price based on the optimal power flow solution result.

[0013] In conjunction with the first aspect above, in one possible implementation, publishing the regional marginal electricity price to the electricity market includes: transmitting the regional marginal electricity price to the market trading platform; using the regional marginal electricity price to update the market clearing price; and publishing the market clearing price as a price signal to market participants.

[0014] In conjunction with the first aspect above, in one possible implementation, the method further includes: collecting and storing historical operating data to form a historical data set; retraining the unsupervised machine learning model using the historical data set at a predetermined update cycle or when the performance of the unsupervised machine learning model deteriorates; and performing current operating mode fingerprint recognition using the retrained unsupervised machine learning model.

[0015] Secondly, a transmission congestion management zoning pricing system considering dynamic topology identification is provided, comprising: a data acquisition and fusion module, a pattern fingerprinting module, a congestion and resilience assessment module, a dynamic zoning optimization module, an optimal power flow calculation module, and a market release module; wherein: The data acquisition and fusion module is used to acquire the physical parameters, topology state parameters, and market information parameters of the power grid, as well as high-frequency synchronous measurement data, to obtain the real-time operation data of the power grid, and to process the real-time operation data to generate fused data. The pattern fingerprint recognition module is used to input the fused data into a preset unsupervised machine learning model that classifies the power grid operating status based on data similarity, and to identify and output the current operating pattern fingerprint of the power grid. The congestion and resilience assessment module is used to analyze key congestion sections in the power grid based on the fingerprint analysis of the current operating mode, and calculate the physical resilience margin of each region by combining high-frequency synchronous measurement data, and generate congestion and resilience assessment results. The dynamic partition optimization module is used to construct a multi-objective optimization algorithm based on the blocking and elasticity assessment results, with the goal of optimizing partition price distortion and internal physical elasticity margin of partitions, and to solve and generate a dynamic partition set. The optimal power flow calculation module is used to run optimal power flow calculation based on the dynamic partition set and generate the partition marginal electricity price for each partition. The market release module is used to release the marginal electricity price of the zone to the electricity market and guide the trading behavior of market participants.

[0016] Compared with the prior art, this application has the following advantages: This application introduces dynamic topology identification and physical resilience margin assessment, enabling the zoned pricing scheme for transmission congestion management to adapt to changes in grid operating conditions in real time. Compared to traditional static zoning methods, this application can quickly identify the grid operating mode based on real-time data and dynamically adjust zone boundaries to accurately match current physical congestion sections. This improves the accuracy and timeliness of zoned pricing signals, allowing market mechanisms to more effectively guide resources to alleviate transmission bottlenecks.

[0017] This application takes the physical resilience margin within the partition as one of the core optimization objectives, thus systematically considering the dynamic security and stability of the power grid when formulating the partitioning scheme. By quantitatively evaluating and optimizing comprehensive resilience indicators, including transient power angle stability and voltage stability, it ensures that each partitioned area has sufficient ability to withstand disturbances, avoiding the formation of security weaknesses due to improper partitioning, thereby improving the operational reliability and security of the entire power system.

[0018] This application constructs a two-layer, multi-objective optimization framework aimed at optimizing price distortion and physical elasticity margin, achieving an effective balance between the two core requirements of economic efficiency and operational safety. This method no longer solely pursues the accuracy of market prices or unilaterally guarantees grid security; instead, it generates a comprehensively optimal dynamic partition set through collaborative optimization. This ensures that the final partition marginal price reflects the true congestion cost while endogenously guaranteeing the system's safety and stability margin, providing more scientific and comprehensive decision support for grid dispatching and operation.

[0019] The method and system proposed in this application constitute a highly automated process from data acquisition, pattern recognition, optimization decision-making to market release. By applying unsupervised machine learning and multi-objective optimization algorithms, the need for manual intervention is reduced, enabling rapid response and intelligent management of transmission congestion. In particular, the introduction of a model self-updating mechanism ensures that the system can continuously learn and adapt to the long-term development and changes of the power grid, guaranteeing the long-term effectiveness and robustness of the regional pricing strategy.

[0020] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A system architecture diagram of a transmission congestion management zone pricing system that takes into account dynamic topology identification is provided for an embodiment of this application; Figure 2 A flowchart illustrating a transmission congestion management zone pricing method that takes into account dynamic topology identification, provided as an embodiment of this application; Figure 3 This is the Pareto front diagram for dynamic partitioning multi-objective optimization provided in the embodiments of this application; Figure 4 This is a schematic diagram illustrating the change of dynamic zoned marginal electricity price over time, provided in an embodiment of this application. Detailed Implementation

[0023] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0024] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0025] The transmission congestion management zone pricing method considering dynamic topology identification provided in this application embodiment can be applied to, for example... Figure 1 In a transmission congestion management zone pricing system 100 that incorporates dynamic topology identification, as shown, Figure 1 As shown, the system includes: a data acquisition and fusion module, a pattern fingerprinting module, a congestion and resilience assessment module, a dynamic partitioning optimization module, an optimal power flow calculation module, and a market release module; The data acquisition and fusion module is used to acquire the physical parameters, topology state parameters, and market information parameters of the power grid, as well as high-frequency synchronous measurement data, to obtain the real-time operation data of the power grid, and to process the real-time operation data to generate fused data. The pattern fingerprint recognition module is used to input the fused data into a preset unsupervised machine learning model that classifies the power grid operating status based on data similarity, and to identify and output the current operating pattern fingerprint of the power grid. The congestion and resilience assessment module is used to analyze key congestion sections in the power grid based on the fingerprint analysis of the current operating mode, and calculate the physical resilience margin of each region by combining high-frequency synchronous measurement data, and generate congestion and resilience assessment results. The dynamic partition optimization module is used to construct a multi-objective optimization algorithm based on the blocking and elasticity assessment results, with the goal of optimizing partition price distortion and internal physical elasticity margin of partitions, and to solve and generate a dynamic partition set. The optimal power flow calculation module is used to run optimal power flow calculation based on the dynamic partition set and generate the partition marginal electricity price for each partition. The market release module is used to release the marginal electricity price of the zone to the electricity market and guide the trading behavior of market participants.

[0026] like Figure 2 As shown, this application provides a transmission congestion management zone pricing method that takes into account dynamic topology identification, including: S1. Obtain the physical parameters, topology state parameters, and market information parameters of the power grid, as well as high-frequency synchronous measurement data, to obtain the real-time operation data of the power grid, and process the real-time operation data to generate fused data; S2. Input the fused data into a preset unsupervised machine learning model that classifies the power grid operating status based on data similarity, and identify and output the current operating mode fingerprint of the power grid; S3. Based on the fingerprint analysis of the current operating mode, analyze the key blocking sections in the power grid, and calculate the physical resilience margin of each region in combination with high-frequency synchronous measurement data to generate blocking and resilience assessment results. S4. Based on the blocking and elasticity assessment results, construct a multi-objective optimization algorithm with the goal of optimizing partition price distortion and physical elasticity margin within partitions, solve for and generate a dynamic partition set. S5. Based on the dynamic partition set, perform optimal power flow calculation to generate the partition marginal electricity price for each partition; S6. The marginal electricity price of the zone is published to the electricity market to guide the trading behavior of market participants.

[0027] It should be noted that by fusing multi-source heterogeneous data, especially high-frequency synchronous measurement data, a comprehensive real-time profile of the power grid's operating status is formed. Then, unsupervised machine learning techniques are used to reduce the dimensionality and pattern of massive operational data, rapidly classifying complex power grid operating conditions into predefined operating mode fingerprints, enabling rapid identification of the power grid's dynamic topology and operating status. Based on this fingerprint, key congestion sections can be analyzed in a targeted manner, and physical elasticity margins can be calculated by combining dynamic stability quantification indicators, thereby comprehensively assessing the power grid's safety and congestion status. This assessment result is the core basis for dynamically dividing the power grid into zones. Through a multi-objective optimization algorithm that balances the authenticity of price signals with the physical security of zones, a dynamic set of zones that changes with the power grid's state is generated. Finally, based on this dynamic partitioning, optimal power flow calculations are performed to generate marginal electricity prices for each zone that accurately reflect the transmission congestion costs between zones, and these prices are then published to the market.

[0028] In one possible implementation of the embodiments of this application, combined with Figure 2 The above S2 can be implemented through the following S21 and S22, which are explained in detail below: S21. Extract features from the fused data to generate a feature vector characterizing the power grid operating status; In some implementations, feature extraction is performed on the fused data to transform high-dimensional, multi-source real-time operational data into a low-dimensional mathematical expression that accurately characterizes the power grid's operating state. The fused data includes steady-state information such as the power grid's topological connections, generator output, node load levels, and line power flow distribution, as well as dynamic information such as voltage phase angle, amplitude, and frequency provided by high-frequency synchronous measurement data. The feature extraction process involves cleaning and normalizing these raw data, calculating derived features such as power transmission margin of key transmission lines, generator power angle difference, and system frequency change rate, and finally combining all these features into a feature vector.

[0029] It should be noted that the eigenvector is a multi-dimensional column vector, denoted as . Each element represents a quantified value of the power grid's operating status in a specific dimension at a given moment.

[0030] For example, at a certain moment, the fused data is processed to generate a feature vector containing 150 features: the first 50 features are the percentage of power flow to transmission capacity on the system's critical transmission lines; the middle 50 features are the voltage phase angle difference between the main generator clusters and the maximum load center; the next 40 features are the aggregated load factor and the proportion of renewable energy generation in key areas; and the last 10 features are the real-time values ​​of the system frequency and its rate of change. For instance, if the feature vector of a certain area is classified into a "transmission channel congestion mode" fingerprint, then the percentage of power flow to capacity on the transmission lines of that vector will be significantly higher than other feature vectors.

[0031] S22. Input the feature vector into an unsupervised machine learning model for cluster analysis to obtain clustering results, and output the clustering results as the fingerprint of the current running mode.

[0032] In some implementations, feature vectors representing the power grid's operating state are input into a pre-trained unsupervised machine learning model for cluster analysis. Cluster analysis is a data mining technique whose core objective is to divide a dataset into several categories or clusters based on the inherent similarity of the data, maximizing the similarity of data points within the same cluster and minimizing the similarity between data points in different clusters. In this application, algorithms such as K-Means, DBSCAN (density-based spatial clustering with noise), or Self-Organizing Maps (SOM) can be used as unsupervised machine learning models. The model receives the current feature vector... Then, the distance or similarity between the vector and each preset cluster center is calculated, and it is classified into the nearest cluster. Each cluster represents a power grid operation mode that recurs in historical data and has typical physical significance, such as peak load mode, new energy power generation mode, and mode after an N-1 fault on a critical line. The final clustering result output by the model, that is, the unique identifier of the cluster to which the current feature vector belongs, is defined as the fingerprint of the current operation mode of the power grid.

[0033] For example, using SOM as an unsupervised machine learning model, this model has been trained on historical operating data with 50 neurons, each representing a specific operating mode. When a feature vector is input into the SOM model: if the vector is mapped to neuron M-07, and the historical mode represented by this neuron is "summer evening peak, high generator power angle difference, and a certain important interconnection line under maintenance," then the output fingerprint of the current operating mode is "M-07." Subsequently, based on the fingerprint M-07, the preset evaluation strategy and parameters will be directly called for more efficient calculation. If the vector is mapped to neuron M-32, the fingerprint is "M-32," and this mode may represent "nighttime off-peak, wind curtailment of renewable energy, and the grid topology in a simple operating state."

[0034] In one possible implementation, combining Figure 2 The above-mentioned S3 can be implemented through the following S31, S32, S33 and S34, which are explained in detail below: S31. Based on the current operating mode fingerprint, determine the set of key transmission branches associated with the mode, perform power flow limit verification on the set of key transmission branches, and identify key blocking sections in the power grid by combining the power flow distribution factor. In some implementations, based on the fingerprint of the current operating mode, a set of key transmission branches closely related to that operating mode is retrieved and identified from a pre-defined mode-branch association library. This association library is pre-built from historical operation analysis and stores transmission lines prone to congestion or significantly impacting system stability under different operating modes. Power flow limit verification is performed on this set of key transmission branches, i.e., using real-time power flow information from fused data to determine whether the power transmission of these branches is approaching their thermal or static stability limits. Simultaneously, by combining power flow distribution factors, such as the power transfer distribution factor (PTDF), the sensitivity of the line power flow in the set of key transmission branches to inter-regional power exchange is analyzed, thereby identifying several key congestion sections composed of multiple transmission lines that play a decisive role in the power balance of the power grid.

[0035] For example, suppose the current operating mode fingerprint is identified as "M-07: Summer evening peak, power output of generator units at the North China sending end is limited". A set of key transmission branches associated with mode M-07 will be retrieved from the association database, including, for example, 10 branches such as "Shanxi-Hebei Interconnection Line A", "Inner Mongolia DC Access Line B", and "Beijing-Tianjin-Tangshan Ring Network C". Power flow limit verification is performed on these 10 branches, revealing that the real-time power flow of Interconnection Line A has reached 95% of its thermal limit. Subsequently, the power transfer distribution factor (PTDF) of these 10 branches is calculated, finding that Interconnection Line A and DC Access Line B have the highest PTDF values, indicating their greatest sensitivity to inter-regional power transactions. Finally, the "Shanxi-East China Power Transmission Main Channel" formed by Interconnection Line A and DC Access Line B is identified as the current key congestion section. This section information will be included in the evaluation results.

[0036] S32. Based on the high-frequency synchronous measurement data, calculate the transient stability index characterizing the power angle stability and the voltage stability margin index characterizing the voltage stability. In some implementations, high-frequency synchronous measurement data, namely real-time voltage and current phasor data collected by a phasor measurement unit (PMU), is used to calculate the physical resilience margin of each region. Physical resilience margin is a comprehensive concept that assesses the grid's ability to withstand disturbances from a dynamic stability perspective. The calculation process consists of two parts. First, calculating the transient stability index characterizing power angle stability. This index can be obtained by real-time monitoring of the power angle difference between key generator units and comparing it with a preset transient stability limit power angle difference. Second, calculating the voltage stability margin index characterizing voltage stability. This index can be calculated based on real-time estimation of the equivalent Thevenin circuit parameters of load nodes using PMU data, and accordingly, the margin between the current operating point and the voltage collapse point can be calculated.

[0037] For example, transient stability index calculations are performed using high-frequency PMU data: real-time monitoring of the real-time power angle difference between the two most distant synchronous generator units, such as unit G1 in the Northwest region and unit G2 in the East China load center. The real-time power angle difference is compared with the preset transient stability limit power angle difference. By comparison, the transient stability margin is obtained. .Will Normalized values ​​are used as transient stability indicators. Voltage stability margin calculation: Key load nodes are selected. Thevenin equivalent impedance of the node can be estimated in real time using PMU measurement data. and equivalent voltage Based on this, the voltage margin between the current operating voltage of the node and the voltage collapse point is calculated. This is the limit point of the PV curve. (The remaining text appears to be incomplete and requires further context.) Normalized and used as a voltage stability margin index .

[0038] S33. The transient stability index and the voltage stability margin index are weighted or coupled to obtain the regional comprehensive elasticity index. In some implementations, to establish a unified measurement standard, transient stability and voltage stability margin indices are weighted or coupled for calculation to obtain a regional comprehensive resilience index. The weighted calculation formula is as follows: ; in, The regional comprehensive elasticity index is a dimensionless normalized value. The higher the value, the greater the physical elasticity margin of the region. The normalized transient stability index is obtained by dividing the real-time power margin by its maximum possible value. The normalized voltage stability margin index is obtained by dividing the real-time voltage stability margin by its maximum value. and These are the corresponding weighting coefficients, summing to 1, and are pre-set based on the stability characteristics and operational importance of different regions. For example, for regions primarily engaged in long-distance power transmission, the weighting coefficient can be appropriately increased. The weight.

[0039] For example, in the "M-07: Summer Evening Peak" operation mode, due to the higher risks associated with long-distance transmission and power angle stability, the preset weights are: and If a certain region is calculated to... and Then the regional comprehensive elasticity index of this region is: The index of 0.74 is used for subsequent multi-objective optimization, reflecting that the physical elasticity margin of this region is moderately high.

[0040] S34. Quantify the comprehensive elasticity index of the key blocking section and region into blocking and elasticity assessment results.

[0041] In some implementations, the identified key congestion section information, such as the power flow level and margin of each line on the section, is integrated with the calculated regional comprehensive resilience index of each area and quantified together into a structured congestion and resilience assessment result.

[0042] For example, the final congestion and resilience assessment results are quantified as structured data objects containing the following: Congestion Information Field: Contains the name of the identified key congestion sections, the real-time margin percentage of the section, and a list of topology switch states associated with the section. Resilience Information Field: Contains the regional comprehensive resilience index for all divided areas in the power grid. Value list. Constraint threshold field: Contains dynamic physical constraint values ​​used in subsequent multi-objective optimization algorithms, for example, specifying that the choke section margin must not be lower than... And the minimum comprehensive elasticity index within the partition must be higher than This structured assessment result serves as the sole input, guiding subsequent efforts to optimize market efficiency while ensuring safety constraints.

[0043] In one possible implementation, combining Figure 2 The above S4 can be implemented through the following S41, S42 and S43, which are explained in detail below: S41. Construct an optimization objective function that includes a first objective function for optimizing partition price distortion and a second objective function for optimizing the physical elasticity margin within the partition; In some implementations, an optimization objective function is constructed. This function consists of two sub-objectives. The first objective function... The objective function is to minimize partitioned price distortion, aiming to ensure that the unified pricing after partitioning reflects the true value of electricity at each node within the partition as accurately as possible. The goal is to maximize the physical resilience margin within each partition, ensuring that each partitioned region possesses sufficient internal stability to withstand certain disturbances. Therefore, this multi-objective optimization problem is formulated as simultaneously optimizing the following two objectives: ; ; in, This represents a dynamic set of partitions to be determined, which divides all nodes of the power grid into several disjoint subsets, i.e., partitions. The first objective function is... It's about partition sets. The smaller the value of the function, the lower the price distortion. Second objective function. Also about The larger the value of the function, the higher the physical elasticity margin of the partition with the worst elasticity in the entire partitioning scheme, which means the better the overall security.

[0044] For example, the first objective function Specifically, it can be designed as follows: the root mean square error (RMSE) between all nodes' LMP and the uniform pricing ZMP of their respective partitions, with the aim of minimizing... Second objective function It can be specifically designed as: a partitioned set In the middle, the comprehensive elasticity index of all regions within the division The goal is to find the minimum value and maximize the minimum value. By maximizing the least resilient partition, the entire partitioning scheme is forced to meet the expected minimum security standards.

[0045] S42. Set node connectivity constraints and blocking constraints as solution constraints for multi-objective optimization algorithms; In some implementations, solution constraints are set for this multi-objective optimization algorithm. These constraints are necessary boundaries to ensure the physical and electrical feasibility of the generated dynamic partition set. They mainly include two types of constraints. The first is node connectivity constraints, which require that in any partitioning scheme... In this system, all nodes belonging to the same partition must be physically connected, meaning they collectively form a connected subgraph. This ensures that each partition is a geographically and electrically continuous whole. The second constraint is the congestion constraint, which guides partitioning based on critical congestion section information. This constraint forces or incentivizes the assignment of nodes at both ends of transmission lines constituting critical congestion sections into different partitions. This constraint ensures that partition boundaries coincide with the physical congestion boundaries of the power grid, thereby enabling partitioned pricing to effectively manage transmission congestion.

[0046] For example, for the identified critical congestion section of the "Jin-Dong Power Transmission Main Channel," a congestion constraint is set as follows: the starting nodes of all connecting lines on this section must be assigned to different partitions than their ending nodes. For instance, node N1 on connecting line A must be assigned to a different partition. Node N2 must be partitioned into a partition. For node connectivity constraints, the constraint algorithm generates new partitions. At that time, any two points within the partition must be verified. Between them, under the current power grid topology, there exists a line that only passes through Transmission paths for internal nodes and branches.

[0047] S43. Solve the first and second objective functions under the constraints to generate a dynamic partition set.

[0048] In some implementations, a multi-objective optimization problem involving a first objective function and a second objective function is solved under constraints of node connectivity and congestion. Since this problem is typically a complex combinatorial optimization problem, it can be solved using genetic algorithms, multi-objective particle swarm optimization, or other heuristic search algorithms. The algorithm iteratively generates, evaluates, and selects different partitioning schemes. The goal is to find a set of optimal solutions that cannot be surpassed by any other solution on all objectives; this set of solutions is called the Pareto optimal solution set. From this solution set, the final solution can be selected as the output dynamic partition set based on the operator's preferences or preset criteria.

[0049] For example, a non-dominated sorting genetic algorithm with an elitist strategy is used to solve the problem. This algorithm generates a large number of new partitioning schemes in each iteration, and based on a set... and Each option is evaluated. The algorithm gradually eliminates solutions that are dominated by other options in both objectives, eventually converging to a Pareto front. The solutions on this front represent the optimal trade-off between economic efficiency and physical security. For example, there might be two options on the front: Option A: , Option B: , Ultimately, based on preset risk preference criteria, the most suitable option will be selected from the frontier and used as the dynamic partition set. Output.

[0050] In one possible implementation, after S43, the transmission congestion management zone pricing method considering dynamic topology identification provided in this application embodiment further includes the following S411 and S412: S411. Obtain the node marginal electricity price and the uniform electricity price within the partition for each node, and calculate the value of the first objective function based on the deviation between the node marginal electricity price and the uniform electricity price within the partition. In some implementations, the primary objective function for optimizing partitioned price distortion is based on the deviation of the unified electricity price after partitioning from the actual nodal price. First, it is necessary to obtain the nodal marginal price for each node. This price is obtained by solving a single optimal power flow calculation across all nodes under the current grid operating conditions, without partitioning. It reflects the marginal cost of adding a unit load to that node, denoted as […]. Meanwhile, for a given zoning scheme, it is necessary to calculate the uniform electricity price within the zoning. This is typically represented by the weighted average marginal price of the loads at all nodes within the zoning, denoted as [example price here]. Based on this, the first objective function is defined as the weighted sum of squared price deviations of nodes within all partitions, aiming to minimize this value. The calculation formula is as follows: ; in, The value of the first objective function represents the total price distortion of the entire partitioning scheme. Indexes representing partitions, Representing the The set of nodes in each partition. This is the index of the node. For nodes The real-time load demand is obtained from the fused data and used as a weighting factor, so that nodes with high load have a greater impact on price distortion. For nodes The marginal electricity price at each node is calculated from the optimal power flow of a primary benchmark. For partitioning The unified electricity price within a zone is calculated as a weighted average of the marginal electricity prices of all nodes within that zone, weighted by load. This function quantifies the price signal loss caused by zone aggregation, providing a basis for evaluating the economics of the zoning scheme.

[0051] For example, suppose a zoning scheme divides the power grid into southern regions. and the northern region Based on baseline calculations, there are nodes within the southern region. of for Yuan / MWh, Node of for Yuan / MWh, Node of for Yuan / MWh. This zone. Unified electricity price Calculated by weighted average Yuan / MWh. At this point, the price distortion in this partition mainly comes from the nodes. Deviation: First objective function The calculated value will significantly include this bias, if If the value is too large, the partitioning scheme will be considered a poorly economical solution in multi-objective optimization.

[0052] S412. Obtain the physical elasticity margin, and calculate the value of the second objective function based on the physical elasticity margin.

[0053] In some implementations, the second objective function used to optimize the physical elasticity margin within a partition is constructed based on the physical elasticity margin. For any given partitioning scheme... This requires calculating an independent physical resilience margin for each of the divided zones. This margin value is the regional comprehensive resilience index. The second objective function adopts the "barrel principle," defined as the value with the smallest physical elasticity margin among all partitions. The optimization objective is to maximize this value, and the calculation formula is as follows: ; in, The value of the second objective function represents the security level of the weakest link in the entire partitioning scheme. This is a partitioned index. For partitioning The regional comprehensive elasticity index, the value of which is based on the region. Calculations based on internal grid conditions and high-frequency synchronous measurement data, taking into account both power angle stability and voltage stability, were performed by maximizing... This can improve the overall security and stability level in partitioned operation mode, ensuring that no partition becomes a potential point of instability. For example... Figure 3 The diagram illustrates the Pareto front obtained by the multi-objective optimization algorithm when solving for a dynamic partition set in this embodiment. This front demonstrates the optimal trade-off between the first and second objective functions. From the front, a solution such as "Final Option B" can be selected based on a preset risk preference criterion. This option provides higher security within an acceptable range of price distortion.

[0054] For example, suppose a multi-objective optimization algorithm is evaluating a three-partition scheme. Based on the calculated regional comprehensive resilience index for different regions under the current operating mode. Regarding the plan It divides the power grid into The corresponding elasticity indices are respectively According to the barrel principle, Regarding the plan It divides the power grid into The corresponding elasticity indices are respectively . Regarding the plan It divides the power grid into The corresponding elasticity indices are respectively . In maximizing Under the given objective, the algorithm will tend to choose the appropriate solution. Because it has no weak points in any of its partitions, it has the highest safety margin, even though its It might not be the lowest.

[0055] In one possible implementation, combining Figure 2 The above S5 can be implemented through the following S51, S52 and S53, which are explained in detail below: S51. Based on the dynamic partition set, obtain the power generator's bid and load demand, and generate power generation cost parameters and load demand parameters; In some implementations, based on a defined set of dynamic partitions, the bidding data of each power generator is obtained from market information parameters and transformed into the generation cost parameters required for the optimal power flow model. This is typically expressed as a quadratic or piecewise linear function of generator output. Simultaneously, the real-time active and reactive power demands of each load node are extracted from the fused data as load demand parameters for the model.

[0056] For example, suppose a dynamic set of partitions Includes the sending-end area, with ample power. Power supply is tight in the receiving and receiving areas. Data transformation will be performed: Power generator quote: Obtain regional generator The price curve is transformed into a cost function in the optimal power flow calculation model. Get District generator The price curve, its cost function A higher quadratic coefficient indicates a higher marginal cost. Load demand parameters: extracted from fused data. Total active load of all nodes in the area ,as well as Total active load of all nodes in the area These values ​​serve as fixed demand inputs in the optimal power flow calculation model. Obtaining these parameters is a necessary prerequisite to ensure that the optimal power flow calculation model accurately simulates the current market competition environment and physical demands.

[0057] S52. With the goal of optimizing the total power generation cost, and under the constraints of grid operation, solve the optimal power flow problem to obtain the optimal power flow solution results; In some implementations, an optimal power flow problem is constructed and solved with the objective of optimizing the total generation cost. Optimal power flow calculation is a power system optimization analysis method whose goal is to find the operating state that optimizes a certain objective function under a series of equality and inequality constraints. In this application, the objective function is set to minimize the total generation cost of the entire power grid, expressed as: ; in, It is a generator The power generation cost function is generated based on the power generator's bid. It is a generator The active power output is the variable to be optimized. This optimization problem must be solved under the following power grid operation constraints: power balance constraint, i.e., the injected power of each node in the power grid equals the outflow power, which ensures the validity of the power flow equations; generator output constraint, i.e., the active and reactive power output of each generator must be within its technically permissible maximum and minimum range; line power flow constraint, i.e., the power transmission of all transmission lines must not exceed their thermal stability limits; node voltage constraint, i.e., the voltage amplitude of all nodes must be maintained within the specified safe range. Based on the dynamic partition set, all nodes belonging to the same partition are regarded as a whole, and the transmission congestion within it is not a hard constraint of the global optimization problem. However, the power flow constraints of lines crossing different partition boundaries, i.e., lines on critical congestion sections, are strictly enforced. By solving this optimal power flow problem, an optimal power flow solution is obtained, which includes the optimal output of each generator, the voltage phase angle and amplitude of each node, and the Lagrange multipliers corresponding to each constraint condition.

[0058] For example, during the solution process, the optimal power flow calculation algorithm attempts to maximize the cost of generators. The output power, and minimize the high cost of generators. The output. However, due to partition boundary constraints, even The district has sufficient low-cost electricity, but it still cannot fully meet the needs. The demand of the region. Therefore, the optimal power flow calculation algorithm will be forced to schedule High-cost generators in the district Exert force to maintain balance The power demand of the zone. The optimal power flow solution will reflect this congestion phenomenon, that is, the power flow across the zone tie line is equal to its maximum capacity, and the resulting Lagrange multiplier will... and A price difference is formed between them.

[0059] S53. Based on the optimal power flow solution results, determine the marginal electricity price for each zone.

[0060] In some implementations, the marginal electricity price of each zone is determined based on the optimal power flow solution. In optimal power flow theory, the Lagrange multiplier associated with the node power balance constraint physically represents the node's marginal electricity price, indicating the minimum generation cost required to increase a unit load at that node. This application uses the node marginal electricity price of a pre-designated reference node within each zone as the zone's marginal electricity price. Therefore, the... The marginal electricity price of each zone This is the reference node for that partition. nodal marginal electricity price This value is obtained directly from the optimal power flow solution.

[0061] For example, suppose the optimal power flow solution results show: District: Due to low-cost generators The marginal cost determines the price in this region, and the reference node for this region is... of for Yuan / megawatt-hour. District: Due to the necessity of starting a high-cost generator Only then can the load be met, and the reference node in this area... of Rise to Yuan / MWh. Will directly determine: Zone Regional marginal electricity price Yuan / MWh; Zone Regional marginal electricity price Yuan / MWh. This Compared to extra The price difference of RMB per megawatt-hour reflects the blocking costs resulting from the strict enforcement of cross-regional blocking constraints.

[0062] In one possible implementation, referring to the diagram, the above S6 can be specifically implemented through the following S61, S62, and S63, which are explained in detail below: S61. Transmit the regional marginal electricity price to the market trading platform; In some implementations, a complete set of calculated marginal electricity prices for each zone, i.e., the specific price values ​​for each zone, is transmitted to the electricity market trading platform through a secure data communication interface.

[0063] It should be noted that this platform is the core information hub for the operation of the electricity market, responsible for handling key business processes such as bidding, clearing and settlement.

[0064] For example, the final determined set of regional marginal electricity prices Through an encrypted API interface, data is pushed to the core clearing server of the market trading platform precisely at the time of market clearing. This transmission process must meet the requirements of low latency and high security to ensure the accuracy and timeliness of market clearing.

[0065] S62. Update the market clearing price using the aforementioned regional marginal electricity price; In some implementations, after receiving this set of authoritative regional marginal electricity prices, the market trading platform uses these prices to update and determine the official market clearing price for the current market cycle. In a market implementing regional pricing, the regional marginal electricity price itself serves as the benchmark price for settling all electricity transactions within that region. Therefore, it essentially formally establishes the result of optimal power flow calculation as a legally and commercially binding settlement basis. For any given region, all electricity transactions by market participants within that region will be financially settled using the market clearing price corresponding to that region.

[0066] For example, the market trading platform receives marginal electricity price of the area Then, immediately enter the current clearing period into the settlement database. The District Market Clearing Price (MCP) field has been updated to... If a certain electricity sales company is operating during that time period... The district purchased The electricity cost will be calculated based on the electricity price. This calculation process ensures that price signals derived from dynamic topology identification and resilient optimization directly impact the economic interests of market participants.

[0067] S63. Issue market clearing prices to market participants as price signals.

[0068] In some implementations, these determined market clearing prices are publicly released to all relevant market participants. The market trading platform broadcasts the market clearing prices for each region as real-time price signals to all market participants, including power generation companies, electricity retailers, and large users, through its official website, application programming interface (API), or dedicated information dissemination system. This dissemination process ensures the openness, transparency, and timeliness of market price information, enabling market participants to assess the economics of their power generation or consumption behavior based on price differences across regions and make adjustments accordingly. Figure 4 The figure illustrates the dynamic changes over time in the marginal electricity prices of two representative zones calculated in this embodiment. (Figure 1) The area showed significantly higher loads during peak daytime hours. The price in the district clearly reflects the real-time cost of cross-district transmission congestion, demonstrating the role of dynamic pricing in guiding the behavior of market players.

[0069] For example, the market trading platform, through its public information disclosure, issues an announcement to all registered market participants within 2 minutes of the price being determined: "Current clearing period, District settlement price ; District settlement price The release of this price signal will immediately prompt... The district's power generation companies are considering increasing output and prompting... Large industrial users in the district are considering responding to price signals by postponing their non-essential load demands, thereby easing their burden autonomously in the next operating cycle. The tight power supply situation in the district demonstrates the role of price signals in guiding market behavior.

[0070] In one possible implementation, after S63, the transmission congestion management zone pricing method considering dynamic topology identification provided in this application embodiment further includes: Collect and store historical operational data to form a historical data set; In some implementations, historical operating data of the power grid is continuously collected and stored. The fused data of each time segment, as well as information such as the fingerprint of the current operating mode identified by the model, are stored in a dedicated historical database, gradually forming a historical data set that is constantly expanding in scale.

[0071] For example, every 5 minutes, the following information at the current moment is packaged into a data record: fused data; model output; and system results. These data records, after being timestamped, are stored in a distributed time-series database, such as Apache Cassandra or InfluxDB. After running continuously for one year, the database will form a historical data set containing approximately 100,000 time-section records, reflecting the complete operating trajectory of the power grid under all operating conditions, including seasonal changes, load fluctuations, maintenance, and faults.

[0072] When the unsupervised machine learning model degrades during a predetermined update cycle, the unsupervised machine learning model is retrained using a historical dataset. In some implementations, the update mechanism is triggered by two conditions. The first is a predetermined update cycle, where a fixed time interval, such as monthly or quarterly, is pre-set based on the speed of power grid development and seasonal variations to automatically initiate the model retraining process. The second is triggered when the performance of the unsupervised machine learning model significantly declines. Because this model is unsupervised learning, its performance cannot be measured by conventional label comparison. Therefore, performance degradation can be judged by monitoring internal evaluation metrics of clustering results, such as the silhouette coefficient or the Calinski-Harabasz index. When these metrics consistently fall below preset thresholds, it indicates that the model's clustering effect is deteriorating. Alternatively, updates can be triggered by external feedback, such as the discovery of a systematic deviation between the identified operating patterns and the actual congestion or resilience conditions.

[0073] For example, the scheduled update cycle is set to be executed automatically at the beginning of each quarter. Additionally, the model's profile coefficient is monitored in real time. If this coefficient is below a threshold for seven consecutive days... When this happens, a performance degradation warning is issued, and the model retraining process is automatically triggered. For example, if a large number of new energy sources are connected to the grid in the new quarter, causing the grid operation status to "drift," the old cluster centers may not be able to classify the new feature vectors well, and the silhouette coefficient will decrease. Based on this, it can be judged that the model is no longer suitable for the current grid operation environment.

[0074] The current operating mode fingerprint recognition is performed using a retrained unsupervised machine learning model.

[0075] In some implementations, once the triggering condition is met, a complete or up-to-date historical dataset is retrieved from the historical database to retrain the unsupervised machine learning model. The retraining process involves re-executing the clustering algorithm using the updated dataset, which may lead to the repositioning of cluster centers or even changes in the number of clusters, thereby generating a new model that better reflects all typical operating states of the current power grid. After training, the generated new model, i.e., the retrained unsupervised machine learning model, is validated by internal metrics to confirm its superior performance compared to the old model, and then replaces the original model. Subsequently, this retrained unsupervised machine learning model will be used to perform subsequent current operating mode fingerprinting tasks and process the real-time incoming fused data.

[0076] For example, after triggering retraining, the SOM model is trained using historical data from the most recent year. The new training may result in a readjustment of the original 50 operational pattern fingerprints; for instance, two old, similar pattern fingerprints may be merged into one, while a new fingerprint specifically representing "sudden increase in summer air conditioning load and localized line maintenance" may be added. After verifying its performance, the new model immediately replaced the old model and was put into use. From that moment on, the real-time incoming fused data will be contained in this... The model, after being retrained with the new fingerprints, will be processed to ensure that the identification results of the current operating mode always match the latest operating characteristics of the power grid.

[0077] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method is applicable to the embodiments of this application as long as it achieves the purpose of this application. The above descriptions are merely exemplary embodiments of this application and should not be construed as limiting the scope of this application.

[0078] All equivalent changes and modifications made in accordance with the teachings of this application shall still fall within the scope of this application. Other embodiments of this application will be readily apparent to those skilled in the art upon consideration of the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary technical means in the art not described in this application.

Claims

1. A method for congestion management zonal pricing considering dynamic topology identification, characterized in that, The method comprises: obtaining physical parameters, topological state parameters and market information parameters of a power grid, and high-frequency synchronous measurement data, to obtain real-time operation data of the power grid, and processing the real-time operation data to generate fusion data; inputting the fusion data into a preset unsupervised machine learning model for classifying power grid operation states according to data similarity, identifying and outputting a current operation mode fingerprint of the power grid; based on the current operation mode fingerprint, analyzing key blocking sections in the power grid, and combining high-frequency synchronous measurement data to calculate physical elasticity margins of each region, to generate blocking and elasticity evaluation results; based on the blocking and elasticity evaluation results, constructing a multi-objective optimization algorithm with the objectives of optimizing zonal price distortion and internal physical elasticity margins of each zone, solving and generating a dynamic zonal set; based on the dynamic zonal set, running an optimal power flow calculation to generate zonal marginal prices of each zone; publishing the zonal marginal prices to the electricity market to guide the transaction behavior of market participants.

2. The power congestion management partition pricing method considering dynamic topology identification according to claim 1, wherein, The identification and output of the current operation mode fingerprint of the power grid comprise: performing feature extraction on the fusion data to generate a feature vector representing the operation state of the power grid; inputting the feature vector into an unsupervised machine learning model for cluster analysis to obtain a cluster result, and outputting the cluster result as the current operation mode fingerprint.

3. The method of claim 1, wherein, The calculation of the physical elasticity margins of each region in combination with the high-frequency synchronous measurement data comprises: based on the current operation mode fingerprint, determining a set of key power transmission branches associated with the mode, performing power flow limit checking on the set of key power transmission branches, and identifying key blocking sections in the power grid in combination with a power flow distribution factor; based on the high-frequency synchronous measurement data, calculating a transient stability index representing power angle stability and a voltage stability margin index representing voltage stability; performing weighted or coupled calculation on the transient stability index and the voltage stability margin index to obtain a regional comprehensive elasticity index; quantifying the key blocking sections and the regional comprehensive elasticity index into the blocking and elasticity evaluation results.

4. The method of claim 1, wherein, The construction of the multi-objective optimization algorithm with the objectives of optimizing zonal price distortion and internal physical elasticity margins of each zone comprises: constructing an optimization objective function comprising a first objective function for optimizing zonal price distortion and a second objective function for optimizing internal physical elasticity margins of each zone; setting node connectivity constraints and blocking constraints as solving constraints of the multi-objective optimization algorithm; solving the first optimization objective function and the second optimization objective function under the solving constraints to generate a dynamic zonal set.

5. The method of claim 4, wherein, The construction of the optimization objective function comprising the first objective function for optimizing zonal price distortion and the second objective function for optimizing internal physical elasticity margins of each zone comprises: obtaining node marginal prices and zonal uniform prices of each node, and calculating the value of the first objective function based on the deviation between the node marginal prices and the zonal uniform prices; obtaining the physical elasticity margins, and calculating the value of the second objective function based on the physical elasticity margins.

6. The method of claim 1, wherein, The optimal power flow calculation comprises: based on the dynamic zonal set, obtaining power supplier bids and load demand to generate power generation cost parameters and load demand parameters; Solve an optimal power flow problem under grid operation constraints to obtain an optimal power flow solution, with the goal of optimizing total generation cost; Determine a zonal marginal price based on the optimal power flow solution.

7. The method of claim 1, wherein, Publishing the zonal marginal price to a power market includes: Transmitting the zonal marginal price to a market trading platform; Updating a market clearing price using the zonal marginal price; Publishing the market clearing price as a price signal to market participants.

8. The power congestion management partition pricing method considering dynamic topology identification according to claim 1, wherein, The method further includes: Collecting and storing historical operation data to form a historical data set; Re-training the unsupervised machine learning model using the historical data set when a predetermined update period or the performance of the unsupervised machine learning model decreases; Performing current operation mode fingerprint identification using the re-trained unsupervised machine learning model.

9. A transmission congestion management partition pricing system that accounts for dynamic topology identification, characterized in that, The system is used for a power transmission congestion management zonal pricing method considering dynamic topology identification according to any one of claims 1-8, and the system includes: A data acquisition and fusion module for obtaining physical parameters, topology state parameters and market information parameters of a power grid, and high-frequency synchronous measurement data, obtaining real-time operation data of the power grid, and processing the real-time operation data to generate fusion data; A mode fingerprint identification module for inputting the fusion data into a preset unsupervised machine learning model for classifying power grid operation states according to data similarity, identifying and outputting a current operation mode fingerprint of the power grid; A congestion and flexibility evaluation module for analyzing key congestion sections in the power grid based on the current operation mode fingerprint, and calculating physical flexibility margins of each region in combination with high-frequency synchronous measurement data to generate congestion and flexibility evaluation results; A dynamic zonal optimization module for constructing a multi-objective optimization algorithm with the goal of optimizing zonal price distortion and zonal internal physical flexibility margins based on the congestion and flexibility evaluation results, solving and generating a dynamic zonal set; An optimal power flow calculation module for running optimal power flow calculation based on the dynamic zonal set to generate zonal marginal prices of each zonal; A market publishing module for publishing the zonal marginal prices to a power market to guide transaction behavior of market participants.

Images