A power system time sequence deduction method based on an improved OPA model

Abstract

The application discloses a power system time sequence deduction method based on an improved OPA model and belongs to the technical field of power system safety simulation. The technical steps of the method are as follows: 1) constructing a three-layer deduction model of a cascading fault inner layer, an intraday time sequence middle layer and an energy blockade outer layer; 2) constructing an input data set based on a power grid topology, power generation information, a load curve and a new energy curve; 3) sequentially simulating dynamic behaviors of a power system under multiple time scales such as short-time faults, intraday operation and long-term energy blockade, and realizing power system full-time sequence deduction from an hour level to a month level by fusing multiple source information such as a power grid topology, a load curve, new energy output, primary energy reserves and supply strategies; and 4) outputting load loss amount, load loss time period and power deduction results. The application can effectively evaluate load loss risks, energy consumption and supply demands of a power system under extreme scenarios, and provide a scientific basis for power system emergency planning and resilience improvement.

Description

Technical Field

[0001] This invention belongs to the field of power system security analysis and simulation technology, specifically involving a power system time series extrapolation method based on an improved OPA model. Background Technology

[0002] Interregional interconnected power grids have developed into one of the most complex man-made industrial networks. Many major power outages in recent years, both domestically and internationally, have been caused by cascading failures in the power grid. Therefore, scholars worldwide are increasingly focusing on the study of cascading failures and the propagation mechanisms of major power outages. They have proposed various models of cascading failures in power grids, among which the OPA model has received the most attention. The OPA (Self-Organized Criticality Theory) model is also known as the "major power outage model" or "self-organized critical power system model." This model is based on the self-organized criticality theory. This theory posits that a complex dynamic system (such as a power grid) will spontaneously evolve to a critical state under the interaction of continuous, slow external "driving forces" (such as load growth) and localized, rapid internal "dissipation" (such as line overload tripping). In this critical state, a small disturbance (such as a line fault) can trigger a series of chain reactions, leading to a large-scale power outage. The OPA model is essentially a theoretical framework and simulation tool that reveals how the power grid spontaneously breeds the risk of major power outages due to its inherent complexity and continuous growth.

[0003] Traditional OPA models, based on the theory of self-organized criticality, simulate line outages and cascading overloads through fast dynamic processes and system capacity and load growth through slow dynamic processes. However, these models do not adequately consider the temporal fluctuations of new energy sources, limitations on primary energy reserves and replenishment, and energy blockade and transfer strategies over long timescales, making it difficult to accurately reflect the temporal evolution behavior of power systems under complex environments.

[0004] The OPA model divides the power grid evolution process into two interacting time scales: "fast dynamics" (simulating fault propagation) and "slow dynamics" (simulating system upgrades). The "fast dynamics" process allows us to understand the related processes of line outages and cascading overloads, while the "slow dynamics" process can simulate increases in line capacity, load levels, and system generation capacity. Combining the "fast dynamics" and "slow dynamics" approaches helps explain the risk of large-scale power outages.

[0005] The OPA model simulates the long-term dynamics of the power grid through a cycle, revealing why, regardless of how much the grid is strengthened, the risk of large-scale blackouts always exists, driven by complexity and continuous growth.

[0006] The core of solving the OPA model lies in the DC power flow optimization problem. The limitations of the traditional OPA model are as follows:

[0007] Static assumptions: It is usually assumed that the load level and power generation output are constant during the fast dynamic process, ignoring intraday and seasonal fluctuations (especially the randomness of wind power and photovoltaic power).

[0008] Ignoring energy constraints: Focusing only on the physical constraints of the power grid itself (line capacity, generator output), without considering the limitations on the storage, consumption, and replenishment of primary energy sources (coal, gas, oil) of power plants.

[0009] Lack of active strategy simulation: The model reflects the self-organizing evolution of the system more, and it is difficult to simulate human active attacks (power disruption) or macro-level energy blockade strategies and their long-term effects.

[0010] Insufficient coupling of time scales: Although fast and slow dynamics are distinguished, the coupling relationship between hourly intraday operations and monthly strategic reserve changes is poorly characterized.

[0011] Therefore, there is an urgent need for a time-series simulation method that can integrate multiple time scales, multiple energy types, and multiple fault modes to improve the simulation credibility and decision support capability of power systems under extreme scenarios. Summary of the Invention

[0012] This invention addresses the shortcomings of existing OPA models by proposing a power system time-series simulation method based on an improved OPA model. By constructing a three-layer simulation structure of "cascading fault inner layer - intraday time-series middle layer - energy blockade outer layer", it enables multi-timescale power system behavior simulation from hourly to monthly levels, thereby improving the simulation credibility and decision support capabilities of power systems under extreme scenarios.

[0013] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0014] In a first aspect, the present invention provides a method for time-series extrapolation of power systems based on an improved OPA model, comprising the following steps:

[0015] S1: Construct a three-layer extrapolation model consisting of an inner layer of cascading faults, a middle layer of intraday time series, and an outer layer of energy blockade;

[0016] S2: Construct an input dataset based on power grid topology, power generation information, load curves, and new energy curves;

[0017] S3: Determine the input parameters and input the input parameters and input dataset into the three-layer inference model, and execute the outer layer, middle layer and inner layer cyclic inference in sequence to realize the multi-time scale power system behavior simulation and obtain the power system time series inference results;

[0018] S4: Output the time-series simulation results of the power system, including the time-series curves of load loss at each node, the total load loss, and changes in energy reserves.

[0019] The technical effects achieved by the above configuration are as follows: This invention expands the traditional OPA model's two-layer "fast dynamic - slow dynamic" framework into a multi-timescale simulation system covering hourly, intraday hourly, and multi-day / monthly scales by constructing a three-layer simulation model consisting of a "cascading fault inner layer—intraday time-series middle layer—energy blockade outer layer." The cascading fault inner layer simulates the propagation process of cascading faults under power attack strategies; the intraday time-series middle layer introduces the time-series changes in load demand and wind and solar power; and the energy blockade outer layer simulates the process of limited primary energy supply and strategic energy transfer. The three-layer model is nested and executed in the order of "outer layer → middle layer → inner layer," realizing full time-series simulation from hourly to monthly scales. This method effectively overcomes the shortcomings of traditional OPA models, such as ignoring the time-series fluctuations of new energy sources, ignoring the limitations of primary energy reserves and supply, and being difficult to simulate human attacks and blockade strategies. It can realistically reflect the long-term evolutionary behavior of the power system under extreme scenarios, providing a scientific basis and decision support for power system emergency planning, strategic energy reserves, and network resilience improvement.

[0020] By outputting the time-series curves of load loss at each node, the total load loss, and changes in energy reserves, this invention provides intuitive and quantitative decision support information for power system emergency planning and resilience enhancement. The time-series curves of load loss at each node can accurately locate vulnerable nodes in the system and the temporal distribution of their load loss, providing a basis for targeted reinforcement; the total load loss reflects the overall level of power outage risk in the system; and changes in energy reserves reveal the dynamics of energy consumption and replenishment at thermal power plants under shutdown scenarios, providing a reference for optimizing strategic energy reserves. These multi-dimensional outputs enable decision-makers to comprehensively assess the risk level and response capabilities of the power system under extreme scenarios.

[0021] Furthermore, in step S1, a three-layer extrapolation model is constructed, consisting of an inner layer of cascading faults, a middle layer of intraday time series, and an outer layer of energy blockade, including:

[0022] Step S1-1: Construct a cascading fault inner-layer simulation model. The cascading fault inner-layer simulation model considers power attack strategies, takes the attacked lines and substations as initial faults, and uses the improved CASCADE model to simulate the cascading fault situation of the power network on an hourly time scale.

[0023] Step S1-2: Construct an intraday time-series mid-level extrapolation model. The intraday time-series mid-level extrapolation model considers the intraday load demand and the time-series changes of new energy sources, simulates the power network operation trend at the hourly time scale, and calculates the primary energy reserve consumption of thermal power plants. The new energy sources include wind power and photovoltaic power.

[0024] Step S1-3: Construct an outer-layer simulation model of energy blockade, which simulates the process of limited energy supply, decreased storage, and strategic energy transfer on multi-day and monthly time scales.

[0025] The technical effects achieved by the above configuration are as follows: Through the independent construction and functional division of the three-layer model, decoupling and coordination of physical processes at different time scales are realized. The inner layer of cascading faults focuses on the chain fault propagation mechanism at the minute to hour level, which can accurately simulate the initial fault and its cascading effects caused by power outage events; the middle layer of intraday time series introduces the intraday fluctuation characteristics of load and renewable energy, enabling the model to reflect the impact of the intermittency of renewable energy output and the peak-valley variation of load on system operation; the outer layer of energy blockade incorporates the constraints of the primary energy supply chain into the power system evolution framework, breaking through the limitation of traditional OPA models that only focus on the physical constraints of the power grid. The organic integration of the three-layer model enables this invention to comprehensively depict the entire process of power system evolution from micro-fault propagation to macro-energy evolution under extreme scenarios.

[0026] Furthermore, in steps S1-3, the primary energy supply restriction includes scenarios where imports of coal, natural gas, and fuel oil are blocked, and the strategic energy transfer includes strategies for supplying primary energy via transportation networks through trains or tankers.

[0027] The technical effects achieved by the above settings are as follows: By clearly defining the specific types of primary energy blockades and strategic energy transfer methods, this invention can simulate the resilience of the power system under extreme scenarios such as energy import disruptions and transportation network damage. The classification of coal, gas, and oil-fired power plants allows the model to distinguish the energy dependence characteristics of different types of power plants; the introduction of transportation network supply strategies such as trains and tanker trucks enables the model to simulate the emergency response process of energy allocation via land transportation under external constraints such as port blockades, providing simulation support for strategic energy reserves and emergency dispatch.

[0028] Furthermore, in step S2, an input dataset is constructed based on the power grid topology, generation information, load curve, and renewable energy curve, including:

[0029] Step S2-1: Obtain and input power network model information including substations, transmission lines, and station-line topology;

[0030] Step S2-2: Obtain and input the maximum load power information of each load substation;

[0031] Step S2-3: Obtain and input typical daily load curves of the entire network load for different months;

[0032] Step S2-4: Obtain and input typical active power output curves for wind power and photovoltaic power in different months.

[0033] The technical effects achieved by the above settings are as follows: By constructing a multi-source input dataset that includes power grid topology, load characteristics, and renewable energy output characteristics, this invention achieves refined modeling of the actual power system operating environment. The introduction of typical daily load curves and renewable energy output curves for different months enables the model to reflect the seasonal variation of load and the temporal fluctuation characteristics of wind and solar power output. This avoids the static assumptions of constant load and constant output used in traditional OPA models, significantly improving the consistency between simulation results and actual operating conditions.

[0034] Further, in step S3, the input parameters are determined, and the input parameters and the input dataset are input into the three-layer inference model. The outer, middle, and inner layers are sequentially cyclically inferred to achieve multi-timescale power system behavior simulation and obtain the power system time-series inference results, including:

[0035] Step S3-1: Determine the input parameters, including the total number of days in the time series simulation, the primary energy import blockade strategy, the initial primary energy reserves of each thermal power plant, and the power attack events specified by the user;

[0036] Step S3-2: Initialize the outer loop variable d=1 for the energy blockade;

[0037] Step S3-3: Initialize the loop variable h=0 in the intraday time series layer;

[0038] Step S3-4: Calculate the power flow of each branch in the power network at the current moment according to the improved maximum flow algorithm, and count the load loss at the current moment;

[0039] Step S3-5: Check if there is a sabotage event at the current moment. If yes, proceed to step S3-6 to perform cascading fault simulation. If no, proceed to step S3-8 to calculate the primary energy reserve consumption.

[0040] Step S3-6: Based on the improved CASCADE model, calculate the cascading faults that will occur in the power network under the initial attack event;

[0041] Step S3-7: Based on the cascading fault set obtained from the deduction, shut down the faulty branch or substation, update the power network model, recalculate the power flow of the power network after the topology change, and count the load loss after the cascading fault.

[0042] Step S3-8: Calculate the primary energy consumption for the current period based on the active power output status of each thermal power plant at the current moment, and subtract the consumption from the current primary energy reserve to update the primary energy reserve status of each thermal power plant.

[0043] Step S3-9: Update the loop variable in the intraday time series, set h=h+1, and check if h is equal to 24. If yes, proceed to step S3-10 to update the outer extrapolation model variable of energy blockade; otherwise, proceed to step S3-4 to calculate the power grid flow at the next moment.

[0044] Step S3-10: Update the outer loop variable of the energy blockade, set d=d+1;

[0045] Step S3-11: Determine whether the current number of days d is greater than the total number of days in the time series simulation. If yes, output the time series simulation results, including the load loss time series curves of each node, the total load loss, and the changes in energy reserves. If no, proceed to step S3-12 to determine whether to implement a single energy replenishment.

[0046] Step S3-12: Determine whether to perform an energy replenishment. If yes, proceed to step S3-13 to update the energy storage. If no, return to step S3-3 to initialize the loop variables in the intraday time series.

[0047] Step S3-13: For thermal power plants, allocate the primary energy import volume according to the proportion of their remaining primary energy storage capacity, update the primary energy storage volume of each thermal power plant, and return to step S3-3 to initialize the loop variables in the intraday time series.

[0048] The technical effects achieved by the above settings are as follows: By establishing a three-layer cyclical simulation mechanism of "outer layer (day) - middle layer (hour) - inner layer (cascade)," this invention realizes full-time-series power system behavior simulation from hourly to monthly levels. The outer layer cycle controls the total number of simulation days, simulating the energy blockade and replenishment process on a long-term scale; the middle layer cycle progresses hourly on a 24-hour basis, simulating the temporal changes of daily load and new energy sources; the inner layer cycle is triggered when a breach event occurs, simulating the rapid propagation process of cascaded faults. The nested execution of the three layers of cycles enables the model to simultaneously capture the coupling effects of short-term fault propagation, medium-term operation scheduling, and long-term energy evolution. The improved maximum flow algorithm replaces the DC optimal power flow of the traditional OPA model, enabling rapid assessment of the power network's limit transmission capacity; the improved CASCADE model accurately simulates the cascading fault process after a breach event. The real-time consumption and replenishment update mechanism of primary energy reserves enables the model to dynamically reflect changes in energy constraints of thermal power plants, providing key support for assessing system resilience under energy blockade scenarios.

[0049] Furthermore, in steps S3-4, the power flow of each branch in the power network at the current moment is calculated based on the improved maximum flow algorithm, and the load shedding at the current moment is statistically analyzed, including:

[0050] Step S3-4-1: Construct a flow network model of the power network, abstracting power plant nodes as source nodes, load substation nodes as sink nodes, transmission lines as edges, and using the thermal stability limit of the lines as the capacity of the edges to form a multi-source and multi-sink network.

[0051] Step S3-4-2: Introduce virtual source node s and virtual sink node t to convert the multi-source multi-sink network into a single-source single-sink network: Add a directed edge (s,g) from the virtual source node s to each power plant node g, with its capacity c(s,g) set to the available power generation capacity of power plant g. Add a directed edge (l,t) from each load node l to the virtual sink t, with its capacity c(l,t) set to the actual active power load demand of load node l at the current moment. ;

[0052] Step S3-4-3: Execute the maximum flow algorithm to find the maximum feasible flow from the virtual source node s to the virtual sink node t. The maximum feasible flow is equal to the minimum cut. The maximum flow algorithm solves for the maximum feasible flow that satisfies the following flow conservation and capacity constraints:

[0053] ;

[0054] ;

[0055] Where f(i,j) means the flow on the edge (i,j) from node i to node j; This means the feasible flow from the virtual source s to the virtual sink t; s means the virtual source, representing the sum of all power plants; t means the virtual sink, representing the sum of all loads; this flow conservation equation indicates that the net flow out of the virtual source s is equal to the net flow into the virtual sink t, and the flow at intermediate nodes remains balanced.

[0056] c(i,j) represents the capacity of edge (i,j); E represents the set of all edges in the network; this capacity constraint equation means that the flow of any edge cannot exceed its capacity.

[0057] Among them, the flow of the edge between the load substation and the virtual sink is the actual power supply at that time point, and the difference between the actual power supply and the actual active load demand is the load loss of the load substation; the flow of the edge between the virtual source and the power plant node is the active power output of the power plant at that time point.

[0058] Step S3-4-4: Identify the location of the minimum cut and determine the system state by judging the type of the minimum cut: when the minimum cut falls in the edge set between the virtual source node and the power plant node, it is determined that the power supply is insufficient; when the minimum cut falls in the edge set of the power network, it is determined that the transmission capacity is limited; when the minimum cut falls in the edge set between the load substation and the virtual sink node, it is determined that the load demand is met.

[0059] Step S3-4-5: Based on the maximum flow calculation results, calculate the load loss of each node. The actual power supply of load node l is equal to the flow rate f(l,t) on edge (l,t). The load loss of load node l is... Calculate using the following formula:

[0060]

[0061] in, This refers to the amount of load loss at load node l at time h. f(l,t) represents the actual active load demand of load node l at time h; f(l,t) represents the flow of the edge from load node l to virtual sink t, i.e., the actual power supply.

[0062] Step S3-4-6: Calculate the total system load loss :

[0063]

[0064] in, This refers to the total load shedding of the system at time h. It means the set of all load nodes in the system;

[0065] Step S3-4-7: Output the flow rate f(i,j) of each branch, the load loss of each load node, the total load loss of the system, and the actual output of each power plant. And the minimum cut position information.

[0066] The technical effects achieved by the above settings are as follows: By introducing virtual source nodes and virtual sink nodes, this invention transforms the multi-source, multi-sink network of a real power system into a single-source, single-sink network solvable by the classic maximum flow problem, enabling mature graph theory algorithms to be directly applied to power system supply capacity assessment. The edge capacity from the virtual source node to the power plant node maps to the available power generation capacity of the power plant, and the edge capacity from the load node to the virtual sink node maps to the real-time load demand, achieving unified modeling of constraints on the generation side and the load side. By identifying the minimum cut position, this invention can quickly determine the state type of the system—insufficient power supply, limited transmission capacity, or satisfied load demand—providing intuitive quantitative indicators for analyzing weak links in the system. The load shedding statistical method based on the maximum flow calculation results can accurately calculate the difference between the actual power supply and demand power of each load node, providing accurate load shedding data for subsequent time series analysis.

[0067] Furthermore, in step S3-4-3, the maximum flow algorithm adopts the Ford-Fulkerson algorithm.

[0068] The technical effects achieved by the above settings are as follows: The Ford-Fulkerson algorithm, as a classic maximum flow solution method, has the advantages of clear algorithmic ideas, simple implementation, and good convergence. By using this algorithm to solve the maximum flow of the virtual network, this invention can achieve high computational efficiency while ensuring computational accuracy, meeting the computational speed requirements of large-scale power system network time-series simulations. The iterative process of finding augmenting links and updating flow in the algorithm ensures that the solution converges to the global optimum, i.e., the maximum transmission capacity of the network.

[0069] Furthermore, in step S3, the multi-timescale power system behavior simulation includes:

[0070] Simulate the cascading failure propagation process on an hourly timescale within the inner layer of the cascading failure;

[0071] The operation trend of the power network is simulated in the intraday time series at the hourly time scale, where the intraday hourly time scale is 24 hours.

[0072] Simulate the energy supply restriction process on a multi-day, monthly timescale outside the energy blockade;

[0073] The three-layer cycle of cascading fault inner layer, intraday time sequence middle layer, and energy blockade outer layer is nested and executed in the order of "outer layer to middle layer to inner layer", realizing full time sequence simulation from hourly to monthly level.

[0074] The technical effects achieved by the above settings are as follows: By clearly defining the time scales and nested execution order of the three-layer model, this invention constructs a hierarchical and logically clear time-series deduction framework. The outer loop advances on a daily basis, simulating the long-term impact of macroeconomic strategies such as energy blockade; the middle loop advances on an hourly basis, capturing the intraday fluctuation patterns of load and new energy sources; the inner loop is triggered when a breach event occurs, simulating the rapid propagation of cascading failures with minute-level precision. The organic coupling of the three time scales enables the model to simultaneously reflect long-term evolution trends, medium-term operating patterns, and short-term failure propagation, achieving a comprehensive characterization of the multi-time-scale behavior of the power system under extreme scenarios.

[0075] Furthermore, in steps S3-6, based on the improved CASCADE model, the cascading faults that will occur in the power network under the initial attack event are calculated, including:

[0076] Step S3-6-1: Load the current power grid state, including the node set V, branch set E, generator output, load demand, and calculate the initial power flow distribution;

[0077] Step S3-6-2: Apply the initial attack fault according to the power attack event specified by the user: if the target is a line, remove the line from the branch set E; if the target is a substation, remove the substation and all its connected lines from the branch set E, and record the initial fault element set F0.

[0078] Step S3-6-3: Set the iteration counter k=1 and enter the cascade propagation iteration loop;

[0079] Step S3-6-4: Based on the current topology, calculate the power flow of each branch using a DC power flow model. The calculation formula is:

[0080] ;

[0081] in, This means the active power of line ij at the k-th iteration; line ij is the line between nodes i and j.

[0082] This refers to the voltage phase angles of nodes i and j at the k-th iteration. This refers to the reactance of line ij;

[0083] This refers to the set of branches after the k-th iteration.

[0084] Step S3-6-5: Calculate the load rate of each operating line. ;in, This refers to the load rate of line ij at the k-th iteration; This refers to the thermal stability limit or transmission capacity of line ij.

[0085] Step S3-6-6: Determine if an overloaded branch exists: If If all operating lines are confirmed to be overloaded, the system is stable, and the process proceeds to step S3-6-10; otherwise, if an overloaded branch exists, the process proceeds to step S3-6-7. This refers to the overload threshold; protection is triggered when the line load rate exceeds this value.

[0086] Step S3-6-7: Determine the branch to be disconnected this time based on the overload protection logic, including disconnecting the line with the highest load rate according to deterministic rules or disconnecting the line according to the tripping probability according to probabilistic rules. Cut off the line;

[0087] in, This means the line tripping probability; α means the sensitivity coefficient of the probabilistic tripping model, and e is the natural constant.

[0088] Step S3-6-8: Cut the selected branches and update the topology, removing the cut branches from the branch set. Remove from the grid and record the set of faulty components; if the grid is split into multiple islands after the disconnection, perform power flow calculations for each island and disconnect some loads according to priority when there is an imbalance between power generation and load in the island;

[0089] Step S3-6-9: Determine the iteration termination condition: If k≥Kmax, then determine that the system has crashed or cannot converge, and jump to step S3-6-10; otherwise, set k=k+1 and return to step S3-6-4 to continue iterating; where Kmax means the maximum number of iterations, used to prevent infinite loops;

[0090] Step S3-6-10: Statistically analyze the cascaded fault results, including the final topology and fault branch sequence. Total load loss The final state of the system and the final output values ​​of each thermal power plant; among which, This refers to the set of faulty components removed during the k-th iteration. This refers to the total load shedding. This refers to the active power load demand of node i. This refers to the actual power supplied to node i.

[0091] The technical effects achieved by the above settings are as follows: The improved CASCADE model, by introducing user-specified power disruption events as initial faults, overcomes the limitation of traditional CASCADE models that only support random faults, enabling this invention to simulate cascading fault processes under targeted disruption scenarios such as military strikes and cyberattacks. The DC power flow model ensures computational efficiency, and the load rate calculation and overload judgment mechanism accurately simulates the operating logic of line protection devices. The flexible configuration of deterministic and probabilistic rules allows the model to adapt to simulation scenarios with different accuracy requirements: deterministic rules are suitable for rapid evaluation, while probabilistic rules can reflect the random behavior characteristics of protection devices. The islanding handling mechanism after disconnection ensures that the model can still perform reasonable power flow calculations and load shedding after system splitting, avoiding computational interruptions caused by disconnection. The setting of iteration termination conditions effectively prevents infinite loops and ensures the convergence of the algorithm.

[0092] Furthermore, in steps S3-8, the consumption of primary energy reserves of the thermal power plant is calculated, including:

[0093] Based on the classification of coal-fired power plants, gas-fired power plants, and oil-fired power plants, calculate the coal consumption, gas consumption, or oil consumption corresponding to the active power output of each type of power plant in the current period.

[0094] The conversion between primary energy consumption and active power output is obtained based on the unit energy consumption coefficient of the power plant.

[0095] The technical effects achieved by the above settings are as follows: By classifying coal-fired, gas-fired, and oil-fired power plants, this invention achieves refined modeling of primary energy consumption. The introduction of a unit energy consumption coefficient allows the model to accurately calculate energy consumption based on the actual output of the power plant, providing an accurate data foundation for subsequent energy reserve updates. This classification approach enables the model to reflect the differences in energy consumption characteristics among different types of power plants, providing support for assessing system resilience under different energy structures. The real-time update mechanism for primary energy reserves allows the model to dynamically reflect changes in the available power generation capacity of thermal power plants under energy blockade scenarios, providing a key function for simulating long-term energy constraints.

[0096] Furthermore, in step S3-13, the primary energy storage of each thermal power plant is updated, including:

[0097] Determine the proportion of each thermal power plant's primary energy surplus storage capacity to the total primary energy surplus storage capacity of all thermal power plants;

[0098] According to the aforementioned proportion, the total amount of allocable primary energy imports will be allocated to each thermal power plant.

[0099] The allocated primary energy imports are added to the current primary energy storage of each thermal power plant to complete the update.

[0100] The technical effects achieved by the above settings are as follows: By allocating primary energy imports according to the proportion of remaining storage capacity, this invention realizes strategic energy supply simulation under energy blockade scenarios. This proportional allocation mechanism simulates the principle of fair allocation under limited supply resources and can reflect the resource allocation logic in actual emergency dispatch. The updated primary energy storage directly affects the available generating capacity of thermal power plants in subsequent time periods, forming a closed-loop feedback mechanism that enables the model to realistically simulate the evolution of the power system under long-term energy constraints.

[0101] Furthermore, in step S4, the output power system time-series simulation results also include the time-series curves of primary energy reserve changes of each thermal power plant, the minimum cut position information of the power grid at each moment, and the cascading fault chain records after each attack event.

[0102] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0103] 1. To address the limitations of traditional OPA models, this invention introduces a three-layer structure (fault inner layer, time-series middle layer, and blockade outer layer), deeply integrating complex real-world factors such as time-series renewable energy output, primary energy supply chain, and human-induced attacks and blockade strategies into the OPA framework. This enables a more realistic and comprehensive assessment of the long-term resilience of power systems under extreme scenarios, and achieves full-process time-series simulation of power systems across multiple time scales and energy types under extreme scenarios.

[0104] 2. This invention integrates multiple factors such as cascading failures, new energy fluctuations, and primary energy reserves and replenishment, making the simulation results more consistent with reality;

[0105] 3. This invention can provide scientific basis and decision support for power system emergency planning, energy strategic reserves, and network resilience enhancement;

[0106] 4. The model structure of this invention is clear and highly scalable, making it suitable for simulation analysis of power systems of different scales and types. Attached Figure Description

[0107] Figure 1 This is a flowchart of the method of the present invention;

[0108] Figure 2 This is a flowchart of the time-series deduction algorithm based on the improved OPA model of this invention;

[0109] Figure 3 This is a schematic diagram of the flow network of a power grid;

[0110] Figure 4A schematic diagram of minimum cut in a power grid flow network under conditions of insufficient power supply;

[0111] Figure 5 A schematic diagram of minimum cut in a power grid flow network under conditions of limited transmission capacity;

[0112] Figure 6 This is a schematic diagram of the minimum cut of the power grid flow network under the condition that the load demand is met. Detailed Implementation

[0113] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0114] Example 1:

[0115] This embodiment provides a power system time-series extrapolation method based on an improved OPA model, such as... Figure 1 As shown, it includes the following steps:

[0116] S1: Construct a three-layer extrapolation model consisting of an inner layer of cascading faults, a middle layer of intraday time series, and an outer layer of energy blockade;

[0117] S2: Construct an input dataset based on power grid topology, power generation information, load curves, and new energy curves;

[0118] S3: Determine the input parameters and input the input parameters and input dataset into the three-layer inference model, and execute the outer layer, middle layer and inner layer cyclic inference in sequence to realize the multi-time scale power system behavior simulation and obtain the power system time series inference results;

[0119] S4: Output the time-series simulation results of the power system, including the time-series curves of load loss at each node, the total load loss, and changes in energy reserves.

[0120] Specifically, in step S1, a three-layer simulation model is constructed, consisting of an inner layer of cascading faults, a middle layer of intraday time series, and an outer layer of energy blockade. Specifically:

[0121] Step S1-1: Constructing a cascading fault inner-layer inference model:

[0122] The inner layer of cascading faults considers power attack strategies, using attacked lines and substations as initial faults, and employs an improved CASCADE model to simulate cascading faults after power network attacks or faults on an hourly timescale.

[0123] Step S1-2: Construct a mid-level extrapolation model for intraday time series of cascading failures:

[0124] The intraday time series simulation considers intraday load demand and the temporal variations of new energy sources such as wind power and photovoltaics, simulating the intraday hourly timescale power network operation trends and calculating the primary energy reserve consumption of coal-fired, gas-fired, and oil-fired power plants. Existing simulation models or software can be used for the simulation process.

[0125] Step S1-3: Construct the outer-layer extrapolation model of cascading failure energy blockade:

[0126] The outer layer of the energy blockade simulates the process of limited primary energy supply, declining storage, and strategic energy transfer over long timescales. Specifically, it simulates multi-day and monthly timescales, primarily reflecting the gradual decline in primary energy storage and the strategic transfer of primary energy via transportation networks (trains, tankers) under the condition of port blockade. The simulation process can utilize existing simulation models or software.

[0127] Specifically, in step S2, an input dataset is constructed based on the power grid topology, power generation information, load curves, and new energy curves. Specifically:

[0128] Step S2-1: Obtain and input the model information of the power network, including substations, transmission lines, and station-line topology.

[0129] Step S2-2: Obtain and input the maximum load power information of each load substation.

[0130] Step S2-3: Obtain and input typical daily load curves of the entire network load for different months.

[0131] Step S2-4: Obtain and input typical active power output curves for wind power and solar power in different months.

[0132] Specifically, in step S3, outer, middle, and inner layer iterative deductions are executed sequentially to achieve multi-timescale power system behavior simulation. Specifically:

[0133] The simulation is set with input parameters such as the total number of days, energy blockade strategy, initial energy reserves, and attack events. It progresses in a cyclical manner, "outer layer → middle layer → inner layer," calculating power flow, identifying faults, and updating energy status hourly. Within each cycle, fault propagation simulation, energy consumption calculation, and replenishment strategy execution are implemented. Figure 2 As shown, the algorithm flow of the time series simulation model is as follows.

[0134] Step S3-1: Input information for time series simulation:

[0135] Determine the total number of days in the time series simulation, whether to implement a blockade on the import of various primary energy sources such as coal, gas, and oil, the initial primary energy reserves of each thermal power plant, and user-specified power attack events, as well as other user-specified input information related to the time series simulation.

[0136] Step S3-2: Initialize the outer loop variable of the energy lockout:

[0137] Initialize the outer loop variable d=1.

[0138] Step S3-3: Initialize the loop variable in the intraday time series layer:

[0139] Initialize the loop variable h=0 in the intraday time series layer.

[0140] Step S3-4: Calculate the power flow of the power network at the current moment:

[0141] Based on the improved maximum flow algorithm, the power flow of each branch in the power network is calculated, and the current load shedding is statistically analyzed, specifically including:

[0142] Step S3-4-1: Construct a flow network model of the power network, abstracting power plant nodes as source nodes, load substation nodes as sink nodes, transmission lines as edges, and using the thermal stability limit of the lines as the capacity of the edges to form a multi-source and multi-sink network.

[0143] Step S3-4-2: Introduce virtual source node s and virtual sink node t to convert the multi-source multi-sink network into a single-source single-sink network: Add a directed edge (s,g) from the virtual source node s to each power plant node g, with its capacity c(s,g) set to the available power generation capacity of power plant g. Add a directed edge (l,t) from each load node l to the virtual sink t, with its capacity c(l,t) set to the actual active power load demand of load node l at the current moment. ;

[0144] Step S3-4-3: Execute the maximum flow algorithm to find the maximum feasible flow from the virtual source node s to the virtual sink node t. The maximum flow algorithm solves for the maximum feasible flow that satisfies the following flow conservation and capacity constraints:

[0145] ;

[0146] ;

[0147] Where f(i,j) means the flow on the edge (i,j) from node i to node j; This means the feasible flow from the virtual source s to the virtual sink t; s means the virtual source, representing the sum of all power plants; t means the virtual sink, representing the sum of all loads; this flow conservation equation indicates that the net flow out of the virtual source s is equal to the net flow into the virtual sink t, and the flow at intermediate nodes remains balanced.

[0148] c(i,j) represents the capacity of edge (i,j); E represents the set of all edges in the network; this capacity constraint equation means that the flow of any edge cannot exceed its capacity.

[0149] Step S3-4-4: Identify the location of the minimum cut and determine the system state by judging the type of the minimum cut: when the minimum cut falls in the edge set between the virtual source node and the power plant node, it is determined that the power supply is insufficient; when the minimum cut falls in the edge set of the power network, it is determined that the transmission capacity is limited; when the minimum cut falls in the edge set between the load substation and the virtual sink node, it is determined that the load demand is met.

[0150] Among them, the flow of the edge between the load substation and the virtual sink is the actual power supply at that time point, and the difference between the actual power supply and the actual active load demand is the load loss of the load substation; the flow of the edge between the virtual source and the power plant node is the active power output of the power plant at that time point.

[0151] Step S3-4-5: Based on the maximum flow calculation results, calculate the load loss of each node. The actual power supply of load node l is equal to the flow rate f(l,t) on edge (l,t). The load loss of load node l is... Calculate using the following formula:

[0152]

[0153] in, This refers to the amount of load loss at load node l at time h. f(l,t) represents the actual active load demand of load node l at time h; f(l,t) represents the flow of the edge from load node l to virtual sink t, i.e., the actual power supply.

[0154] Step S3-4-6: Calculate the total system load loss :

[0155]

[0156] in, This refers to the total load shedding of the system at time h. It means the set of all load nodes in the system;

[0157] Step S3-4-7: Output the flow rate f(i,j) of each branch, the load shedding amount of each load node, the total system load shedding amount, and the actual output of each power plant as well as the information of the minimum cut position.

[0158] Specifically, in step S3-4-3, the maximum flow algorithm is executed to solve the maximum feasible flow from the virtual source point s to the virtual sink point t, which specifically includes the following sub-steps:

[0159] Step S3-4-3-1: Initialize the flow rate of all edges in the network , and set the current feasible flow ;

[0160] Step S3-4-3-2: Use the breadth-first search method in the residual network to find an augmenting path p from the virtual source point s to the virtual sink point t;

[0161] Among them, the residual network means a network composed of all forward edges that satisfy and reverse edges that satisfy f(j,i)>0; the augmenting path means a path from the source point s to the sink point t, and all forward edges on the path satisfy f(i,j)<c(i,j), and all reverse edges satisfy f(j,i)>0;

[0162] Step S3-4-3-3: If there is an augmenting path p, determine the maximum flow rate that can be increased on this augmenting path :

[0163]

[0164] Among them, means the maximum flow rate that can be increased on the current augmenting path; means the set of forward edges on the augmenting path p; means the set of reverse edges on the augmenting path p; means the remaining capacity of the forward edge; f(j,i) means the current flow rate of the reverse edge;

[0165] Step S3-4-3-4: Update the flow rate along the augmenting path: For , update ; For , update ; At the same time, update the current feasible flow ;

[0166] Step S3-4-3-5: Repeat steps S3-4-3-2 to step S3-4-3-4 until there is no augmenting path from the virtual source point s to the virtual sink point t;

[0167] Step S3-4-3-6: The algorithm terminates, and the current feasible flow is the maximum feasible flow And satisfy:

[0168]

[0169] in, This refers to the maximum feasible flow from the virtual source node s to the virtual sink node t; This refers to the minimum cut capacity of the network. It means the set of nodes that can be reached in the residual network starting from the virtual source node s; This refers to the set of the remaining nodes in the network; the equation indicates that the maximum flow equals the minimum cut.

[0170] Step S3-4-3-7: Output the maximum feasible flow flow rate of each side and minimum cut set .

[0171] Through maximum flow calculation, the flow of the edge between the load substation and the virtual sink is the actual power supply at that point in time. The difference between the power supply and the demand power is the load shedding of the corresponding load substation. The flow of the edge between the virtual source and the power plant node is the active power output of the power plant at that point in time. For coal-fired, gas-fired, and oil-fired power plants, the primary energy consumption for the corresponding period can be calculated based on this active power output, and the primary capacity reserve of the power plant can be updated.

[0172] Maximum Flow Algorithm Principle:

[0173] The maximum flow minimum cut problem is a classic combinatorial programming problem in operations research.

[0174] Definition 1: Maximum Flow Minimum Cut Problem

[0175] The minimum cut maximum flow problem can be described as follows: In a network with capacity, given a starting point *s* and a ending point *t*, the maximum flow that the network can handle between these two points at any given time is the maximum flow. Define a graph G(V, E) containing nodes and arcs, where V is the set of nodes and E is the set of edges. Assume that all arcs in the network have weights, and define the weight as the arc's capacity *C*. The network graph also contains an initial flow *F*. This results in a network with both capacity and flow, denoted as G(V, E, C, F), where the initial flow of any arc in the network does not exceed its capacity. Define the flow *f*. st Let f be the flow from source s to sink t in network G(V, E, C, F). st Given a network G, the feasible flow with the largest flow among all feasible flows satisfying the given conditions is called the maximum flow, denoted as . In the above formula The edges are non-saturated edges, satisfying The edges are saturated edges. The definition of feasible flow satisfies capacity conservation, which is reflected in two aspects: firstly, all flow from source node s reaches sink node t; secondly, all nodes other than source node s and sink node t only pass through flow without storing flow.

[0176] Lemma 1: Arbitrary cut sets in capacity networks In feasible flow .

[0177] Lemma 2: The relationship between any feasible flow and its corresponding cut set in a capacity network is: .

[0178] Definition 2: Augmenting chains and their conditions:

[0179] Let the positive direction be from the source node s to the sink node t, and let p be a chain in the network with a starting point s and an ending point t. The chain contains several nodes, forward edges, and reverse edges. All forward edges on the augmenting chain p are non-saturated edges, and all reverse edges are non-zero edges. The necessary and sufficient condition for the existence of an augmenting chain in the network is that the feasible flow in the network is not a maximum flow.

[0180] Theorem 1: In a network, the maximum flow... Equal to minimum cut .

[0181] The basic idea of ​​the Ford-Fulkerson algorithm is to find augmenting paths under the constraint of flow conservation, and the algorithm terminates when no augmenting path can be found.

[0182] However, power networks have a graph structure, where power plants and substations can be abstracted as nodes of the grid, and transmission lines as edges. Power flow is transmitted from generator nodes to load nodes via these edges. However, algorithms like the Ford-Fulkerson algorithm are typically only applicable to maximum flow problems in single-source-single-sink networks, while real-world power networks usually contain numerous power plants and load substations. Clearly, each power plant is a source of injected power, while each load substation is a sink; thus, the power network is a multi-source-multi-sink network.

[0183] Therefore, the Ford-Fulkerson algorithm cannot be directly applied to the power network supply capacity assessment problem required by this invention. To address this, virtual source points and virtual sink points are introduced. Virtual edges are added between the virtual source points and each power plant node to establish a connection. These virtual edges are defined as unidirectional, allowing power flow only from the virtual source point to the power plant node, and their capacity is equivalent to the available generating capacity of the corresponding power plant. Similarly, virtual edges are added between the virtual sink points and each load substation to establish a connection. Likewise, these virtual edges are defined as unidirectional, allowing power flow only from the load substation to the virtual sink point, and their capacity is equivalent to the actual power demand of each load substation at each time point. Specifically, as follows... Figure 3 As shown.

[0184] Figure 3 middle, This represents the available generating capacity of power plant G1, and also the capacity of the one-way edge from the virtual source point to this power plant node. This represents the power capacity of the transmission line between hub substations T1 and T4, which is used as the capacity of this bidirectional side. This represents the active power load demand of the load substation L1 at a certain moment, and is used as the capacity of the one-way side from the load substation to the virtual load point.

[0185] By introducing virtual source and virtual sink points, the traditional Ford-Fulkerson algorithm becomes suitable for the power grid supply capacity assessment problem required in this project. It is worth noting that, for this problem, the algorithm results may yield three different outcomes, explained in detail below:

[0186] The minimum cut lies in the set of edges between the virtual source node and the power plant node, such as... Figure 4 As shown, this means that at this point in time, the total available generating capacity is less than the load demand, i.e., there is a shortage of power supply. Figure 4 ).

[0187] The minimum cut lies within the set of edges in the power network, such as... Figure 5 As shown, this means that at this point in time, due to insufficient power transmission capacity of the power grid, the power transmission at the minimum cutoff section is limited. This situation often occurs when the power grid is deliberately attacked, or when some key substations or transmission lines are out of service. Figure 5 ).

[0188] The minimum cut lies in the set of edges between the load substation and the virtual sink, such as... Figure 6 As shown, this occurs when the maximum current in the power network equals the total load demand, meaning the power supply can meet the load demand. Figure 6 ).

[0189] Through maximum flow calculation, the flow of the edge between the load substation and the virtual sink is the actual power supply at that point in time. The difference between the power supply and the demand power is the load shedding of the corresponding load substation. The flow of the edge between the virtual source and the power plant node is the active power output of the power plant at that point in time. For coal-fired, gas-fired, and oil-fired power plants, the primary energy consumption for the corresponding period can be calculated based on this active power output, and the primary capacity reserve of the power plant can be updated.

[0190] Step S3-5: Determine if a sabotage event exists at the current moment:

[0191] Check if there is a sabotage event at the current moment. If so, deduce the cascading failure situation. If not, calculate the consumption of primary energy reserves of coal-fired, gas-fired, and oil-fired power plants.

[0192] Step S3-6: Simulate cascading failure scenarios:

[0193] Based on the improved Cascade model, the cascading failures that will occur in the power network under the initial attack event are calculated.

[0194] Step S3-6-1: Load the current power grid state, including the node set V, branch set E, generator output PG, and load demand PD, and calculate the initial power flow distribution;

[0195] Step S3-6-2: Apply the initial attack fault according to the power attack event specified by the user: if the target is a line, remove the line from the branch set E; if the target is a substation, remove the substation and all its connected lines from the branch set E, and record the initial fault element set F0.

[0196] Step S3-6-3: Set the iteration counter k=1 and enter the cascade propagation iteration loop;

[0197] Step S3-6-4: Based on the current topology, calculate the power flow of each branch using a DC power flow model. The calculation formula is:

[0198] ;

[0199] in, This means the active power of line ij at the k-th iteration; line ij is the line between nodes i and j.

[0200] This refers to the voltage phase angles of nodes i and j at the k-th iteration. This refers to the reactance of line ij;

[0201] This refers to the set of branches after the k-th iteration.

[0202] Step S3-6-5: Calculate the load rate of each operating line. ;in, This refers to the load rate of line ij at the k-th iteration; This refers to the thermal stability limit or transmission capacity of line ij.

[0203] Step S3-6-6: Determine if an overloaded branch exists: If If all operating lines are confirmed to be overloaded, the system is stable, and the process proceeds to step S3-6-10; otherwise, if an overloaded branch exists, the process proceeds to step S3-6-7. This refers to the overload threshold; protection is triggered when the line load rate exceeds this value.

[0204] Step S3-6-7: Determine the branch to be disconnected this time based on the overload protection logic, including disconnecting the line with the highest load rate according to deterministic rules or disconnecting the line according to the tripping probability according to probabilistic rules. Cut off the line;

[0205] in, This means the line tripping probability; α means the sensitivity coefficient of the probabilistic tripping model, and e is the natural constant.

[0206] Step S3-6-8: Cut the selected branches and update the topology, removing the cut branches from the branch set. Remove from the grid and record the set of faulty components; if the grid is split into multiple islands after the disconnection, perform power flow calculations for each island and disconnect some loads according to priority when there is an imbalance between power generation and load in the island;

[0207] Step S3-6-9: Determine the iteration termination condition: If k≥Kmax, then determine that the system has crashed or cannot converge, and jump to step S3-6-10; otherwise, set k=k+1 and return to step S3-6-4 to continue iterating; where Kmax means the maximum number of iterations, used to prevent infinite loops;

[0208] Step S3-6-10: Statistically analyze the cascaded fault results, including the final topology and fault branch sequence. Total load loss The final state of the system and the final output values ​​of each thermal power plant; among which, This refers to the set of faulty components removed during the k-th iteration. This refers to the total load shedding. This refers to the active power load demand of node i. This refers to the actual power supplied to node i.

[0209] Step S3-7: Based on the cascading fault set obtained from the deduction, shut down the faulty branch / substation, update the power network model, recalculate the power flow of the power network after the topology change, and count the load loss after the cascading fault.

[0210] Step S3-8: Calculate the primary energy reserve consumption of coal-fired, gas-fired, and oil-fired power plants:

[0211] Based on the current active power output of each thermal power plant, calculate the consumption of primary energy sources such as coal, gas, and oil during the current period, and subtract the aforementioned consumption from the current primary energy reserves to update the primary energy reserve status of each thermal power plant.

[0212] Step S3-9: Update the loop variable in the intraday time series:

[0213] Update the loop variable in the intraday time series, setting h = h + 1. Check if h equals 24; if yes, update the loop variable in the outer energy blockade layer; otherwise, calculate the power flow of the power grid at the current moment.

[0214] Step S3-10: Update the loop variable of the outer layer of energy lockout:

[0215] Update the outer loop variable of the energy blockade, setting d = d + 1.

[0216] Step S3-11: Determine if the total duration of the time series deduction model has been reached:

[0217] Determine if the number of days exceeds the total number of days. If yes, output the time series simulation results; otherwise, determine whether to implement an energy replenishment.

[0218] Step S3-12: Determine whether to perform a primary energy replenishment:

[0219] Determine whether to implement primary energy replenishment. If yes, update the primary energy storage of each thermal power plant; otherwise, initialize the loop variables in the intraday time series.

[0220] Step S3-13: Update the primary energy storage of each thermal power plant:

[0221] For thermal power plants, the import volume of primary energy sources such as coal, gas, and oil is allocated according to the proportion of their remaining primary energy storage capacity. The primary energy storage of each thermal power plant is updated. After the update, the loop variables in the intraday time series are initialized, and the time series simulation results are output.

[0222] Specifically, in step S4, the power system time-series simulation results are output. Specifically:

[0223] Output the cumulative load loss, load loss period and power, etc.

[0224] The output power system time-series simulation results also include the time-series curves of primary energy reserve changes for each thermal power plant, the minimum cut position information of the power grid at each moment, and the cascading fault chain records after each attack event.

[0225] Example 2:

[0226] Taking a provincial power grid as an example, its topology, generator parameters, typical daily load curves, and wind / solar power output curves are obtained. The simulation duration is set to 30 days, simulating a coal import blockade event with an initial coal reserve of 70% of the normal value. Attack events targeting critical transmission lines are set on days 5 and 15.

[0227] According to the process described in this invention, cyclical simulations are performed sequentially at the outer (daily), middle (hourly), and inner (fault cascading) levels. During the simulation, power flow, fault sets, and energy reserves are updated in real time, and strategic energy replenishment is performed on days 10 and 20. After the simulation, the load loss time-series curves for each node, the total load loss, and changes in energy reserves are output, providing a reference for power grid emergency dispatch and energy allocation.

[0228] Key data for this embodiment are shown in Table 1:

[0229] Table 1. Schematic diagram of the deduction data

[0230]

[0231] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied 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.

[0232] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0233] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0234] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0235] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A power system time-series extrapolation method based on an improved OPA model, characterized in that, Includes the following steps: S1: Construct a three-layer extrapolation model consisting of an inner layer of cascading faults, a middle layer of intraday time series, and an outer layer of energy blockade; S2: Construct an input dataset based on power grid topology, power generation information, load curves, and renewable energy curves; S3: Determine the input parameters and input the input parameters and input dataset into the three-layer inference model, and execute the outer layer, middle layer and inner layer cyclic inference in sequence to realize the multi-time scale power system behavior simulation and obtain the power system time series inference results; S4: Output the power system time-series simulation results, including the load loss time-series curves of each node, the total load loss, and changes in energy reserves; In step S1, a three-layer extrapolation model is constructed, consisting of an inner layer of cascading failures, a middle layer of intraday time series analysis, and an outer layer of energy blockade. This includes: Step S1-1: Construct a cascading fault inner-layer simulation model. The cascading fault inner-layer simulation model considers power attack strategies, takes the attacked lines and substations as initial faults, and uses the improved CASCADE model to simulate the cascading fault situation of the power network on an hourly time scale. Step S1-2: Construct an intraday time-series mid-level extrapolation model. The intraday time-series mid-level extrapolation model considers the intraday load demand and the time-series changes of new energy sources, simulates the power network operation trend at the hourly time scale, and calculates the primary energy reserve consumption of thermal power plants. The new energy sources include wind power and photovoltaic power. Step S1-3: Construct an outer-layer simulation model of energy blockade, which simulates the process of limited energy supply, decreased storage, and strategic energy transfer on multi-day and monthly time scales; The steps to improve the CASCADE model for simulating cascading failures include: Step S3-6-1: Load the current power grid state, including the node set V, branch set E, generator output, load demand, and calculate the initial power flow distribution; Step S3-6-2: Apply the initial attack fault according to the power attack event specified by the user: if the target is a line, remove the line from the branch set E; if the target is a substation, remove the substation and all its connected lines from the branch set E, and record the initial fault element set F0. Step S3-6-3: Set the iteration counter k=1 and enter the cascade propagation iteration loop; Step S3-6-4: Based on the current topology, calculate the power flow of each branch using a DC power flow model; Step S3-6-5: Calculate the load rate of each operating line. ; Step S3-6-6: Determine if an overloaded branch exists: If If all operating lines are confirmed to be overloaded, the system is stable, and the process proceeds to step S3-6-10; otherwise, if an overloaded branch exists, the process proceeds to step S3-6-7. This refers to the overload threshold. Step S3-6-7: Determine the branch to be disconnected this time according to the overload protection logic, including disconnecting the line with the highest load rate according to deterministic rules or disconnecting the line according to the tripping probability according to probabilistic rules; Step S3-6-8: Cut off the selected branch and update the topology. Remove the cut-off branch from the branch set and record the set of faulty components. Step S3-6-9: If k≥Kmax, then determine that the system has crashed or cannot converge, and jump to step S3-6-10; otherwise, set k=k+1 and return to step S3-6-4 to continue iterating; where Kmax means the maximum number of iterations. Step S3-6-10: Statistically analyze the cascading fault results, including the final topology, fault branch sequence, total load loss, final system state, and final output of each thermal power plant.

2. The power system time series extrapolation method based on the improved OPA model according to claim 1, characterized in that, In steps S1-3, the primary energy supply restriction includes scenarios where imports of coal, natural gas, and fuel oil are blocked, and the strategic energy transfer includes strategies for supplying primary energy via transportation networks by train or tanker truck.

3. The power system time series extrapolation method based on the improved OPA model according to claim 2, characterized in that, In step S2, an input dataset is constructed based on the power grid topology, generation information, load curve, and renewable energy curve, including: Step S2-1: Obtain and input power network model information including substations, transmission lines, and station-line topology; Step S2-2: Obtain and input the maximum load power information of each load substation; Step S2-3: Obtain and input typical daily load curves of the entire network load for different months; Step S2-4: Obtain and input typical active power output curves for wind power and photovoltaic power in different months.

4. The power system time series extrapolation method based on the improved OPA model according to claim 3, characterized in that, In step S3, the input parameters are determined, and the input parameters and input dataset are input into the three-layer inference model. The outer, middle, and inner layers are sequentially cyclically inferred to simulate the behavior of the power system at multiple time scales, obtaining the power system time-series inference results, including: Step S3-1: Determine the input parameters, including the total number of days in the time series simulation, the primary energy import blockade strategy, the initial primary energy reserves of each thermal power plant, and the power attack events specified by the user; Step S3-2: Initialize the outer loop variable d=1 for the energy blockade; Step S3-3: Initialize the loop variable h=0 in the intraday time series layer; Step S3-4: Calculate the power flow of each branch in the power network at the current moment according to the improved maximum flow algorithm, and count the load loss at the current moment; Step S3-5: Check if there is a sabotage event at the current moment. If yes, proceed to step S3-6 to perform cascading fault simulation. If no, proceed to step S3-8 to calculate the primary energy reserve consumption. Step S3-6: Based on the improved CASCADE model, calculate the cascading faults that will occur in the power network under the initial attack event; Step S3-7: Based on the cascading fault set obtained from the deduction, shut down the faulty branch or substation, update the power network model, recalculate the power flow of the power network after the topology change, and count the load loss after the cascading fault. Step S3-8: Based on the active power output status of each thermal power plant at the current moment, calculate the primary energy consumption for the current period and subtract the consumption from the current primary energy reserve to update the primary energy reserve status of each thermal power plant. Step S3-9: Update the loop variable in the intraday time series, set h=h+1, and check if h is equal to 24. If yes, proceed to step S3-10 to update the outer extrapolation model variable of energy blockade; otherwise, proceed to step S3-4 to calculate the power grid flow at the next moment. Step S3-10: Update the outer loop variable of the energy blockade, set d=d+1; Step S3-11: Determine whether the current number of days d is greater than the total number of days in the time series simulation. If yes, output the time series simulation results, including the load loss time series curves of each node, the total load loss, and the changes in energy reserves. If no, proceed to step S3-12 to determine whether to implement a single energy replenishment. Step S3-12: Determine whether to perform an energy replenishment. If yes, proceed to step S3-13 to update the energy storage. If no, return to step S3-3 to initialize the loop variables in the intraday time series. Step S3-13: For thermal power plants, allocate the primary energy import volume according to the proportion of their remaining primary energy storage capacity, update the primary energy storage volume of each thermal power plant, and return to step S3-3 to initialize the loop variables in the intraday time series.

5. The power system time series extrapolation method based on the improved OPA model according to claim 4, characterized in that, In steps S3-4, based on the improved maximum flow algorithm, the power flow of each branch in the power network at the current moment is calculated, and the load shedding at the current moment is statistically analyzed, including: Step S3-4-1: Construct a flow network model of the power network, abstracting power plant nodes as source nodes, load substation nodes as sink nodes, transmission lines as edges, and using the thermal stability limit of the lines as the capacity of the edges to form a multi-source and multi-sink network. Step S3-4-2: Introduce virtual source node s and virtual sink node t to convert the multi-source multi-sink network into a single-source single-sink network: Add a directed edge (s,g) from the virtual source node s to each power plant node g, with its capacity c(s,g) set to the available power generation capacity of power plant g. Add a directed edge (l,t) from each load node l to the virtual sink t, with its capacity c(l,t) set to the actual active power load demand of load node l at the current moment. ; Step S3-4-3: Execute the maximum flow algorithm to find the maximum feasible flow from the virtual source node s to the virtual sink node t. The maximum feasible flow is equal to the minimum cut. The maximum flow algorithm solves for the maximum feasible flow that satisfies the following flow conservation and capacity constraints: ； ; Where f(i,j) means the flow on the edge (i,j) from node i to node j; This means the feasible flow from the virtual source s to the virtual sink t; s means the virtual source, representing the sum of all power plants; t means the virtual sink, representing the sum of all loads; this flow conservation equation indicates that the net flow out of the virtual source s is equal to the net flow into the virtual sink t, and the flow at intermediate nodes remains balanced. c(i,j) represents the capacity of edge (i,j); E represents the set of all edges in the network; this capacity constraint equation means that the flow of any edge cannot exceed its capacity. Among them, the flow of the edge between the load substation and the virtual sink is the actual power supply at that time point, and the difference between the actual power supply and the actual active load demand is the load loss of the load substation; the flow of the edge between the virtual source and the power plant node is the active power output of the power plant at that time point. Step S3-4-4: Identify the location of the minimum cut and determine the system state by judging the type of the minimum cut: when the minimum cut falls in the edge set between the virtual source node and the power plant node, it is determined that the power supply is insufficient; when the minimum cut falls in the edge set of the power network, it is determined that the transmission capacity is limited; when the minimum cut falls in the edge set between the load substation and the virtual sink node, it is determined that the load demand is met. Step S3-4-5: Based on the maximum flow calculation results, calculate the load loss of each node. The actual power supply of load node l is equal to the flow rate f(l,t) on edge (l,t). The load loss of load node l is... Calculate using the following formula: ； in, This refers to the amount of load loss at load node l at time h. f(l,t) represents the actual active load demand of load node l at time h; f(l,t) represents the flow of the edge from load node l to virtual sink t, i.e., the actual power supply. Step S3-4-6: Calculate the total system load loss : ； in, This refers to the total load shedding of the system at time h. It means the set of all load nodes in the system; Step S3-4-7: Output the flow rate f(i,j) of each branch, the load loss of each load node, the total load loss of the system, and the actual output of each power plant. And the minimum cut position information.

6. The power system time-series extrapolation method based on the improved OPA model according to claim 5, characterized in that, In step S3-4-3, the maximum flow algorithm uses the Ford-Fulkerson algorithm.

7. The power system time series extrapolation method based on the improved OPA model according to claim 6, characterized in that, In step S3, the multi-timescale power system behavior simulation includes: Simulate the cascading failure propagation process on an hourly timescale within the inner layer of the cascading failure; The operation trend of the power network is simulated in the intraday time series at the hourly time scale, where the intraday hourly time scale is 24 hours. Simulate energy supply constraints on a multi-day, monthly timescale outside the energy blockade layer.

8. The power system time series extrapolation method based on the improved OPA model according to claim 4, characterized in that, Steps S3-8 involve calculating the primary energy reserve consumption of thermal power plants, including: Based on the classification of coal-fired power plants, gas-fired power plants, and oil-fired power plants, calculate the coal consumption, gas consumption, or oil consumption corresponding to the active power output of each type of power plant in the current period. The conversion between primary energy consumption and active power output is obtained based on the unit energy consumption coefficient of the power plant.

9. The power system time series extrapolation method based on the improved OPA model according to claim 8, characterized in that, In step S3-13, the primary energy storage of each thermal power plant is updated, including: Determine the proportion of each thermal power plant's primary energy surplus storage capacity to the total primary energy surplus storage capacity of all thermal power plants; According to the aforementioned proportion, the total amount of allocable primary energy imports will be allocated to each thermal power plant. The allocated primary energy imports are added to the current primary energy storage of each thermal power plant to complete the update.

Images