A power grid overload control method and system considering backup power supply device

Abstract

The present application belongs to the technical field of power system operation control, and relates to a power grid overload control method and system considering backup power automatic switching, which comprises the following steps: obtaining dispatching control system operation data and constructing a fault set; performing power flow calculation on the fault set to determine an overload element set and calculate a power transmission distribution factor; establishing a backup power automatic switching action library to generate a candidate action set, calculating candidate action influence degree and newly added out-of-limit risk indicators and sorting to form a backup power automatic switching set; solving a load shedding strategy with minimum power loss output of a correction control model when the state code indicates that there is still an overload element; generating a switch breaker action list according to the backup power automatic switching set and the load shedding strategy, recalculating and evaluating the overload element set after writing back the operation data. The technical scheme of the present application is suitable for fault overload disposal scenarios, can realize consistent decision-making and closed-loop checking of load transfer and load shedding, and keeps the out-of-limit elimination efficiency and power loss controllable.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, and specifically relates to a power grid overload control method and system that takes into account automatic transfer switching. Background Technology

[0002] In existing technologies, power grid overload control typically relies on the coordinated operation of dispatch control systems and automatic safety devices. This involves verifying real-time operating data, identifying overloaded components, and triggering pre-set handling logic to maintain the safety margin of lines and main transformers under typical fault conditions and prevent the escalation of cascading risks. However, existing overload handling procedures have some significant shortcomings in the consistency of load transfer measures and load shedding.

[0003] In practical applications, overload handling often relies on fixed-cycle switching logic or manually confirmed handling sequence. The order of load transfer and load shedding, the scope of action, and the constraint boundaries depend on experience. This can easily lead to a weak match between the handling action and the actual power flow response of the system, resulting in unstable overall performance in terms of overload elimination speed and power outage impact control, and may also lead to unnecessary expansion of the power outage range.

[0004] Therefore, it is evident that existing technologies often suffer from problems such as weak overall consistency in the coordinated control of load transfer and load shedding during fault overload handling, and limited control over the impact of power outages. These are the shortcomings of existing technologies.

[0005] In view of this, it is very necessary to provide a power grid overload control method and system that takes into account automatic switching of backup power, so as to solve the above-mentioned defects in the prior art. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of the prior art, namely, the weak overall consistency of load transfer and load shedding coordination control in fault overload handling and the limited control level of power outage impact, by providing a power grid overload control method and system that takes into account automatic backup switching, so as to solve the above-mentioned technical problems.

[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, this application provides a power grid overload control method that takes into account automatic backup switching, comprising: Real-time power grid operation data is obtained from the real-time operation data source of the power grid dispatch and control system, and a fault set is constructed. Power flow calculations are performed on each fault set to determine the set of overloaded components and to calculate the power transmission distribution factor. Establish a backup automatic transfer action library and generate a candidate action set. Calculate the impact of the candidate actions on the set of overloaded components using the power transmission distribution factor and construct a new over-limit risk index. Sort the candidate actions according to their impact and the new over-limit risk index. Verify the candidate actions in turn based on the comprehensive ranking results and incorporate them into the backup automatic transfer set. Obtain the power grid topology after the backup automatic transfer set is activated and output the status code. When the status code indicates that there are still overloaded components, the grid correction control model is solved based on the grid topology after the automatic transfer switch set action, and the load shedding strategy with the minimum power outage loss is output. Based on the automatic transfer actions and load shedding strategies in the automatic transfer set, a list of switch and circuit breaker actions is generated, and the overload component set is re-evaluated after writing back the real-time power grid operation data.

[0008] By adopting the above technical solution, the automatic transfer action evaluation mechanism is linked with the correction control model that minimizes power outage losses and forms a closed-loop recalculation. This achieves consistency in the timing and scope of load transfer and load shedding during overload handling. It can ensure the efficiency of over-limit elimination while keeping the impact of power outages controllable, thus meeting the requirements for stronger overall consistency of collaborative control and a higher level of control over the impact of power outages during fault overload handling.

[0009] Specifically, firstly, a set of typical fault conditions is formed based on online measurements, topology, and operating mode information from the dispatch control system and kept synchronized with the real-time status, ensuring that subsequent handling is carried out within the current grid state. Then, power flow verification is performed on each condition to identify the range of exceedances. Simultaneously, a power transmission distribution factor is introduced to quantify the correlation between key injection changes and overload response, providing a calculable sensitivity basis for optimal action selection. Based on this, a backup automatic transfer action library is constructed and a candidate action set is generated. The contribution of candidate actions to overload reduction and the newly added exceedance risk indicators are combined to form a... The inclusion rules are sortable, iterative, and terminateable, ensuring that load transfer prioritizes its role within risk-controlled boundaries. When transfer resources are exhausted and over-limits still exist, the load shedding is further refined with the goal of minimizing power outage losses, ensuring that the shedding location and scale match system safety constraints and are consistent with previous transfer results. Finally, the backup automatic transfer set and load shedding strategy are mapped to an executable list of switch and circuit breaker actions, and the updated operating data is written back for recalculation and verification, ensuring that the handling results are consistent with the power flow response and avoiding unnecessary expansion of the power outage range due to action and response mismatch.

[0010] Preferably, the steps of acquiring real-time power grid operation data and constructing a fault set include: Acquire power grid topology data, switch location information, and measurement data, perform timestamp consistency verification, and output data quality identifiers; A fault set is generated based on power grid topology data. The fault set includes line fault items and fault items of parallel lines on the same tower. Within the fault set, bind a corresponding set of out-of-limit criteria to each fault item.

[0011] By adopting the above technical solution, relying on timestamp consistency verification and data quality identification, real-time operating data has clear usable boundaries and reduces the probability of misjudgment. It achieves unified management of fault scenario coverage and limit-breaking criteria binding, which can enhance the accuracy and traceability of overload identification.

[0012] Preferably, the step of calculating the power transfer distribution factor includes: A DC power flow linearization model is established based on the grid topology data and branch parameters verified by timestamp consistency, and a reference bus is selected. The branch parameters include line parameters and transformer parameters. Construct a sensitivity mapping from changes in bus active power injection to changes in branch active power flow and solidify it as a power transmission distribution factor; For the branches corresponding to the overload component set, sub-maps are extracted from the power transmission distribution factor and used for candidate action evaluation. The branches include line branches and transformer branches.

[0013] By adopting the above technical solution, the response relationship between key injection and branch power flow is solidified based on the sensitivity mapping of DC power flow linearization, so that the candidate action evaluation has a unified standard and a fast calculation basis, which can reduce the overhead of repeated calculations and improve the stability of the ranking results.

[0014] Preferably, the step of calculating the impact of candidate actions on the set of overloaded components using the power transfer distribution factor includes: The candidate action is equivalent to a combination of a decrease in active power injection at the source bus and an increase in active power injection at the receiving bus, forming an equivalent injection change. The contribution of the equivalent injection change to the power flow of each element in the overloaded element set is calculated based on the power transmission distribution factor and then synthesized to obtain the power flow change. The influence degree is obtained by correlating the power flow change with the overload margin of the overload component set.

[0015] By adopting the above technical solution, the power migration direction of the transfer action is characterized by equivalent injection change and the power flow change is synthesized, so that the influence degree and the over-limit margin can be calculably correlated, and the contribution to load reduction can be quantitatively measured, which can improve the pertinence and interpretability of action selection.

[0016] As a preferred approach, the steps of constructing new limit-crossing risk indicators and comprehensively ranking candidate actions based on their impact and the new limit-crossing risk indicators include: A new limit violation risk index is constructed based on the change in component margin corresponding to the limit violation criterion set caused by candidate actions, and the impact degree and the new limit violation risk index are normalized. The normalized impact and the normalized new risk index for exceeding the limit are linearly combined according to preset weights to obtain a comprehensive score. The candidate actions are then ranked according to the comprehensive score to determine the comprehensive ranking result.

[0017] By adopting the above technical solution and combining it with the normalization trade-off mechanism of new limit violation risk and impact, the ranking can take into account both the limit violation elimination effect and the boundary of associated risks, so as to achieve the comparability and scale independence of the comprehensive score, reduce the probability of new limit violation triggering and enhance decision robustness.

[0018] As a preferred option, the steps of sequentially verifying candidate actions based on the comprehensive ranking results and incorporating them into a backup automatic transfer set, obtaining the power grid topology after the backup automatic transfer set is activated, and outputting the status code include: Candidate actions are selected sequentially based on the comprehensive ranking results and a rapid verification is performed. The rapid verification includes consistency verification of the load reduction direction of the overloaded component set and threshold verification of newly added over-limit risk indicators. For candidate actions that pass the rapid verification, perform precise verification and incorporate them into the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set is activated. The precise verification adopts the over-limit criterion set with the same caliber as the power flow calculation. When a new over-limit occurs or the power flow does not converge in the precise verification, the most recently incorporated candidate action is revoked and the candidate action is marked as disabled. At the same time, the status code is output and the iteration continues until the termination condition is met.

[0019] By adopting the above technical solution, a two-level verification of rapid verification and precise verification is used, and abnormal actions are marked for cancellation and disabling. This ensures that the iterative inclusion process maintains convergent and controllable boundaries, avoids the spread of inconsistent load reduction directions and non-convergence states, and improves online processing efficiency.

[0020] Preferably, when the status code indicates that overloaded components still exist, the steps of solving the grid correction control model based on the grid topology after the backup automatic transfer set operation and outputting the load shedding strategy with the minimum power outage loss include: An optimization model is constructed when the output status code indicates that there are still overloaded components. The optimization model includes a power outage loss function based on load importance stratification and load shedding ratio. Establish power balance constraints, branch thermal stability constraints, voltage constraints, and topology constraints, and use the grid topology after the automatic transfer switch set is activated as the input to the optimization model; When the optimization model is not feasible, a constraint relaxation is introduced and a penalty term is added to the power outage loss function. The load shedding strategy with the minimum default is output and the status code is updated.

[0021] By adopting the above technical solution, the location and proportion of load shedding are solved by combining the power outage loss function and safety constraints. When it is not feasible, a relaxation penalty is introduced to obtain the minimum default result, so that the strategy remains executable within the safety boundary and can minimize the impact of power outages in scenarios where over-limits still exist.

[0022] Preferably, the steps of generating a list of circuit breaker actions and re-evaluating the overload component set after writing back the real-time power grid operating data include: The actions of the switch and circuit breaker corresponding to the automatic transfer set and the actions of the switch and circuit breaker corresponding to the load shedding strategy are uniformly mapped into a structured action list. The structured action list includes object identifier, action type, action sequence and verification identifier. Based on the structured action list, write-back information is generated and the real-time power grid operation data is updated. Based on the updated real-time power grid operation data, power flow calculation is performed again and the overload component set and status code are updated.

[0023] By adopting the above technical solution, relying on the object identification and sequence verification of the structured action list, the transfer and load shedding actions are unified into an executable sequence and written back for recalculation and verification, so that the processing results form a closed-loop verification and status synchronization, which can improve the consistency of execution and reduce the risk of misoperation.

[0024] Secondly, this application also provides a power grid overload control system that takes into account automatic backup switching, comprising: The data construction unit is used to obtain real-time power grid operation data from the real-time operation data source of the power grid dispatch and control system and construct a fault set. The power flow assessment unit is used to perform power flow calculations on each fault set, determine the set of overloaded components, and calculate the power transmission distribution factor. The action screening unit is used to establish a backup automatic transfer action library and generate a candidate action set. It uses the power transmission distribution factor to calculate the impact of the candidate actions on the set of overloaded components and constructs a new over-limit risk index. The candidate actions are comprehensively sorted according to the impact and the new over-limit risk index. Based on the comprehensive sorting results, the candidate actions are checked in turn and incorporated into the backup automatic transfer set. The grid topology after the backup automatic transfer set is activated is obtained and the status code is output. The correction control unit, when the status code indicates that there are still overloaded components, solves the grid correction control model based on the grid topology after the automatic transfer switch set action and outputs the load shedding strategy with the minimum power outage loss. The action closed-loop unit is used to generate a list of switch and circuit breaker actions based on the automatic transfer actions and load shedding strategies in the automatic transfer set, and re-evaluate the overload component set after writing back the real-time operation data of the power grid.

[0025] Preferably, the action filtering unit includes: The fast verification subunit is used to select candidate actions in sequence according to the comprehensive sorting results and perform fast verification. Fast verification includes consistency verification of the load reduction direction of the overload component set and threshold verification of newly added over-limit risk indicators. The fine-check sub-unit is used to perform precise checks on candidate actions that have passed the fast check and incorporate them into the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set has been activated. The precise check adopts the over-limit criterion set with the same caliber as the power flow calculation. The rollback sub-unit cancels the most recently included candidate action and marks it as disabled when a new limit violation or power flow non-convergence occurs during the precise verification. The state sub-unit is used to output the state code and drive the action filtering unit to continue iterating until the termination condition is met.

[0026] As can be seen from the above technical solutions, the present invention has the following advantages: This application provides a power grid overload control method and system that takes into account automatic transfer switching. By linking the automatic transfer switching action evaluation mechanism with the correction control model that minimizes power outage losses and forming a closed-loop recalculation, the timing and scope of load transfer and load shedding in overload handling are consistent. This ensures the efficiency of overload elimination while keeping the impact of power outages controllable, and meets the requirements of stronger overall consistency of collaborative control and higher level of control over the impact of power outages in fault overload handling. Attached Figure Description

[0027] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a flowchart of a power grid overload control method that takes into account automatic backup switching, provided by the present invention; Figure 2 This is a schematic diagram of a power grid overload control system that includes automatic switching on standby, provided by the present invention.

[0029] The system comprises: 1. Data construction unit; 2. Trend evaluation unit; 3. Action screening unit; 4. Correction and control unit; and 5. Action closed-loop unit. Detailed Implementation

[0030] Various embodiments of this disclosure are described more fully below with reference to the accompanying drawings. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0031] In the following, the terms “comprising” or “may include”, which may be used in various embodiments of this disclosure, indicate the presence of the disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. Furthermore, as used in various embodiments of this disclosure, the terms “comprising,” “having,” and their cognates are intended only to indicate a particular feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more combinations of the foregoing.

[0032] It should be noted in advance that, in order to facilitate a clear and accurate description of the technical solutions in the embodiments of this application, the following is a brief explanation of some terms and related technologies involved in the embodiments of this application: 1. Automatic Transfer Switching: This usually refers to the automatic switching of backup power. When the main power supply fails, is undervoltage, or is disconnected, the automatic control function automatically disconnects the failed power supply and switches to the backup power supply according to the protection and interlocking logic. It often includes elements such as voltage and frequency verification, synchronous or asynchronous switching criteria, interlocking conditions and action time limits, which are used to ensure the continuity of power supply to important loads and the safety and certainty of the handling process.

[0033] 2. Power Transfer Distribution Factor (PTDF): Used to characterize the incremental distribution relationship of active power flow in each branch when the injected power from the bus is transferred between the source and receiver under a given network and operating mode. Essentially, it is a sensitivity coefficient based on the linearized power flow assumption and is often used for congestion identification, limit verification, and rapid assessment of the impact of control actions on branch over-limits.

[0034] 3. DC power flow linearization model: A linear approximation model that simplifies the AC power flow equation. It usually uses active power and phase angle as the main variables and ignores the effects of resistance and reactive power. Under the assumption that the voltage amplitude is approximately constant and the phase angle difference is small, the power flow calculation is transformed into a linear solution, thereby supporting rapid verification, sensitivity analysis and batch scenario evaluation in large-scale networks.

[0035] To address the issues that load transfer and shedding in fault overload handling often rely on fixed-round switching logic or manual confirmation sequence, resulting in weak coordination consistency, insufficient matching between handling actions and power flow response, unstable overall performance of limit elimination and power outage impact control, and the potential for unnecessary expansion of the power outage range, it is difficult to meet the actual needs of controllable constraints on the impact of power outages while ensuring the safety margin of lines and main transformers under typical fault conditions. This application discloses a power grid overload control method and system that takes into account automatic backup switching. By introducing a fault set-driven verification mechanism based on real-time operational data, an action impact quantification mechanism supported by power transmission distribution factors, and an iterative optimization mechanism for backup automatic transfer (SALT) to address new over-limit risks, and by connecting the grid correction control model with the minimum outage loss and the closed-loop evaluation mechanism of write-back recalculation after the SALT set terminates, collaborative decision-making and executable action list generation under a unified boundary can be achieved for SALT and load shedding. This ensures that the sequence and scope of actions during the over-limit elimination process remain consistent with the actual power flow response of the system. As a result, while ensuring the efficiency of overload handling, the impact of power outages is kept at a controllable level, further reducing unnecessary expansion of the outage scope and improving the overall consistency of fault overload handling and operational safety assurance capabilities.

[0036] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments 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.

[0037] like Figure 1 As shown, this embodiment provides a power grid overload control method that takes into account automatic backup switching, including: Step S1: Obtain real-time power grid operation data from the real-time operation data source of the power grid dispatch and control system and construct a fault set; Step S2: Perform power flow calculations on each fault set to determine the set of overloaded components and calculate the power transmission distribution factor; Step S3: Establish a backup automatic transfer action library and generate a candidate action set. Calculate the impact of the candidate actions on the set of overloaded components using the power transmission distribution factor and construct a new over-limit risk index. Sort the candidate actions according to their impact and the new over-limit risk index. Verify the candidate actions in turn based on the comprehensive sorting results and incorporate them into the backup automatic transfer set. Obtain the power grid topology after the backup automatic transfer set is activated and output the status code. Step S4: When the status code indicates that there are still overloaded components, solve the grid correction control model based on the grid topology after the automatic transfer switch set action and output the load shedding strategy with the minimum power outage loss; Step S5: Generate a list of switch and circuit breaker actions based on the automatic transfer actions and load shedding strategies in the automatic transfer set, and re-evaluate the overload component set after writing back the real-time power grid operation data.

[0038] This embodiment employs online modeling based on real-time operational data from the power grid dispatch and control system, along with a fault set-driven power flow verification method. This ensures that the overload identification and handling boundaries are updated synchronously with the operating mode, maintaining consistency and traceability. By introducing a power transmission distribution factor, the power flow redistribution effect caused by candidate operations is quantified. Furthermore, by combining newly added over-limit risk indicators, candidate backup automatic transfer actions are ranked and iteratively incorporated. This ensures that load transfer actions, under controlled risk conditions, prioritize effective reduction contributions to overload components and maintain the consistency of the action sequence. Finally, by introducing a method to minimize power outage losses when the backup automatic transfer set terminates and over-limit conditions still exist... The target-oriented power grid correction control solution mechanism coordinates the scope and scale of load shedding under safety constraints while taking into account the controllability of outage impacts. By mapping the standby automatic transfer set and load shedding strategy into an executable list of switch and circuit breaker actions and performing write-back and recalculation verification, the handling actions are consistent with the actual power flow response of the system and the probability of unnecessary expansion of the outage scope is reduced. Overall, in the process of handling fault overload, the mechanism achieves coordinated and consistent decision-making and closed-loop evaluation of load transfer and load shedding, improves the stability of over-limit elimination and the ability to maintain the safety margin of lines and main transformers, and meets the operational requirements of higher control levels of outage impacts.

[0039] The above steps will be specifically described below based on the embodiments of this application.

[0040] In step S1, the core task is to complete the reliable acquisition and consistency verification of topology and measurement within the same online section, and on this basis, form a fault set covering N-1 and faults of parallel lines on the same tower. At the same time, configure a set of over-limit judgment criteria with a unified caliber for each fault item, so that subsequent power flow calculation and action verification always use the same data caliber and judgment boundary.

[0041] In this embodiment, real-time power grid operation data needs to be obtained from the real-time operation data source of the power grid dispatch and control system. It should be noted that the real-time power grid operation data can come from two channels: periodic sampling and event triggering. Furthermore, to avoid interference from the transient process after fault clearing on consistency verification, in this embodiment, data sampling is limited to the steady-state window after fault clearing, which can be 5.0s to 10.0s. For missing branch active power flow and branch reactive power flow, when the data quality identifier is still available, a minimum change interpolation method under power balance constraints can be used to fill in the missing measurements, i.e., to make minimal corrections to the missing measurements without changing the bus power balance; this interpolation is only used to form continuous input and is not used to change the judgment boundary of the over-limit criterion set.

[0042] Specifically, the system can acquire power grid topology data, switch location information, and measurement data. Power grid topology data describes the connection relationship between buses and branches; switch location information determines the connectivity status of cross-sections; and measurement data includes bus voltage amplitude, branch active power flow, and branch reactive power flow. Subsequently, a timestamp consistency check is performed on the three types of data. Using a unified sampling time as a benchmark, data with a time deviation exceeding a preset consistency threshold is marked as unusable, and a data quality identifier is output after the check. For example, the preset consistency threshold for measurement data is 1.0s, and the preset consistency threshold for switch location information is 2.0s. When the proportion of missing key measurements exceeds 5% or the difference between repeated uploads of the same bus voltage amplitude exceeds 0.02 pu, the data quality identifier is set to unusable, and that cross-section is not included in subsequent calculations.

[0043] In some embodiments of this application, the fault set construction stage is entered when the data quality is identified as available. In these embodiments, a combination of branch enumeration and parallel-connection enumeration on the same tower is used to generate a fault set based on power grid topology data. This fault set includes line fault items and parallel-connection line fault items on the same tower. Line fault items are enumerated by exiting each line branch individually, while parallel-connection line fault items are enumerated by exiting the entire set of parallel-connection branches on the same tower simultaneously. During enumeration, the switch and circuit breaker status is updated synchronously to obtain the post-fault topology. Based on this, a verification boundary is configured for each fault item, and a corresponding over-limit criterion set is bound to it. This over-limit criterion set includes branch thermal stability over-limit criteria and voltage over-limit criteria, and includes the thermal stability over-limit criteria for transformer branches when transformer branches exist at the cross-section. For example, the branch thermal stability over-limit criterion is based on the branch's apparent power not exceeding its thermal stability capacity, and the voltage over-limit criterion is based on the bus voltage amplitude not exceeding the upper limit and not lower than the lower limit.

[0044] It should be further explained that during fault enumeration, the same object identification system is used for line fault items and fault items of lines on the same tower. The object identification includes voltage level, substation name, bay number, and switch / circuit breaker number to ensure that the action list generation stage can be unambiguously mapped to the target switch / circuit breaker object. Meanwhile, the set of over-limit criteria in this embodiment is expressed as a triple of "criteria type-threshold-duration". The criteria type includes branch thermal stability over-limit and bus voltage over-limit, the threshold includes branch thermal stability capacity and upper and lower limits of bus voltage amplitude, and the duration is used to distinguish between transient and steady-state over-limit. For example, the steady-state over-limit duration threshold is 3.0s.

[0045] At this point, step S1 has completed the acquisition and consistency verification of online cross-section data, and formed a fault set covering line faults and faults of parallel lines on the same tower, as well as a set of over-limit judgment criteria. This constitutes a data preparation system that uses data quality identification to control the entry point, fault enumeration to limit the evaluation boundary, and the judgment set to unify the verification caliber, providing a reliable input basis for subsequent fault-by-fault power flow calculation and action evaluation.

[0046] In step S2, the core task is to solve the power flow after the fault for each fault condition in the fault set and identify the set of overloaded components. At the same time, a power transmission distribution factor is established for quickly assessing the impact of actions, so that candidate actions can obtain the branch active power flow change without repeatedly solving the AC power flow and support sorting and rapid verification.

[0047] In this embodiment, power flow calculation is required for each fault item in the fault set. The power flow calculation uses AC power flow solver to output the bus voltage amplitude and branch power flow. Under the over-limit criterion set, the over-limit objects are identified, thereby determining the overload element set. To facilitate subsequent evaluation, the overload element set is distinguished between line branches and transformer branches according to the object identifier, and the current active power flow, current reactive power flow, thermal stability capacity, and over-limit margin of each overload element are stored.

[0048] Specifically, the determination of the overload component set is based on the over-limit criterion set, where the branch thermal stability over-limit is determined by the criterion that the branch apparent power does not exceed the branch thermal stability capacity. This can be achieved using:

[0049] Among them, the branch road has meritorious trend Indicates a branch The main current is meritorious, while the secondary current is ineffective. Indicates a branch Reactive power flow; branch thermal stability capacity Indicates a branch The allowable apparent power limit is set. When the apparent power of a branch exceeds its thermal stability capacity, the branch is included in the overload component set, and its over-limit margin is calculated as the denominator for subsequent impact calculations. Bus voltage over-limit is determined using a dual criterion of the upper and lower limits of bus voltage amplitude, and over-limit objects are also included in the constrained component set during the risk index construction phase.

[0050] Furthermore, when the set of overloaded components is not empty, the power transmission distribution factor is calculated to support rapid action assessment. The modeling process first establishes a DC power flow linearization model based on grid topology data and branch parameters verified by timestamp consistency, then selects the reference bus, where branch parameters include line parameters and transformer parameters. For example, the DC power flow linearization model uses the active power injection vector from the bus. The represented bus active power injection and bus phase angle vector The bus voltage phase angle relationship can be written as:

[0051] Branch active power flow vector The represented branch active power flow and bus phase angle vector The bus voltage phase angle relationship can be written as:

[0052] Wherein, the node admittance matrix The branch-node admittance matrix is ​​constructed from the equivalent reactance between buses. It is constructed from branch reactance and branch origin and destination busbars. The busbar phase angle vector is eliminated by the above two equations. We can obtain:

[0053] Based on this, a sensitivity mapping from changes in bus active power injection to changes in branch active power flow can be constructed, and this sensitivity mapping can be solidified into a power transmission distribution factor matrix. It can be written as:

[0054] Among them, the power transfer distribution factor matrix elements Indicates busbar Changes in active power injection on branch The sensitivity to changes in active power flow is within a certain range. .

[0055] In some embodiments of this application, during the solution of the DC power flow linearization model, the selection of a reference bus is used to fix the bus phase angle reference and eliminate matrix singularities. For example, the reference bus is selected as the bus corresponding to the bus object identifier with the largest generating capacity. Meanwhile, to ensure the numerical stability of the matrix solution, the nodal admittance matrix is ​​solved... When performing the inverse mapping, sparse decomposition can be used to solve the linear equation system instead of explicit inversion, i.e., solving column by column using unit basis vectors. This yields the equivalent inverse mapping column vector. Furthermore, if the post-fault topology causes the system to split into multiple islands, the power transmission distribution factor matrix is ​​preferentially constructed within the island with the highest load. The sensitivity mapping is represented, and candidate action sets that are not in the main island are marked as unavailable to avoid cross-island failure of action evaluation.

[0056] Furthermore, to reduce the scale of subsequent evaluations, sub-mappings are extracted from the power transmission distribution factor for the branches corresponding to the overloaded component set, where the overloaded object is located. These sub-mappings are then used in the action evaluation phase to evaluate candidate actions. The branches include line branches and transformer branches. For example, when the overloaded component set contains 8 line branches and 2 transformer branches, the sub-mappings only retain the corresponding 10 branches in the power transmission distribution factor matrix. The row vectors in the algorithm are used to keep the multiply-accumulate size of a single evaluation within the range that can be computed online.

[0057] Thus far, step S2 has completed the fault-by-fault power flow solution and the identification of overloaded component sets, and formed a sensitivity assessment system that uses power transmission distribution factor to support rapid incremental evaluation and sub-mapping to compress the computational scale, providing a quantitative basis for subsequent candidate action impact calculation, risk indicator construction and ranking iteration.

[0058] In step S3, the core task is to form a backup automatic transfer action library and a candidate action set. The power transmission distribution factor is used to quickly calculate the branch power flow changes of the sub-actions, thereby constructing the impact degree and the newly added over-limit risk index and forming a sortable comprehensive score. At the same time, a two-level mechanism of fast verification and precise verification is used to realize the rollback disabling and status code output, so as to complete the iterative inclusion and obtain the backup automatic transfer set under controllable risk.

[0059] In this embodiment, a backup automatic transfer switch (ATS) operation library can be established first. This library stores the source bus object identifier, receiving bus object identifier, maximum transferable active power, interlock constraints, and topological preconditions for each ATS device. Based on this library, a candidate operation set can be generated. The formation of the candidate operation set must adhere to section constraints; an operation is only included in the candidate operation set when both the source bus and the receiving bus are energized and the interlock and blocking conditions are met. For example, the maximum transferable active power is set to 10MW, and a minimum transfer power threshold of 1MW is configured in the operation library to avoid invalid operations.

[0060] Furthermore, to reduce the number of backoffs, a topology-action feasibility prediction model can be introduced as a pre-screening step before the candidate action set enters the scoring process. The model's input is a graph structure based on the post-fault topology. Nodes represent bus object identifiers, and their feature vectors include bus voltage amplitude, bus active power injection, and bus load active power. Edges represent branch object identifiers, and their feature vectors include branch reactance and branch thermal stability capacity. The action input mainly consists of the source bus object identifier, the receiving-end bus object identifier, and transferable active power. The model can use a graph attention layer to perform neighborhood aggregation and output the action feasibility probability. Normalized weights for graph attention It can be written as:

[0061] Among them, attention score Indicates busbar With busbar The correlation score between the current topology and its operating state. Indicates busbar With neighboring bus The relevance score between the current topology and its operating state, and the set of adjacent buses. Indicates busbar The set of neighboring buses in the post-fault topology. During the training phase, the pass / fail status of the exact check is used as the label. Using binary cross-entropy loss It can be written as:

[0062] Among them, the number of samples This represents the number of training samples. For example, the training samples consist of 1000 historical cross-sections and 50 candidate actions per cross-section, with a sample size of 50,000, a learning rate of 0.001, a batch size of 256, 50 training epochs, 64 hidden dimensions, and 4 attention heads. This is achieved by applying active power to the bus load. Data augmentation samples are constructed using random perturbations and topology perturbations that randomly disconnect or reconnect non-critical branches. Based on this, during the online phase, actions with a feasibility probability below 0.20 are directly removed to reduce subsequent rollback overhead.

[0063] It should be further explained that the network structure of the topology-action feasibility prediction model in this embodiment adopts a two-layer graph attention layer and a one-layer action fusion multilayer perceptron. The first-layer graph attention layer performs neighborhood aggregation on the node feature vectors to extract local topological influences. The second-layer graph attention layer further aggregates based on the output of the first layer to extract cross-segment coupling influences. Subsequently, the node representations of the source bus and the receiving bus are concatenated and then concatenated with the scalar features of the action-transferable active power before being input into the multilayer perceptron to output the action feasibility probability. For example, the multilayer perceptron contains three fully connected layers, with hidden layer dimensions of 128, 64, and 32 respectively, ReLU activation function, and Sigmoid output layer to obtain the action feasibility probability. To avoid class imbalance, this embodiment downsamples samples that have not passed the precise verification to make the ratio of positive to negative samples close to 1:1, and uses the validation set AUC as the early stopping index, with an early stopping patience value of 5.

[0064] Furthermore, after the candidate action set is pre-screened, the impact and risk index calculation stage begins. In this stage, the power transmission distribution factor can be used to calculate the impact of candidate actions on the overloaded component set. Specifically, equivalent injection modeling is performed for each candidate action, which can be equivalent to a decrease in active power injection at the source bus and an increase in active power injection at the receiving bus, thus forming an equivalent injection change. Based on this, the contribution of the equivalent injection change to the power flow of each component in the overloaded component set is calculated based on the power transmission distribution factor, and then the power flow contributions are synthesized to obtain the power flow change. For example, the first... Each action is related to the branch. Active power flow change It can be written as:

[0065] Among them, the action can transfer active power. This indicates the active power transferred by the action, and the source bus number. This indicates the sequence number corresponding to the source bus object identifier, and the receiving bus sequence number. Indicates the sequence number corresponding to the receiving bus object identifier, and the branch sequence number. This indicates the sequence number corresponding to the overloaded branch object identifier. Based on this, the power flow change is correlated with the over-limit margin of the overloaded component set to calculate the impact degree. In the embodiments of this application, the impact degree... Measured by the ratio of load reduction contribution to over-limit margin, it can be written as:

[0066] Among them, the overload branch set This represents the set of branches corresponding to the overloaded component set, and the over-limit margin. This represents the difference between the branch's thermal stability capacity and its power flow. Desirable Used to avoid a denominator of zero.

[0067] Furthermore, to constrain the cost of new out-of-limit actions, it is necessary to construct an indicator for the risk of new out-of-limit actions. Specifically, this indicator can be formed based on the changes in component margins corresponding to the out-of-limit criterion set caused by candidate actions. Added risk indicators for exceeding limits. The cumulative measure of margin degradation for all constrained components can be written as:

[0068] Among them, the set of constrained elements This includes line branches, transformer branches, and busbar objects covered by the over-limit criterion set, as well as changes in component margin. Indicates the first Component margin before and after each action The difference.

[0069] In some embodiments of this application, after obtaining the impact degree and the newly added limit violation risk indicator, the impact degree and the newly added limit violation risk indicator can be normalized, and the normalized impact degree can be... and the newly added over-limit risk indicators after normalization A linear combination based on preset weights yields a sortable comprehensive score. Normalization and scoring can be expressed as:

[0070]

[0071] Among them, the overall score Indicates the overall priority of actions, with preset weights. and satisfy For example, take and .

[0072] In some embodiments of this application, when synthesizing power flow changes, an incremental superposition method can be used for the power flow changes of the same branch after multiple actions have been accumulated. This involves summing the active power flow changes of the branch caused by the included actions before calculating the new over-limit margin, ensuring consistency between the sorting process and the precise verification method. Simultaneously, to avoid an abnormally amplified impact due to an extremely small over-limit margin for a particular branch, a lower bound constraint can be added to the over-limit margin; for example, the lower bound constraint is set to 0.5MW. Furthermore, the normalization of impact and risk indicators can employ extreme value normalization within the candidate action set, and the normalization should be applied when the range is less than... In this case, the normalized value is set directly to 0 to avoid numerical amplification.

[0073] Furthermore, the candidate actions are ranked comprehensively based on a combined score formed by impact degree and newly added over-limit risk indicators, and the comprehensive ranking result is determined. Specifically, in the inclusion phase, candidate actions can be selected sequentially according to the comprehensive ranking result and a rapid verification is performed. The first part of the rapid verification is to verify the consistency of the load reduction direction of the overload component set, with the change in active power flow of the branch satisfying... As a criterion for determining the load reduction direction, a negative threshold is also introduced. ,when If the condition is not met, the load reduction is deemed insufficient; for example, a negative threshold is used. Take -2.0MW. The second part of the quick verification is the threshold verification of the newly added over-limit risk indicators. For example, the risk threshold is taken as 0.10. When the risk indicator If the risk threshold is exceeded, the action is directly rejected. For candidate actions that pass the rapid verification, a more precise verification can be performed and they are included in the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set is activated. The precise verification adopts the same set of over-limit criteria as the power flow calculation, that is, the same set of over-limit criteria as in step S2 is used to determine the new over-limit and check whether the power flow converges.

[0074] When a new limit violation or power flow non-convergence occurs during precise verification, the most recently included candidate action can be revoked. After revocation, the candidate action is marked as disabled, and a disabling flag is written at the action library level to prevent it from being selected again. At the end of each iteration, a status code is output. In this embodiment of the application, the status code must at least distinguish between no overload status code, overload residual status code, infeasible status code, and data unavailable status code.

[0075] For example, the status codes can be integers. A status code of 0 indicates that the overload component set is empty and no new overloads have occurred; a status code of 1 indicates that the overload component set is still not empty but candidate actions can still be evaluated; a status code of 2 indicates that a new overload has occurred and a backoff has been triggered; a status code of 3 indicates that the power flow is not converging and a backoff has been triggered; and a status code of 4 indicates that the candidate action set is exhausted or the number of consecutive backoffs has reached its limit. The termination condition can be determined jointly by the status code and the overload component set. When the status code is 0, the process terminates directly; when the status code is 4, the process terminates and enters the subsequent correction control solution branch. Simultaneously, the termination condition can be used to determine whether to continue execution. If the termination condition is not met, iteration continues until the termination condition is met; when the termination condition is met, iterative iterations are incorporated into the backup self-starting set. For example, the termination conditions include an empty overload component set, exhaustion of the candidate action set, reaching the upper limit of consecutive backoffs, and failure to meet the risk threshold.

[0076] Thus, step S3 establishes an impact calculation and risk indicator assessment system driven by the power transmission distribution factor. Through comprehensive scoring and ranking, a two-level mechanism of rapid verification and precise verification is established to create an iterative link driven by back-off disabling and status codes. This enables the backup self-starting set to achieve a controlled balance between load reduction benefits and new over-limit risks, providing clear action results and constraint boundaries for subsequent correction control models.

[0077] In step S4, the core task is to construct and solve the grid correction control model based on the grid topology after the automatic transfer switch set action when the status code indicates that there are still overloaded components, so as to obtain the load shedding strategy with the minimum power outage loss; at the same time, when the optimization model is not feasible, the executable load shedding strategy with the minimum default is given by constraint relaxation and penalty terms and the status code is refreshed to ensure that the overload can be eliminated within the safety constraints.

[0078] In this embodiment, when the output status code indicates that overloaded components still exist, it is necessary to enter the correction control solution stage, at which point the power grid correction control model needs to be solved. Specifically, an optimization model can be constructed when the output status code indicates that overloaded components still exist, and the power outage loss is taken into account. The target output load shedding strategy is to minimize the load. In this embodiment, the optimization model includes a power outage loss function based on load importance stratification and load shedding ratio, wherein the power outage loss function is based on the node power outage ratio. The load shedding ratio and the active power of the nodal load are represented. The active power of the bus load represented is the independent variable, and the objective function can be written as:

[0079] Among them, the power outage loss function Using a quadratic form function that crosses zero, we can use:

[0080] Where, constant and Determined by load importance stratification, with more important loads corresponding to larger loads. To inhibit resection.

[0081] It should be noted that, in the embodiments of this application, the load objects are divided into three layers: important loads, general loads, and interruptible loads, and the layering result is mapped to a constant in the power outage loss function. For example, critical loads correspond to Typical load correspondence Interruptible load corresponding This creates a strong penalty for critical loads. The load shedding ratio can be discretized, for example, with a step size of 0.05, and converted into a mixed integer form inside the solver. This ensures that the output load shedding strategy can be directly mapped to the object-level load shedding actions in the switch and circuit breaker action list.

[0082] Furthermore, regarding the constraint system, power balance constraints, branch thermal stability constraints, and voltage constraints can be established, along with topology constraints. The grid topology after the automatic transfer switch (ATS) operation is used as the input to the optimization model. Topology constraints ensure that corrective control actions do not introduce unacceptable islanding or loop network interruption conflicts. Specifically, these constraints can include maintaining main grid connectivity, preventing simultaneous disconnection of critical channels, and satisfying the interlocking constraints of circuit breakers. For example, connectivity constraints are applied to critical substation buses, requiring that after load shedding, the bus still maintains at least one effective connection to the main grid to avoid creating power supply blind spots.

[0083] For example, power balance constraints include active power balance constraints. and reactive power balance constraints It can be written as:

[0084]

[0085] Among them, the active power output of the generator Indicates busbar The generator has active power output and reactive power output. Indicates busbar The generator has no reactive power output. Indicates busbar Active power at the load, Indicates busbar reactive power at load, Indicates the connection with the busbar In the current topology, the set of directly connected adjacent buses, and the active power flow of branches between buses. reactive power flow between the busbar and the branch line Determined by the power flow equations.

[0086] For example, the branch thermal stability constraint can adopt the constraint form that the branch apparent power does not exceed the branch thermal stability capacity, where the branch apparent power is calculated from the branch active power flow and reactive power flow given in step S2. The voltage constraint and outage ratio constraint can be written as:

[0087]

[0088] Among them, bus voltage amplitude Indicates busbar Voltage amplitude at the location, upper limit of bus voltage amplitude Lower limit of bus voltage amplitude For boundary parameters, take, for example, and .

[0089] Furthermore, when the optimization model is infeasible, a relaxation mechanism can be implemented, introducing constraint relaxation amounts and adding a penalty term to the power outage loss function to suppress the degree of default. In the embodiments of this application, relaxation amounts can be introduced into thermal stability constraints and voltage constraints, and a linear penalty is applied to the relaxation amounts in the objective function to obtain the minimum default solution.

[0090] Specifically, when introducing constraint relaxation, relaxation amounts can be introduced for branch thermal stability constraints and voltage constraints respectively, and an upper bound can be set for each relaxation amount to avoid excessive default. For example, the upper bound of the branch thermal stability relaxation amount is 0.02 times the branch thermal stability capacity, and the upper bound of the voltage relaxation amount is 0.01 pu. The penalty term coefficient can be calibrated according to the power outage loss coefficient of important loads, so that within the feasible range, small-scale shelving of general loads is preferred rather than replacing safety constraints with large-scale default.

[0091] Based on this, a load shedding strategy with minimal default can be output and the status code updated. For example, when the thermal stability capacity of a critical branch is insufficient, resulting in no feasible solution, a 1% short-term default of the thermal stability constraint can be allowed through relaxation, and the penalty term can be set to twice the cost of shedding the critical load, so as to prioritize maintaining the power supply range while eliminating the overload as soon as possible.

[0092] Thus, step S4 completes the fallback link from the overload residual state to the optimization solution of the correction control, forming a load shedding strategy generation system that quantifies the cost with the power outage loss function, defines the safety boundary with the power balance and thermal stability voltage topology constraints, and ensures that the output of infeasible scenarios is guaranteed by the relaxation penalty mechanism. This provides executable control input for the subsequent implementation of the action list and closed-loop recalculation.

[0093] In step S5, the core task is to transform the automatic transfer set and load shedding strategy into a unified, verifiable and rewritable list of switch and circuit breaker actions, and after rewriting, to recalculate the power flow to verify the overload elimination effect, and finally update the overload component set and status code to form a closed-loop handling result.

[0094] In this embodiment, a list of switch / circuit breaker actions can be generated based on each automatic transfer switch (ATS) action in the ATS set and the load shedding strategy, and the two types of actions can be structured separately. Specifically, the switch / circuit breaker actions corresponding to the ATS set and the switch / circuit breaker actions corresponding to the load shedding strategy can be uniformly mapped to a structured action list. The structured action list fields include object identifier, action type, action sequence, and verification identifier. The object identifier is used to uniquely indicate the target switch / circuit breaker object, the action type is used to indicate whether it is opening, closing, or holding, the action sequence is used to constrain the execution order, and the verification identifier is used to mark whether the preconditions of the action are met and whether the consistency check is passed after the action is executed.

[0095] It should be noted that the sequence of actions in the structured action list is used to reflect the engineering logic of load shedding after automatic transfer of backup power, and at the same time to ensure that the opening and closing operations within the same bay meet the requirements of opening before closing and anti-misoperation interlocking; the verification identifier in this application embodiment includes a pre-topology verification identifier and a post-power flow verification identifier. The pre-topology verification identifier is used to check whether the source bus and the receiving bus meet the allowable conditions configured in the action library, and the post-power flow verification identifier is used to record whether the recalculation after writing back converges and whether any new over-limits occur.

[0096] Based on this, write-back information can be generated according to the structured action list, and the write-back information can be written into the section data interface of the dispatch control system to write back the real-time operation data of the power grid. After the write-back, the section data is refreshed once, thereby updating the real-time operation data of the power grid.

[0097] Furthermore, after the write-back is completed, a closed-loop recalculation is performed. This involves recalculating the power flow based on the updated real-time grid operation data on the updated cross-section, determining the overloaded objects according to the overload criterion set to re-evaluate the overloaded component set, and finally writing the recalculation results back to the state variables and updating the overloaded component set and status code. For example, when the standby automatic transfer set transfers 10MW and the load shedding strategy cuts off 3MW, the recalculation shows that the apparent power of all overloaded branches is less than the branch thermal stability capacity. The status code is then updated to a no-overload status code, and an action list execution success flag is output.

[0098] At this point, step S5 completes the structured implementation of the control strategy into the action list, and verifies the overload elimination effect under the same caliber through write-back and recalculation, forming a closed-loop processing system that constrains execution with action sequence and verification identifier, confirms the results with closed-loop power flow recalculation, and drives subsequent business processing with status code updates.

[0099] In summary, this method verifies the consistency of timestamps between topology and measurement data within the same online section and constructs a fault set covering both line faults and faults on parallel towers. It then combines AC power flow results and power transmission distribution factors to quantify the load reduction contribution of candidate backup automatic transfer actions to the overload component set. Furthermore, it drives the sorting iteration and two-level verification and rollback disabling based on a joint score of impact degree and newly added overload risk indicators. Finally, when overload remains, it solves the load shedding strategy under thermal stability, voltage, and topology constraints with the outage loss function as the objective and outputs a structured action list for closed-loop recalculation. This approach can eliminate overloads more quickly and reduce outage losses and outage range while controlling newly added overload and non-convergence risks. It reduces reliance on manual handling and repeated trial operations, improves the consistency, interpretability, and online stability of overload handling, and ultimately enhances the safety margin of power grid operation and power supply reliability.

[0100] It should be noted that, although the embodiments in this application are based on... Figure 1The steps are described sequentially, but this does not mean that the steps must be performed in a strict order. The reason this embodiment follows this order is... Figure 1 The order in which each step is described is intended to facilitate understanding of the technical solutions of the embodiments of this application by those skilled in the art. In other words, the step numbers are only used to distinguish different steps and do not constitute a limitation on the execution order of the steps; the specific execution order of each step can be appropriately adjusted according to actual needs, functional requirements, and the inherent logic in actual application scenarios.

[0101] In this embodiment, a power grid overload control method considering automatic backup switching is applied to an overload handling scenario of a regional power grid. This regional power grid includes a 220kV main grid, 110kV supply areas, and several substations with dual power supply. The real-time operation data source of the dispatch control system provides topology, switches, and measurements with a refresh cycle of 2 seconds, and a state estimation refresh cycle of 10 seconds.

[0102] For ease of explanation, the overload handling process following a typical fault is selected as an example of a complete implementation process: at the sampling time, the following is taken: At that time, 220kV line A tripping occurred, and the line condition was estimated after the fault. Meritorious trend The thermal stability capacity of the line is An overload occurred.

[0103] The complete implementation process may include the following steps: Step 1: The data output from the real-time operating data source of the power grid dispatch and control system enters the unified data construction link and forms a cross-sectional input that can be used for verification and decision-making.

[0104] In this embodiment, data from the real-time operating data source of the power grid dispatch and control system is first aggregated. The aggregated content covers power grid topology data, switch location information, and measurement data. Then, timestamp consistency verification is performed on the three types of data to obtain data quality identifiers. Among them, the power grid topology data represents the connection relationships of primary equipment such as buses, lines, and transformers; the switch position information represents the open / closed status of circuit breakers and disconnectors; the measurement data represents active power, reactive power, voltage, and power flow measurements; and the data quality label... This indicates the value of the timestamp consistency check result.

[0105] For example, at the sampling time Within the corresponding state estimation period, the maximum deviation between the timestamp of the power grid topology data and the timestamp of the measurement data is: The preset consistency threshold is set to Therefore, data quality labeling Valid states are taken. To avoid misjudgments caused by single-point measurement anomalies, missing data completion and out-of-bounds clamping are performed synchronously on the measurement data. Missing data completion uses linear interpolation of two adjacent state estimation periods, and out-of-bounds clamping uses truncation based on the upper and lower limits of the equipment's rated range.

[0106] Subsequently, a fault set was generated based on the power grid topology data. Fault set This includes line fault items and fault items of lines running on the same tower, and is included in the fault set. The inner part contains a set of over-limit criteria bound to each fault item. Among them, the fault set This represents the set of fault items used for N-1 and N-2 verification of parallel tower operation, and the set of over-limit criteria. Indicates the fault item A set of rules for determining limits related to line thermal stability, main transformer thermal stability, and node voltage.

[0107] For example, this embodiment constructs a total of line fault items for 220kV and 110kV lines. The items include a total of fault items for parallel lines on the same tower. Item. A set of out-of-limit criteria bound to each fault item. It should include at least line thermal stability criteria, main transformer thermal stability criteria, and node voltage criteria. The line thermal stability criteria adopt... The form, node voltage criterion adopts form.

[0108] Step 2: Perform power flow verification on each fault set and form an overload component set and power transmission distribution factor for subsequent action impact assessment.

[0109] In this embodiment, the fault set Power flow calculations are performed one by one. The calculations use AC power flow, consistent with the dispatch control system, as the accurate verification standard, while a DC power flow linearization model is used for rapid evaluation. For each fault item... Below, based on the set of limits-crossing criteria. Identify and summarize the overloaded components to obtain the overloaded component set. Among them, the overload component set This represents the set of line branches and transformer branches that exceed the limit under the current target fault item.

[0110] Determining the set of overload components Subsequently, a DC power flow linearization model was established based on the grid topology data and branch parameters that had undergone timestamp consistency verification, and a reference bus was selected. For example, in the fault item Below, overload component collection Includes lines With the main change total Each component. For the circuit. The corresponding sensitivity row vector was extracted. ,in , This is used for incremental evaluation of subsequent load transfer actions.

[0111] Step 3: Construct a candidate action evaluation link based on the load transfer mechanism of the backup automatic transfer action and form a backup automatic transfer set.

[0112] In this embodiment, a backup automatic switching action library is first established. And generate a set of candidate actions Among them, the backup self-starting action library This represents the set of load transfer actions that can be triggered by the automatic transfer switch (ATS), and is also known as the candidate action set. This represents the set of executable actions selected under the constraints of the current topology, interlocks, and available power supply.

[0113] To improve the candidate action set To assess the feasibility and ranking quality of candidate actions, this embodiment introduces a topology-action feasibility discrimination model before evaluating the candidates' actions. This method is used to pre-screen candidate actions without increasing the number of power flow solutions. The model input includes the power grid diagram structure, bus state features, and action dual endpoint features; the model output is the action feasibility probability. Among them, the topology-action feasibility discrimination model This represents a binary classification model based on graph attention mechanism, where the action feasibility probability is... Indicates candidate actions The predicted probability value that satisfies the conditions of no new over-limits and power flow convergence under the current cross-section.

[0114] In this embodiment, the power grid diagram structure adopts a graph representation with nodes as buses and edges as branches. The node feature vector includes bus voltage magnitude, bus voltage phase angle, bus active power injection, bus reactive power injection, and bus load importance encoding. The edge feature vector includes branch reactance, branch thermal stability capacity, and branch current load rate. To enable the model to explicitly perceive the semantics of "the action transfers load from the source bus to the receiving bus," additional action role encoding is performed for the source bus and the receiving bus, and the action transfer power is encoded. After normalization, it is used as the input for action intensity features. To improve the generalization ability to topological perturbations, edge dropping enhancement is performed on the graph structure during the training phase, Gaussian perturbation enhancement is performed on the node features, and load scaling enhancement is performed on the historical cross-sections. The edge dropping rate is taken as... The standard deviation of node disturbances is taken as The load scaling range is taken as follows Among them, the transfer power This indicates the scale of load transfer for the candidate action.

[0115] Model The network structure consists of Layer diagram attention layer Fully connected layer and It consists of 1 sigmoid output layer, and each graph attention layer uses 1 sigmoid output layer. Each attention head has a hidden dimension. The hidden dimension of the fully connected layer is taken The training loss function is the binary cross-entropy loss function, and the optimizer is AdamW. The initial learning rate is [value missing]. Batch size The number of training rounds is taken Here, the binary cross-entropy loss function represents the classification loss with feasible labels as the supervision signal, and the batch size represents the number of samples used for each parameter update.

[0116] The training samples are generated from offline simulation: power flow calculations are performed on historical operational sections and combinations of fault items, and a precise AC power flow check is performed for each candidate action. Samples with "power flow convergence without new limit violations" are marked as positive, while samples with "power flow non-convergence or new limit violations" are marked as negative. For example, this embodiment constructs a sample size of [number missing]. The sample size is approximately [number], with a positive and negative sample ratio of approximately [percentage]. To mitigate the impact of class imbalance on training, oversampling is applied to positive samples, and class weights are introduced into the loss function. The positive class weights are set to... .

[0117] After pre-screening, the power transfer distribution factor matrix is ​​used to analyze the remaining candidate actions. Calculate candidate actions for the set of overloaded components The impact and the construction of new limits-crossing risk indicators. For any candidate action The candidate action is equivalent to a combination of a decrease in active power injection at the source bus and an increase in active power injection at the receiving bus, forming an equivalent injection change. It can be written as:

[0118] in, Indicates candidate actions For branch roads The incremental impact of the active current, Indicates the source bus number of the candidate action. Indicates the receiving bus number of the candidate action. This represents the transfer power of the candidate action.

[0119] To establish a sortable relationship between power flow changes and over-limit margins, a branch over-limit margin is defined. And calculate the impact based on this. It can be written as:

[0120] Among them, the excess margin of branch roads Indicates a branch Capacity margin, Indicates a branch Thermal stability capacity, Indicates a branch The current trend of positive influence Indicates candidate actions The overall load reduction contribution of the overload component assembly This represents a numerical stability constant used to avoid a denominator of 0.

[0121] Simultaneously, a new out-of-limit risk indicator is constructed based on the changes in component margins corresponding to the out-of-limit criterion set caused by candidate actions. And the degree of influence With newly added risk indicators for exceeding limits Perform normalization. This can be written as:

[0122] Among them, new indicators for exceeding limits risk have been added. Indicates candidate actions The newly introduced measure of exceeding limits This represents the set of components that are not currently exceeding limits but require monitoring. Indicates candidate actions After the function of the component Capacity margin.

[0123] The normalized impact value and the normalized new risk index for exceeding limits are linearly combined according to preset weights to obtain the comprehensive score. The ranking results are determined based on the comprehensive score. This can be written as:

[0124] Among them, the overall score Indicates candidate actions The sorting score, Indicating the degree of influence The normalized value, This indicates the addition of new indicators for exceeding limits. The normalized value, Indicates the weight of influence. This represents the risk weight.

[0125] For example, in the fault item Below, model right Output the feasible probability of each candidate action. Using a preset feasible probability threshold Retain Each candidate action is entered into the sorting process. For one of the candidate actions... Transfer power is taken Source bus The receiving end busbar takes Substitute and get This characterizes the effect of the action on the line. It has a clear contribution to the direction of load reduction.

[0126] After sorting, candidate actions are selected sequentially according to the comprehensive sorting results and a rapid verification is performed. The rapid verification includes consistency checks on the load reduction direction of the overloaded component set and threshold checks on newly added over-limit risk indicators. To ensure the rapid verification has an executable scope, this embodiment introduces a load reduction direction consistency threshold. With risk threshold Among them, the load unloading direction consistency threshold Risk threshold is used to limit the proportion of components in an overloaded component set that must contribute to load reduction. This is used to limit the allowable upper limit of newly added over-limit risk indicators. In this embodiment, the load reduction direction consistency threshold... Pick Risk threshold Pick When the candidate action is... At least Components satisfy and At that time, the candidate action will proceed from quick verification to precise verification.

[0127] For candidate actions that pass the rapid verification, a precise verification is performed. The precise verification uses a set of out-of-limit criteria consistent with the power flow calculation. If a new out-of-limit event or power flow non-convergence is detected during the precise verification, the most recently included candidate action is revoked and marked as disabled, and a status code is output. The iteration continues until the termination condition is met. To maintain the determinism of the iteration, this embodiment adopts a strategy of "recalculating the cross-section and recalculating the sorting after each action is included," that is, after each action is included, the topology and load distribution are written back and recalculated. The remaining candidate actions are then updated and sorted based on impact and risk indicators. Among these, status codes... This indicates the status of the iteration process and the termination result.

[0128] For example, candidate actions After accurate verification, it will be included in the backup self-transfer collection. After the action is carried out, the line Meritorious trend Descending to Then, the second-ranked candidate action... If the bus voltage falls below the lower limit during precise verification, the candidate action is cancelled. Mark it as disabled, status code Update to "Action Regression" status and continue iterating. Continue to include candidate actions. After that, the line The tide of merit further declined to It is still higher than the thermal stability capacity of the line. Therefore, the termination condition is determined as "the backup automatic transfer unit terminates and there are still overloaded components", status code Updated to "Overloaded components still exist" status.

[0129] Step 4: When the backup automatic transfer set terminates and overloaded components still exist, solve the power grid correction control model and output the load shedding strategy that minimizes power outage losses.

[0130] In this embodiment, when the status code When overloaded components still exist, an optimization model is constructed and solved using the node load shedding ratio as the decision variable. The optimization model includes a power outage loss function based on load importance stratification and load shedding ratio, while establishing power balance constraints, branch thermal stability constraints, voltage constraints, and topology constraints, and incorporating the automatic transfer switch set. The topology after the action is used as input to the optimization model.

[0131] The objective function for power outage losses can be written as:

[0132] in, This represents the target value for comprehensive power outage losses. Indicates the number of busbars. Represents a node The load shearing ratio, Represents a node The power outage loss function.

[0133] In this embodiment, the power outage loss function is constructed using a quadratic form consistent with the load importance hierarchy. It can be written as:

[0134] in, Represents a node The corresponding quadratic loss coefficient, Represents a node The corresponding first-order loss coefficient, Represents a node Active load.

[0135] The power balance constraint can be written as:

[0136]

[0137] in, Represents a node The active power balance residual, Represents a node The reactive power balance residual, Represents a node The generator's active power output, Represents a node The generator's reactive power output, Represents a node Active load, Represents a node reactive load, Represents a node To the node The contributing current of the relevant branch roads, Represents a node To the node The reactive power flow of the relevant branch lines.

[0138] Inequality constraints must include at least upper and lower limits for load shedding ratio, upper and lower limits for voltage, and branch thermal stability constraints. This can be written as:

[0139] in, Represents a node Voltage amplitude, Represents a node Lower voltage limit Represents a node Voltage limit Indicates a branch The unproductive current.

[0140] For example, in this embodiment, the upper limit of voltage is taken as The lower limit of voltage is taken as A three-level coefficient configuration is used for load importance stratification: for low-importance load nodes, a coefficient of 1 is applied. For medium-importance load nodes, take For high-importance load nodes, take .

[0141] In preparation for self-injection collection After the action, the line Still overloaded Solving the optimization model yields the nodes. Load shearing ratio ,node Load shearing ratio The load shedding ratio for the remaining nodes is Under this load shedding strategy, the line The active current has decreased to And the voltage constraint remains Within the range.

[0142] When the optimization model is not feasible, this embodiment introduces a constraint relaxation amount. A penalty term is added to the power outage loss function to output the load shedding strategy with the minimum default and update the status code. Among them, the constraint slack amount This represents the set of relaxed variables for branch thermal stability and voltage constraints.

[0143] Step 5: Generate an action list and write back the cross-section to conduct a closed-loop review to confirm overload elimination and output an executable sequence.

[0144] In this embodiment, based on the backup self-connection set A list of switch / circuit breaker actions is generated based on the load shedding strategy. After writing back the real-time grid operation data, the overload component set is re-evaluated. To ensure the output can be directly implemented, the switch / circuit breaker actions corresponding to the standby automatic transfer set and the switch / circuit breaker actions corresponding to the load shedding strategy are uniformly mapped into a structured action list. The structured action list includes object identifier, action type, action sequence, and verification identifier. Subsequently, write-back information is generated based on the structured action list, and the real-time grid operation data is updated. Based on the updated real-time grid operation data, power flow calculation is re-performed, and the overload component set and status codes are updated. .

[0145] For example, this embodiment will set up a self-propelled collection. Actions included Mapped to a combined command of "closing the 110kV bus tie circuit breaker" and "opening the feeder circuit breaker," this load shedding strategy... Mapped to nodes The tripping commands for the two low-importance feeders are sequenced according to load importance from low to high. After execution, the switch position information and measurement data are written back, and AC power flow verification is performed again to obtain the line status. Load rate by Descending to Main change Load rate by Descending to status code Updated to "Overload Elimination" status, closed-loop review complete.

[0146] Through the complete implementation process described above, this method achieves rapid pre-screening of feasible actions by evaluating the impact of backup automatic transfer candidate actions driven by power transmission distribution factors and ranking newly added over-limit risk constraints, combined with a topology-action feasibility discrimination model based on graph attention mechanism. Under the iterative inclusion, precise verification and backoff mechanism, a backup automatic transfer set without adding new over-limits is formed. At the same time, when overloaded components still exist, a grid correction control model including power balance, branch thermal stability, voltage and topology constraints is introduced to output the load shedding strategy with the minimum power outage loss. In fault overload scenarios, it can prioritize the use of load transfer to reduce the degree of overload and accurately control the necessary load shedding scale, reduce unnecessary expansion of the power outage impact range, and improve the stability, determinism and engineering feasibility of overload elimination.

[0147] It should be understood that the step numbers identified by "Step 1, Step 2" and other similar forms in the above embodiments are only used to distinguish different steps and do not limit the steps to be executed in the order of these numbers. The specific execution order of each step can be adjusted according to its functional requirements and the inherent logic in the actual application scenario. The above step numbers should not be interpreted as a limitation on the implementation process of the embodiments of this application.

[0148] like Figure 2 As shown, the following is an embodiment of a grid overload control system with automatic transfer switch (ATS) provided by this disclosure. This grid overload control system with ATS and the grid overload control methods with ATS in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the grid overload control system with ATS, please refer to the embodiments of the grid overload control methods with ATS described above.

[0149] Based on the same concept, another embodiment of this application provides a power grid overload control system that takes into account automatic backup switching, including: Data construction unit 1 is used to obtain real-time power grid operation data from the real-time operation data source of the power grid dispatch and control system and construct a fault set; Power flow assessment unit 2 is used to perform power flow calculations on each fault set, determine the set of overloaded components, and calculate the power transmission distribution factor. Action screening unit 3 is used to establish a backup automatic transfer action library and generate a candidate action set. It uses the power transmission distribution factor to calculate the impact of the candidate actions on the set of overloaded components and constructs a new over-limit risk index. The candidate actions are comprehensively sorted according to the impact and the new over-limit risk index. Based on the comprehensive sorting results, the candidate actions are checked in turn and incorporated into the backup automatic transfer set. The grid topology after the backup automatic transfer set is activated is obtained and the status code is output. When the status code indicates that there are still overloaded components, the correction control unit 4 solves the grid correction control model based on the grid topology after the automatic transfer switch set action and outputs the load shedding strategy with the minimum power outage loss. The action closed-loop unit 5 is used to generate a list of switch and circuit breaker actions based on each automatic transfer action and load shedding strategy in the automatic transfer set, and re-evaluate the overload component set after writing back the real-time operation data of the power grid.

[0150] In some embodiments of this application, the action filtering unit 3 includes: The fast verification subunit is used to select candidate actions in sequence according to the comprehensive sorting results and perform fast verification. Fast verification includes consistency verification of the load reduction direction of the overload component set and threshold verification of newly added over-limit risk indicators. The fine-check sub-unit is used to perform precise checks on candidate actions that have passed the fast check and incorporate them into the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set has been activated. The precise check adopts the over-limit criterion set with the same caliber as the power flow calculation. The rollback sub-unit cancels the most recently included candidate action and marks it as disabled when a new limit violation or power flow non-convergence occurs during the precise verification. The state sub-unit is used to output the state code and drive the action filtering unit 3 to continue iterating until the termination condition is met.

[0151] The above-disclosed embodiments are merely preferred embodiments of the present invention, but the present invention is not limited thereto. Any non-creative variations that can be conceived by those skilled in the art, as well as any improvements and modifications made without departing from the principles of the present invention, should fall within the protection scope of the present invention.

Claims

1. A power grid overload control method considering automatic transfer switching, characterized in that, include: Real-time power grid operation data is obtained from the real-time operation data source of the power grid dispatch and control system, and a fault set is constructed. Power flow calculations are performed on each fault set to determine the set of overloaded components and to calculate the power transmission distribution factor. Establish a backup automatic transfer action library and generate a candidate action set. Calculate the impact of the candidate actions on the set of overloaded components using the power transmission distribution factor and construct a new over-limit risk index. Sort the candidate actions according to their impact and the new over-limit risk index. Verify the candidate actions in turn based on the comprehensive ranking results and incorporate them into the backup automatic transfer set. Obtain the power grid topology after the backup automatic transfer set is activated and output the status code. When the status code indicates that there are still overloaded components, the grid correction control model is solved based on the grid topology after the automatic transfer switch set action, and the load shedding strategy with the minimum power outage loss is output. Based on the automatic transfer actions and load shedding strategies in the automatic transfer set, a list of switch and circuit breaker actions is generated, and the overload component set is re-evaluated after writing back the real-time power grid operation data.

2. The grid overload control method considering automatic transfer switch as described in claim 1, characterized in that, The steps for acquiring real-time power grid operation data and constructing a fault set include: Acquire power grid topology data, switch location information, and measurement data, perform timestamp consistency verification, and output data quality identifiers; A fault set is generated based on power grid topology data. The fault set includes line fault items and fault items of parallel lines on the same tower. Within the fault set, bind a corresponding set of out-of-limit criteria to each fault item.

3. The grid overload control method considering automatic transfer switch as described in claim 2, characterized in that, The steps for calculating the power transfer distribution factor include: A DC power flow linearization model is established based on the grid topology data and branch parameters verified by timestamp consistency, and a reference bus is selected. The branch parameters include line parameters and transformer parameters. Construct a sensitivity mapping from changes in bus active power injection to changes in branch active power flow and solidify it as a power transmission distribution factor; For the branches corresponding to the overload component set, sub-maps are extracted from the power transmission distribution factor and used for candidate action evaluation. The branches include line branches and transformer branches.

4. The power grid overload control method considering automatic transfer switch as described in claim 3, characterized in that, The steps for calculating the impact of candidate actions on the set of overloaded components using the power transfer distribution factor include: The candidate action is equivalent to a combination of a decrease in active power injection at the source bus and an increase in active power injection at the receiving bus, forming an equivalent injection change. The contribution of the equivalent injection change to the power flow of each element in the overloaded element set is calculated based on the power transmission distribution factor and then synthesized to obtain the power flow change. The influence degree is obtained by correlating the power flow change with the overload margin of the overload component set.

5. The power grid overload control method considering automatic transfer switch as described in claim 2, characterized in that, The steps for constructing new limits-crossing risk indicators and comprehensively ranking candidate actions based on their impact and the new limits-crossing risk indicators include: A new limit violation risk index is constructed based on the change in component margin corresponding to the limit violation criterion set caused by candidate actions, and the impact degree and the new limit violation risk index are normalized. The normalized impact and the normalized new risk index for exceeding the limit are linearly combined according to preset weights to obtain a comprehensive score. The candidate actions are then ranked according to the comprehensive score to determine the comprehensive ranking result.

6. The grid overload control method considering automatic transfer switch as described in claim 5, characterized in that, The steps include: verifying candidate actions sequentially based on the comprehensive ranking results and incorporating them into a backup automatic transfer set; obtaining the power grid topology after the backup automatic transfer set is activated; and outputting the status code. Candidate actions are selected sequentially based on the comprehensive ranking results and a rapid verification is performed. The rapid verification includes consistency verification of the load reduction direction of the overloaded component set and threshold verification of newly added over-limit risk indicators. For candidate actions that pass the rapid verification, perform precise verification and incorporate them into the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set is activated. The precise verification adopts the over-limit criterion set with the same caliber as the power flow calculation. When a new over-limit occurs or the power flow does not converge in the precise verification, the most recently incorporated candidate action is revoked and the candidate action is marked as disabled. At the same time, the status code is output and the iteration continues until the termination condition is met.

7. The power grid overload control method considering automatic transfer switch as described in claim 6, characterized in that, When the status code indicates that overloaded components still exist, the steps for solving the grid correction control model based on the grid topology after the automatic transfer switch set operation and outputting the load shedding strategy that minimizes power outage losses include: An optimization model is constructed when the output status code indicates that there are still overloaded components. The optimization model includes a power outage loss function based on load importance stratification and load shedding ratio. Establish power balance constraints, branch thermal stability constraints, voltage constraints, and topology constraints, and use the grid topology after the automatic transfer switch set is activated as the input to the optimization model; When the optimization model is not feasible, a constraint relaxation is introduced and a penalty term is added to the power outage loss function. The load shedding strategy with the minimum default is output and the status code is updated.

8. The power grid overload control method considering automatic transfer switch as described in claim 7, characterized in that, The steps for generating a list of switch and circuit breaker actions, writing back real-time power grid operating data, and reassessing the overload component set include: The actions of the switch and circuit breaker corresponding to the automatic transfer set and the actions of the switch and circuit breaker corresponding to the load shedding strategy are uniformly mapped into a structured action list. The structured action list includes object identifier, action type, action sequence and verification identifier. Based on the structured action list, write-back information is generated and the real-time power grid operation data is updated. Based on the updated real-time power grid operation data, power flow calculation is performed again and the overload component set and status code are updated.

9. A power grid overload control system that takes into account automatic transfer switching, characterized in that, include: The data construction unit is used to obtain real-time power grid operation data from the real-time operation data source of the power grid dispatch and control system and construct a fault set. The power flow assessment unit is used to perform power flow calculations on each fault set, determine the set of overloaded components, and calculate the power transmission distribution factor. The action screening unit is used to establish a backup automatic transfer action library and generate a candidate action set. It uses the power transmission distribution factor to calculate the impact of the candidate actions on the set of overloaded components and constructs a new over-limit risk index. The candidate actions are comprehensively sorted according to the impact and the new over-limit risk index. Based on the comprehensive sorting results, the candidate actions are checked in turn and incorporated into the backup automatic transfer set. The grid topology after the backup automatic transfer set is activated is obtained and the status code is output. The correction control unit is used to solve the grid correction control model based on the grid topology after the automatic transfer switch set has been activated and output the load shedding strategy that minimizes power outage losses when the status code indicates that there are still overloaded components. The action closed-loop unit is used to generate a list of switch and circuit breaker actions based on the automatic transfer actions and load shedding strategies in the automatic transfer set, and re-evaluate the overload component set after writing back the real-time operation data of the power grid.

10. The power grid overload control system considering automatic transfer switch as described in claim 9, characterized in that, The action filtering unit includes: The fast verification subunit is used to select candidate actions in sequence according to the comprehensive sorting results and perform fast verification. Fast verification includes consistency verification of the load reduction direction of the overload component set and threshold verification of newly added over-limit risk indicators. The fine-check sub-unit is used to perform precise checks on candidate actions that have passed the fast check and incorporate them into the backup automatic transfer set to obtain the power grid topology after the backup automatic transfer set has been activated. The precise check adopts the over-limit criterion set with the same caliber as the power flow calculation. The rollback subunit is used to revoke the most recently included candidate action and mark the candidate action as disabled when a new limit violation or power flow non-convergence occurs during the precise verification. The state sub-unit is used to output the state code and drive the action filtering unit to continue iterating until the termination condition is met.

Images