A heterogeneous resource multi-scale spatio-temporal coordination vigilance state prevention control method

Abstract

The application discloses a heterogeneous resource multi-scale space-time coordination alert state prevention control method, comprising the following steps: step 1, constructing a basic scene physical model, and reading coarse-grained photovoltaic output prediction data; step 2, determining an upper basic optimization target, and solving an ESS basic power / energy plan and a preventive topology reconstruction scheme based on a set upper target function; and step 3, constructing a lower real-time adjustment model, and solving an SOP active / reactive power exchange instruction and an ESS charging / discharging fine adjustment instruction based on an optimization target function. The application aims at the problem of insufficient resource coordination, constructs a double-layer framework of 'upper plan-lower correction', decouples the response characteristics of mechanical and electronic resources, creates a static safety space and reserves a margin through reconstruction in the upper slow scale, relies on SOP space power transfer and ESS time energy transfer to smooth high-frequency fluctuations in the lower fast scale, and realizes space-time complementation of heterogeneous resources.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network optimization control technology, and relates to a warning state prevention and control method for multi-scale spatiotemporal coordination of heterogeneous resources. Background Technology

[0002] Faced with multi-scale weather events, existing single-mode regulation methods have exposed inherent physical limitations: slow mechanical resources (such as network reconfiguration), while possessing global optimization capabilities, are slow to act and cannot track minute-level fluctuations; while fast continuous resources such as SOPs and ESSs are limited by capacity boundaries and investment costs, making it extremely uneconomical or even infeasible to rely solely on them to withstand several hours of peak backflow. Control strategies limited to a single time scale or a single resource are showing signs of fatigue, necessitating the exploration of efficient collaborative mechanisms for heterogeneous resources. Existing technical solutions mainly include the following three categories: 1) optimizing the power flow distribution of the distribution network through discrete network reconfiguration technology to improve the system's capacity to support distributed renewable energy; 2) utilizing smart soft switches (SOPs) for continuous active power regulation and reactive power compensation, thereby effectively improving the system's voltage distribution and power supply flexibility; 3) utilizing the flexible scheduling of energy storage systems and the coordinated configuration of remote control switches and SOPs to significantly enhance the recovery capability and resilience of the distribution network under fault scenarios.

[0003] The shortcomings of existing technologies are as follows:

[0004] 1) Existing research focuses on the independent control of a single resource or optimization within a single time scale. Few studies have been able to deeply reveal the deep spatiotemporal coupling mechanism between discrete network topology (which determines the static security boundary) and continuous power electronic equipment (which determines the dynamic response performance) across multiple time scales.

[0005] 2) Some two-layer or collaborative optimization frameworks treat the collaboration of heterogeneous resources as purely mathematical coupling, containing a large number of discrete variables, resulting in low computational efficiency and sometimes difficulty in convergence. At the same time, they fail to truly achieve effective decoupling of device response characteristics across multiple spatiotemporal scales at the physical level, making it difficult to fully coordinate the interaction between long-term static topological boundaries and short-term dynamic power regulation. Summary of the Invention

[0006] The purpose of this invention is to provide a vigilance-state prevention and control method for multi-scale spatiotemporal coordination of heterogeneous resources. To address the problem of insufficient resource coordination, a two-layer framework of "upper-level planning - lower-level correction" is constructed to decouple the response characteristics of mechanical and electronic resources. The upper-level slow scale creates a static safety space and reserves margin through reconstruction; the lower-level fast scale relies on SOP spatial power transfer and ESS time energy translation to smooth high-frequency fluctuations, thereby achieving spatiotemporal complementarity of heterogeneous resources.

[0007] To achieve the above objectives, the solution of the present invention is:

[0008] A method for alert-state prevention and control of heterogeneous resources with multi-scale spatiotemporal coordination includes the following steps:

[0009] Step 1: Construct a basic scenario physical model and read coarse-grained prediction data of photovoltaic power output;

[0010] Step 2: Determine the upper-level basic optimization objective, and solve the ESS basic power / energy plan and preventive topology reconfiguration scheme based on the set upper-level objective function;

[0011] Step 3: Construct the lower-level real-time adjustment model, and obtain the SOP active / reactive power exchange command and ESS charge / discharge fine-tuning command based on the optimization objective function.

[0012] The specific process of step 1 above is as follows:

[0013] Minute-level baseline data for each photovoltaic node is generated based on historical or predicted sunshine curves;

[0014] Aggregate the minute sets corresponding to the upper-level hourly time periods to obtain the hourly photovoltaic output under the basic scenario;

[0015] The hourly photovoltaic power output of the corresponding event is obtained by superimposing the deviation of typical variable weather events on the hourly photovoltaic processing under the basic scenario.

[0016] A bell-shaped curve was used to simulate the baseline for photovoltaic power output prediction, and Gaussian noise was added to simulate the corresponding event data.

[0017] The above-mentioned typical events of variable weather include two types of events.

[0018] The sudden drop in cloud cover occurred during the high-irradiance period at midday when there were no clouds, resulting in long-term, large-capacity reverse power transmission from photovoltaic systems.

[0019] Strong light peak feedback occurs during normal photovoltaic power output, caused by uneven cloud movement resulting in minute-level sudden drops and high-frequency fluctuations in output.

[0020] The specific process of step 2 above is as follows:

[0021] Step 21: Set an upper-level objective function with the core objective of minimizing the overall network voltage deviation and branch load balance, while optimizing the basic scenario.

[0022] Step 22: For photovoltaic events in the predicted data, with remedial feasibility as a constraint, it is required that under the same topology, ESS and SOP still have sufficient physical control space and voltage does not exceed the limit under disturbance.

[0023] Step 23: Take into account the feasibility of remediation of basic scenario objectives and events, and obtain the optimal preventive topology reconfiguration scheme and ESS basic power / energy plan for future time periods by solving mixed integer linear programming.

[0024] Step 24: Fix the solved discrete topology state as a hard boundary, and send the planned hourly power and basic charging and discharging power of the ESS to the lower layer as a reference.

[0025] In step 23 above, the basic scenario objectives and the remedial feasibility constraints of each event are replicated and uniformly modeled according to the scenario, and an open-source solver is used to solve for the globally optimal or feasible solutions within a given optimal gap.

[0026] The specific process of step 3 above is as follows:

[0027] Step 31: Based on real-time minute-level photovoltaic data, establish the SOP real-time power transfer model and the ESS dynamic power adjustment model;

[0028] Step 32: In the objective function, set differentiated weights, prioritize calling SOP for spatially low-cost power exchange, and then call ESS for time-energy shifting.

[0029] Step 33: Introduce terminal state constraints, designed so that after a period of drastic fine-tuning, the lower-level ESS will approach the upper-level SOC plan value as closely as possible at the end of the time period.

[0030] Step 34: Construct the lower-level optimization objective function. The primary goal is to minimize the minute-level voltage deviation, followed by achieving smooth tracking of the upper-level ESS power plan and minimizing equipment adjustment costs.

[0031] Step 35: Within each hour, continuously solve the lower-level convex optimization problem in minute increments, and output the active / reactive switching instructions of SOP and the charging / discharging fine-tuning instructions of ESS in real time.

[0032] After adopting the above solution, the beneficial effects of the present invention are:

[0033] This invention decouples the response characteristics of slow-speed machinery and fast-speed power electronic equipment. The upper layer utilizes hourly-level reconstruction to construct a static safety domain and reserve margins; the lower layer relies on SOP spatial transfer and ESS time shift for minute-level dynamic correction, fully leveraging the physical potential that "topology determines safety boundaries, and power electronics determine dynamic performance." At the same time, it abandons the traditional iterative algorithms between upper and lower layers that are prone to oscillations. The upper layer ensures global optimality and defense margins, while the lower layer, relying on the upper layer boundary, smooths photovoltaic fluctuations and tracks the planned trajectory through minute-level convex optimization, fully meeting the timeliness requirements of online control. Attached Figure Description

[0034] Figure 1This is the overall framework diagram of the present invention;

[0035] Figure 2 These are simulated photovoltaic power output curves under two types of weather events;

[0036] Figure 3 This is the actual photovoltaic power output curve;

[0037] Figure 4 It is a top-level preventative topology scheme;

[0038] Figure 5 This is the result of lower-level scrolling optimization. Detailed Implementation

[0039] The technical solution and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0040] like Figure 1 As shown, this invention provides a vigilance-state prevention and control method for multi-scale spatiotemporal coordination of heterogeneous resources, comprising the following steps:

[0041] Step 1: Construct the basic scenario physical model. Read the coarse-grained prediction data of photovoltaic output (hourly level), establish the radial topology virtual network model of the distribution network, linearized power flow constraints, and physical component constraints for SOP capacity and ESS charging and discharging;

[0042] Step 2: Determine the upper-level basic optimization objective, and solve for the ESS basic power / energy plan and preventive topology reconfiguration scheme based on the set upper-level objective function; the specific process includes:

[0043] Step 21: Establish upper-level basic optimization objectives. Set an upper-level objective function with the core objective of minimizing the overall network voltage deviation and branch load balance, while simultaneously performing basic scenario optimization.

[0044] Step 22, ensure the feasibility of the execution event. For photovoltaic events in the predicted data, remedial feasibility constraints are embedded, requiring that under the same topology, ESS and SOP still have sufficient physical control space and voltage does not exceed the limit under disturbance.

[0045] Step 23: Solve for the upper-level hourly optimization scheme. Taking into account the basic scenario objectives and the feasibility of remedial measures for events, use mixed-integer linear programming to solve for the optimal "preventive topology reconfiguration scheme" and "ESS basic power / energy plan" for future time periods.

[0046] Step 24, solidify the topology and distribute the plan. The discrete topology state (switch on / off) solved by the upper layer is fixed as a hard boundary, and the hourly planned power of ESS (SOC plan) and the basic charging and discharging power are distributed to the lower layer as a reference benchmark.

[0047] Step 3: Construct the lower-level real-time adjustment model, and obtain the SOP active / reactive power switching command and ESS charge / discharge fine-tuning command based on the optimization objective function; the specific process includes:

[0048] Step 31: Construct the lower-level real-time adjustment model. Receive real-time minute-level photovoltaic data and establish the SOP real-time power transfer model and the ESS dynamic power adjustment model.

[0049] Step 32: Define the heterogeneous resource differentiation characteristics. In the objective function, set differentiation weights. Since the SOP lifetime loss is extremely low, prioritize calling the SOP for extremely low-cost spatial power exchange, and then call the ESS for time-energy shifting to achieve optimal loss allocation.

[0050] Step 33: Apply recursive feasibility terminal constraints. Introduce terminal state constraints: after undergoing drastic fine-tuning for one hour, the lower-level ESS must approach the planned SOC value set by the upper level as close as possible at the end of the time period to prevent power depletion and ensure feasibility for the next time period.

[0051] Step 34: Establish the lower-level multi-dimensional tracking objective. Construct the lower-level optimization objective function: First, ensure the minimum minute-level voltage deviation (smoothing fluctuations); second, achieve smooth tracking of the upper-level ESS power plan; and minimize equipment adjustment costs.

[0052] Step 35, Real-time Scrolling Optimization and Instruction Generation. Within each hour, the convex optimization problem is continuously solved in minute increments, and the active / reactive power switching instructions of the SOP and the charging / discharging fine-tuning instructions of the ESS are output in real time.

[0053] For ease of subsequent model description, the main symbols in this paper have the following meanings.

[0054] The parameters include:

[0055] , Improve the charging / discharging efficiency of energy storage systems;

[0056] Used as a weather event identifier;

[0057] , These are the upper and lower limits of energy storage capacity;

[0058] For large M auxiliary parameters;

[0059] , For nodes Active / reactive power of load;

[0060] , For nodes Photovoltaic active / reactive power output;

[0061] , The upper limit of active / reactive power transmitted by line (i, j);

[0062] , This represents the maximum charging / discharging power for energy storage.

[0063] The SOP can support the upper limit of reactive power;

[0064] , For line resistance and reactance;

[0065] SOP capacity;

[0066] The rated apparent power of the branch circuit;

[0067] , , These are the upper and lower limits of the node voltage and the nominal voltage, respectively.

[0068] Variables include:

[0069] For line (i, j) in The on / off state at any given time, 1 indicates closed and 0 indicates open;

[0070] For virtual streams, The virtual power emitted by the source node;

[0071] , The active / reactive power of the branch circuit;

[0072] Node voltage;

[0073] , Inject nodes for SOP Active / reactive power;

[0074] , Active power for charging / discharging energy storage, , This corresponds to the charge / discharge state;

[0075] The active power output from energy storage to nodes. In terms of energy storage state, Power is adjusted in real time for energy storage.

[0076] In step 1, the flexible power distribution system alert state scenario involved in this invention mainly addresses variable weather events, considering two typical types of events:

[0077] 1) Sudden drop in cloud cover: This often occurs during the high-irradiance period at noon when there are no clouds, and is characterized by long-term, large-capacity reverse power transmission from photovoltaic systems.

[0078] 2) Peak return of strong light: occurs during normal photovoltaic power output, caused by uneven cloud movement, resulting in minute-level sudden drops and high-frequency fluctuations in power output.

[0079] The two types of disturbances differ significantly in time scale and physical characteristics (slow evolution on an hourly scale and drastic abrupt changes on a minute-scale), and also exhibit deep spatiotemporal coupling in actual operation. On the one hand, long-term extreme value backfeeding processes are often superimposed with short-term, high-frequency cloud disturbances; on the other hand, network topology reconfiguration implemented for slow time scales fundamentally alters the system's impedance distribution and electrical distance, thereby profoundly affecting the power regulation boundaries and response performance of power electronic devices (SOP & ESS) on fast time scales.

[0080] Data construction instructions:

[0081] Data is used to build a minute-level photovoltaic forecast baseline. Based on this, firstly, minute-level baseline data for each photovoltaic node is generated based on historical or predicted sunshine curves, and small Gaussian noise can be superimposed to simulate prediction errors; then, according to the upper-level hourly time periods... Corresponding minute set By aggregating the data, we obtain the hourly photovoltaic output under the basic scenario:

[0082]

[0083] Based on this, weather event parameters are introduced. ,in The moment the event begins. , , These are the climbing, holding, and recovery times, respectively. The relative amplitude, For spatial coverage mode. Event shape function. Used to describe the minute-level timing process of climbing-holding-recovery; after adding event bias, the event... The hourly photovoltaic output is as follows:

[0084]

[0085] For cloud cover drop events, let's assume... Represents a node Is it subject to spatial mode? Coverage, with a negative deviation:

[0086]

[0087] For strong light peak return events, the impact is usually manifested as a global or large-scale positive deviation, which can be removed from the coverage index and written as:

[0088]

[0089] The above formula can be used to consistently map minute-level weather events to hourly nominal scenes and event scenes that can be called upon at the upper level; then, the start time, persistence structure, amplitude intensity, and spatial pattern are calibrated and combined in a limited manner to form an event library. This serves as a constraint to ensure the feasibility of remedial measures for subsequent higher-level events.

[0090] like Figure 2 As shown, a bell-shaped curve is used to simulate the photovoltaic output prediction baseline, and Gaussian noise with a standard deviation of 3% is added to the data. Then, two types of event data are generated through simulation.

[0091] SOP uses a model that ignores VSC power loss and applies a polyhedral approximation to the capacity constraint, as shown below:

[0092] (1)

[0093] (2)

[0094] (3)

[0095] (4)

[0096] Equation (3) approximates the capacity constraint (2) as a regular L-gon, representing a set of linear constraints.

[0097] The necessary radial constraints for topology reconstruction are implemented using a virtual network model, as follows:

[0098] (5)

[0099] (6)

[0100] (7)

[0101] (8)

[0102] The power flow constraint adopts a linearized DistFlow model, combined with SOP and network reconstruction variables, as follows:

[0103] (9)

[0104] (10)

[0105] (11)

[0106] in, ( Similarly), it represents the output of photovoltaic and energy storage.

[0107] Safety constraints limit the range of branch power and node voltage as follows:

[0108] (12)

[0109] (13)

[0110] (14)

[0111] The distributed generation (DG) systems connected to the distribution network discussed in this invention mainly include photovoltaic (PV) and energy storage systems (ESS). Among them, PV mainly refers to rural rooftop PV or small-scale distributed PV at the user level, whose output is not controllable and serves as input parameters for the model. The output of the energy storage system is a controllable variable. ), as the model decision variable. The energy storage system adopts a general energy storage model, as follows:

[0112] (15)

[0113] (16)

[0114] (17)

[0115] (18)

[0116] (19)

[0117] (20)

[0118] The upper-level model built in step 2 ensures the feasibility of scheduling under basic scenarios and events.

[0119] Slow-scale, hourly: Decision network topology ( ) and basic energy plans for energy storage ( Then, typical alert-state events are introduced, combined with subsequent event remediation feasibility assurance constraints, to ensure the safety margin for normal operation of the basic scenario while leaving room for lower-level fluctuation response.

[0120] at this time, In the basic scenario ( The photovoltaic output is calculated using hourly output values ​​derived from minute-level forecast baselines. Define a set of variables: The optimized model for the basic scenario is as follows:

[0121] (twenty one)

[0122] The upper-level model must not only calculate the appropriate network topology and energy storage plan, but also reserve feasible margins for rapid adjustment of SOP / ESS on short timescales in the lower level. This invention introduces event remediation feasibility guarantee constraints under an event library: for any typical event, without changing the already determined basic plan and network topology in the upper level, there must exist a set of remediation decisions (recourse variables) under the event conditions to keep power flow and voltage constraints feasible.

[0123] At this point, define typical events. The following set of remedial variables: Photovoltaic power output uses simulated events. Photovoltaic output under certain conditions Then the following constraints are introduced:

[0124] Current flow under event conditions:

[0125] (twenty two)

[0126] ESS recovery capability under event conditions :

[0127] (twenty three)

[0128] (twenty four)

[0129] (25)

[0130] (26)

[0131] (27)

[0132] The above together characterize the ESS power-energy feasible region under event conditions.

[0133] SOP remedial capabilities under event conditions :

[0134] (28)

[0135] (29)

[0136] (30)

[0137] Under certain circumstances, voltage stability must be strictly guaranteed, with the following hard voltage constraints:

[0138] (31)

[0139] The feasibility of event remediation is coupled to the basic scenario through topological variables, with the following coupling constraints:

[0140] (32)

[0141] Among them, topological variables It does not distinguish between the nominal scenario and the event conditions; the purpose is to make the event... The same physical topology as the base scenario must be used so that the topology optimized at the upper layer can simultaneously take into account the target optimization under the base scenario and the feasibility of remediation for event library coverage.

[0142] Regarding the objective function, the main considerations are voltage deviation and branch load balancing. The voltage and load balancing targets include both basic scenario components and event-remedial variables. The limitation that energy storage cannot be charged and discharged simultaneously is incorporated into the objective function as a penalty term. This is the penalty parameter. As shown below:

[0143] (33)

[0144] Solving the upper-level model:

[0145] The upper-level model includes discrete variables such as network topology switching variables and energy storage charging and discharging states, as well as continuous variables such as power flow, voltage, and SOP / ESS output. The objective function may also include branch load square terms, node voltage deviation terms, and penalty function terms. Therefore, the upper-level model usually manifests as a MIQP with integer variables or a linearized MILP. Its non-convexity mainly comes from integer decision-making, logical coupling, and partial variable product constraints.

[0146] Before solving, standard convexity transformation and equivalent transformation can be performed: a linearized DistFlow power flow model is used to replace the AC power flow nonlinear terms; the SOP capacity circle constraint is transformed into a linear inequality using an in-polygon approximation; the product of binary and continuous variables is treated with Big M constraints, indicator constraints, or McCormick envelopes; an auxiliary variable is introduced to transform the absolute voltage deviation into a supremum; the quadratic objective retains its convex quadratic form, or piecewise linearization is used to further transform MIQP into MILP. After the above processing, the continuous relaxation part consists of linear constraints and a positive semidefinite quadratic objective, which can be treated as a convex relaxation by the mixed integer optimization framework.

[0147] In practice, the basic scenario constraints and the remedial feasibility constraints of each event in the event library can be replicated and modeled uniformly according to the scenario, and solved directly using commercial solvers such as Gurobi and CPLEX or open-source solvers with mixed-integer optimization capabilities. The solver obtains the global optimum or a feasible solution within a given optimal gap through branch and bound, cutting plane, and heuristic search, and outputs the upper-level topology state. Energy storage basic power plan and energy trajectory This serves as input for lower-level scrolling optimization.

[0148] In step 3, the lower-level model implements real-time rolling fluctuation smoothing.

[0149] Fast-scale, minute-level: Given a defined topology and energy storage power plan at the upper layer, determine the real-time power adjustment for the Start-up Plan (SOP) and the power adjustment amount for energy storage. Track the upper-layer plan while simultaneously addressing minute-level high-frequency fluctuations and voltage limit exceedance risks. The variable set is: The optimal topology derived from the upper-level model ( ), energy storage capacity baseline ( Energy storage power plan ( The upper-level slow-timescale energy storage and charging plan is assumed to remain unchanged within the lower-level minute scale, i.e. The lower-level model is as follows:

[0150] (34)

[0151] (35)

[0152] (36)

[0153] (37)

[0154] (38)

[0155] (39)

[0156] In the objective function of the lower-level model, it is generally set to Since the lifespan loss of SOP is extremely low, we should prioritize the use of SOP's power balancing capability and minimize the use of ESS. Equations (36)-(37) indicate that while ESS is tracking and adjusting power in real time, it is necessary to prevent the power from being depleted or exceeding the energy limit; Equation (38) restricts the lower-level energy storage adjustment from exceeding the upper and lower limits of the inverter power; Equation (39) is an end constraint that ensures that the lower-level adjustment is executed as close as possible to the upper-level plan and retains a certain margin at the last moment, while introducing slack variables and penalty terms in the objective function in the constraint. The constraint is "softened" by using a deviation penalty coefficient. If force majeure (extreme cloud cover) occurs, temporary deviation is allowed, ensuring that a solution exists. This constraint guarantees the recursive feasibility of the framework.

[0157] Solving the lower-level model:

[0158] The lower-level model executes on a rolling basis on a minute-by-minute scale during the real-time operation phase. Its inputs are the topology state, energy storage baseline, and power plan already determined by the upper-level model. Therefore, the lower-level model no longer optimizes discrete variables such as network interruptions. It only needs to decide on the SOP active / reactive power adjustment and real-time energy storage adjustment under a fixed topology to track the upper-level plan and smooth out minute-by-minute photovoltaic fluctuations.

[0159] Under the above conditions, the lower-level model is a convex optimization problem: the linearized DistFlow power flow constraint, SOP power balance constraint, ESS energy recursion constraint, upper and lower limits of charge / discharge power, upper and lower limits of voltage, and soft constraints of terminal energy can all be written as affine equations or inequalities; the SOP capacity constraint is also a linear constraint if a polygonal approximation is used; the objective function is usually composed of voltage deviation, penalty for ESS deviating from the upper-level plan, SOP / ESS adjustment cost, and slack variable penalty, and can be set as a linear weighted absolute value form or a positive semidefinite quadratic form. Therefore, this problem can fall into convex optimization categories such as LP, QP, or SOCP.

[0160] During the solution process, the latest minute-level photovoltaic output is read in each scrolling window, while the upper-level given value remains fixed. , and Then, the solver can be called to solve the problem. The solution result only executes the control instructions for the current minute or the current short period of time. Subsequently, the status is updated based on the latest measurement value and the next rolling optimization is initiated, thus taking into account real-time performance, feasibility, and tracking of the upper-level plan.

[0161] The following simulation analysis of the application effect of the present invention will be conducted with reference to specific embodiments. A real-world scenario (taken from a city in southeastern China) is used within the IEEE-33 node network. The actual photovoltaic daily power generation curve is shown below. Figure 3As shown, during the periods of 10:00-12:00 and 14:00-16:30, there were small-scale strong light return events and continuous cloud cover events that caused sudden drops in temperature.

[0162] Figure 4 The preventative topology scheme derived from the upper layer is shown. As can be seen, in the initial stage, the tie switch of branch 35 (24-28) is closed, and the sectionalizing switch of branch 21 (2-22) is disconnected, transferring some heavily loaded nodes to branches after node 5. In the evening stage, the tie switch between branches 34 (11-21) is closed, and the sectionalizing switch between 7 and 8 is disconnected, balancing the load distribution across the entire network. The total number of switch operations across the entire network during the optimization period is 8, which is within a reasonable range and will not cause excessive lifespan loss to the switching equipment.

[0163] Figure 5 The results of rolling optimization within a day under the above preventative topology scheme are shown. It can be seen that the SOP and ESS power were adjusted in real time according to the fluctuations in photovoltaic output. Correspondingly, the energy storage SOC was also adjusted, and quickly returned to the upper-level plan after the event ended. The voltage curve for the final day shows that the system voltage was effectively controlled throughout the entire process.

[0164] like Figure 5 As shown in (a) and (c), the blue shading is... Figure 3 The two typical events correspond to the following events: 10:00-12:00 corresponds to a small-amplitude strong light peak return, and 14:00-16:30 corresponds to a continuous cloud obstruction and sudden drop event. Figure 5 In (a), the orange curve represents the transfer power curve of SOP over 24 hours in the lower-level optimization results. Figure 5 In (c), the solid line (Lower ESS real-time power) represents the lower-level ESS power plan curve, and the dashed line (Upper ESS power plan) represents the upper-level ESS power plan curve. Figure 5 As shown in (a) and (c), the power output of SOP and ESS was adjusted in real time according to the fluctuations in photovoltaic output. Outside the event time window, SOP, due to its better performance, still undertook the power transfer during normal operation, and the power output of ESS coincided with the upper-level planning curve when there were no events, which is consistent with reality.

[0165] Figure 5In (b), the solid line (Lower ESS SOC) represents the real-time SOC state curve of energy storage in the lower-level optimization results, and the dashed line (Upper ESS SOC plan) represents the planned SOC curve of energy storage in the upper-level optimization results. The portion between the two represents the SOC adjustment made by energy storage within a short time window based on the real-time situation of events. Figure 5 (b) also made corresponding adjustments within the window where the event occurred, and quickly returned to the upper-level plan after the event ended, matching the behavior of the ESS real-time power curve.

[0166] Figure 5 (d) The solid line (Node voltage) in the figure represents a typical node voltage curve in which the voltage condition in the system is most likely to exceed the limit. As can be seen from the figure, the system voltage fluctuates slightly when the event occurs, but it is effectively controlled and does not exceed the limit, and is close to 1.00.

[0167] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0168] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0169] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for alert-state prevention and control of heterogeneous resources with multi-scale spatiotemporal coordination, characterized in that... Includes the following steps: Step 1: Construct a basic scenario physical model and read coarse-grained prediction data of photovoltaic power output; Step 2: Determine the upper-level basic optimization objective, and solve the ESS basic power / energy plan and preventive topology reconfiguration scheme based on the set upper-level objective function; Step 3: Construct the lower-level real-time adjustment model, and obtain the SOP active / reactive power exchange command and ESS charge / discharge fine-tuning command based on the optimization objective function.

2. The method as described in claim 1, characterized in that: The specific process of step 1 is as follows: Minute-level baseline data for each photovoltaic node is generated based on historical or predicted sunshine curves; Aggregate the minute sets corresponding to the upper-level hourly time periods to obtain the hourly photovoltaic output under the basic scenario; The hourly photovoltaic power output of the corresponding event is obtained by superimposing the deviation of typical variable weather events on the hourly photovoltaic processing under the basic scenario. A bell-shaped curve was used to simulate the baseline for photovoltaic power output prediction, and Gaussian noise was added to simulate the corresponding event data.

3. The method as described in claim 2, characterized in that: The typical events of variable weather include two categories. The sudden drop in cloud cover occurred during the high-irradiance period at midday when there were no clouds, resulting in long-term, large-capacity reverse power transmission from photovoltaic systems. Strong light peak feedback occurs during normal photovoltaic power output, caused by uneven cloud movement resulting in minute-level sudden drops and high-frequency fluctuations in output.

4. The method as described in claim 1, characterized in that: The specific process of step 2 is as follows: Step 21: Set an upper-level objective function with the core objective of minimizing the overall network voltage deviation and branch load balance, while optimizing the basic scenario. Step 22: For photovoltaic events in the predicted data, with remedial feasibility as a constraint, it is required that under the same topology, ESS and SOP still have sufficient physical control space and voltage does not exceed the limit under disturbance. Step 23: Take into account the feasibility of remediation of basic scenario objectives and events, and obtain the optimal preventive topology reconfiguration scheme and ESS basic power / energy plan for future time periods through mixed integer linear programming. Step 24: Fix the solved discrete topology state as a hard boundary, and send the planned hourly power and basic charging and discharging power of the ESS to the lower layer as a reference.

5. The method as described in claim 4, characterized in that: In step 23, the basic scenario objectives and the remedial feasibility constraints of each event are replicated and uniformly modeled according to the scenario, and an open-source solver is used to solve for the globally optimal or feasible solutions within a given optimal gap.

6. The method as described in claim 1, characterized in that: The specific process of step 3 is as follows: Step 31: Based on real-time minute-level photovoltaic data, establish the SOP real-time power transfer model and the ESS dynamic power adjustment model; Step 32: In the objective function, set differentiated weights, prioritize calling SOP for spatially low-cost power exchange, and then call ESS for time-energy shifting. Step 33: Introduce terminal state constraints, designed so that after a period of drastic fine-tuning, the lower-level ESS will approach the upper-level SOC plan value as closely as possible at the end of the time period. Step 34: Construct the lower-level optimization objective function. The primary goal is to minimize the minute-level voltage deviation, followed by achieving smooth tracking of the upper-level ESS power plan and minimizing equipment adjustment costs. Step 35: Within each hour, continuously solve the lower-level convex optimization problem in minute increments, and output the active / reactive switching instructions of SOP and the charging / discharging fine-tuning instructions of ESS in real time.

Images