Power distribution network mutual aid operation control method and system under topology reconstruction
By constructing a multi-scale topology-aware state tensor and dual-domain optimization, topology switching candidates are generated, and multi-scenario power flow evolution and protection coordination simulations are performed. This solves the problem of insufficient accuracy in the mutual assistance response of the distribution network and achieves more efficient mutual assistance operation control.
Patent Information
- Application Number
- CN202512028027.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-04-07
AI Technical Summary
Existing topology reconfiguration methods lack sufficient accuracy in mutual assistance response in distribution networks, cannot effectively cope with complex operating conditions, and lack comprehensive perception of multi-dimensional network states and adaptability to multiple scenarios.
By constructing a multi-scale topology-aware state tensor, generating topology switching candidates, performing multi-scenario power flow evolution and protection coordination simulations, introducing a mutual assistance contribution matrix, constructing a supply-demand complementary relationship graph, calculating the operating index set, and adopting a dual-domain optimization approach of local sensitivity and global trade-offs to output a mutual assistance operation control strategy.
It improves the accuracy of the mutual assistance operation and control of the distribution network, and enhances its adaptability and mutual assistance effect to complex operating conditions.
Smart Images

Figure CN121813394A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of smart distribution networks, and in particular to a method and system for mutual assistance operation control of distribution networks under topology reconfiguration. Background Technology
[0002] With the integration of a high proportion of renewable energy into the distribution network, the uncertainty and volatility of system operation have increased dramatically. How to dynamically adjust the network topology to achieve mutual support between supply and demand resources and improve system resilience has become a key challenge for smart distribution network operation and control. Currently, methods mainly rely on topology switching based on static rules or reconfiguration methods based on single-objective optimization, adjusting the network structure through predefined operating scenarios or real-time power flow optimization. However, these existing methods largely depend on limited operational information for decision-making, lacking a comprehensive understanding of the network's multi-dimensional state and failing to fully consider the actual contributions of inter-node supply and demand complementarity and adaptability to multiple scenarios. This results in insufficient adaptability and limited mutual support effects of topology reconfiguration strategies when dealing with complex operating conditions.
[0003] At present, the mutual assistance operation control of distribution networks under topology reconfiguration has the technical problem of insufficient accuracy of mutual assistance response. Summary of the Invention
[0004] This application provides a method and system for distribution network mutual assistance operation control under topology reconfiguration. It collects real-time information from multiple sources in the distribution network, integrates switch connectivity and line congestion to construct a multi-scale topology-aware state tensor, triggers topology reconfiguration requirements based on this tensor, generates multiple topology switching candidates, performs multi-scenario power flow evolution and protection coordination simulations on these candidates, filters out feasible candidate topologies, introduces a mutual assistance contribution matrix, constructs a node supply-demand complementarity graph, calculates the operation index set of each candidate topology, constructs a multi-objective evaluation function, and outputs a mutual assistance operation control strategy through dual-domain optimization of local sensitivity and global trade-offs within the feasible candidate topology set. This solves the technical problem of insufficient mutual assistance response accuracy in existing distribution network mutual assistance operation control under topology reconfiguration, achieving the technical effect of improving the accuracy of distribution network mutual assistance operation regulation.
[0005] This application provides a method for mutual assistance operation control of distribution networks under topology reconfiguration, including: acquiring multi-source real-time information on node voltage, power flow distribution, line impedance, load forecast, and renewable energy output forecast of the distribution network; constructing a multi-scale topology-aware state tensor by integrating switch connectivity and line congestion; triggering topology reconfiguration demand based on the multi-scale topology-aware state tensor to generate multiple sets of topology switching candidates; performing multi-scenario power flow evolution and protection coordination simulation on the multiple sets of topology switching candidates to construct a set of feasible candidate topologies; introducing a mutual assistance contribution matrix into the set of feasible candidate topologies; constructing a supply-demand complementarity relationship diagram between nodes based on load forecast and renewable energy output forecast; calculating the node voltage recovery index, line load balancing index, and renewable energy absorption improvement rate under each candidate topology to establish a set of operation indices; constructing a multi-objective evaluation function based on the mutual assistance contribution matrix and the set of operation indices; performing dual-domain optimization based on local sensitivity and global trade-offs on the set of feasible candidate topologies; and outputting a mutual assistance operation control strategy.
[0006] In a possible implementation, after constructing a multi-objective evaluation function based on the mutual contribution matrix and the operating index set, the following processing is performed on the feasible candidate topology set based on a dual-domain optimization of local sensitivity and global trade-offs: After constructing the objective evaluation function, a cross-node index perturbation propagation tensor is constructed using the multi-dimensional gradient changes of the operating index set. The index perturbation propagation tensor is used to characterize the propagation strength and diffusion path of local branch state changes on the global supply and demand complementarity relationship in any feasible candidate topology; the perturbation absorption capacity of the corresponding feasible candidate topology is calculated using the index perturbation propagation tensor to generate a local adjustable margin score characterizing the local adjustment controllability; the local adjustable margin score is fused with the mutual contribution matrix at the tensor level to generate a global mutual control field representing the optimal power flow transfer path of the entire network; under the constraints of the global mutual control field, cross-domain coupling optimization of the feasible candidate topology set is performed, and dual-domain optimization is executed.
[0007] In a possible implementation, under the constraints of the global mutual support control field, cross-domain coupling optimization of the feasible candidate topology set is performed, and dual-domain optimization is performed, including the following processes: gradient calculation is performed on each feasible candidate topology based on the multi-objective evaluation function to form a target sensitivity vector for quantifying the direction and magnitude of the influence of each operating index on the comprehensive objective; the target sensitivity vector is projected onto the local adjustable margin score to obtain the target-oriented adjustable weight of the local domain; the target sensitivity vector is projected onto the global mutual support control field to obtain the target-oriented power flow redistribution bias of the global domain; using the target-oriented adjustable weight and the power flow redistribution bias as driving signals, the dynamic evolution of the feasible candidate topology is performed by constructing a joint iterative update rule for the local and global domains, and dual-domain optimization is performed.
[0008] In a possible implementation, the following processing is performed: The joint iterative update rule includes: in each iteration step, based on the target-oriented adjustability weight of the local domain, the state adjustment amount of sensitive branches is weighted and amplified or suppressed according to the weight size to form a local domain update step size; based on the power flow redistribution bias of the global domain, the supply and demand flow directions between nodes are directionally offset to construct a global domain update step size; the local domain update step size and the global domain update step size are fused according to a preset coupling coefficient to generate a comprehensive update vector for indicating the current topology adjustment direction, and feasible candidate topology iterative update management is performed based on the comprehensive update vector.
[0009] In a possible implementation, based on the multi-scale topology-aware state tensor, the topology reconfiguration requirement is triggered by performing the following processing: extracting multi-dimensional variation features of node voltage offset rate, power flow direction change rate, and line congestion degree from the multi-scale topology-aware state tensor, and constructing a topology disturbance-sensitive surface to characterize topology stability; calculating the cross-scale topology risk integral based on the topology disturbance-sensitive surface, and triggering the topology reconfiguration requirement based on the topology risk integral.
[0010] In a possible implementation, the topology reconstruction requirement is triggered based on the topology risk score, and the following processing is performed: it is determined whether the topology risk score meets the adaptive reconstruction threshold; if the topology risk score meets the adaptive reconstruction threshold, the maximum contributing connected path corresponding to the topology risk score is identified, and the topology reconstruction requirement is triggered.
[0011] In a possible implementation, multi-scenario power flow evolution and protection coordination simulations are performed on the multiple sets of topology switching candidates to construct a feasible candidate topology set. The following processing is then performed: a multi-dimensional simulation scenario space is generated based on load forecasting, renewable power output forecasting, and fault disturbance modes; a scenario vector group is constructed, which is used to drive the dynamic evolution of power flow for each topology switching candidate under different operating states; in each scenario, serialized power flow calculations are performed on the topology switching candidates, and a power flow evolution sequence tensor is constructed based on the time gradient of power flow changes; the power flow evolution sequence tensor is used to trigger a protection action consistency check, and a protection coordination consistency score is output; multiple sets of topology switching candidates are screened based on the protection coordination consistency score to construct a feasible candidate topology set.
[0012] In a possible implementation, multiple sets of topology switching candidates are screened based on the protection coordination consistency score, and the following processing is performed: the protection coordination consistency score is fused with the power flow evolution sequence tensor to construct a topology feasibility determination tensor; based on the topology feasibility determination tensor, the steady-state feasibility, transient feasibility, and protection coordination reliability of topology switching under multiple scenarios are quantified, and multiple sets of topology switching candidates are screened.
[0013] In possible implementations, the following processing is performed: consistency verification includes comparing the action criteria of relay protection actions with the time synchronization relationship of power flow evolution trajectory, identifying whether there are overstepping actions, false actions, or failures to act, and constructing a protection coordination consistency score.
[0014] This application also provides a distribution network mutual assistance operation control system under topology reconfiguration, including: a multi-scale topology-aware state tensor construction module, used to acquire multi-source real-time information on node voltage, power flow distribution, line impedance, load forecast, and renewable energy output forecast of the distribution network, and construct a multi-scale topology-aware state tensor by integrating switch connectivity and line congestion; a topology reconfiguration demand triggering module, used to trigger topology reconfiguration demand based on the multi-scale topology-aware state tensor and generate multiple sets of topology switching candidates; and a simulation module, used to perform multi-scenario power flow evolution and protection on the multiple sets of topology switching candidates. The system employs a coordinated simulation approach to construct a set of feasible candidate topologies. An operational index set establishment module introduces a mutual contribution matrix into the feasible candidate topologies, constructs a supply-demand complementarity graph between nodes based on load forecasting and renewable energy output forecasting, calculates the node voltage recovery index, line load balancing index, and renewable energy absorption improvement rate for each candidate topology, and establishes an operational index set. A dual-domain optimization module constructs a multi-objective evaluation function based on the mutual contribution matrix and the operational index set, performs dual-domain optimization based on local sensitivity and global trade-offs on the feasible candidate topologies, and outputs a mutual-aid operation control strategy.
[0015] The proposed method and system for distribution network mutual assistance operation control under topology reconfiguration, as outlined in this application, firstly acquires multi-source real-time information on node voltage, power flow distribution, line impedance, load forecast, and renewable energy output forecast of the distribution network. By integrating switch connectivity and line congestion, a multi-scale topology-aware state tensor is constructed. Then, based on this multi-scale topology-aware state tensor, topology reconfiguration requirements are triggered, generating multiple sets of topology switching candidates. Next, multi-scenario power flow evolution and protection coordination simulations are performed on these candidates to construct a set of feasible candidate topologies. A mutual assistance contribution matrix is then introduced into this set. Based on load forecast and renewable energy output forecast, a supply-demand complementarity graph between nodes is constructed. The node voltage recovery index, line load balancing index, and renewable energy absorption improvement rate are calculated for each candidate topology, establishing a set of operation indices. Finally, a multi-objective evaluation function is constructed based on the mutual assistance contribution matrix and the set of operation indices. Then, a dual-domain optimization based on local sensitivity and global trade-offs is performed on the feasible candidate topologies to output a mutual assistance operation control strategy. Through the above process, the method and system proposed in this application achieve the technical effect of improving the accuracy of the mutual assistance operation and control of the distribution network. Attached Figure Description
[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.
[0017] Figure 1 A flowchart illustrating the distribution network mutual assistance operation control method under topology reconfiguration provided in this application embodiment.
[0018] Figure 2 A schematic diagram of the structure of the distribution network mutual assistance operation control system under topology reconfiguration provided in the embodiments of this application.
[0019] Figure labeling: Multi-scale topology-aware state tensor construction module 10, topology reconstruction requirement triggering module 20, simulation module 30, running index set establishment module 40, dual-domain optimization module 50. Detailed Implementation
[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.
[0021] This application provides a method for distribution network mutual assistance operation control under topology reconfiguration, such as... Figure 1 As shown, the method includes: Step S100: Obtain multi-source real-time information on node voltage, power flow distribution, line impedance, load forecast, and renewable energy output forecast of the distribution network. Construct a multi-scale topology-aware state tensor by integrating switch connectivity and line congestion.
[0022] Specifically, node voltage is collected by smart meters and PMU (Phase Quantity Measurement Unit) devices deployed at each node to determine if the node voltage exceeds limits. Node voltage is a direct indicator for assessing power quality and system stability. Active power, reactive power, and current of the lines are collected by SCADA systems or PMUs to determine if the lines are overloaded, identify weak points in the system, and provide initial values for power flow calculations. Line impedance is an inherent parameter of the line, determined during system modeling, and is used for power flow calculations and network equivalent modeling, forming the basis for constructing the network topology.
[0023] Based on historical load data, weather data, and date type, time series analysis or machine learning methods are used to predict the load demand of each node over a future period. Load demand is a prerequisite for forward-looking scheduling and topology optimization. For example, if a surge in load is predicted for a certain area tomorrow, topology reconfiguration needs to be considered in advance to increase power supply capacity.
[0024] For renewable energy power generation units such as photovoltaic and wind power, based on numerical weather forecasts and historical power output data, a physical mechanism-based model combined with a machine learning model is used to predict the volatility and uncertainty of their future power output, so as to better absorb renewable energy in topology reconfiguration and reduce wind and solar curtailment.
[0025] The real-time status of all sectionalizing switches and tie switches in the distribution network is organized into a switch status matrix, where closed / open is represented by 1 / 0. This matrix determines the network topology. Line congestion is defined as the ratio of the current power flow value of a line to its rated capacity. By calculating the congestion of each line, a line congestion vector is obtained. The node voltage vector, power flow distribution matrix, line impedance matrix, switch status matrix, load prediction vector, renewable energy output prediction vector, and line congestion vector obtained from the above collection and calculation are used to construct a high-dimensional, multi-scale topology-aware state tensor based on their characteristics at different scales such as node level, line level, region level, and time level, through tensor decomposition or direct stacking. For example, a possible tensor dimension could be [number of nodes, number of lines, time steps, number of features], where the number of features includes information such as voltage, power, switch status, and congestion. This tensor comprehensively characterizes the operating status and topology of the distribution network in the current and predicted future period, providing a unified data input for topology reconfiguration decisions.
[0026] Step S200: Based on the multi-scale topology-aware state tensor, trigger the topology reconstruction requirement and generate multiple sets of topology switching candidates.
[0027] Specifically, based on the multi-scale topology-aware state tensor constructed in step S100, the topology risk integral is calculated using the methods described in steps S210 and S220. When this integral value exceeds a preset adaptive reconfiguration threshold, it is considered that the current network operating state has significant risks, such as voltage instability, severe line overload, and easy fault propagation. These risks need to be mitigated or eliminated by changing the network topology, thereby triggering the topology reconfiguration requirement.
[0028] Once a reconfiguration requirement is triggered, the system employs graph search algorithms such as depth-first search or breadth-first search, combined with the switch layout and network constraints of the distribution network, to find all possible new topologies achievable by changing a finite number of switch states. For each potential new topology, the switches requiring operation and their order of operation are recorded, forming a topology switching candidate scheme. To ensure computational efficiency, the number of generated candidate schemes can be limited according to certain heuristic rules, such as prioritizing tie switches connecting heavily loaded and lightly loaded areas, thereby avoiding combinatorial explosion. For example, the system may generate multiple different topology switching candidates such as "close switch K1, open switch K2" and "close switch K3, open switch K4".
[0029] In one possible implementation, based on the multi-scale topology-aware state tensor triggering the topology reconfiguration requirement, step S200 further includes step S210, extracting multi-dimensional variation features of node voltage offset rate, power flow direction abrupt change rate, and line congestion degree from the multi-scale topology-aware state tensor, and constructing a topology disturbance-sensitive surface to characterize topology stability. Specifically, the difference between the actual voltage and the rated voltage of each node is calculated, and then divided by the rated voltage to obtain the voltage offset rate, which directly reflects the quality of the node voltage. The power flow evolution sequence tensor is analyzed, and by comparing the power flow direction of adjacent time steps to determine whether it is positive or negative, the number or frequency of changes in the line power flow direction per unit time is calculated to obtain the power flow direction abrupt change rate. Frequent power flow direction abrupt changes may indicate system instability or oscillations. As described in step S100, the ratio of the current power flow value of the line to its rated capacity is calculated to obtain the line congestion degree. A convolutional neural network (CNN) can be used to extract features from the multi-scale topology-aware state tensor. The CNN's convolutional kernel can automatically learn local correlation features in the data, thereby efficiently extracting the aforementioned key variation features and their combination patterns from the high-dimensional tensor. Alternatively, traditional signal processing methods, such as sliding window and Fourier transform, can be combined with statistical analysis to extract features.
[0030] The extracted multidimensional variation features, including voltage offset rate, power flow direction abrupt change rate, and line congestion, are used as coordinates in a high-dimensional space. Each sample point represents a specific system state. Using these sample points in the high-dimensional space, methods such as radial basis function (RBF) interpolation, polynomial interpolation, or Gaussian process regression are employed to construct a continuous topological disturbance-sensitive surface in the high-dimensional space. Each point on this surface corresponds to a potential system operating state. The height or curvature of the surface can be defined as the system's sensitivity to disturbances or its risk value. For example, a steeper region on the surface indicates that the system state at that point is highly sensitive to small disturbances, making it prone to large state changes or even instability.
[0031] Step S220: Calculate the cross-scale topological risk integral based on the topological disturbance-sensitive surface, and trigger the topological reconfiguration requirement based on the topological risk integral. Specifically, on the topological disturbance-sensitive surface constructed in step S210, an integration region is defined. This region can be determined based on the position of the current system state point on the surface, a preset risk threshold, and the system's safety boundary. For example, the region around the current state point where the surface height exceeds a certain risk threshold can be defined as the integration region. Within the integration region, a risk density function is defined, the value of which is proportional to the surface height, i.e., the disturbance sensitivity. Using numerical integration methods, such as Monte Carlo integration or Gaussian quadrature, the risk density function within the integration region is integrated. The integration result is the topological risk integral, which quantifies the accumulated risk that the current system state may fall into an unsafe operating state due to its own fluctuations or external disturbances over a future period. The calculated topological risk integral is compared with a preset adaptive reconfiguration threshold. If the integral value is greater than or equal to the threshold, the current risk level of the system is considered unacceptable, and measures need to be taken to reduce the risk, thereby triggering the topological reconfiguration process.
[0032] In one possible implementation, step S220, which triggers a topology reconfiguration requirement based on the topology risk integral, further includes step S221, determining whether the topology risk integral meets the adaptive reconfiguration threshold. Specifically, an initial static threshold is set based on the system's historical fault data, safe operation procedures, and economic assessment. To improve the adaptability and accuracy of the threshold, an adaptive adjustment mechanism is introduced, employing fuzzy logic-based adjustment. Fuzzy rules are defined, such as lowering the threshold when the load level is high and renewable energy output fluctuates greatly; and lowering the threshold when the system is under maintenance and has low redundancy. Based on the fuzzy input of the current operating condition, an adjusted threshold is output through fuzzy inference. A lower and upper limit constraint is set for the adaptive threshold to ensure basic system safety and avoid unnecessary frequent reconfigurations. The real-time topology risk integral value calculated in step S220 is compared with the reconfiguration threshold determined by the adaptive mechanism at the current moment. If the topology risk integral ≥ the adaptive reconfiguration threshold, it is determined that the requirement is met, and a reconfiguration action is triggered; otherwise, it is determined that the requirement is not met, and the current topology structure is maintained.
[0033] Step S222: If the topology risk integral meets the adaptive reconfiguration threshold, the maximum contributing connected path corresponding to the topology risk integral is identified, triggering the topology reconfiguration requirement. Specifically, when the topology risk integral meets the reconfiguration threshold, the network element or path that contributes the most to the risk integral is identified. The contribution is defined as the extent to which a small change in the state of a branch or node causes a change in the topology risk integral. This can be calculated by taking the partial derivative of the risk integral function with respect to variables such as branch admittance or node injected power, i.e., sensitivity analysis. Based on the calculated contribution of branches or nodes to the risk, a path search algorithm is used to find one or more connected paths in the distribution network topology from high-risk sources to relatively safe areas or areas that can provide support. The branches on these paths contribute the most to the overall risk and are the key factors leading to excessive system risk. For example, sensitivity analysis reveals that the overload of lines L1, L2, and L3 contributes the most to the risk integral, and they constitute a critical power supply path from load center A to power source B; therefore, this path is identified as the maximum contributing connected path. Once the connecting path with the greatest contribution is identified, the areas and objects that need to be adjusted are clear. That is, the power flow distribution needs to be changed by operating the switches on or around the path to reduce the burden on the critical branches and thus reduce system risk.
[0034] Step S300: Perform multi-scenario power flow evolution and protection coordination simulation on the multiple sets of topology switching candidates to construct a feasible candidate topology set.
[0035] Specifically, according to step S310, a multi-dimensional simulation scenario space covering load forecasting, renewable energy output forecasting, and fault disturbance modes is generated. For each topology switching candidate scheme, in each generated scenario, sequential power flow calculations are performed using power system simulation software, including steady-state power flow calculations and transient simulations considering dynamic processes. During the simulation, the changes in key electrical quantities over time, such as node voltage and line current / power, are recorded to form a power flow evolution sequence tensor.
[0036] In the simulation platform, based on the actual relay protection configuration of the distribution network, such as instantaneous overcurrent protection, overcurrent protection, and differential protection, corresponding protection device models are built, and protection settings and operating time limits are set. Simultaneously with the power flow evolution simulation, the operating logic of the relay protection device is simulated. When the electrical quantities in the simulation reach the protection's operating conditions, the protection model issues a trip or close command.
[0037] Using the method in step S330, the time synchronization between the theoretical criteria for relay protection operation and the actual power flow evolution trajectory is compared to check for inconsistencies such as cascading tripping, protection maloperation, or failure to operate, and a score is assigned. Screening criteria are set, such as: in all simulation scenarios, node voltage deviation is within the allowable range, and the load rate of lines and equipment does not exceed the rated value; the protection coordination consistency score is higher than a certain set value, ensuring the correct and reliable operation of the protection system; the system can remain stable in transient simulations, and oscillations such as power angle and voltage can decay rapidly. Based on the above criteria, all topology switching candidate schemes are evaluated one by one. Candidate schemes that meet these constraints in all or most simulation scenarios are included in the feasible candidate topology set.
[0038] In one possible implementation, multi-scenario power flow evolution and protection coordination simulations are performed on the multiple sets of topology switching candidates to construct a feasible candidate topology set. Step S300 further includes step S310, generating a multi-dimensional simulation scenario space based on load forecasting, renewable energy output forecasting, and fault disturbance modes, and constructing a scenario vector group. The scenario vector group is used to drive the dynamic evolution of power flow for each topology switching candidate under different operating states. Specifically, the scenario dimensions and parameters are determined. The load dimension input is the load forecast curve of each node, and fluctuation ranges such as ±5% and ±10% of the forecast value can be used as different load level scenarios. The renewable energy dimension input is the output forecast curve of each photovoltaic power station and wind farm. Combined with the uncertainty of its forecast, scenarios with different output levels or different fluctuation modes are generated. The fault disturbance dimension defines the possible fault types, fault locations, and fault durations. Fault types include three-phase short circuits, single-phase ground faults, and two-phase short circuits, and fault durations include transient faults and permanent faults. By utilizing historical operating data of the power distribution network, clustering algorithms are used to group similar operating conditions into one class, and the center or typical representative of each class can serve as a simulation scenario. All parameters of each generated simulation scenario, such as load values at each node, output values of each renewable energy source, fault type, and location coding, are quantified and organized into a multi-dimensional vector. All generated scenario vectors together constitute a scenario vector group.
[0039] Step S320: In each scenario, serialized power flow calculations are performed on the topology switching candidates, and a power flow evolution sequence tensor is constructed based on the time gradient of power flow changes. Specifically, an appropriate simulation time step is set according to simulation requirements; for example, the transient process focuses on the millisecond level, while the medium- to long-term dynamics focus on the second or minute level. For each topology switching candidate and each scenario vector, the system state at the current moment or the start of the scenario is used as the initial condition. At each time step, based on the current network topology, injected power, and the dynamic model of the components, power flow calculation algorithms such as the Newton-Raphson method and the fast decomposition method, or transient stability calculation programs, are used to solve for the system state quantities at the end of that time step. The above process is repeated until the simulation period ends, thereby obtaining the complete time series of the changes of each electrical quantity over time.
[0040] The large amount of time series data obtained from serialized power flow calculations is organized in a structured manner. For example, for a simulation containing N nodes, M lines, and T time steps, its output data can be organized into a tensor with dimensions [time step T, number of nodes N, number of lines M, and types of electrical quantities S]. Each element in the tensor represents the value of the s-th type of electrical quantity at the n-th node and m-th line at the t-th time step.
[0041] Step S330: The power flow evolution sequence tensor triggers a protection action consistency check, outputting a protection coordination consistency score. The consistency check includes comparing the action criteria of the relay protection actions with the time synchronization relationship of the power flow evolution trajectory, identifying whether overstepping actions, maloperations, or failures to operate occur, and constructing a protection coordination consistency score. Specifically, in the power flow evolution simulation model, a corresponding relay protection model is connected to each protected object such as each line and each bus. These models contain specific action criteria, such as current setting values, action time limits, and impedance boundaries. At each time step of the serialized power flow calculation, the simulated electrical quantity at that moment is input into the corresponding protection model. The protection model determines in real time whether the action conditions are met based on its internal logic. If the protection model determines that the action conditions are met, the action time and action type of the protection are recorded.
[0042] Compare the protection action with the actual fault situation reflected by the power flow evolution trajectory, and compare whether the actual action time of the protection model is consistent with the theoretical time when the protection should act after the fault occurs; check whether there is overstepping action, that is, the lower level protection does not act while the upper level protection acts first; check whether there is false action, that is, the protection acts when the system is running normally or when a fault outside the protection range occurs, or fails to act, that is, the protection does not act when a fault occurs within the protection range.
[0043] Based on the above verification results, quantitative scoring rules are formulated. For example: no incoordination, i.e., correct action, no skipping steps, no erroneous action or refusal to act, gets full marks; a slight inconsistency in the timing of an action within the allowable error range deducts X points; a skipping step, erroneous action, or refusal to act deducts Y points, where Y > X; different deduction weights are set according to the severity and frequency of incoordination events.
[0044] The coordination of protection actions for a topology switching candidate is comprehensively evaluated across all simulation scenarios, and an overall protection coordination consistency score is calculated according to the scoring rules. This score reflects the degree of matching and reliability performance between the protection system and the system's dynamic behavior under the given topology scheme.
[0045] Step S340: Based on the protection coordination consistency score, multiple sets of topology switching candidates are screened to construct a feasible candidate topology set. Specifically, comprehensive screening indicators are set, including protection coordination consistency score, power flow constraint satisfaction, transient stability, and network loss. Power flow constraint satisfaction refers to whether node voltages are within acceptable ranges and line power flows exceed their rated capacity in all simulation scenarios, which can be quantified as a pass rate. Transient stability refers to whether the system experiences power angle instability, voltage collapse, or other phenomena during simulation, which can be judged by analyzing whether the power angle swing curve converges and whether the voltage can recover. Network loss refers to the total network loss of the system calculated in steady-state scenarios, serving as an economic indicator. A minimum acceptable threshold is set for each screening indicator. For example: protection coordination consistency score ≥ 85 points; node voltage pass rate ≥ 99%; all line load rates ≤ 105%; no instability phenomena in steady-state simulation. The performance of each topology switching candidate scheme in all simulation scenarios is compared one by one with the above-set thresholds. When a topology switching candidate solution meets the threshold requirements of all screening indicators in all or a specified proportion, such as more than 95%, simulation scenarios, it is determined to be feasible and added to the set of feasible candidate topologies.
[0046] In one possible implementation, multiple sets of topology switching candidates are screened based on the protection coordination consistency score. Step S340 further includes step S341, which fuses the protection coordination consistency score with the power flow evolution sequence tensor to construct a topology feasibility determination tensor. Specifically, the protection coordination consistency score is integrated into the original power flow evolution sequence tensor as a new dimension or a new feature channel. For example, if the original power flow evolution sequence tensor is [T, N, M, S], after fusion, for a simulation containing Z scenarios, the dimensions of the topology feasibility determination tensor can become [number of scenarios Z, time step T, number of nodes N, number of lines M, electrical quantity type S]. The newly added feature channel values at all time steps, all nodes, lines, and electrical quantity types are filled with the protection coordination consistency score for that scenario.
[0047] Step S342: Based on the topology feasibility determination tensor quantization, the steady-state feasibility, transient feasibility, and protection coordination reliability of topology switching under multiple scenarios are evaluated, and multiple sets of topology switching candidate screening are performed. Specifically, for the power flow evolution sequence tensor under each scenario, indicators such as node voltage amplitude and line load rate under steady-state conditions are extracted. The number of voltage-overrunning nodes, the degree of overrunning, and the duration are calculated; the number of overloaded lines, the overload multiple, and the duration are also calculated. This information is combined into a steady-state feasibility index using a weighted summation method, for example: Steady-state feasibility = 1 - (α × weighted sum of voltage overruns + β × weighted sum of line overloads), where α and β are weighting coefficients. The closer the steady-state feasibility value is to 1, the better the steady-state feasibility.
[0048] Analyze the power angle swing curve, voltage recovery curve, and frequency change curve during the transient process. Calculate the maximum power angle swing amplitude, maximum voltage drop amplitude and recovery time, maximum frequency deviation and recovery time, etc. A transient feasibility index can also be constructed, for example: Transient Feasibility = γ × (1 - Maximum Power Angle Swing Amplitude / Critical Swing Amplitude) + δ × (1 - Maximum Voltage Drop Amplitude / Rated Voltage) + ε × (Recovery Time / Standard Recovery Time), where γ, δ, and ε are weighting coefficients, and each index needs to be normalized.
[0049] By utilizing the protection coordination consistency score information in the topology feasibility determination tensor, this score is directly normalized and used as a protection coordination reliability index. For example: Protection coordination reliability = Protection coordination consistency score / Full score.
[0050] For each topology switching candidate, calculate the average or minimum values of its steady-state feasibility index, transient feasibility index, and protection coordination reliability index across all simulation scenarios. Set minimum acceptable threshold values for each of these three quantified feasibility indicators. When a topology switching candidate scheme's comprehensive feasibility assessment results across all simulation scenarios meet or exceed the set threshold values, it is ultimately determined to be feasible and retained in the feasible candidate topology set.
[0051] Step S400: Introduce a mutual assistance contribution matrix to the feasible candidate topology set, construct a supply and demand complementarity relationship diagram between nodes based on load forecasting and renewable energy output forecasting, calculate the node voltage recovery index, line load balance index and renewable energy absorption improvement rate under each candidate topology, and establish an operation index set.
[0052] Specifically, based on load forecasting and renewable energy output forecasting, the net injected power of each node over a future period is analyzed. The net injected power of a node is calculated by subtracting the forecasted load demand from the node's forecasted renewable energy output. A positive result indicates that the node is a power surplus node with potential power supply capacity; a negative result indicates that the node has a power deficit and is a potential power recipient. Mutual assistance paths between nodes are identified. For each power deficit node, all surplus nodes that can potentially provide power are identified, and the corresponding power supply paths are determined. There may be multiple such paths. Based on this, the mutual assistance contribution is calculated. The mutual assistance contribution reflects the contribution of a surplus node to a deficit node through a specific path in improving the latter's voltage quality, alleviating line congestion, etc. This contribution can be quantified using methods such as sensitivity analysis or small-disturbance simulation, and is positively correlated with indicators such as the magnitude of node voltage increase and the degree of decrease in critical line load rate. The mutual assistance contributions between all nodes are organized into a mutual assistance contribution matrix.
[0053] A graph depicting the supply and demand complementarity among nodes is constructed, with distribution network nodes as vertices. If a surplus node has an effective mutual assistance path to a deficit node and its mutual assistance contribution exceeds a set threshold, a directed edge is established between the two nodes. The weight of the edge corresponds to the magnitude of the mutual assistance contribution. This graph presents the potential power support relationships and their capacity strength within the network.
[0054] A set of operational indices is calculated to comprehensively evaluate the system's operational status. Among them, the node voltage recovery index measures the speed and extent to which node voltages recover to normal levels after a fault or disturbance. The line load balancing index is obtained by calculating the dispersion of the load rate across the entire network, reflecting the uniformity of load distribution; the closer the value is to 1, the more balanced and economically reliable the line operation. The renewable energy absorption improvement rate is obtained by comparing the total renewable energy absorption of the current candidate topology with that of the baseline topology; a positive improvement rate indicates that the current topology is more conducive to renewable energy absorption. Combining these indices constitutes the operational index set for the candidate topology, reflecting various key performance aspects of the system operation.
[0055] Step S500: After constructing a multi-objective evaluation function based on the mutual assistance contribution matrix and the operating index set, perform dual-domain optimization based on local sensitivity-global trade-off in the feasible candidate topology set, and output the mutual assistance operation control strategy.
[0056] Specifically, a comprehensive multi-objective evaluation function is constructed to mathematically integrate and quantify several core objectives of distribution network operation, such as improving voltage quality, balancing line loads, maximizing renewable energy absorption, and enhancing overall mutual support capability. These objectives are transformed into a series of sub-functions, corresponding to maximizing the overall level of node voltage recovery index, maximizing the line load balancing index, maximizing the renewable energy absorption improvement rate, and maximizing the global mutual support capability calculated based on the mutual support contribution matrix. Using methods such as the analytic hierarchy process (AHP) and considering decision-makers' preferences, appropriate weights are assigned to each sub-objective, and a unified comprehensive evaluation function is formed through weighted summation. The goal is to find the optimal solution from all feasible candidate topologies that maximizes this comprehensive evaluation value.
[0057] A dual-domain optimization strategy combining local and global perspectives is adopted. In the local domain optimization, each candidate topology is analyzed, and the impact of minor changes in branches or nodes in the network on various operational indicators is evaluated. This helps identify key sensitive points and weak links in the topology, thereby judging the stability and local adjustment potential of the topology. In the global domain optimization, the optimal solution for the comprehensive evaluation function is sought among all candidate topologies. The core of dual-domain optimization lies in feeding the results of local analysis into global decision-making. For example, a solution with a high global score may need adjustment or careful consideration in decision-making if it shows excessive sensitivity to disturbances in local sensitivity analysis, thus ensuring that the final solution is both globally optimal and has good local robustness.
[0058] Finally, after determining the target topology scheme with the best overall performance through dual-domain optimization, a specific mutual assistance operation control strategy is output, including the explicit sequence of switching operation instructions required to achieve the optimal topology, as well as the predicted values of key operating indicators under the topology and the mutual assistance contribution matrix, providing comprehensive decision support for dispatchers.
[0059] In one possible implementation, after constructing a multi-objective evaluation function based on the mutual contribution matrix and the set of operating indices, and performing dual-domain optimization based on local sensitivity and global trade-offs on the feasible candidate topology set, step S500 further includes step S510: after constructing the objective evaluation function, a cross-node index perturbation propagation tensor is constructed using the multidimensional gradient changes of the set of operating indices. This index perturbation propagation tensor is used to characterize the propagation strength and diffusion path of local branch state changes on the global supply-demand complementarity relationship in any feasible candidate topology. Specifically, some key indicators sensitive to system state changes, such as node voltage and line power flow, are selected from the set of operating indices. Under a given candidate topology, a small perturbation is applied to each key component in the network, and the changes in the values of each key operating indicator before and after these perturbations are observed. By calculating the rate of change of each indicator relative to the perturbation, the gradient of the indicator with respect to the component parameter is obtained, reflecting the magnitude of the influence of the component perturbation on a specific indicator. Based on this gradient information, a multidimensional data structure of the index perturbation propagation tensor is constructed. This tensor records the impact of applying a unit perturbation to any node or branch on various operational indicators of all other nodes in the network. By analyzing the numerically significant elements in this tensor, the potential paths for the perturbation to propagate from the source to other locations in the network can be inferred, thus systematically characterizing the propagation patterns and impact range of local perturbations across the entire network.
[0060] Step S520: Calculate the disturbance absorption capacity of the local domain using the index disturbance propagation tensor and corresponding feasible candidate topologies, generating a local adjustability margin score to characterize the controllability of local adjustments. Specifically, the distribution network is divided into several relatively independent local regions. For each local region, the disturbance absorption capacity is quantified using the index disturbance propagation tensor constructed in the previous step. Specifically, this assesses whether the main impact of disturbances originating from within the region is confined to the region itself or more easily spreads to other external regions. If the impact of the disturbance is more confined within the region, it indicates a strong ability to absorb and attenuate internal disturbances; conversely, it indicates that the disturbance is easily spread and the region's absorption capacity is weak. This capability is quantified by methods such as calculating the ratio of internal to external impacts. Based on the calculated disturbance absorption capacity, and combined with the abundance of adjustable resources within the region, a local adjustability margin score is generated for that region. This score comprehensively reflects the region's potential and ability to absorb disturbances and maintain stable operation by mobilizing its own resources when facing internal or external disturbances. The higher the score, the stronger the region's self-adjustment ability and the better its robustness.
[0061] Step S530 involves tensor-level fusion of the local adjustable margin score and the mutual assistance contribution matrix to generate a global mutual assistance control field representing the optimal power flow transfer path across the entire network. Specifically, the mutual assistance contribution matrix, characterizing the inherent mutual assistance potential between nodes, is fused with the local adjustable margin score, characterizing the adjustment capability of each node itself. This modifies and enhances the original mutual assistance contribution relationship. For example, if a node's original contribution to another node is high, but its own adjustable margin is low, making it unable to flexibly provide support, then the actual mutual assistance control capability between them will be reduced; conversely, if both nodes have high adjustable margins, the actual mutual assistance control capability will be enhanced. The resulting matrix is the global mutual assistance control field, a force field covering the entire network and indicating the direction of power flow. The field strength defines the magnitude and direction of the control force from any node to another node. By analyzing the distribution of this field, the optimal power flow transfer paths across the entire network can be identified, namely, those power support channels that can most effectively utilize the mutual assistance capabilities and local adjustment potential of each node to achieve the global optimization goal.
[0062] Step S540: Under the constraints of the global mutual support control field, cross-domain coupling optimization of the feasible candidate topology set is performed, and dual-domain optimization is carried out. Specifically, under the guidance of the newly constructed global mutual support control field, a dual-domain optimization process is performed. At the global optimization level, the global mutual support control field provides clear guidance for the search direction, prioritizing topology schemes that can strengthen or utilize the optimal mutual support path identified in the field. At the local optimization level, using the previously calculated local sensitivity information and adjustability margin score, the better schemes found by global optimization are fine-tuned, for example, adjusting key components that have a large impact on global performance and high local adjustment potential. This is an iterative process: first, a global search is performed to find potential schemes, then local fine-tuning is performed to improve their robustness and performance, and then the fine-tuned schemes are fed back into the global search space. This process is repeated until the overall performance of the found schemes no longer significantly improves, and finally, a mutual support operation control strategy that balances global optimality and local robustness under the constraints of the global mutual support control field is output.
[0063] In one possible implementation, under the constraints of the global mutual support control field, cross-domain coupling optimization of the feasible candidate topology set is performed, and dual-domain optimization is performed. Step S540 further includes step S541, which calculates the gradient of each feasible candidate topology based on the multi-objective evaluation function to form a target sensitivity vector for quantifying the direction and magnitude of the influence of each operating index on the comprehensive objective. Specifically, for each feasible candidate topology scheme, its target sensitivity vector is calculated. The specific method is to fine-tune the key decision variables in the topology one by one and observe the change in the comprehensive evaluation function value caused by each fine-tuning. The magnitude and sign of this change are the sensitivity of the decision variable to the overall objective. Arranging the sensitivities of all key decision variables in order forms the target sensitivity vector. This vector indicates which switches to adjust and how to adjust them for the current topology to most effectively improve the overall operating performance of the system.
[0064] Step S542: Project the target sensitivity vector onto the local adjustability margin score to obtain the target-oriented adjustability weights of the local domain. Specifically, assign a priority weight to each switching operation to guide the order of local adjustments. This is done by binding each switching operation to its primary affected local region, thereby obtaining the local adjustability margin score for that region. Multiply the absolute value of each switching operation's target sensitivity by the local adjustability margin score of its region to obtain an initial priority value. The priority value uses absolute values, meaning that the primary focus is on the strength of the operation's impact on the global target; the adjustment direction is recorded by the sign of the sensitivity. That is, an operation with a significant impact on the global target will have a higher priority if its region is flexible and resource-rich; conversely, even if an operation is important, its priority will be lowered if it is located in a rigid region with weak adjustment capabilities. Finally, normalize the initial priority values of all operations so that their sum is one, thus forming a local domain target-oriented adjustability weight vector. This weight vector indicates which important and feasible switching operations should be prioritized during subsequent local fine-tuning.
[0065] Step S543: Project the target sensitivity vector onto the global mutual assistance control field to obtain the global domain target-oriented power flow redistribution bias. Specifically, evaluate whether each switching operation conforms to the strategic direction indicated by the global mutual assistance control field in guiding power flow. Specifically, firstly, interpret the ideal power flow tendency of the network from the global mutual assistance control field, i.e., identify which nodes are strong power sources and which nodes are urgent power sinks. Then, for each switching operation, analyze and predict how its execution will change the power flow of all lines in the network. Compare this power flow change with the expectation of the global mutual assistance control field. If a switching operation can significantly enhance the power flow on the path from a critical power source node to a critical power sink node, then a large positive bias value is assigned; if the operation weakens the power flow on these critical paths or exacerbates unreasonable reverse power flow, then a negative bias value is assigned; operations with little impact on critical mutual assistance paths have bias values close to zero. Finally, the bias values calculated for all switching operations together constitute the global domain target-oriented power flow redistribution bias vector. This vector identifies which operations are positive gain operations that align with the overall direction of global optimization and should be prioritized in the global optimization process.
[0066] Step S544: Using the target-oriented adjustability weight and power flow redistribution bias as driving signals, a joint iterative update rule for the local and global domains is constructed to perform dynamic evolution of feasible candidate topologies and dual-domain optimization. The joint iterative update rule includes: in each iteration step, based on the target-oriented adjustability weight of the local domain, the state adjustment amount of sensitive branches is weighted and amplified or suppressed according to the weight size to form a local domain update step size; based on the power flow redistribution bias of the global domain, the supply and demand flow directions between nodes are directionally offset to construct a global domain update step size; the local domain update step size and the global domain update step size are fused according to a preset coupling coefficient to generate a comprehensive update vector for indicating the current topology adjustment direction, and feasible candidate topologies are iteratively updated and managed based on the comprehensive update vector.
[0067] Specifically, the results of the aforementioned steps are integrated to drive the dynamic optimization of the topology scheme. First, a joint iterative update rule is constructed: on the one hand, based on the target-oriented adjustability weights of the local domain, an update step size focusing on fine-tuning of local operations is generated, prioritizing switching operations with high weights that can improve the overall objective for fine-tuning; on the other hand, based on the target-oriented power flow redistribution bias of the global domain, an update step size focusing on strategic guidance is generated, prioritizing switching operations with positive and large bias values to promote the redistribution of power flow towards the ideal mode. A coupling coefficient is introduced to balance the relative importance of local and global updates. Through the coupling coefficient, these two update step sizes, originating from local and global perspectives respectively, are merged into a unified comprehensive update vector. Based on this joint rule, an iterative optimization process is executed. Starting from the initial topology, in each iteration, the various sensitivities and bias vectors of the current topology are recalculated, and an update step size is generated accordingly and applied to the topology. Then, the performance of the new topology is evaluated. This process repeats itself, enabling the topology scheme to simultaneously respond to the needs of local fine-tuning and the guidance of global strategic direction, thereby continuously evolving towards better overall performance until a convergent optimal solution is found, outputting the final mutually supportive operation control strategy.
[0068] This application's embodiments collect real-time information from multiple sources in the distribution network, integrate switch connectivity and line congestion to construct a multi-scale topology-aware state tensor, trigger topology reconfiguration requirements based on this tensor and generate multiple topology switching candidates, perform multi-scenario power flow evolution and protection coordination simulations on these candidates, screen out feasible candidate topologies, introduce a mutual assistance contribution matrix, construct a node supply and demand complementarity relationship graph, calculate the operating index set of each candidate topology, construct a multi-objective evaluation function, and output mutual assistance operation control strategies through dual-domain optimization of local sensitivity and global trade-offs in the feasible candidate topology set. This solves the technical problem of insufficient mutual assistance response accuracy in existing distribution network mutual assistance operation control under topology reconfiguration, and achieves the technical effect of improving the accuracy of distribution network mutual assistance operation regulation.
[0069] In the above text, refer to Figure 1 The method for distribution network mutual assistance operation control under topology reconfiguration according to embodiments of the present invention is described in detail. Next, reference will be made to... Figure 2 This invention describes a distribution network mutual assistance operation control system under topology reconfiguration according to an embodiment of the present invention.
[0070] The distribution network mutual assistance operation control system under topology reconfiguration according to embodiments of the present invention is used to solve the technical problem of insufficient mutual assistance response accuracy in existing distribution network mutual assistance operation control under topology reconfiguration, thereby achieving the technical effect of improving the accuracy of distribution network mutual assistance operation regulation. The distribution network mutual assistance operation control system under topology reconfiguration includes: a multi-scale topology sensing state tensor construction module 10, a topology reconfiguration demand triggering module 20, a simulation module 30, an operation index set establishment module 40, and a dual-domain optimization module 50.
[0071] The multi-scale topology-aware state tensor construction module 10 is used to acquire multi-source real-time information on node voltage, power flow distribution, line impedance, load forecast, and renewable energy output forecast of the distribution network. By integrating switch connectivity and line congestion, a multi-scale topology-aware state tensor is constructed. The topology reconfiguration demand triggering module 20 is used to trigger topology reconfiguration demand based on the multi-scale topology-aware state tensor and generate multiple sets of topology switching candidates. The simulation module 30 is used to perform multi-scenario power flow evolution and protection coordination simulation on the multiple sets of topology switching candidates and construct a set of feasible candidate topologies. The operation index set establishment module 40 is used to introduce a mutual assistance contribution matrix into the set of feasible candidate topologies, construct a supply and demand complementarity relationship diagram between nodes based on load forecast and renewable energy output forecast, calculate the node voltage recovery index, line load balance index, and renewable energy absorption improvement rate under each candidate topology, and establish an operation index set. The dual-domain optimization module 50 is used to construct a multi-objective evaluation function based on the mutual assistance contribution matrix and the operation index set, and then perform dual-domain optimization based on local sensitivity-global trade-offs on the set of feasible candidate topologies to output a mutual assistance operation control strategy.
[0072] The detailed description of the specific configuration of the dual-domain optimization module 50 is explained as follows: As mentioned above, after constructing a multi-objective evaluation function based on the mutual contribution matrix and the operating index set, the dual-domain optimization module 50 can further include: an index perturbation propagation tensor construction unit, which, after constructing the objective evaluation function, uses the multi-dimensional gradient changes of the operating index set to construct a cross-node index perturbation propagation tensor. The index perturbation propagation tensor is used to characterize the effect of local branch state changes on global supply and demand complementarity in any feasible candidate topology. The relationship propagation strength and diffusion path; the disturbance absorption capacity calculation unit is used to calculate the disturbance absorption capacity of the corresponding feasible candidate topology in the local domain using the index disturbance propagation tensor, and generate a local adjustable margin score to characterize the local adjustment controllability; the tensor-level fusion unit is used to perform tensor-level fusion of the local adjustable margin score and the mutual assistance contribution matrix to generate a global mutual assistance control field to represent the optimal power flow transfer path of the entire network; the cross-domain coupling optimization unit is used to perform cross-domain coupling optimization of the feasible candidate topology set under the constraints of the global mutual assistance control field, and perform dual-domain optimization.
[0073] Specifically, under the constraints of the global mutual assistance control field, cross-domain coupling optimization of the feasible candidate topology set is performed, and dual-domain optimization is carried out. The cross-domain coupling optimization unit may further include: a gradient calculation subunit for performing gradient calculation on each feasible candidate topology based on the multi-objective evaluation function to form a target sensitivity vector for quantifying the direction and magnitude of the influence of each operating index on the comprehensive objective; a first projection subunit for the target sensitivity vector for projecting the target sensitivity vector onto the local adjustable margin score to obtain the target orientation adjustable weight in the local domain; a second projection subunit for projecting the target sensitivity vector onto the global mutual assistance control field to obtain the global domain target orientation power flow redistribution bias; and a dynamic evolution subunit for using the target orientation adjustable weight and power flow redistribution bias as driving signals, and performing dynamic evolution of the feasible candidate topology by constructing a joint iterative update rule for the local and global domains, thereby performing dual-domain optimization.
[0074] The dynamic evolution subunit may further include: the joint iterative update rule includes: in each iteration step, based on the target-oriented adjustability weight of the local domain, the state adjustment amount of sensitive branches is weighted and amplified or suppressed according to the weight size to form a local domain update step size; based on the power flow redistribution bias of the global domain, the supply and demand flow directions between nodes are directionally offset to construct a global domain update step size; the local domain update step size and the global domain update step size are fused according to a preset coupling coefficient to generate a comprehensive update vector for indicating the current topology adjustment direction, and feasible candidate topology iterative update management is performed based on the comprehensive update vector.
[0075] The detailed description of the specific configuration of the topology reconfiguration requirement triggering module 20 is as follows: As mentioned above, the topology reconfiguration requirement triggering module 20 can further include: a multi-dimensional change feature extraction unit for extracting multi-dimensional change features of node voltage offset rate, power flow direction change rate and line congestion degree from the multi-scale topology sensing state tensor, and constructing a topology disturbance sensitive surface for characterizing topology stability; and a topology risk integral calculation unit for calculating the cross-scale topology risk integral based on the topology disturbance sensitive surface, and triggering the topology reconfiguration requirement based on the topology risk integral.
[0076] The topology risk integral calculation unit, which triggers the topology reconstruction requirement based on the topology risk integral, may further include: a judgment subunit for judging whether the topology risk integral meets the adaptive reconstruction threshold; and a topology reconstruction requirement triggering subunit for identifying the maximum contributing connected path corresponding to the topology risk integral and triggering the topology reconstruction requirement if the topology risk integral meets the adaptive reconstruction threshold.
[0077] The specific configuration of the simulation module 30 is described in detail below: As mentioned above, the simulation module 30 performs multi-scenario power flow evolution and protection coordination simulation on the multiple sets of topology switching candidates to construct a feasible candidate topology set. The simulation module 30 may further include: a scenario vector group construction unit for generating a multi-dimensional simulation scenario space based on load prediction, renewable output prediction, and fault disturbance mode, and constructing a scenario vector group, which is used to drive the dynamic evolution of power flow of each topology switching candidate under different operating states; a serialized power flow solution unit for performing serialized power flow solution on the topology switching candidates in each scenario, and constructing a power flow evolution sequence tensor based on the time gradient of power flow changes; a protection action consistency verification unit for triggering protection action consistency verification using the power flow evolution sequence tensor and outputting a protection coordination consistency score; and a multiple topology switching candidate screening unit for screening multiple topology switching candidates based on the protection coordination consistency score to construct a feasible candidate topology set.
[0078] The process of selecting multiple topology switching candidates based on the protection coordination consistency score can be further divided into: a topology feasibility determination tensor construction subunit for fusing the protection coordination consistency score with the power flow evolution sequence tensor to construct a topology feasibility determination tensor; and a multiple topology switching candidate selection execution subunit for quantifying the steady-state feasibility, transient feasibility, and protection coordination reliability of topology switching under multiple scenarios based on the topology feasibility determination tensor, and executing the selection of multiple topology switching candidates.
[0079] The protection action consistency verification unit may further include: consistency verification includes comparing the action criteria of relay protection actions with the time synchronization relationship of power flow evolution trajectory, identifying whether there are over-level actions, false actions or failures to act, and constructing a protection coordination consistency score.
[0080] The distribution network mutual assistance operation control system under topology reconfiguration provided in the embodiments of the present invention can execute the distribution network mutual assistance operation control method under topology reconfiguration provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0081] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.
[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A method for mutual assistance operation control of distribution networks under topology reconfiguration, characterized in that, The method includes: The system acquires multi-source real-time information on node voltage, power flow distribution, line impedance, load forecasting, and renewable energy output forecasting of the distribution network. By integrating switch connectivity and line congestion, it constructs a multi-scale topology-aware state tensor. Based on the multi-scale topology-aware state tensor, triggering the topology reconstruction requirement, multiple sets of topology switching candidates are generated. Multi-scenario power flow evolution and protection coordination simulations are performed on the multiple sets of topology switching candidates to construct a feasible candidate topology set; A mutual assistance contribution matrix is introduced into the feasible candidate topology set. Based on load forecasting and renewable energy output forecasting, a supply and demand complementarity relationship diagram between nodes is constructed. The node voltage recovery index, line load balance index and renewable energy absorption improvement rate under each candidate topology are calculated to establish an operation index set. After constructing a multi-objective evaluation function based on the mutual assistance contribution matrix and the set of operating indices, a dual-domain optimization based on local sensitivity-global trade-off is performed on the feasible candidate topology set to output a mutual assistance operation control strategy.
2. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 1, characterized in that, After constructing a multi-objective evaluation function based on the mutual contribution matrix and the set of operating indices, the optimization in the feasible candidate topology set is performed in a dual-domain manner based on local sensitivity-global tradeoffs, including: After constructing the target evaluation function, the multidimensional gradient changes of the running index set are used to construct the cross-node index perturbation propagation tensor. The index perturbation propagation tensor is used to characterize the propagation strength and diffusion path of local branch state changes on the global supply and demand complementarity relationship in any feasible candidate topology. The perturbation propagation tensor of the index is used to perform the calculation of the perturbation absorption capacity of the local domain in the corresponding feasible candidate topology, and a local adjustability margin score is generated to characterize the local adjustability controllability. The local adjustable margin score and the mutual assistance contribution matrix are fused in tensor mode to generate a global mutual assistance control field that represents the optimal power flow transfer path of the entire network. Under the constraints of the global mutual regulation field, cross-domain coupling optimization of feasible candidate topologies is performed, and dual-domain optimization is carried out.
3. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 2, characterized in that, Under the constraints of the global mutual regulation field, cross-domain coupled optimization of the feasible candidate topology set is performed, and dual-domain optimization is performed, including: Based on the multi-objective evaluation function, gradient calculation is performed on each feasible candidate topology to form a target sensitivity vector used to quantify the direction and magnitude of the influence of each operational indicator on the comprehensive target. The target sensitivity vector is projected onto the local adjustability margin score to obtain the target orientation adjustability weight in the local domain. Project the target sensitivity vector onto the global mutual assistance control field to obtain the global domain target-oriented power flow redistribution bias; Using the target-oriented adjustability weight and power flow redistribution bias as driving signals, the dynamic evolution of feasible candidate topologies is executed by constructing a joint iterative update rule for the local and global domains, and dual-domain optimization is performed.
4. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 3, characterized in that, The joint iterative update rules include: In each iteration step, based on the target-oriented adjustable weights of the local domain, the state adjustment amount of the sensitive branch is weighted and amplified or suppressed according to the weight size to form the local domain update step size. Based on the power flow redistribution bias of the global domain, the supply and demand flow between nodes is directionally offset to construct the global domain update step size; The local domain update step size and the global domain update step size are fused according to a preset coupling coefficient to generate a comprehensive update vector that indicates the current topology adjustment direction. Feasible candidate topology iterative update management is then performed based on the comprehensive update vector.
5. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 1, characterized in that, The topology reconstruction requirement triggered by the multi-scale topology-aware state tensor includes: Multidimensional variation features of node voltage offset rate, power flow direction change rate and line congestion degree are extracted from the multi-scale topology sensing state tensor to construct a topology disturbance sensitive surface for characterizing topology stability. Calculate the cross-scale topological risk integral based on the topological perturbation-sensitive surface, and trigger the topological reconstruction requirement based on the topological risk integral.
6. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 5, characterized in that, The topology reconfiguration requirement is triggered based on the topology risk score, including: Determine whether the topology risk integral meets the adaptive reconstruction threshold; If the topology risk integral meets the adaptive reconstruction threshold, the maximum contributing connected path corresponding to the topology risk integral is identified, triggering the topology reconstruction requirement.
7. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 1, characterized in that, Multi-scenario power flow evolution and protection coordination simulations are performed on the multiple sets of topology switching candidates to construct a feasible candidate topology set, including: A multi-dimensional simulation scenario space is generated based on load forecasting, renewable power output forecasting, and fault disturbance modes. A scenario vector group is constructed, which is used to drive the dynamic evolution of power flow of each topology switching candidate under different operating states. In each scenario, the topology candidate switching candidate is serialized and the power flow solution is calculated. The power flow evolution sequence tensor is constructed based on the temporal gradient of the power flow change. The protection action consistency check is triggered using the power flow evolution sequence tensor, and a protection coordination consistency score is output. Based on the protection coordination consistency score, multiple sets of topology switching candidates are screened to construct a feasible candidate topology set.
8. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 7, characterized in that, Based on the protection coordination consistency score, multiple sets of topology switching candidates are screened, including: The protection coordination consistency score is fused with the power flow evolution sequence tensor to construct a topology feasibility determination tensor; Based on the topology feasibility determination tensor quantization, the steady-state feasibility, transient feasibility, and protection coordination reliability of topology switching under multiple scenarios are evaluated, and multiple sets of topology switching candidate screening are performed.
9. The distribution network mutual assistance operation control method under topology reconfiguration as described in claim 7, characterized in that, Consistency verification includes comparing the action criteria of relay protection actions with the time synchronization relationship of power flow evolution trajectory, identifying whether there are overstepping actions, false actions, or failures to act, and constructing a protection coordination consistency score.
10. A distribution network mutual assistance operation control system under topology reconfiguration, characterized in that, The system is used to implement the distribution network mutual assistance operation control method under topology reconfiguration as described in any one of claims 1-9, and the system includes: The multi-scale topology-aware state tensor construction module is used to acquire multi-source real-time information on node voltage, power flow distribution, line impedance, load forecasting, and renewable energy output forecasting of the distribution network. By integrating switch connectivity and line congestion, a multi-scale topology-aware state tensor is constructed. The topology reconstruction requirement triggering module is used to trigger topology reconstruction requirements based on the multi-scale topology-aware state tensor and generate multiple sets of topology switching candidates. The simulation module is used to perform multi-scenario power flow evolution and protection coordination simulation on the multiple sets of topology switching candidates, and to construct a feasible candidate topology set; The operation index set establishment module is used to introduce a mutual assistance contribution matrix into the feasible candidate topology set, construct a supply and demand complementarity relationship diagram between nodes based on load forecasting and renewable energy output forecasting, calculate the node voltage recovery index, line load balance index and renewable energy absorption improvement rate under each candidate topology, and establish the operation index set. The dual-domain optimization module is used to construct a multi-objective evaluation function based on the mutual assistance contribution matrix and the operating index set, and then perform dual-domain optimization based on local sensitivity-global trade-off in the feasible candidate topology set to output a mutual assistance operation control strategy.
