Island reconstruction method of virtual power plant driven by impact power spectrum

Abstract

The application relates to the field of power systems and discloses a virtual power plant island reconstruction method driven by impact power spectrum. The method aims to solve the optimization reconstruction problem of the existing virtual power plant in the island state. The method collects the operation data of the distributed power sources in the virtual power plant, calculates the impact power spectrum characteristics of each power source, and performs island division and reconstruction optimization based on the impact power spectrum characteristics. Through the technical scheme, the power balance and economic optimization of the virtual power plant in the island operation mode can be realized, and the power supply reliability and operation efficiency of the system are improved.

Description

Technical Field

[0001] This application belongs to the field of power system technology, and in particular relates to the technology of post-disaster resilience recovery of distribution networks and collaborative dispatch of virtual power plants, specifically to a virtual power plant islanding reconfiguration method driven by the impact power spectrum. Background Technology

[0002] Frequent extreme disasters such as typhoons, earthquakes, and ice storms can easily lead to the loss of power to the main power grid and large-scale line faults, causing prolonged power outages for critical loads and seriously affecting the resilience of the power grid and public safety. Post-disaster recovery by utilizing local distributed power sources, energy storage devices, and switchable interconnection switches to divide the distribution network into several self-sufficient islands and restore power to the lost loads is one of the core methods to enhance the post-disaster resilience of the distribution network.

[0003] Existing technologies related to post-disaster recovery of power distribution networks can be broadly categorized as follows: First, using intelligent optimization algorithms (such as particle swarm optimization, ant colony optimization, and graph neural networks) to solve network reconstruction schemes after a fault, defining islanded operation ranges with the goal of maximizing the recovery of lost loads and minimizing the number of switching operations; Second, using virtual power plants (VPPs) as aggregation layers to uniformly evaluate and coordinate the scheduling of distributed resources such as photovoltaic, wind power, biomass power generation, and battery energy storage, thereby enhancing their ability to participate in post-disaster power supply recovery; Third, employing hybrid energy storage systems (HESS) composed of batteries and supercapacitors, utilizing the high power density and long cycle life of supercapacitors to handle high-frequency power fluctuations or bus voltage stabilization tasks, and utilizing the high energy density of batteries to handle continuous energy supply tasks; Fourth, adopting multi-timescale scheduling methods, using fine-grained time-segmentation in the early stages of recovery for rapid response, and coarse-grained time-segmentation in the middle and later stages of recovery to reduce computational burden.

[0004] However, existing technologies still have the following shortcomings in the rapid recovery of critical loads after a disaster: First, existing recovery and dispatch schemes usually treat energy storage as a homogeneous energy resource, failing to distinguish between the step-pickup power demand generated when critical loads transition from shutdown to power restoration and the persistent energy gap demand in the later stages of recovery. This forces batteries to discharge at high rates to support the impact power in the early stages of recovery, limiting the scale of load that can be restored in the early stages and accelerating battery degradation. Second, in existing battery-supercapacitor hybrid energy storage schemes, the division of labor between supercapacitors and batteries is usually only determined at the equipment or bus level (such as frequency division based on power change frequency or segmented control based on bus voltage change rate), without considering whether the supercapacitor can effectively deliver power to specific critical load nodes along available paths when the distribution network topology changes after a fault. In other words, there is a lack of consideration for impedance distance, path... The quantitative evaluation of topology reachability factors such as path flow margin and node voltage margin means that the high power advantage of supercapacitors may not be effectively transformed into support for critical loads due to excessive electrical distance between the nodes or limited flow of candidate paths. Third, in existing schemes, battery degradation protection mostly relies on adding degradation costs related to throughput to the objective function for ex-post economic penalties, without establishing direct constraint feedback between the actual output of supercapacitors and the allowable discharge rate and ramp rate of batteries. The protection mechanism between energy storage has not yet formed a closed loop. Fourth, the time period division of existing multi-timescale scheduling schemes is mostly a fixed division set by experience, which fails to dynamically adjust according to the intensity of disaster impact, the rate of change of impact, and the real-time energy margin of power-type energy storage. Unnecessary fine-grained division is maintained even in periods with low disaster impact intensity or tight energy of power-type energy storage, resulting in waste of computing resources or insufficient energy storage protection.

[0005] In summary, existing technologies lack a technical solution that can decouple the impact power demand for post-disaster critical load recovery from the continuous energy demand, and couple the decoupling result with energy storage topology accessibility evaluation, islanding reconfiguration, energy storage protection, and recovery period division in a closed loop. Therefore, it is urgent to propose a virtual power plant islanding reconfiguration method driven by the impact power spectrum to improve the post-disaster critical load recovery capability of the distribution network and reduce the risk of energy storage degradation without increasing the energy storage capacity. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies, such as the lack of decoupling between impact power and continuous energy demand due to homogenized energy storage processing, the lack of topology accessibility criteria for the division of labor between supercapacitors and batteries, the failure to form a closed loop between battery degradation protection and energy storage division of labor, and the fixed recovery period division method that cannot adapt to the intensity of disaster impact. This invention provides a virtual power plant islanding reconfiguration method driven by the impact power spectrum. This method solves the technical problem of how to decouple the impact power demand caused by critical load recovery from the continuous energy demand under the condition of grid disconnection after extreme disasters in the distribution network, without increasing the energy storage capacity. It involves embedding the decoupling results into supercapacitor topology accessibility evaluation, early critical load recovery and islanding root node selection, battery protection constraints, and recovery period division, thereby improving the critical load recovery rate in the first hour after a disaster and reducing battery degradation during high-rate discharge.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A virtual power plant islanding reconfiguration method driven by impact power spectrum includes the following steps: Step S1: Obtain the faulty branches, switch status, node load demand, load importance and line parameters of the post-disaster distribution network, as well as the operating parameters of the photovoltaic power generation unit, wind power generation unit, biomass power generation unit, battery energy storage unit and supercapacitor unit in the virtual power plant; Step S2: Based on the gap between the node load demand and the available output of photovoltaic power generation units, wind power generation units, and biomass power generation units, calculate the disaster criticality for each time period. Combined with the support capacity indicators of resources other than battery energy storage units and supercapacitor units, use a method combining CRITIC objective weighting and TOPSIS proximity to determine the support capacity priority of each resource in each time period. The disaster criticality is used to determine the values ​​of safety weight and economic weight in the load reduction penalty, and the support capacity priority is used to bias the selection of the islanded root node. Step S3: Calculate the disaster impact power demand of each node in each time period by weighted summing of the three parts: the step pick-up impact power generated by the recovery of critical loads, the power supply gap of basic resources, and the short-term fluctuation of new energy output, and obtain the disaster impact power spectrum. Step S4: Compare the disaster impact power demand of the node during the time period with the power increment that the battery energy storage unit can provide during the time period according to its ramp rate. Take the excess part as the impact power component undertaken by the supercapacitor unit. Take the part of the difference between the node load demand and the available output of basic resources, excluding the impact power component, as the continuous energy component undertaken by the battery energy storage unit. Step S5: Calculate the rapid support reachability coefficient from the supercapacitor unit to each node based on the product of four factors: the support capacity of the supercapacitor unit, the network impedance distance to each node, the power flow margin of the candidate path, and the voltage margin. If and only if there exists a supercapacitor unit whose rapid support reachability coefficient to a certain node is not less than a preset threshold, determine that the node is qualified for early recovery of critical loads. Based on this, generate early recovery constraints for critical loads, supercapacitor power reservation constraints that are dynamically adjusted according to the intensity of disaster impact, and island root node selection bias based on support capacity priority and rapid support reachability coefficient. Step S6: Based on the actual support of the supercapacitor unit, dynamically correct the rate discharge constraint and ramp constraint of the battery energy storage unit. The larger the actual support of the supercapacitor unit, the smaller the allowed rate discharge increment of the battery energy storage unit and the larger the upper limit of power change in adjacent time periods; and construct a battery degradation cost including throughput degradation, high rate degradation and power jump degradation. Step S7: Based on the disaster impact intensity, the rate of change of disaster impact intensity and the energy margin of the supercapacitor unit, determine whether the preset fine-grained division conditions are met, and determine the duration of each recovery period accordingly, so that the division of the initial recovery period is finer than the division of the middle and late recovery periods. Step S8: Establish and solve the joint islanding reconfiguration optimization model: Set radial topology constraints, single-commodity virtual flow connectivity constraints, and branch power flow and node voltage constraints for the distribution network; set early recovery constraints for critical loads generated in Step S5, supercapacitor power reservation constraints, and islanding root node selection bias, as well as battery energy storage unit constraints corrected in Step S6; construct an objective function that includes load reduction penalties weighted by the disaster criticality and load importance, battery degradation costs, resource operating costs, switching operation costs, and accessibility risk penalties mentioned in Step S6, and give a reward on the objective function to nodes where resources with higher support capacity priority are located as islanding root nodes; solve the joint islanding reconfiguration optimization model to obtain the branch opening and closing status, islanding root nodes, node load recovery amounts, and output commands of the supercapacitor units and battery energy storage units for each time period, and control the operation of switching equipment, supercapacitor units, and battery energy storage units in the distribution network according to the output commands.

[0008] Furthermore, it also includes: during the configuration phase, gradually increasing the configuration capacity of the supercapacitor unit, calculating the marginal increase in the critical load recovery rate in the first hour after the disaster as the configuration capacity increases, and when the marginal increase is less than a preset threshold, determining the configuration capacity at this time as the configuration capacity of the supercapacitor unit to guide the planning and configuration of the supercapacitor unit.

[0009] Preferably, the calculation of disaster criticality in step S2 adopts a continuous function form, and at the same time integrates the disaster impact intensity information defined in step S3, so that the value of the safety weight can respond to the indicators of both disaster criticality and disaster impact intensity, thereby improving the matching degree between the weight setting and the disaster situation.

[0010] Preferably, the disaster impact power spectrum aggregation index in step S3 is reused in the supercapacitor reservation constraint in step S5 and the adaptive time scale criterion in step S7 to achieve the unification of node-level impact identification and system-level impact intensity evaluation, and avoid the deviation of key load recovery strategy caused by single-dimensional evaluation.

[0011] Preferably, the rapid support reachability coefficient in step S5 adopts a multi-path optimization strategy. When there are multiple feasible power supply paths between the supercapacitor unit and the critical load, the reachability coefficient of each path is calculated and the optimal value of each path is substituted into the calculation. This reflects the role of network redundancy in improving reachability and enhances the reliability of early recovery decisions for critical loads.

[0012] Preferably, the impedance distance attenuation factor in step S5 adopts a negative exponential function form. The larger the impedance distance, the smaller the attenuation factor. The power flow margin of the candidate path adopts the minimum value of the ratio of the capacity margin of each branch. The voltage margin of the candidate path adopts the maximum value of the voltage deviation from the median of each node. The three factors together with the support capability evaluation value constitute the four product factors of the rapid support reachability coefficient. If any factor is too low, it will significantly reduce the reachability coefficient, reinforcing the design concept that high power does not equal effective support.

[0013] Preferably, in step S6, the battery degradation cost model replaces the high-rate degradation term and the power jump degradation term with other model forms that reflect the nonlinear degradation law of the battery. For example, a cycle life calculation model based on rainflow counting method or a lithium battery calendar degradation and cycle degradation coupling model based on semi-empirical formula can be used, as long as the degradation model can receive the actual support amount of the supercapacitor unit as a correction input and maintain a closed-loop relationship.

[0014] Preferably, the adaptive time scale division in step S7 is extended to a multi-level setting, including three levels: fine-grained, medium-grained, and coarse-grained. The time period of each level corresponds to a different interval of the disaster impact intensity. The level switching logic is consistent with the preset conditions, and only the binary classification threshold is extended to a multi-level threshold to improve the flexibility of time scale adjustment.

[0015] Preferably, the adaptive time scale criterion in step S7 further introduces an online update mechanism based on the rolling time domain. After each decision period is completed during the recovery process, the duration of subsequent periods is recalculated and the optimization model is resolved based on the actual disaster impact intensity, the actual remaining energy of the supercapacitor unit, and the load recovery progress, in order to cope with uncertainties such as disaster evolution, fluctuations in new energy output, and changes in equipment status.

[0016] Preferably, the joint islanding reconfiguration optimization model in step S8 adopts a strategy of using the second-order cone relaxation branch power flow model and the linearized branch power flow model in parallel. First, the initial reconfiguration scheme is obtained by quickly solving the linearized model. Then, the second-order cone relaxation model is substituted to check whether the voltage and current meet the nonlinear power flow constraints. If the check fails, the over-limit constraint is added to the linearized model as a cutting plane feedback and solved again. The iteration continues until the requirements of linearization efficiency and nonlinear feasibility are met simultaneously.

[0017] Preferably, when the scale of the distribution network is large, which leads to a significant increase in the solution scale of the mixed integer linear programming model, step S8 adopts a distributed solution method to decompose the global optimization problem into multiple sub-region optimization problems for parallel solution, or adopts a column generation method to implicitly enumerate the candidate island topology set to reduce the dimensionality of integer variables, thereby significantly shortening the computation time while ensuring the solution quality.

[0018] Preferably, the rapid support reachability coefficient in step S5 introduces an online power flow update mechanism. Before solving each decision period, the power flow margin and voltage margin of each branch are updated based on the power flow results of the previous period. This allows the reachability coefficient to dynamically reflect the impact of island topology changes and load recovery processes on the supercapacitor support reachability, avoiding deviations between static calculations and actual operating conditions.

[0019] Preferably, the supercapacitor unit adopts a carbon-based supercapacitor device, which has the characteristics of high power density, long cycle life and good low temperature performance, which matches the functional positioning of the impact power support of the present invention.

[0020] Preferably, the configuration capacity of the supercapacitor unit is gradually increased during the configuration phase. The marginal increase in the critical load recovery rate in the first hour after the disaster is calculated as the configuration capacity increases. When the marginal increase is less than a preset threshold, the configuration capacity at this time is determined as the configuration capacity of the supercapacitor unit. This is used to guide the planning and configuration of the supercapacitor unit, avoiding insufficient capacity configuration that would prevent the impact power from being effectively absorbed, and also avoiding over-configuration that would lead to investment waste.

[0021] This invention also provides a virtual power plant islanding reconfiguration system driven by impact power spectrum, including a data acquisition module, a resource evaluation module, an impact power spectrum construction module, a demand decomposition module, a reachability evaluation module, a battery protection module, an adaptive time scale module, an optimization solution module, and a control execution module. Each module is used to implement the functions corresponding to steps S1 to S8 of the above method. The output of the impact power spectrum construction module is processed by the demand decomposition module and then input to the reachability evaluation module and the battery protection module. The output of the reachability evaluation module is also fed back to the battery protection module and the adaptive time scale module. The outputs of the battery protection module and the adaptive time scale module are jointly input to the optimization solution module, forming a closed-loop collaborative structure of disaster impact identification, energy storage function division, topology reachability support, energy storage protection feedback, and time scale adaptation.

[0022] The present invention also provides a computer device and a computer-readable storage medium for implementing the above-described method in a programmatic manner.

[0023] Compared with the prior art, the present invention has the following beneficial effects: 1. By decomposing the disaster impact power spectrum into the impact power component carried by supercapacitors and the continuous energy component carried by batteries, the two types of energy storage can assume roles that match their own power / energy characteristics. Without increasing the total energy storage capacity, the critical load recovery rate in the first hour after a disaster can be increased from about 29.7% of the baseline of undifferentiated dispatch to more than 80%, and the overall load recovery rate can be improved simultaneously. The recovery effect is supported by measured data. 2. By introducing a rapid support reachability factor for supercapacitors, impedance distance, path power flow margin, and voltage margin are incorporated into the evaluation of supercapacitors' support capability for critical loads. This avoids misjudging supercapacitors as effective support resources due to excessive electrical distance or limited path power flow at the nodes where they are located. Compared to sorting by equipment rated power alone, this method can more accurately screen energy storage resources that can truly support critical loads, improving the reliability of island root node selection and early recovery decisions for critical loads. 3. By feeding back the actual support of the supercapacitor to the battery's rate discharge constraint, ramp-up constraint, and degradation cost model, a protective closed loop is formed between energy storage systems. Compared with the scheme that does not distinguish between impact / continuous demand under the same energy configuration, it can significantly reduce the battery's high-rate discharge peak and cycle degradation cost while maintaining a similar recovery level. 4. The CRITIC-TOPSIS dynamic resource evaluation mechanism driven by disaster criticality can dynamically adjust the safety and economic weights and resource support priorities according to the evolution of the disaster situation, compared with the fixed weight ranking method, thereby improving the matching degree between resource allocation decisions and the actual disaster situation. 5. The adaptive time scale division driven by the disaster impact intensity, impact change rate and supercapacitor energy margin can, compared with the fixed time period division method set by experience, ensure the accuracy of rapid response in the early stage of recovery, while automatically switching to coarse-grained division in the period when the disaster impact weakens or the energy of power storage becomes tight, thus taking into account both recovery effect and calculation efficiency. 6. The above-mentioned technical features are synergistically coupled and interlocked in the same joint island reconfiguration optimization model. The overall technical effect of "simultaneously improving the first-hour recovery rate and reducing battery degradation without increasing the energy storage capacity" is something that cannot be achieved by using any one of the existing virtual power plant resource evaluation technology, battery-supercapacitor hybrid energy storage technology, distribution network island reconfiguration technology or multi-timescale scheduling technology alone. Attached Figure Description

[0024] Figure 1 This is an overall flowchart of the virtual power plant islanding reconfiguration method driven by the impact power spectrum of the present invention; Figure 2 This is a schematic diagram illustrating the principle of disaster impact power spectrum construction and impact / sustained demand decomposition in this invention; Figure 3 This is a schematic diagram of the constituent elements of the achievable coefficient of rapid support for supercapacitors in this invention; Figure 4 This is a block diagram of the closed-loop feedback relationship for correcting the protection constraints of the battery energy storage unit based on the actual support amount of the supercapacitor in this invention. Figure 5 This is a schematic diagram of the adaptive time scale criterion driven by disaster impact intensity in this invention; Figure 6 This is a block diagram of the virtual power plant islanding reconfiguration system driven by the impact power spectrum of the present invention. Figure 7 This is a schematic diagram of post-disaster distribution network island topology reconstruction in an embodiment of the present invention; Figure 8 This is a comparison chart of recovery rates in ablation scenarios according to embodiments of the present invention; Figure 9 This is a comparison chart of the battery energy storage protection effects in embodiments of the present invention; Figure 10 This is a diagram showing the marginal benefit of supercapacitor configuration capacity in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only one preferred embodiment of this invention and are only used to explain this invention. They do not limit the scope of protection of this invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0026] Example 1: As Figures 1-10 As shown, Figure 1 As shown, a virtual power plant islanding reconfiguration method driven by the impact power spectrum includes steps S1 to S8. For ease of description, the sets and variables involved in this embodiment are first uniformly explained: Let the distribution network node set be... The set of branches is The set of faulty branches is The set of critical load nodes is The set of recovery periods is The virtual power plant integrates basic resources (photovoltaic power generation units, wind power generation units, biomass power generation units, etc.) other than battery energy storage units and supercapacitor units. The battery energy storage unit is assembled as The supercapacitor unit is assembled as ;node During the period The load demand is denoted as Basic resources During the period The available output is denoted as .

[0027] In step S1, the set of post-disaster fault branches is obtained through the unified data interface between the distribution network dispatch automation system and the virtual power plant energy management system. Current status of each tie switch and section switch, and load demand at each node. Its importance level and the resistance of each branch Reactance and capacity limit And the available output of photovoltaic power generation units, wind power generation units, and biomass power generation units within the virtual power plant. The data includes the rated power, capacity, current state of charge, and ramp rate of the battery energy storage unit, as well as the rated power, capacity, and current remaining energy of the supercapacitor unit. The data is preprocessed by timestamp alignment, missing value imputation, and normalization to form a standardized dataset for use in subsequent steps.

[0028] In step S2, the equivalent load deficit for each time period is first calculated: (1) In the formula, For positive part operators; For nodes During the period Load demand; Basic resources During the period Available output; For distribution network nodes; This is the set of recovery periods. Based on this, the disaster criticality is calculated: (2) In the formula, To avoid small positive numbers with a denominator of zero, and The larger the value, the more severe the gap between the available power output of the distribution network and the load demand during that period. This is based on the severity of the disaster. The level range in which the load reduction penalty is located determines the safety weight in the load reduction penalty. Economic weight : (3) In the formula, , As a preset level threshold, this embodiment takes... , As an optional further implementation, the security weights can also be in the form of a continuous function: (4) In the formula, The disaster impact intensity defined in step S3, The coupling coefficient is used to ensure that the weights respond simultaneously to both the severity of the disaster and the intensity of the disaster impact. ; Minimum security weight; This is the maximum security weight.

[0029] After obtaining the safety / economic weights, this embodiment uses a combination of CRITIC objective weighting and TOPSIS proximity to calculate the support capability priority of dispatchable resources other than battery energy storage units and supercapacitor units (such as load shelving, demand response resources, and non-black start units). The response time, ramp rate, continuous power supply time, and topological correlation with critical loads of each resource are used as evaluation indicators. The indicator matrix is ​​first normalized, and then the objective weights are calculated according to the comparative strength and conflict of the indicators using the CRITIC method. ( (assign indicator numbers), and then calculate each resource Euclidean distance to positive and negative ideal solutions , Prioritize the support capabilities: (5) In the formula, The range of values ​​is The larger the value, the more resources are available. During the period The higher the priority of the support capability; The bias used in step S8 for selecting the root node of the island will be applied. (Refer to...) Figure 3 The accessibility logic shown above applies to the supercapacitor unit, which also calculates its support capability evaluation value using the same method. ( ), which is one of the constituent factors of the rapid support reachability coefficient in step S5.

[0030] In step S3, such as Figure 2 As shown, this embodiment will use nodes During the period Disaster impact power demand The weighted sum is constructed as follows: (6) In the formula, The preset weighting coefficients are used in this embodiment. , , ; For nodes The step pickup impulse power generated when switching from a shutdown state to power restoration is calculated as follows: (7) In the formula, For nodes During the period Should power be restored? In order to restore power, For nodes During the period relative time period From power loss to power supply; For nodes The apportionment of the power supply gap in the region's basic resources at the node level: (8) In the formula, For nodes Basic resources such as photovoltaic, wind, and biomass that can be allocated can be used for power output; For nodes The fluctuation in the output of the connected new energy source between adjacent time periods.

[0031] The three parts correspond to "the instantaneous impact triggered by the recovery action", "the persistent power shortage", and "the short-term disturbance caused by the uncertainty of new energy output". The weighted sum of the three parts yields a unified disaster impact power spectrum that can be used for subsequent decomposition. .

[0032] Furthermore, to characterize the intensity of disaster impact at the entire distribution network level, the following aggregated index is defined: (9) The supercapacitor reservation constraint in step S5 and the adaptive time scale criterion in step S7 are reused to achieve the unification of node-level impact identification and system-level impact intensity evaluation.

[0033] In step S4, such as Figure 2 As shown, this embodiment decomposes the disaster impact power spectrum into the impact power component borne by the supercapacitor and the continuous energy component borne by the battery, based on the available ramp-up capability of the battery energy storage unit. Let the node... Electrically related battery energy storage unit collection (i.e., topologically, nodes can be reached directly or via the shortest path) A collection of battery energy storage units that provide power during a specific time period. The incremental polymerization power that can be provided by the ramp rate is ,in Battery energy storage unit The rate of ascent, For time period The duration. Based on this, the impact power components are calculated: (10) That is, when the power demand due to a disaster exceeds the ramp-up response capability of the battery energy storage unit, the excess is borne by the supercapacitor unit. The continuous energy component is the portion of the power supply gap in the node's basic resources excluding the impact power component: (11) and The supercapacitor unit and the battery energy storage unit are respectively used as inputs to steps S5 and S6, driving them to assume roles that match their own power and energy characteristics. This is one of the core features that distinguishes this invention from existing technologies that treat energy storage as a homogeneous energy resource.

[0034] In step S5, such as Figure 3 As shown, this embodiment uses a supercapacitor unit. To node The rapid support reachability coefficient is constructed as the product of the following four factors. The first factor is the support capacity evaluation value obtained in step S2. ; The second term is the network impedance distance attenuation factor, let's assume... Supercapacitor Unit To node Given the set of branches traversed by the candidate path, the impedance distance is: (12) In the formula, The reactance reduction factor is taken as the impedance distance attenuation factor. The greater the impedance distance, the smaller the attenuation factor; The third item is the power flow margin of the candidate path: (13) In the formula, branch road Maximum capacity branch road During the period The trend; The fourth item is the voltage margin of the candidate path: (14) In the formula, For the nodes the path passes through During the period voltage amplitude, , These are the upper and lower limits of the node voltage, respectively.

[0035] Considering the above four factors, the rapid support attainment coefficient is: (15) The range of values ​​is Any factor that is too low will significantly reduce the reachability coefficient, reflecting the design principle that "high power does not equal effective support." Nodes are defined accordingly. During the period Critical load early recovery eligibility criteria: (16) In the formula, This is a preset reachability threshold. The critical load early recovery constraint is: for any... ,node During the period The load recovery amount shall not exceed each of the requirements. The sum of the power that the supercapacitor cells can provide, weighted by the availability factor, and when Time node During the period It does not qualify for early recovery via supercapacitor units. Furthermore, this embodiment is based on the intensity of the disaster impact. Set dynamic power / energy reservation constraints for supercapacitor units: (17) In the formula, Supercapacitor Unit During the period The remaining energy, For its rated capacity, To reserve a coefficient, It is a normalization constant; the higher the intensity of the disaster impact, the higher the proportion of energy that the supercapacitor unit is required to reserve, in order to ensure its impact response capability in subsequent periods.

[0036] Rapid support reach coefficient Priority of support capabilities (including) Together, they constitute the island root node selection bias: Among the candidate root nodes, feeder segments containing nodes with high support capacity priority and high critical load reachability coefficient are preferentially selected as island root nodes. This bias will be reflected as a reward term in the objective function of step S8.

[0037] In step S6, such as Figure 4 As shown, this embodiment feeds back the actual support of the supercapacitor unit to the protection constraints of the battery energy storage unit, forming a closed loop. Let's assume the battery energy storage unit... Electrically related supercapacitor unit collection is Define the battery rate discharge correction factor: (18) In the formula, Supercapacitor Unit During the period The actual discharge power, For its rated power, , These are the preset upper and lower limits of the correction coefficient; the higher the actual supported capacity of the supercapacitor unit relative to its rated power, the better. The closer The smaller the allowable rate of discharge increment for the battery energy storage unit, the better; conversely, when the actual support capacity of the supercapacitor unit is zero, the better. Take the maximum value Based on this, the rate discharge constraint of the battery energy storage unit is constructed: (19) In the formula, Battery energy storage unit It bears the upper limit of the reference power corresponding to the continuous energy component. This refers to the upper limit of the surge power that can be temporarily added when the supercapacitor unit is insufficient. Similarly, the upper limit of the change in discharge power of the battery energy storage unit between adjacent time periods is also adjusted according to the actual support capacity of the supercapacitor unit: (20) In the formula, As a ramp-up correction factor, the larger the actual support capacity of the supercapacitor unit, the greater the upper limit of the allowable power change of the battery energy storage unit. In other words, the more impact power the supercapacitor unit shares, the smaller the power jump pressure the battery needs to withstand. Based on this, a battery degradation cost is constructed, comprising three parts: throughput degradation, high-rate degradation, and power jump degradation. (twenty one) In the formula, Battery energy storage unit During the period The charging power; For its rated power, , , C b These are the three degradation cost coefficients for throughput, high rate, and power jump, respectively.

[0038] Compared to using only the degradation cost that is linearly related to throughput, the high-rate degradation term and the power jump degradation term in Equation (21) can more accurately reflect the physical law of accelerated aging of the battery under impact conditions, and together with the constraint corrections in Equations (18) and (20), they form an active protection closed loop for the battery.

[0039] In step S7, such as Figure 5 As shown, this embodiment determines whether to use fine-grained or coarse-grained recovery period division based on the disaster impact intensity, its rate of change in adjacent time periods, and the collective average energy margin of the supercapacitor units. The collective average energy margin of the supercapacitor units is defined as follows: (twenty two) The duration of the time period is determined according to the following criteria: (twenty three) In the formula, The preset high intensity threshold, The preset rate of change threshold, This is a preset low margin threshold. In this embodiment, we take... , The fine-grained time period is 15 minutes, which matches the characteristic time scale of critical load step pick-up impact and rapid evolution of disaster, and is used to update the output commands of supercapacitor units and battery energy storage units in a timely manner to avoid power and energy exceeding limits; the coarse-grained time period is 1 hour, which corresponds to the middle and late stage of recovery when the disaster is stable and energy balance is the main focus, and reduces the solution scale of the optimization model while ensuring scheduling accuracy.

[0040] The three triggering conditions in Equation (23) correspond to three scenarios that require encrypted decision-making frequency: “the disaster impact itself is strong”, “the disaster situation is changing rapidly”, and “power-type energy storage energy is becoming tight and more refined scheduling is needed to avoid exceeding the limit”. The three are connected by “or” logic, that is, if any one of the conditions is met, fine-grained division is adopted, so as to realize the adaptive adjustment of the granularity of the recovery period according to the disaster situation and energy storage status, which is different from the experience division method of fixed 15 minutes in the early stage of recovery and fixed 1 hour thereafter.

[0041] In step S8, a joint island reconstruction optimization model is established on a unified time-scale grid, which is divided into the following four sub-steps.

[0042] Step S8-1: Set topology and power flow related constraints: Introducing branch opening and closing state variables ( Indicates a branch (Closed) and node energized state variables The single-product virtual flow method is used to ensure connectivity within isolated islands and prevent the formation of loops: (twenty four) In the formula, branch road Virtual stream on For sufficiently large positive numbers, The root node marker for the isolated island (at the root node) Take the larger positive value, and zero at non-root nodes). branch road Virtual streams on Represented by node For the beginning and the node (For the terminal branch). For nodes During the period The energized state variable, taking a value of 1 to represent the node During the period The device is in a powered-on state; a value of 0 indicates that it is not powered on. The summation index indicates that all nodes are traversed. Branches for end nodes The first node Accordingly, This indicates traversing all nodes. Branches for the first node end node Equation (24) ensures that each energized node has a unique virtual upstream path, thus implying a radial and connected topology; for faulty branches Forced Branch power flow and node voltage constraints are handled using a linearized branch power flow model: (25) In the formula, For nodes During the period The amount of load reduction, For nodes The sum of the net injected power from basic resources, battery energy storage units, and supercapacitor units. , Branch roads Branch roads During the period Transmitted active power, branch road During the period Transmitted reactive power For nodes During the period The load demand (i.e., the aforementioned) In the node index (values ​​at time) , They are nodes ,node During the period voltage amplitude, , These represent the lower and upper limits of the allowable node voltage, respectively. , , Branch roads The resistance, reactance, and capacity upper limit (consistent with the branch parameters obtained in step S1), Equation (25) second equation when the branch is disconnected ( When the voltage drop constraint is relaxed by using a large M constant, the voltage drop constraint can be relaxed. As a further alternative implementation, a second-order cone-relaxed branch power flow model can be used in parallel to check the feasibility of voltage and current based on the linear model of equation (25). The two can verify each other, taking into account both solution efficiency and physical accuracy.

[0043] Step S8-2: Set the constraints generated in steps S5 and S6: The key load early recovery constraint in step S5, the supercapacitor power / energy reservation constraint in equation (17), and the rate discharge constraint in step S6, the ramp constraint in equation (19), the state of charge / remaining energy constraint of the supercapacitor unit and the battery energy storage unit are all included in the constraint set; in addition, a no-load charging constraint is set to avoid consuming power to charge the energy storage during periods when the load itself is reduced: (26) In the formula, Indicates time period Is there load reduction? , It is a sufficiently large positive number; For nodes During the period The load reduction amount, and the amount in equation (25) For the same variable (the difference is that here it is...) (As a node index, the same applies below); , These are battery energy storage units Supercapacitor Unit During the period The charging power, of which The variable with the same name as the variable in equation (21) is the same variable.

[0044] Step S8-3: Construct the objective function: (27) (28) In the formula, For nodes The importance weight of the load, The time period described in step S2 The criticality of the disaster For nodes During the period The load reduction amount (which is the same variable as the load reduction amount in equations (25) and (26)). Basic resources During the period Actual output power (not exceeding its available output power) ), , These are the operating cost coefficients for basic resources and battery energy storage units, respectively. branch road The switching operation cost coefficient, , Branch roads During the period Compared to the previous period The open and closed states, and the absolute value of the difference between them are used to measure the branch. Switching actions between adjacent time periods Battery energy storage unit During the period The degradation cost generated internally, i.e., the degradation cost described in equation (21). corresponding time period The amount, For those who do not meet the early reinstatement eligibility criteria ( The critical loads still face the risk of accessibility penalties for impact power demand. For nodes Prioritize the support capabilities of the resources connected to the node (take the resources connected to the node) or (maximum value) Choose a reward coefficient for the root node. The bias of "nodes with high support capability priority and high reachability coefficient are more suitable as the root node of an island" is reflected in the form of rewards. For load reduction penalties, This is the cost of battery degradation. The cost term for the switching operation. The objective function formed by equations (27) and (28), together with equations (24)-(26) and the constraints of steps S5 and S6, constitute a mixed-integer linear programming (MILP) model.

[0045] Step S8-4: Solve the above model using a commercial or open-source solver (such as Gurobi, CPLEX, SCIP, etc.) to obtain the branch opening and closing states for each time period. Isolated root node, node load recovery amount And the output commands of the supercapacitor unit and the battery energy storage unit. , The dispatch center generates switch opening and closing commands and energy storage output commands based on the above solution results and sends them to the distribution network automation terminal and the virtual power plant energy management system for execution.

[0046] As a further implementation method in the planning phase, this embodiment also provides a method for configuring the capacity of supercapacitor cells: Let This indicates that the total configuration capacity of the supercapacitor unit is At that time, based on the critical load recovery rate obtained from steps S1 to S8 above, the configured capacity is gradually increased and the marginal increase is calculated: (29) when ( When the preset marginal improvement threshold is reached, the capacity at this time will be... The configuration capacity of the supercapacitor unit is determined to guide the planning and configuration of the supercapacitor unit, so as to avoid insufficient capacity configuration that would result in the inability to effectively absorb the impact power, and also to avoid over-configuration that would lead to investment waste.

[0047] To verify the effectiveness of the method of this invention, this embodiment provides a case study on an improved IEEE 33-node distribution network feeder. The case study uses the IEEE 33-node distribution network as the basic feeder and adds virtual power plant resources, fault branches, tie switches, and critical load settings in a post-disaster islanding recovery scenario. Specific basic conditions are shown in Table 1, and the typical post-disaster islanding topology is as follows: Figure 7 As shown.

[0048] Table 1. Basic parameters of Example 1

[0049] Based on the parameters in Table 1, step S2 calculates the disaster criticality during the initial stage of the disaster (18:00-19:00). The high-level range of equation (3) is included, and the safety weight is 0.95. The weights of the four indicators obtained by CRITIC weighting are: response time 0.244, ramp rate 0.266, continuous power supply time 0.288, and topology correlation 0.202. The TOPSIS proximity of each resource in the early stage of the disaster is shown in Table 2. Among them, the two supercapacitor units have the highest proximity, indicating that their fast response and high ramp rate characteristics are preferentially identified in the high disaster criticality state.

[0050] Table 2 Priority of Resource Support Capacity in the Initial Stage of Disaster

[0051] According to steps S3-S4, the step pick-up impact caused by the recovery of critical loads, the power supply gap of basic resources, and the fluctuation of new energy sources together form the disaster impact power spectrum. When the impact power demand exceeds the ramp-up increment that the battery energy storage unit can provide within the corresponding period, the excess is decomposed into an impact power component borne by the supercapacitor unit, and the remaining basic resource gap is borne by the battery energy storage unit for continuous energy supply. This decomposition allows for an explicit distinction between the high-power short-term demand in the initial stage of recovery and the subsequent continuous energy supply demand in the model.

[0052] The rapid support availability factor from the two supercapacitor units to the critical load node was calculated according to step S5, and the results are shown in Table 3. As can be seen from Table 3, simply having similar connection locations does not necessarily guarantee support availability; impedance distance, path power flow margin, and voltage margin all jointly determine whether the requirements are met. Early restoration of eligibility.

[0053] Table 3. Availability Factors for Rapid Support from Supercapacitors to Critical Load Nodes

[0054] Based on Table 3, nodes 8, 7, and 14 are covered by CSC-1, nodes 29 and 32 are covered by CSC-2, and node 4 is covered by both CSC-1 and CSC-2. Therefore, all six critical load nodes have at least one fast support path that meets the threshold. This result directly supports the critical load early recovery constraint, supercapacitor reservation constraint, and island root node bias in step S5. According to step S7, the adaptive time scale criterion calculates 18:00, 18:15, 18:30, 18:45, 19:00, and 20:00 using 15-minute fine-grained or continuous fine-grained solutions, and switches to 1-hour coarse-grained solutions after 21:00; the average energy margin of the supercapacitor remains above 0.965, and the low margin protection condition is not triggered.

[0055] To verify the contributions of each technical module, this embodiment sets up six ablation scenarios as shown in Table 4. Each scenario is solved under the same load, fault, and resource access conditions, with only the use of disaster criticality-driven evaluation, supercapacitor functional division, and multi-timescale recovery period division being changed.

[0056] Table 4 Ablation Scenario Settings

[0057] Table 5 and Figure 8 The recovery rate, overall cost, and battery protection metrics for six scenarios are presented. The complete solution S5 achieves a critical load recovery rate of 80.1% in the first hour, significantly higher than S0's 29.7%, S2's 67.7%, S3's 63.5%, and S4's 47.7%; the overall load recovery rate reaches 83.6%, and the overall cost is 11,313 yuan, a 40.3% reduction compared to S0.

[0058] Table 5 Summary of ablation scenario simulation results

[0059] From Table 5 and Figure 8 It is evident that the S5 reduces battery degradation costs from 193 yuan to 145 yuan compared to the S0, a decrease of 24.9%; compared to energy battery solutions such as the S4, the battery degradation cost decreases from 219 yuan to 145 yuan, a decrease of 33.8%, and the battery throughput decreases from 1459 kWh to 969 kWh, a decrease of 33.6%. This indicates that the effect of the S5 does not come from simply increasing energy storage capacity, but from the supercapacitor's ability to withstand high power surges in short periods, and the feedback protection provided by this capacity against battery rate, ramp-up, and degradation constraints.

[0060] From the perspective of module contributions, S1 improved the critical load recovery rate in the first hour by 19.3 percentage points compared to S0, indicating that resource assessment driven by disaster criticality has an independent contribution; S5 improved by 12.4 percentage points compared to S2, indicating that fixed weights cannot fully respond to the critical load recovery needs in the early stages of a disaster; S5 improved by 16.6 percentage points compared to S3, indicating that fine-grained scheduling at the 15-minute level in the first hour helps to capture step-picking shocks and battery ramp-up transitions; S5 improved by 32.4 percentage points compared to S4, indicating that the power characteristics of supercapacitors cannot be simply replaced by equivalent-energy batteries. These results collectively demonstrate that there is a synergistic effect among disaster impact identification, energy storage function division, topology reachability support, energy storage protection feedback, and time-scale adaptation.

[0061] During the recovery process of the complete S5 solution, the island topology and root node are dynamically adjusted as the disaster situation evolves. In the initial recovery phase (18:00-19:00), the optimized model forms two main islands with node 8 and node 29 as root nodes. These two nodes are equipped with both battery storage units and supercapacitor units, and their TOPSIS proximity ranking is the top two. For example... Figure 7 As shown, at 19:00, the typical topology closes tie switches 8-21 and 25-29, connecting the feeder segments isolated due to permanent faults; nodes 8 and 29 act as root nodes, providing rapid support to nearby critical loads. During the recovery phase, resource nodes such as nodes 3, 14, 25, and 30, which are equipped with biomass power generation units, temporarily participate in the role of root nodes; in the later recovery phase, as photovoltaic output recovers, the isolated nodes gradually merge. This topology evolution demonstrates the adaptive response capability of the method of this invention to changes in disaster situation, resource availability, and load fluctuations.

[0062] As a verification step during the configuration phase, this embodiment also scans the total configured power of the supercapacitor, and the results are shown in Table 6 and... Figure 10 As shown, the critical load recovery rate in the first hour increases monotonically with the increase of supercapacitor capacity, but the marginal benefit gradually decreases: the benefit is relatively high in the 0-1400kW range, and it reaches a saturation inflection point of about 95% near 2800kW. When it continues to increase to 11200kW, the critical load recovery rate in the first hour approaches 100%, but the marginal increase drops to 0.6 percentage points.

[0063] Table 6. Supercapacitor Capacitance Scan Results

[0064] To verify the correctness of the simulation platform's behavior before the introduction of supercapacitor expansion, this embodiment also reproduced the original distributed resource recovery method without supercapacitors, and the results are shown in Table 7. Configuration A is the complete original method, configuration B removes the comprehensive resource evaluation, configuration C uses fixed weights, and configuration D uses a single 1-hour time scale. The reproduction results show that after removing the comprehensive evaluation in configuration B, the overall recovery rate drops to 68.8%, and the overall cost rises to 21,950 yuan, indicating that distributed resource scheduling without unified evaluation leads to insufficient utilization of rapid response capacity. The order of configurations A, C, and D also conforms to the expectations of the original method, indicating that the basic recovery logic of the simulation platform is reproducible.

[0065] Table 7. Reproduction results of the original baseline method

[0066] The above graphical verification results demonstrate that the disaster criticality assessment, impact / sustained demand decomposition, supercapacitor rapid support accessibility, battery protection feedback, and adaptive time scale division described in this invention all have clear data support. In particular, compared to S0, S5 increases the first-hour critical load recovery rate from 29.7% to 80.1% without increasing the total energy storage capacity, while simultaneously reducing battery degradation costs from 193 yuan to 145 yuan. This fully illustrates that the innovative purpose of this invention is to simultaneously optimize post-disaster recovery effectiveness and reduce battery degradation risk through impact power spectrum-driven energy storage functional division and topology accessibility constraints.

[0067] like Figure 6 As shown, corresponding to the above method, this embodiment also provides a virtual power plant islanding reconfiguration system driven by the impulse power spectrum, including: The system comprises the following modules: a data acquisition module for executing step S1; a resource evaluation module for executing step S2; an impact power spectrum construction module for executing step S3; a demand decomposition module for executing step S4; an accessibility evaluation module for executing step S5; a battery protection module for executing step S6; an adaptive time scale module for executing step S7; an optimization solution module for executing step S8; and a control execution module for generating and issuing switching action commands and energy storage output commands based on the output of the optimization solution module. The output of the impact power spectrum construction module (…) , After processing by the demand decomposition module, the resulting impact power components are... Input accessibility evaluation module, continuous energy component The actual support capacity of the supercapacitor unit output from the accessibility evaluation module is input into the battery protection module; the output of the accessibility evaluation module is also related to the disaster impact intensity. The energy margin of the supercapacitor unit is fed back to the adaptive time-scale module; the outputs of the battery protection module and the adaptive time-scale module are ultimately merged into the optimization solution module for joint solution. These modules can be implemented in hardware such as field-programmable gate arrays, digital signal processors, microcontrollers, or general-purpose computers, or as software functional modules running on a distribution network dispatch automation system or a virtual power plant energy management system. The modules exchange data via internal data buses or industrial Ethernet.

[0068] Example 2

[0069] This embodiment provides several alternative implementation methods based on Embodiment 1 to demonstrate the flexibility of the scope of protection of this invention.

[0070] Firstly, the adaptive time scale criterion described in equation (23) can be expanded to multiple levels with more than two levels: for example, setting three levels: fine-grained (15 minutes), medium-grained (30 minutes), and coarse-grained (1 hour), based on the intensity of the disaster impact. The different intervals correspond to different time periods of different gears. The gear switching logic is the same as that of equation (23), except that the binary classification threshold is extended to a multi-level threshold.

[0071] Secondly, the branch power flow model described in Equation (25) can be used in parallel with the second-order cone relaxation branch power flow (SOC) model: firstly, the initial reconstruction scheme is obtained by quickly solving the linear model of Equation (25), and then the voltage and current are substituted into the SOC relaxation model to check whether they meet the nonlinear power flow constraints. If the check fails, the over-limit branch / node constraints given by the SOC model are added to the model described in Equation (25) as cutting plane feedback and solved again. This process is iterated until the requirements of linearization efficiency and nonlinear feasibility are met simultaneously.

[0072] Third, candidate paths between supercapacitor units and critical loads. Not limited to a single path, when there are multiple feasible power supply paths in the distribution network, the index corresponding to each candidate path can be calculated separately, and the optimal value of each path can be substituted into equation (15) to reflect the effect of network redundancy on accessibility.

[0073] Fourth, the supercapacitor unit described in this invention is not limited to carbon-based supercapacitor devices, but may also employ other types of double-layer capacitors or hybrid supercapacitor devices; as a preferred embodiment, the supercapacitor unit adopts a carbon-based supercapacitor device, which has the characteristics of high power density, long cycle life and good low-temperature performance, and matches the functional positioning of the impact power support of this invention.

[0074] Fifth, the calculation method for the support capability priority in step S2 is not limited to CRITIC-TOPSIS. It can also be replaced by other multi-index comprehensive evaluation methods such as the Analytic Hierarchy Process (AHP) and the entropy weight method, as long as it can output the resource support capability priority that can be used to bias the selection of the island root node.

[0075] Sixth, the high-rate degradation term and power jump degradation term in the battery degradation cost model described in equation (21) can also be replaced with other model forms that reflect the nonlinear degradation law of the battery, such as the cycle life calculation model based on the rainflow counting method or the lithium battery calendar degradation-cycle degradation coupling model based on the semi-empirical formula. As long as the degradation model can receive the actual support amount of the supercapacitor unit as the correction input, the closed-loop relationship that the larger the support amount of the supercapacitor unit, the lower the battery degradation risk assessment value can be maintained.

[0076] Seventh, the method described in this invention is not limited to the IEEE 33-node distribution network, but can also be applied to larger-scale distribution networks (such as IEEE 69-node, IEEE 123-node, or actual engineering distribution network feeders). In this case, the set dimension in each step increases accordingly, and the solver can use distributed solving or column generation and other acceleration methods to ensure solution efficiency. The disaster fault scenario is not limited to the three branch permanent faults in Example 1, but can be extended to different locations, different numbers of fault combinations, and different disaster intensity levels. The technical process of the method in this invention remains consistent.

[0077] Eighth, in the supercapacitor unit capacity configuration stage, the marginal improvement scanning described in Equation (29) is not limited to gradually increasing the total configuration capacity, but can also simultaneously optimize the access location of the supercapacitor unit: for several candidate access nodes, calculate the marginal improvement of the recovery rate in the first hour and the marginal decrease of battery degradation cost under different access schemes according to Equation (29), and select the access location and configuration capacity combination that makes the overall benefits optimal.

[0078] Ninth, the adaptive time scale criterion in step S7 can be further introduced with an online update mechanism based on the rolling time domain: during the recovery process, after each decision period is completed, the duration of subsequent periods is recalculated and the optimization model is resolved based on the actual disaster impact intensity, the actual remaining energy of the supercapacitor unit and the load recovery progress, in order to cope with uncertainties such as disaster evolution, fluctuations in new energy output and changes in equipment status.

[0079] Tenth, in the joint island reconfiguration optimization model of step S8, when the scale of the distribution network is large (such as IEEE 69 nodes, IEEE 123 nodes or actual engineering distribution networks) which leads to a significant increase in the solution scale of the mixed integer linear programming model, a distributed solution method (such as the Alternating Direction Multiplier Method ADMM) can be used to decompose the global optimization problem into multiple sub-region optimization problems for parallel solution, or a column generation method can be used to implicitly enumerate the candidate island topology set to reduce the dimension of integer variables, thereby significantly shortening the computation time while ensuring the solution quality.

[0080] Eleventh, the rapid support from supercapacitor units to critical loads can reach a coefficient. Not limited to the static topology calculation method used in this embodiment, an online power flow update mechanism can also be introduced: before solving each decision period, the power flow margin of each branch is updated according to the power flow results of the previous period. And voltage margin, so that the reachability coefficient dynamically reflects the impact of island topology changes and load recovery process on the accessibility of supercapacitor support.

[0081] In summary, this invention extends the functional division of supercapacitor units and battery energy storage units from the equipment level to the topology accessibility level by constructing and decomposing the disaster impact power spectrum. It also feeds back the actual support effect of the supercapacitor units to the battery protection constraints and recovery period division. In the same joint islanding reconfiguration optimization model, it realizes closed-loop coordination of disaster impact identification, energy storage functional division, topology accessibility support, energy storage protection feedback and time scale adaptation. It can effectively improve the recovery capability of critical loads in the distribution network after disasters and reduce the risk of battery degradation without increasing the energy storage capacity. It has good engineering application value and promotion prospects.

Claims

1. A virtual power plant islanding reconfiguration method driven by impact power spectrum, characterized in that, Includes the following steps: Step S1: Obtain the operating parameters of the post-disaster distribution network and the operating parameters of each distributed resource in the virtual power plant. The operating parameters include node load demand, tie switch and sectional switch status, electrical parameters of each branch, and the predicted available output of the distributed resources. Step S2: Calculate the equivalent load deficit for each time period based on the node load demand and the predicted available output of the distributed resources, and then calculate the disaster criticality for each time period; Step S3: Determine the weight configuration for each time period based on the disaster criticality, and calculate the priority of the support capacity of schedulable resources using a multi-index comprehensive evaluation method; Step S4: Construct the disaster impact power demand of each node within the time period as a weighted sum of three parts: step pick-up impact power, basic resource power supply gap and short-term fluctuation of new energy output, and calculate the aggregation index based on the weighted sum. Step S5: Based on the available ramping capability of the battery energy storage unit, decompose the disaster impact power spectrum into the impact power component borne by the supercapacitor unit and the continuous energy component borne by the battery energy storage unit. Step S6: Evaluate the rapid support accessibility of the supercapacitor unit to the critical load, determine the early recovery eligibility of the critical load based on the rapid support accessibility coefficient, and set dynamic power and energy reservation constraints for the supercapacitor unit. Step S7: Incorporate the actual support amount of the supercapacitor unit into the protection constraints of the battery energy storage unit to form a closed-loop correction of the battery rate discharge constraint and ramping constraint, and calculate the battery degradation cost model. Step S8: Establish a joint island reconfiguration optimization model, set topology constraints and power flow constraints, and embed the dynamic power and energy reservation constraints and the closed-loop corrected constraints, and solve for the branch opening and closing status, island root node, node load recovery amount and energy storage output command for each time period.

2. The method according to claim 1, characterized in that, In step S1, timestamp alignment is performed on multi-source heterogeneous data to unify data from different sources and with different sampling frequencies to the same time reference. Data preprocessing also includes missing value interpolation and normalization. For data missing due to communication interruption or sensor failure, a linear interpolation method based on adjacent time period data is used to fill the missing data. For parameters with large differences in numerical range, a range normalization method is used to map each parameter to a preset interval.

3. The method according to claim 1, characterized in that, The equivalent load deficit is: In the formula, For positive part operators; For nodes During the period Load demand; Basic resources During the period Available output; For the set of distribution network nodes; For the set of recovery periods; The severity of the disaster is as follows: In the formula, To avoid small positive numbers with a denominator of zero, .

4. The method according to claim 3, characterized in that, The multi-index comprehensive evaluation method adopts a combination of CRITIC objective weighting and TOPSIS proximity. The evaluation indexes include response time, ramp rate, continuous power supply time, and topology correlation with critical loads. The priority of the support capabilities is as follows: In the formula, The range of values ​​is , , For each resource The Euclidean distance to the positive and negative ideal solutions.

5. The method according to claim 4, characterized in that, The step pickup impact power is: In the formula, For nodes The step pickup impulse power generated when switching from a shutdown state to power restoration; For nodes During the period Should power be restored? In order to restore power, For nodes During the period relative time period From power loss to power supply; For nodes The apportionment of the power supply gap in the region's basic resources at the node level: In the formula, For nodes It can be allocated to the available output of basic resources.

6. The method according to claim 1, characterized in that, The impact power component is the larger of the difference between the node's impact power demand and the battery energy storage unit's ramp response capability and zero; the continuous energy component is the node's basic resource power supply gap minus the impact power component; the impact power component is borne by the supercapacitor unit in a high-power output mode for a short period of time, and the continuous energy component is continuously supplied by the battery energy storage unit in a relatively stable manner.

7. The method according to claim 1, characterized in that, The rapid support reachability coefficient is constructed as the product of four factors: support capacity evaluation value, network impedance distance attenuation factor, candidate path power flow margin, and candidate path voltage margin. The network impedance distance attenuation factor is a negative exponential function of impedance distance. The power flow margin is the minimum ratio of the capacity margin of each branch of the candidate path to the upper limit of capacity. The voltage margin is the minimum ratio of the voltage margin of each node of the candidate path to the allowable range. Whether a node is qualified for early recovery of critical loads is determined based on whether the rapid support reachability coefficient reaches a preset threshold.

8. The method according to claim 1, characterized in that, The correction coefficient for the battery rate discharge constraint is determined based on the ratio of the actual support amount of the supercapacitor unit to its rated power; the correction for the battery ramp constraint is determined based on the actual support amount of the supercapacitor unit; the battery degradation cost model includes throughput degradation term, high rate degradation term, and power jump degradation term.

9. The method according to claim 1, characterized in that, The determination of whether to use fine-grained time period division is based on the rate of change of the aggregated index in adjacent time periods and the average energy margin of the supercapacitor unit group; the aggregated index is the weighted sum of the disaster impact power demand of all nodes, with the weight being the importance of the node load; the average energy margin of the supercapacitor unit group is the average of the ratio of the remaining energy to the rated capacity of each supercapacitor unit; when the aggregated index exceeds the high intensity threshold, or the rate of change exceeds the rate of change threshold, or the energy margin is lower than the low margin threshold, fine-grained time period division is used, otherwise coarse-grained time period division is used.

10. The method according to claim 1, characterized in that, The topology constraints employ a virtual flow method to ensure connectivity within the island and prevent the formation of loops. The objective function includes load reduction penalties, battery degradation costs, resource operation costs, switching operation costs, reachability risk penalties, and island root node selection rewards. The solution results of the mixed-integer linear programming model include branch opening and closing states, island root nodes, node load recovery amounts, and energy storage output commands for each time period.

Images