Switching operation dispatching method based on main and auxiliary device integration coordination
By constructing switching dispatch paths and optimizing objective functions, the problem of insufficient dynamic adaptability of distribution network operation status in switching dispatch was solved, and the efficiency, safety and stability of switching operations were improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID ZHEJIANG ELECTRIC POWER CO LTD NINGBO POWER SUPPLY CO
- Filing Date
- 2026-05-12
- Publication Date
- 2026-06-09
AI Technical Summary
In existing switching and dispatching technologies, the dynamic adaptability of the distribution network operation status is insufficient, and risk assessment is mostly static, which leads to unstable operation and low dispatching efficiency during switching operations.
Construct allocation paths both within and outside the set, combine a two-layer optimization objective function to generate initial and backup allocation paths adapted to switching allocation needs, and optimize switching operations using dynamic penalty factors and abnormal risk relationships to ensure overall safety and stability.
Improve the response efficiency and adaptability of switching operations, enhance the operating condition adaptability and operational stability of the main distribution network, avoid blind spots in the coverage of cross-regional path risks, and ensure the reliability and timeliness of backup dispatch paths.
Smart Images
Figure CN122178359A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of switching and dispatching technology, and in particular to a switching and dispatching method based on integrated main and dispatching coordination. Background Technology
[0002] Switching operations are standardized electrical operations performed in power system dispatching and operation. These operations involve switching on and off electrical equipment such as circuit breakers, disconnectors, grounding switches, high and low voltage side switches of main transformers, and line tie switches to achieve grid topology switching, equipment state transitions, power supply switching, and load path transfer.
[0003] Integrated main grid and distribution network coordinated switching is a technical system developed under the background of smart grid to solve the problems of low switching operation efficiency, high safety risks and insufficient power supply reliability caused by the separate control of the main grid and distribution network in traditional grid. Its core is to realize the data connection, model fusion, decision coordination and control linkage between the main grid EMS and the distribution network DMS.
[0004] Existing integrated main and distribution switching and dispatching technologies are mostly unidirectional drive-type collaborative architectures, only targeting main grid-side switching scenarios such as main grid transformer outages and busbar adjustments, and formulating distribution network-side load transfer and adaptation strategies to achieve distribution network coordinated control under main grid operation.
[0005] Existing switching and dispatching technologies lack dynamic adaptability to the operating status of the distribution network, and risk assessments are mostly static assessments. They cannot be flexibly adjusted according to real-time operating data and load fluctuations of the distribution network, which leads to problems such as unstable operation and low dispatching efficiency during switching operations. Summary of the Invention
[0006] This application addresses the technical problems of insufficient adaptability to operating conditions and insufficient overall security in existing switching and dispatching technologies. It provides a switching and dispatching method based on integrated main and distribution coordination. By constructing a first dispatching path between various distribution networks within a set and a second dispatching path between sets, and combining a two-layer optimization objective function, it synchronously outputs an initial dispatching path adapted to the switching and dispatching requirements, as well as a backup dispatching path adapted to the initial dispatching path with the lowest abnormal risk. While meeting real-time operating condition requirements, it ensures overall power transfer redundancy and improves the overall security of switching operations.
[0007] To achieve the above technical objectives, this application provides a technical solution: a switching dispatching method based on integrated main and distribution coordination, comprising the following steps: dividing the entire distribution network according to the power supply area type and load characteristics corresponding to each distribution network, and constructing several distribution network sets; constructing a first dispatching path within the distribution network set and a second dispatching path between distribution network sets based on the inter-regional interconnection lines between distribution networks, and constructing dispatching path constraints based on main and distribution load constraints; constructing anomaly risk relationships based on historical switching anomaly events and historical anomaly operation data of each power supply area, and obtaining dynamic penalty factors for each power supply area based on the anomaly risk relationships; constructing an upper-level optimization objective function of the shortest manifold based on the first dispatching path and the second dispatching path combined with dispatching path constraints; constructing a lower-level optimization objective function of the lowest uncertainty based on the dynamic penalty factor and the heterogeneity of the distribution network set; constructing a switching dispatching model based on the upper-level optimization objective function and the lower-level optimization objective function; and outputting a switching dispatching strategy including the initial dispatching path and the backup dispatching path based on the switching dispatching model and the real-time operation data of each power supply area in response to switching operation requirements.
[0008] Furthermore, the step of constructing abnormal risk relationships based on historical switching anomaly events and historical abnormal operation data of each power supply area, and obtaining dynamic penalty factors for each power supply area based on abnormal risk relationships, includes: taking each power supply area as a network node, constructing directed edges according to the first dispatch path and the second dispatch path; obtaining the number of historical switching associations between two power supply areas and the number of abnormal events at each level in the historical switching associations based on the historical switching anomaly events and historical abnormal operation data of each power supply area; constructing directed edge weights based on the ratio of the number of historical switching associations, the number of abnormal events at each level, and the preset level of abnormal weights; obtaining direct abnormal relationships and transmitted abnormal relationships between power supply areas based on the directed edges and directed edge weights; and outputting the dynamic penalty factors for each power supply area based on the direct abnormal relationships and transmitted abnormal relationships.
[0009] Furthermore, the construction of directed edge weights based on the ratio of historical switching association counts, the number of abnormal events at each level, and preset level abnormal weights includes: obtaining the corresponding preset level prior weights based on the abnormality type of historical switching abnormal events; performing normalization mapping on the historical abnormal operation data of all historical switching abnormal events to obtain adjustment coefficients; obtaining preset level weights for abnormal events at each level using the adjustment coefficients and preset level prior weights; calculating the weighted abnormal count using the preset level weights and the number of abnormal events at each level; and constructing directed edge weights based on the ratio of the weighted abnormal count to the historical switching association counts.
[0010] Furthermore, the step of obtaining the direct and propagated anomaly relationships between power supply areas based on directed edges and directed edge weights includes: if there is a directed edge between two power supply areas, then constructing a direct anomaly relationship based on the directed edge weight corresponding to the directed edge; using a restarted random walk algorithm, taking any power supply area as the starting node, calculating the steady-state access probability to reach the other power supply areas based on the directed edge weights, and obtaining the propagated anomaly relationships between the power supply areas.
[0011] Furthermore, the method of using the restart random walk algorithm, taking any power supply area as the starting node, and calculating the steady-state access probability of reaching each of the other power supply areas based on the directed edge weights to obtain the transmission anomaly relationship between the power supply areas includes: for each power supply area, calculating the sum of the directed edge weights of all its outgoing edges, using the ratio of the directed edge weight of each outgoing edge to the sum of the directed edge weights as the one-step transition probability of the power supply area along that outgoing edge, and constructing a transition probability matrix; performing random walk iterations on the transition probability matrix based on a preset restart probability until convergence to obtain a steady-state access probability vector; and using the probability values between the starting node and each arriving node in the steady-state access probability vector as the transmission anomaly relationship between the corresponding power supply areas.
[0012] Furthermore, the upper-level optimization objective function for constructing the shortest manifold based on the first allocation path and the second allocation path combined with allocation path constraints includes: constructing the upper-level optimization objective function for the shortest manifold based on the first allocation path and the second allocation path combined with allocation path constraints, with the minimum number of allocation devices and the shortest allocation path length as optimization objectives.
[0013] Furthermore, the lower-level optimization objective function based on the dynamic penalty factor and the heterogeneity of the distribution network set to construct the lowest uncertainty includes: seeking the standby distribution network based on the initial dispatch path output by the upper-level optimization objective function; calculating the uncertainty of the standby distribution network based on the dynamic penalty factor, the heterogeneity of the distribution network set, and the time-series fluctuation of the distribution network load; and constructing the lower-level optimization objective function with the lowest uncertainty.
[0014] Furthermore, the step of dividing the entire distribution network according to the power supply area type and load characteristics corresponding to each distribution network and constructing several distribution network sets includes: obtaining the power supply area type label and the daily load curve within a preset time period corresponding to each distribution network; performing preset feature extraction on the daily load curves of different distribution networks to obtain the corresponding load feature vector; and performing cluster analysis on the entire distribution network according to the power supply area type label and the load feature vector to construct several distribution network sets.
[0015] Furthermore, the step of performing cluster analysis on the entire distribution network based on power supply area type labels and load feature vectors to construct several distribution network sets includes: performing a first clustering of the entire distribution network based on power supply area type labels to obtain a first distribution network set; performing a second clustering of the first distribution network set based on load feature types to obtain a second distribution network set; and performing a third clustering of the second distribution network set based on load feature vectors to obtain a third distribution network set.
[0016] Furthermore, the second allocation path includes at least a first-level path corresponding to the first distribution network set, a second-level path corresponding to the second distribution network set, and a third-level path corresponding to the third distribution network set; wherein, the heterogeneity of the distribution network set is calculated based on the number of first-level paths, the number of second-level paths, and the number of third-level paths combined with a preset quantity weight.
[0017] The beneficial effects of this application are as follows: 1. By utilizing the upper-level optimization objective function with the shortest manifold as the optimization direction, an initial dispatch path adapted to the switching operation requirements is generated, ensuring the timeliness of load transfer and overall dispatch efficiency in conventional switching scenarios. Simultaneously, by utilizing the lower-level optimization objective function with the lowest uncertainty as the optimization direction, and combining historical abnormal operation data and the heterogeneity of the distribution network set, a hot standby path with the lowest operational risk is pre-generated before the execution of the initial dispatch path. By synchronously outputting the initial dispatch path and the standby dispatch path, the response efficiency and adaptability to switching operation requirements are improved. Furthermore, the standby dispatch path allows for rapid and flexible replenishment when the initial path experiences local power flow exceedances or abnormal load disturbances, improving the operational adaptability, stability, and overall security of the main distribution network's switching dispatch.
[0018] 2. Directed edges are generated through the first and second dispatch paths to reflect the possible switching interactions in the entire main distribution network system. Then, directed edge weights are constructed based on the historical switching association frequency, the anomaly level during historical switching associations, and the number of anomalies, reflecting the direct and transmitted anomaly impacts between power supply areas. The dynamic penalty factor for each power supply area is solved by combining the direct and transmitted anomaly relationships with the directed edge weights. This avoids the blind spots of risk coverage in single cross-regional paths, adapts to the risk coupling and disturbance propagation characteristics in the context of full-domain cross-regional dispatch scenarios, and makes the quantification results of the dynamic penalty factor more closely match the actual switching operation conditions, providing accurate risk constraints for the lower-level minimum uncertainty optimization objective function.
[0019] 3. Convert the equipment in the standby dispatch path from cold standby to hot standby, that is, pre-set the tie switches and transfer equipment involved in the standby dispatch path to hot standby closed standby state to improve the response efficiency of interlocking anomalies.
[0020] 4. In the initial state responding to switching operation requirements, the preset quantity weight of the first-level paths is less than that of the second-level paths, and the preset quantity weight of the second-level paths is less than that of the third-level paths. Based on the prior principle that higher heterogeneity corresponds to lower operational risk, this ensures that the backup dispatch path from the third-level path receives the optimal evaluation in the absence of real-time feedback, while avoiding the risk of no solution or excessive electrical distance, thus ensuring the reliability of the backup dispatch path. After executing the initial dispatch path, the preset quantity weights are adjusted according to the fluctuation level of the real-time operating data of each distribution network on the initial dispatch path. The higher the similarity of the fluctuation level of the real-time operating data of distribution networks of different power supply area types, the larger the preset quantity weight of the first-level paths; the smaller the fluctuation level of the real-time operating data, the larger the preset quantity weight of the second-level paths. This ensures dispatch safety while avoiding excessive heterogeneity and improving dispatch adaptability. Attached Figure Description
[0021] Figure 1 This is a flowchart illustrating the switching and dispatching method based on integrated main and distribution coordination in this application.
[0022] Figure 2 This is a schematic diagram illustrating the construction process of the dynamic penalty factor in the switching dispatching method based on integrated main and distribution coordination in this application. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0024] like Figure 1 As shown in the first embodiment of this application, the switching and dispatching method based on integrated main and distribution coordination includes the following steps: The entire distribution network is divided according to the power supply area type and load characteristics corresponding to each distribution network, and several distribution network sets are constructed. The first dispatch path within the distribution network set and the second dispatch path between distribution network sets are constructed based on the inter-regional interconnection lines between distribution networks, and dispatch path constraints are constructed based on the main distribution load constraints. Anomaly risk relationships are constructed based on historical switching anomaly events and historical abnormal operation data of each power supply area, and dynamic penalty factors for each power supply area are obtained based on the anomaly risk relationships. Based on the first and second allocation paths and the allocation path constraints, a higher-level optimization objective function for the shortest manifold is constructed. A lower-level optimization objective function with minimum uncertainty is constructed based on the dynamic penalty factor and the heterogeneity of the distribution network set; A switching operation model is constructed using the upper-level optimization objective function and the lower-level optimization objective function. In response to switching operation requirements, the switching dispatch strategy, which includes the initial dispatch path and the backup dispatch path, is output based on the switching dispatch model and the real-time operation data of each power supply area.
[0025] In this embodiment, an initial dispatch path adapted to the switching operation requirements is generated using the upper-level optimization objective function with the shortest manifold as the optimization direction, ensuring the timeliness of load transfer and overall dispatch efficiency in conventional switching scenarios. Simultaneously, a lower-level optimization objective function with the lowest uncertainty as the optimization direction, combined with historical abnormal operation data and the heterogeneity of the distribution network set, pre-generates a hot standby path with the lowest operational risk before the initial dispatch path is executed. By synchronously outputting the initial dispatch path and the standby dispatch path, the response efficiency and adaptability to switching operation requirements are improved. Furthermore, the standby dispatch path allows for rapid and flexible replenishment when the initial path experiences local power flow exceedances or abnormal load disturbances, improving the operational adaptability, stability, and overall security of the main distribution network's switching dispatch.
[0026] Specifically, the entire distribution network is divided according to the power supply area type and load characteristics corresponding to each distribution network, and several distribution network sets are constructed, including: Obtain the power supply area type label and daily load curve within a preset time period for each distribution network; Pre-defined features are extracted from the daily load curves of different distribution networks to obtain the corresponding load feature vectors; Cluster analysis is performed on the entire distribution network based on the power supply area type label and load feature vector to construct several distribution network sets.
[0027] The power supply area types include at least industrial areas, commercial areas, residential areas, and rural power supply areas. Continuous data collection results of electricity consumption from each distribution network feeder within a preset time period are obtained. These continuous data collection results are then preprocessed with noise reduction and normalization to eliminate dimensional differences and abnormal data interference, resulting in the acquisition of daily load curves.
[0028] Load characteristics include at least the daily average load, daily load rate, daily minimum load rate, peak-to-valley difference rate, average load during peak hours, average load during valley hours, load differences during holidays, and seasonal load sensitivity.
[0029] Extract all load characteristics from the daily load curves of all distribution networks and fuse them to construct a load feature vector.
[0030] At this point, during the clustering process, the initial clustering is first performed according to the power supply area type label to obtain the initial clusters, and then the initial clusters are clustered again according to the load feature vector to obtain the distribution network set.
[0031] Based on the inter-regional interconnection lines between distribution networks, a first dispatch path within the distribution network set and a second dispatch path between distribution network sets are constructed, and dispatch path constraints are constructed based on the main distribution load constraints, including: The inter-regional connection lines between distribution networks within the same distribution network set are used as the first dispatch path; The inter-regional connection lines between different distribution network sets are used as the second dispatch path; Dispatch path constraints are constructed based on the rated capacity of the main transformer and the bus voltage threshold.
[0032] By using inter-regional interconnection lines within and between distribution network sets, a first distribution path and a second distribution path are constructed in layers. The first distribution path is a transfer path between homogeneous distribution networks, and the second distribution path is a transfer path between heterogeneous distribution networks. These simultaneously form a complete network-wide transfer candidate path. Furthermore, distribution path constraints are constructed based on the rated capacity of the main transformer and the bus voltage threshold to ensure the safety of the distribution path.
[0033] It is understandable that inter-regional interconnection lines are power lines that connect different power supply areas or power supply zones, consisting of conductors, towers, and interconnection switches used to control the on / off state of the two distribution network areas.
[0034] like Figure 2 As shown, anomaly risk relationships are constructed based on historical switching anomaly events and historical abnormal operation data of each power supply area, and dynamic penalty factors for each power supply area are obtained based on these anomaly risk relationships, including: Each power supply area is treated as a network node, and directed edges are constructed based on the first dispatch path and the second dispatch path; Based on the historical switching anomaly events and historical abnormal operation data of each power supply area, the number of historical switching associations between two power supply areas and the number of anomaly events at each level in the historical switching associations are obtained. Directed edge weights are constructed using the ratio of the number of historical power outage associations, the number of abnormal events at each level, and the preset abnormal weights at each level. The direct and propagated anomaly relationships between power supply areas are obtained based on directed edges and their weights. The dynamic penalty factor for each power supply area is output based on the direct anomaly relationship and the conducted anomaly relationship.
[0035] In this embodiment, directed edges are generated through the first and second dispatch paths to reflect the possible relationships of switching interactions in the entire main distribution network system. Then, directed edge weights are constructed based on the historical switching association frequency, the anomaly level during historical switching associations, and the number of anomalies, reflecting the direct and transmitted anomaly impacts between power supply areas. The dynamic penalty factor for each power supply area is solved by combining the direct and transmitted anomaly relationships with the directed edge weights, avoiding blind spots in the coverage of single cross-regional path risks. This adapts to the risk coupling and disturbance propagation characteristics in the full-domain cross-regional dispatch scenario, making the quantification results of the dynamic penalty factor more closely match the actual switching operation conditions, and providing accurate risk constraints for the lower-level minimum uncertainty optimization objective function.
[0036] Specifically, the directed edge weights are constructed using the ratio of the number of historical power outage associations, the number of abnormal events at each level, and the preset abnormal weights at each level. Obtain the corresponding preset level prior weight based on the anomaly type of historical switching anomaly events; Perform a normalized mapping on the historical abnormal operation data of all historical switching abnormal events to obtain adjustment coefficients; The preset level weights of abnormal events at each level are obtained by adjusting the coefficients and using preset level prior weights; The weighted number of anomalies is calculated using preset hierarchical weights and the number of anomalies at each level. The weighted number of anomalies is then used to construct directed edge weights based on the ratio of the weighted number of anomalies to the number of historical power outage associations.
[0037] The anomaly types include at least cascading overload, voltage exceeding limits, protection malfunction, and secondary power transfer. The preset level prior weights can be set based on the perceived severity of the anomaly. In this embodiment, the preset level prior weight for cascading overload is set to 0.4, for voltage exceeding limits to 0.3, for protection malfunction to 0.2, and for secondary power transfer to 0.1.
[0038] Historical abnormal operation data should include at least the peak target feeder load rate, maximum bus voltage deviation, number of protection trips, and secondary power transfer intervals. All historical abnormal operation data are normalized and mapped to adjustment coefficients within the (0, 1] range, ensuring that the more severe the abnormality of the same type, the larger the adjustment coefficient. For example, among abnormalities involving interlocking overloads, the adjustment coefficient for an interlocking overload with a feeder load rate of 95% is larger than that for an interlocking overload with a feeder load rate of 82%.
[0039] Specifically, for historical abnormal operation data of chained overloads, a normalized mapping is performed as follows: ; in, This indicates the adjustment factor for cascading overload. This indicates the peak load rate of the target feeder. This indicates the maximum preset target feeder load rate. This indicates the minimum preset target feeder load rate.
[0040] In this embodiment, 80%, The value is 120%. In other embodiments, the value can also be set according to actual needs.
[0041] For historical abnormal operating data involving voltage exceeding limits, a normalized mapping is performed as follows: ; in, This indicates the adjustment factor for voltage exceeding the limit. This indicates the maximum deviation of the bus voltage. This indicates the maximum preset bus voltage offset. This indicates the minimum preset bus voltage offset.
[0042] In this embodiment, It is 15%. The value is 7%. In other embodiments, the value can also be set according to actual needs.
[0043] To protect historical abnormal operation data that has been malfunctioning, a normalized mapping is performed as follows: ; in, This indicates the adjustment factor for protection against malfunctions. Indicates the number of times the protection mechanism tripped. This indicates logarithmic normalization, which converts the number of protection trips from a linear scale to a logarithmic scale.
[0044] For historical abnormal operation data of secondary transfer, the normalization mapping is performed as follows: ; in, This indicates the adjustment factor for secondary supply. Indicates the time interval between secondary power transfers. This indicates the maximum preset time interval for secondary power transfer. This indicates the minimum preset time interval for secondary power transfer.
[0045] In this embodiment, It is 7200s. The duration is 300 seconds. In other embodiments, the duration can also be set according to actual needs.
[0046] The preset level weights for each level of abnormal events are obtained by adjusting the coefficients and using preset level prior weights: ; in, Indicates the preset hierarchical weights. Indicates the adjustment factor. This indicates the preset hierarchical prior weight. The adjustment factor is the adjustment factor for interlocking overload, protection malfunction, or secondary transfer.
[0047] The weighted average number of anomalies is calculated based on preset hierarchical weights and the number of anomalies at each level: ; in, Indicates the power supply area To the power supply area The weighted number of anomalies, Indicates the first Preset hierarchical weights for each level of abnormal time. Indicates the total number of levels.
[0048] The weights of the directed edges are constructed using the ratio of the weighted number of anomalies to the number of historical power outage associations: ; in, Indicates the power supply area To the power supply area The weight of the directed edge. Indicates the power supply area To the power supply area The number of historical power outage correlations This represents a very small positive number to prevent the denominator from being 0.
[0049] The direct and propagated anomaly relationships between power supply areas are obtained based on directed edges and their weights, including: If there is a directed edge between two power supply areas, a direct anomaly relationship is constructed based on the directed edge weight corresponding to that directed edge. By using the restart random walk algorithm, taking any power supply area as the starting node, the steady-state access probability of reaching other power supply areas is calculated based on the directed edge weights, and the transmission anomaly relationship between each power supply area is obtained.
[0050] In this embodiment, a direct anomaly relationship is constructed by having directed edges, and a restart random walk algorithm is used to reflect the multi-order transmission relationship with the steady-state access probability, ensuring that the dynamic penalty factor reflects both local direct risk and global transmission risk.
[0051] Specifically, using the restarted random walk algorithm, starting from any power supply area, the steady-state access probability of reaching other power supply areas is calculated based on the directed edge weights, and the transmission anomaly relationships between the power supply areas are obtained, including: For each power supply area, calculate the sum of the directed edge weights of all its outgoing edges. Use the ratio of the directed edge weight of each outgoing edge to the sum of the directed edge weights as the one-step transition probability of the power supply area along that outgoing edge, and construct a transition probability matrix. Based on the preset restart probability, perform random walk iterations on the transition probability matrix until convergence, and obtain the steady-state access probability vector; The probability values between the starting node and each arriving node in the steady-state access probability vector are used as the transmission anomaly relationship between the corresponding power supply areas.
[0052] Understandably, since abnormal risk relationships are constructed with directed edges, the steady-state access probability between two power supply areas has directional asymmetry. For example, the steady-state access probability from the first power supply area to the second power supply area and the steady-state access probability from the second power supply area to the first power supply area are calculated independently, thus reflecting the directional characteristics of power transfer risks caused by topology, load characteristics, and equipment differences in the actual power grid.
[0053] By normalizing the weights of the directed edges from each power supply area, a transition probability matrix is constructed using the ratio of the weight of each directed edge to the sum of the weights of all directed edges. This matrix ensures that the transition probabilities match the coupling strength differences of cross-regional switching anomalies, allowing for the assessment of the ease of risk propagation along different connection paths. Combined with a preset restart probability, a random walk iteration is performed on the transition probability matrix to simulate the natural decay characteristics of grid switching anomaly disturbances, ensuring the accuracy and adaptability of the steady-state access probability. Furthermore, the steady-state access probability between two power supply areas is used as the basis for assessing their anomaly propagation relationship, quantifying the cascading risks of global switching operation anomalies, and improving the reliability of global dispatching.
[0054] The preset restart probability can be set according to the number of power supply areas. In this embodiment, the preset restart probability is the reciprocal of the total number of power supply areas.
[0055] Furthermore, by combining the preset anomaly fusion weights with the direct anomaly relationships and the transmitted anomaly relationships, dynamic penalty factors are output for each power supply area. Direct interference and anomaly transmission are simultaneously fused, taking into account both local direct risks and global indirect risks, thereby improving the accuracy of risk probability.
[0056] In this embodiment, the preset abnormal fusion weight for direct abnormal relationships is 0.8, and the preset abnormal fusion weight for transmitted abnormal relationships is 0.2.
[0057] The upper-level optimization objective function for constructing the shortest manifold based on the first and second allocation paths and allocation path constraints includes: Based on the first and second allocation paths and the allocation path constraints, the upper-level optimization objective function of the shortest manifold is constructed with the goal of minimizing the number of allocation devices and the shortest allocation path length.
[0058] In this embodiment, the dual optimization objectives of minimizing the number of dispatching devices and minimizing the dispatching path length are used as the objective function of the upper-level optimization. This achieves the solution of the dispatching path with the shortest manifold. Under the premise of satisfying the dispatching path constraints, the upper-level optimization can quickly solve for the physically optimal initial dispatching path, providing an initial dispatching scheme that meets the timeliness requirements for switching operations.
[0059] Specifically, the upper-level optimization objective function is: ; in, Represents manifold values; Z represents the weight of the number of equipment to be allocated; Z represents the set of directed edges formed by the first allocation path and the second allocation path. Indicates the power supply area To the power supply area The allocation path; Indicates whether to select a power supply area. To the power supply area The allocation path, The time indicates the selection of the power supply area. To the power supply area The allocation path, Indicates that no power supply area is selected. To the power supply area The allocation path; Indicates the power supply area To the power supply area The number of dispatching devices along the dispatching route; Indicates the weight of the allocated path length; Indicates the power supply area To the power supply area The length of the dispatch path.
[0060] In this embodiment, , , In the process of solving the objective function at the upper level, the allocation path constraint needs to be satisfied.
[0061] In other cases, allocation path constraints also include power flow balance constraints.
[0062] The lower-level optimization objective function with minimum uncertainty, based on the dynamic penalty factor and the heterogeneity of the distribution network set, includes: The initial allocation path output by the upper-level optimization objective function is used to find the backup distribution network; The uncertainty of the standby distribution network is calculated based on the dynamic penalty factor, the heterogeneity of the distribution network set, and the time-series fluctuation of the distribution network load. Construct the lower-level optimization objective function with the lowest uncertainty.
[0063] In this embodiment, the input of the upper-level optimization objective function is used as the input of the lower-level optimization objective function. Using each distribution network in the initial allocation path as a benchmark, the remaining distribution networks associated with it are sought as backup distribution networks. The dynamic penalty factor of the backup distribution networks is retrieved, and the heterogeneity of the distribution network set is quantified by the number of distribution networks participating in the backup. The degree of load time-series fluctuation of the distribution network is obtained by comparing the real-time load data of the backup distribution network under the current time series with the load data under the corresponding historical time series. The uncertainty of the backup distribution network is then calculated. Specifically, the larger the dynamic penalty factor, the greater the uncertainty of the backup distribution network; the greater the heterogeneity of the distribution network set, the smaller the uncertainty of the backup distribution network; and the greater the load time-series fluctuation of the distribution network, the greater the uncertainty of the backup distribution network.
[0064] Specifically, the lower-level optimization objective function is: ; in, The uncertainty of the standby distribution network is represented by K; K represents the set of all power supply areas that can be kept energized by at least one path from the initial dispatch path. Indicates the power supply area To the power supply area The allocation path; Indicates the weight of the dynamic penalty factor; Indicates the power supply area To the power supply area The dynamic penalty factor value of the allocation path; This represents the maximum value of the dynamic penalty factor; Weights representing the degree of heterogeneity of the distribution network set; Indicates the power supply area With power supply area The degree of heterogeneity of the distribution network set; Weights representing the degree of time-series fluctuations in distribution network load; Indicates the power supply area under time sequence t Real-time load data; express Power supply area under timing Load data; This represents the historical time series corresponding to time series t; .
[0065] In this embodiment, if the distribution networks corresponding to the two power supply areas in the allocation path belong to different distribution network sets, their distribution network set heterogeneity is considered to be 1; otherwise, their distribution network set heterogeneity is 0. Historical time series corresponding to the current time series can be extracted based on historical dates or based on historical months with the highest environmental similarity. The initial allocation path distribution network is used as a benchmark to filter associated backup distribution networks, ensuring consistent optimization between upper and lower layers and close topological association. The common topological boundary between upper and lower layer optimizations is fixed, ensuring that the search for backup allocation paths always revolves around the initial allocation path. This ensures that the two layers of optimization are consistent in the solution space and have close topological association, avoiding the fragmentation of the solution space caused by each layer acting independently. Furthermore, an optimization function is constructed with the minimum uncertainty as the sole objective. Under the premise of satisfying hard constraints such as main transformer capacity and voltage, the combination with the lowest risk is selected from the standby distribution network to form a standby dispatch path. The abnormal risks associated with historical switching and the difference between the current load and the historical load are used to obtain the instability of the standby distribution network operating conditions. The regional load stability is offset by the number of distribution network sets to improve the reliability of the standby dispatch path.
[0066] Furthermore, in response to the switching operation requirements, the initial dispatch path is output based on the upper-level optimization objective function of the switching dispatch model and the real-time operation data of each power supply area to execute the initial switching dispatch adapted to the switching operation requirements. At the same time, the backup dispatch path is output based on the lower-level optimization objective function of the switching dispatch model, and the equipment in the backup dispatch path is converted from cold standby to hot standby. That is, the tie switches and transfer equipment involved in the backup dispatch path are preset to the hot standby closed standby state to improve the response efficiency of interlocking anomalies.
[0067] In other cases, based on the differences in electricity consumption characteristics among power supply area types, load feature extraction is performed by category. For different power supply area types, corresponding preset features are extracted to avoid feature distortion and similarity calculation deviations caused by using a uniform feature extraction standard across the entire region. Specifically, preset feature extraction is performed on the daily load curves of different distribution networks based on power supply area type labels to obtain the corresponding load feature vectors, including: For the daily load curve corresponding to the industrial area, load features are extracted based on preset features such as daily average load, daily load rate, daily minimum load rate, and peak-valley difference rate to obtain a load feature vector; For the daily load curve corresponding to the commercial area, load features are extracted based on preset features such as peak-valley difference rate, average load during peak hours, average load during valley hours, and load differences during holidays, to obtain a load feature vector; For the daily load curve corresponding to the residential area, load features are extracted based on the preset features of daily average load, peak period average load, valley period average load, and seasonal load sensitivity to obtain the load feature vector; For the daily load curve corresponding to the rural power grid area, load features are extracted based on preset characteristics such as daily minimum load rate, seasonal load sensitivity, and load differences during holidays to obtain load feature vectors.
[0068] At this point, cluster analysis is performed on the entire distribution network based on the power supply area type label and load feature vector, constructing several distribution network sets including: Based on the power supply area type label, the entire distribution network is clustered once to obtain the first distribution network set; The first distribution network set is clustered a second time based on the load characteristic type to obtain the second distribution network set. The second distribution network set is clustered three times based on the load feature vector to obtain the third distribution network set.
[0069] First, a clustering is performed based on the power supply area type to achieve preliminary zoning and isolation of distribution networks for different power consumption types, avoiding inaccurate division benchmarks caused by differences in cross-type load characteristics. Then, a second clustering is performed on the initial clustering results based on load characteristic types to unify the feature extraction types of distribution networks in the same cluster and eliminate data comparison bias caused by differential feature screening. Finally, a third clustering is performed based on the similarity of load feature vectors to ensure that the distribution networks in the third distribution network set maintain a high degree of homogeneity in terms of regional attributes, feature rules, and load change patterns.
[0070] In this embodiment, the second dispatch path includes at least a first-layer path corresponding to the first distribution network set, a second-layer path corresponding to the second distribution network set, and a third-layer path corresponding to the third distribution network set.
[0071] At this point, the difference from Example 1 is that the uncertainty of the standby distribution network, calculated based on the dynamic penalty factor, the heterogeneity of the distribution network set, and the time-series fluctuation of the distribution network load, includes: The heterogeneity of the distribution network set is calculated based on the number of paths in the first layer, the second layer, and the third layer, combined with preset weights.
[0072] In this embodiment, in the initial state responding to the switching operation requirement, the preset quantity weight of the first-layer path is less than the preset quantity weight of the second-layer path, and the preset quantity weight of the second-layer path is less than the preset quantity weight of the third-layer path. Based on the prior principle that higher heterogeneity leads to lower operational risk, this ensures that the backup dispatch path from the third-layer path can obtain the optimal evaluation in the absence of real-time feedback, while avoiding the risk of no solution or excessive electrical distance, thus ensuring the reliability of the backup dispatch path. After executing the initial dispatch path, the preset quantity weight is adjusted according to the fluctuation degree of real-time operating data of each distribution network on the initial dispatch path. When the similarity of the fluctuation degree of real-time operating data of distribution networks of different power supply area types is higher, the preset quantity weight of the first-layer path is larger; when the fluctuation degree of real-time operating data is smaller, the preset quantity weight of the second-layer path is larger. This ensures dispatch safety while avoiding excessive heterogeneity and improving dispatch adaptability.
[0073] Correspondingly, in this embodiment, the heterogeneity of the distribution network sets among the third distribution network sets is higher than that among the second distribution network sets, and the heterogeneity of the distribution network sets among the second distribution network sets is higher than that among the first distribution network sets.
[0074] Specifically, the number of paths at the first, second, and third levels represents the total number of physical interconnection topologies that can participate in switching and power transfer scheduling between different distribution network sets at the corresponding level. Multiple dispatch paths can be formed between any two distribution network sets at the same level using multiple independent interconnection lines and ring network backup channels; therefore, the number of paths at each level is a positive integer greater than or equal to 1.
[0075] The preset quantity weight range is (0,1], and the sum of the preset quantity weights of the first layer path quantity, the second layer path quantity, and the third layer path quantity is 1.
[0076] In some feasible embodiments, in the initial state responding to the switching operation requirement, the preset quantity weight of the first-level path is 0.1, the preset quantity weight of the second-level path is 0.25, and the preset quantity weight of the third-level path is 0.65. At this time, if there are multiple first-level paths between two power supply areas (i.e., the two power supply areas belong to different first distribution network sets), the heterogeneity of the distribution network set is the product of 0.1 and the number of first-level paths; if there are multiple second-level paths between two power supply areas (i.e., the two power supply areas belong to different second distribution network sets), the heterogeneity of the distribution network set is the product of 0.25 and the number of second-level paths; if there are multiple third-level paths between two power supply areas (i.e., the two power supply areas belong to different third distribution network sets), the heterogeneity of the distribution network set is the product of 0.65 and the number of third-level paths.
[0077] Understandably, in actual calculations, the heterogeneity of a distribution network set is the product of the number of paths at the corresponding level, the preset weight of the corresponding level, and the truth value of the cross-set judgment. The case where the power supply area belongs to the same distribution network set is included using the cross-set judgment truth value. That is, if the distribution networks corresponding to two power supply areas belong to different distribution network sets, their cross-set judgment truth value is considered to be 1; otherwise, it is 0. When there are multiple paths at different levels between two power supply areas, the heterogeneity of the distribution network set is the weighted sum of the paths at each level. Furthermore, the calculated heterogeneity of the distribution network set is mapped to the interval [0,1] to facilitate uncertainty calculation.
[0078] By maximizing the preset quantity weight of the third-layer path in the initial state, the third-layer path is made the preferred path. This can avoid excessive cross-regional and cross-type transfers and reduce the possibility of synchronous anomalies after the transfer to a certain extent.
[0079] The specific embodiments described above are preferred embodiments of the switching and dispatching method based on the integrated coordination of main and auxiliary circuits in this application, and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape and structure of this application are within the protection scope of this application.
Claims
1. A switching and dispatching method based on integrated main and distribution coordination, characterized in that: Includes the following steps: The entire distribution network is divided according to the power supply area type and load characteristics corresponding to each distribution network, and several distribution network sets are constructed. The first dispatch path within the distribution network set and the second dispatch path between distribution network sets are constructed based on the inter-regional interconnection lines between distribution networks, and dispatch path constraints are constructed based on the main distribution load constraints. Anomaly risk relationships are constructed based on historical switching anomaly events and historical abnormal operation data of each power supply area, and dynamic penalty factors for each power supply area are obtained based on the anomaly risk relationships. Based on the first and second allocation paths and the allocation path constraints, a higher-level optimization objective function for the shortest manifold is constructed. A lower-level optimization objective function with minimum uncertainty is constructed based on the dynamic penalty factor and the heterogeneity of the distribution network set; A switching operation model is constructed using the upper-level optimization objective function and the lower-level optimization objective function. In response to switching operation requirements, the switching dispatch strategy, which includes the initial dispatch path and the backup dispatch path, is output based on the switching dispatch model and the real-time operation data of each power supply area.
2. The switching and dispatching method based on integrated main and distribution coordination as described in claim 1, characterized in that: The process of constructing anomaly risk relationships based on historical switching anomaly events and historical abnormal operation data for each power supply area, and obtaining dynamic penalty factors for each power supply area based on these anomaly risk relationships, includes: Each power supply area is treated as a network node, and directed edges are constructed based on the first dispatch path and the second dispatch path; Based on the historical switching anomaly events and historical abnormal operation data of each power supply area, the number of historical switching associations between two power supply areas and the number of anomaly events at each level in the historical switching associations are obtained. Directed edge weights are constructed using the ratio of the number of historical power outage associations, the number of abnormal events at each level, and the preset abnormal weights at each level. The direct and propagated anomaly relationships between power supply areas are obtained based on directed edges and their weights. The dynamic penalty factor for each power supply area is output based on the direct anomaly relationship and the conducted anomaly relationship.
3. The switching and dispatching method based on integrated main and distribution coordination as described in claim 2, characterized in that: The construction of directed edge weights based on the ratio of historical switching frequency, the number of abnormal events at each level, and the preset abnormal weights at each level includes: Obtain the corresponding preset level prior weight based on the anomaly type of historical switching anomaly events; Perform a normalized mapping on the historical abnormal operation data of all historical switching abnormal events to obtain adjustment coefficients; The preset level weights of abnormal events at each level are obtained by adjusting the coefficients and using preset level prior weights; The weighted number of anomalies is calculated using preset hierarchical weights and the number of anomalies at each level. The weighted number of anomalies is then used to construct directed edge weights based on the ratio of the weighted number of anomalies to the number of historical power outage associations.
4. The switching and dispatching method based on integrated main and distribution coordination as described in claim 3, characterized in that: The method of obtaining the direct anomaly relationship and the propagation anomaly relationship between power supply areas based on directed edges and directed edge weights includes: If there is a directed edge between two power supply areas, a direct anomaly relationship is constructed based on the directed edge weight corresponding to that directed edge. By using the restart random walk algorithm, taking any power supply area as the starting node, the steady-state access probability of reaching other power supply areas is calculated based on the directed edge weights, and the transmission anomaly relationship between each power supply area is obtained.
5. The switching and dispatching method based on integrated main and distribution coordination as described in claim 4, characterized in that: The method of using a restarted random walk algorithm, taking any power supply area as the starting node, calculates the steady-state access probability of reaching other power supply areas based on the directed edge weights, and obtains the transmission anomaly relationship between the power supply areas, includes: For each power supply area, calculate the sum of the directed edge weights of all its outgoing edges. Use the ratio of the directed edge weight of each outgoing edge to the sum of the directed edge weights as the one-step transition probability of the power supply area along that outgoing edge, and construct a transition probability matrix. Based on the preset restart probability, perform random walk iterations on the transition probability matrix until convergence, and obtain the steady-state access probability vector; The probability values between the starting node and each arriving node in the steady-state access probability vector are used as the transmission anomaly relationships between the corresponding power supply areas.
6. The switching and dispatching method based on integrated main and distribution coordination as described in claim 1, characterized in that: The upper-level optimization objective function for constructing the shortest manifold based on the first and second allocation paths combined with allocation path constraints includes: Based on the first and second allocation paths and the allocation path constraints, the upper-level optimization objective function of the shortest manifold is constructed with the goal of minimizing the number of allocation devices and the shortest allocation path length.
7. The switching and dispatching method based on integrated main and distribution coordination as described in claim 1, characterized in that: The lower-level optimization objective function with the lowest uncertainty, constructed based on the dynamic penalty factor and the heterogeneity of the distribution network set, includes: The initial allocation path output by the upper-level optimization objective function is used to find the backup distribution network; The uncertainty of the standby distribution network is calculated based on the dynamic penalty factor, the heterogeneity of the distribution network set, and the time-series fluctuation of the distribution network load. Construct the lower-level optimization objective function with the lowest uncertainty.
8. The switching and dispatching method based on integrated main and distribution coordination as described in claim 1, characterized in that: The process of dividing the entire distribution network according to the power supply area type and load characteristics corresponding to each distribution network, and constructing several distribution network sets, includes: Obtain the power supply area type label and daily load curve within a preset time period for each distribution network; Pre-defined features are extracted from the daily load curves of different distribution networks to obtain the corresponding load feature vectors; Cluster analysis is performed on the entire distribution network based on the power supply area type label and load feature vector to construct several distribution network sets.
9. The switching and dispatching method based on integrated main and distribution coordination as described in claim 8, characterized in that: The method of clustering the entire distribution network based on power supply area type labels and load feature vectors to construct several distribution network sets includes: Based on the power supply area type label, the entire distribution network is clustered once to obtain the first distribution network set; The first distribution network set is clustered a second time based on the load characteristic type to obtain the second distribution network set. The second distribution network set is clustered three times based on the load feature vector to obtain the third distribution network set.
10. The switching and dispatching method based on integrated main and distribution coordination as described in claim 9, characterized in that: The second dispatch path includes at least a first-level path corresponding to the first distribution network set, a second-level path corresponding to the second distribution network set, and a third-level path corresponding to the third distribution network set; The heterogeneity of the distribution network set is calculated based on the number of paths in the first layer, the second layer, and the third layer, combined with a preset weight.