Load power balance regulation and control method based on knowledge graph and random sequence optimization

Abstract

The invention discloses a load power balance regulation and control method and device based on knowledge graph and random sequence optimization, a medium and a product, and relates to the field of power system dispatching, and the method comprises the steps: constructing a power grid topology knowledge graph according to power grid physical equipment and corresponding topology connection and operation constraints; performing probability distribution fitting on the power output of the renewable energy according to the output data to obtain the output of the renewable energy; obtaining a plurality of typical output scenes; taking the state of the power grid physical equipment after fault isolation as an initial state, and constructing a random Markov decision process; searching in a state space by using an improved heuristic search algorithm by taking an initial state as a starting point, minimizing an accumulated cost function as a target, taking a hard constraint as a constraint condition and taking a topological distance as a heuristic function to obtain an optimal operation sequence; verifying the optimal operation sequence to obtain a final operation sequence; the power of the power grid is regulated and controlled; according to the invention, the fault response speed can be improved, and rapid adaptive decision in a complex scene can be realized.

Description

Technical Field

[0001] This application relates to the field of power system dispatching, and in particular to a load power balance control method, equipment, medium, and product based on knowledge graph and random sequence optimization. Background Technology

[0002] With the rapid development of ultra-high-voltage power grids and the large-scale grid connection of renewable energy sources such as wind and solar power (reaching over 50% in some areas), the operating characteristics of the power grid have undergone fundamental changes. Renewable energy is highly random and volatile, and its output is closely related to meteorological conditions. Frequent extreme weather events (ice storms, strong winds, low radiation) cause drastic fluctuations in renewable energy output within a short period, even resulting in significant drops. When such a complex system fails, the rapid isolation of the fault point by protection devices inevitably disrupts the original power balance, causing power outages in some areas or loss of load for certain generators.

[0003] The process from the occurrence of a fault to the eventual restoration of power typically involves several key time phases: millisecond-level relay protection action (<100ms), second-level automatic stabilization device response (<1s), minute-level manual decision-making (1-5min), and hour-level reconfiguration scheme (>10min). In the second-to-minute phase following fault isolation, dispatchers need to: quickly determine the scope of the power outage and affected equipment; assess the power deficit based on current generation output and load demand; select the optimal solution from numerous possible restoration paths; and coordinate the timing of multiple dispatch actions. The complexity of this decision-making process increases exponentially with the expansion of the power grid.

[0004] However, the effectiveness of control measures is often conditional when extreme weather causes fluctuations in renewable energy output. A load dispatch scheme that works when winds are of moderate intensity may fail when winds suddenly weaken, leading to insufficient power generation. Traditional methods relying on human experience and text queries struggle to generate safe and economical solutions within a limited timeframe.

[0005] Specifically, existing methods face the following limitations: Rule-based and experience-based methods require dispatchers to meticulously examine possible operational sequences based on power system operating procedures and accumulated accident response plans. This approach relies heavily on personnel skill, and as system complexity increases, humans struggle to process massive amounts of information quickly. When using single numerical optimization algorithms (genetic algorithms, particle swarm optimization, etc.) to solve load dispatching or generator coordination problems, most neglect system topology constraints, potentially resulting in solutions that violate interlocking conditions. They also struggle to handle the randomness of renewable energy output and have long convergence times, failing to meet the decision-making time requirement of less than 500 milliseconds. Furthermore, these methods typically optimize only for specific operating conditions, making it difficult to fully assess the adaptability of solutions to different output scenarios. Machine learning end-to-end models use deep learning or large models for end-to-end policy learning. Although the decision-making speed is fast, the decision-making process is a "black box" mode, making it difficult to explain the generation logic of operation suggestions to dispatchers. It is also difficult to effectively encode and guarantee the hard requirements such as physical constraints and safety regulations of the power system. At the same time, model training relies on a large amount of high-quality labeled historical case data, which is often difficult to obtain under certain rare or extreme operating conditions.

[0006] Therefore, based on the above problems, there is an urgent need to provide a load power balance control method based on knowledge graph and random sequence optimization, which can improve fault response speed and realize rapid adaptive decision-making in complex scenarios. Summary of the Invention

[0007] The purpose of this application is to provide a load power balance control method, device, medium and product based on knowledge graph and random sequence optimization, which can improve fault response speed and realize rapid adaptive decision-making in complex scenarios.

[0008] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a load power balance regulation method based on knowledge graphs and random sequence optimization, including: A power grid topology knowledge graph is constructed based on the power grid physical devices and their corresponding topology connections and operational constraints; the power grid topology knowledge graph uses power grid physical devices as nodes and topology connections and operational constraints as edges; The power output of renewable energy is obtained by fitting a probability distribution to the power output data; and several typical power output scenarios are obtained based on the power output data; the power output data includes: historical wind power data or weather forecast data; Based on the power grid topology knowledge graph and the renewable energy output, a stochastic Markov decision process is constructed with the state of the power grid physical equipment after fault isolation as the initial state. The state space of the stochastic Markov decision process includes: circuit breaker state, bus voltage, generator output, load consumption, and renewable energy output; the action space includes topology operation, load scheduling, and generator regulation; and the state transition function includes topology update, power adjustment, and power flow calculation after the action is executed. Starting from the initial state, aiming to minimize the cumulative cost function, using hard constraints as conditions, and topological distance as a heuristic function, an improved heuristic search algorithm is employed to search the state space and obtain the optimal operation sequence. The hard constraints include power balance constraints, bus voltage constraints, line capacity constraints, generator constraints, ramping constraints, interlocking constraints, and the N-1 safety criterion. The improved heuristic search algorithm includes: a best-first search method or... algorithm; The optimal operation sequence is verified based on the multiple typical power output scenarios to obtain the final operation sequence; and the power grid is regulated using the final operation sequence.

[0009] Optionally, the step of constructing a power grid topology knowledge graph based on power grid physical devices and their corresponding topology connections and operational constraints specifically includes: Using formula Constructing a power grid topology knowledge graph ; in, For the set of vertices, , Let be the set of edges. , It is a set of physical equipment nodes of the power grid. For the set of constraint nodes, It is a set of topologically related edges. To run the set of constraint edges.

[0010] Optionally, the step of fitting a probability distribution of the power output of renewable energy based on the output data to obtain the renewable energy output specifically includes: Using formula Probability distribution fitting is performed on the power output of renewable energy sources; in, for Renewable energy output at all times It can be a normal distribution, a Beta distribution, or a Weibull distribution. The mean of the distribution reflects the average output level. The standard deviation reflects the magnitude of force fluctuation.

[0011] Optionally, the above describes several typical power output scenarios based on renewable energy output, specifically including: Based on renewable energy output, multiple typical output scenarios are obtained using the percentile method, Gaussian quadrature method, or empirical method; and each typical output scenario is assigned a corresponding probability weight.

[0012] Optionally, the step of constructing a stochastic Markov decision process based on the power grid topology knowledge graph and the renewable energy output, with the state of the power grid physical equipment after fault isolation as the initial state, specifically includes: Using formula Determine the state space of power grid physical devices ; Using formula Determine the operating space of power grid physical equipment ; Using formula Determine the state transition function; in, This is the circuit breaker state matrix. The bus voltage vector. To provide power to the generator, For load consumption, Contribute to renewable energy For topological actions, For load dispatching actions, For engine adjustment actions, This refers to the state of the power grid physical equipment after the state transition. This is the state transition function. for The state at any given moment, for Momentary actions for Renewable energy output at all times Optionally, the step of using an improved heuristic search algorithm to search the state space, starting from the initial state, aiming to minimize the cumulative cost function, using hard constraints as constraints, and using topological distance as a heuristic function, to obtain the optimal operation sequence, specifically includes: Using formula Determine the cumulative cost function ;in, This is the weighting coefficient for power imbalance. For power imbalance, The weighting coefficient for the weighted load loss is... For weighted load loss, This is a weighting coefficient for operational complexity. For operational complexity, To restore the time weighting coefficient, This refers to the recovery time.

[0013] Using formula Determine the power balance constraints; where, For the first The active power output of the generator, For the first The active power consumption of a load, For the first The actual output of each renewable energy site For power grid system losses, The total load requirement of the power grid system; Using formula Determine the bus voltage constraints; where For the first Rated voltage of each busbar For the first Voltage amplitude of each busbar; Using formula Determine the line capacity constraints; among which, For the first The transmission power of each line, For the first The thermal stability limit of the line; Using formula Determine generator constraints; where, For the first The maximum active power output of the generator. For the first The minimum active power output of the generator; Using formula Determine the climbing constraints; among which, For the first The change in the active power output of the generator. For time intervals, For the first The maximum ramp rate of the generator; Using formula Determine the heuristic function ;in, The topological distance weighting coefficient. This represents the topological distance from the currently de-energized busbar to the nearest energized busbar. These are the weighting coefficients for the operands. Operands that have already been executed. This refers to the state space of the power grid physical equipment.

[0014] Optionally, the step of verifying the optimal operation sequence based on the multiple typical output scenarios to obtain the final operation sequence specifically includes: The optimal operation sequence is then applied to multiple typical power output scenarios to obtain the corresponding topological relationships and load states. Based on the topology and load conditions, determine the number of typical output scenarios that satisfy the hard constraints; and determine the success rate; the success rate is the ratio of the number of typical output scenarios that satisfy the hard constraints to the total number of typical output scenarios. Based on the success rate, the final operation sequence is obtained.

[0015] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described load power balance regulation method based on knowledge graph and random sequence optimization.

[0016] Thirdly, this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described load power balance regulation method based on knowledge graph and random sequence optimization.

[0017] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described load power balance regulation method based on knowledge graph and random sequence optimization.

[0018] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a load power balance regulation method, device, medium, and product based on knowledge graph and random sequence optimization. It constructs a power grid topology knowledge graph based on the physical equipment of the power grid and its corresponding topological connections and operational constraints. Through structured semantic association, it achieves computability and reasoning of knowledge. It fits the probability distribution of renewable energy power output based on output data to obtain renewable energy output. Based on the power grid topology knowledge graph and renewable energy output, and using the state of the power grid physical equipment after fault isolation as the initial state, it constructs a stochastic Markov Decision Process (MDP) to improve the adaptability of the scheme to extreme weather and renewable energy fluctuations. Starting from the initial state, with the goal of minimizing the cumulative cost function, using hard constraints as constraints, and topological distance as a heuristic function, it employs an improved heuristic search algorithm to search in the state space, ensuring that all operation sequences satisfy topological constraints and safety regulations. The optimal operation sequence is verified according to multiple typical output scenarios to obtain the final operation sequence, which is then used to regulate the power grid. This enables the application to respond quickly to faults and improves the feasibility and robustness of the scheme. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a load power balance control method based on knowledge graph and random sequence optimization in one embodiment of this application. Figure 2 A flowchart illustrating a load power balance regulation method based on knowledge graph and random sequence optimization, provided as an embodiment of this application; Figure 3 This is a flowchart illustrating the determination of multiple typical power output scenarios in one embodiment of this application; Figure 4 This is a flowchart of a pruning strategy in one embodiment of this application; Figure 5 This is a comparison diagram between the method in one embodiment of this application and a conventional method. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] In one exemplary embodiment, such as Figure 1 and Figure 2 As shown, a load power balance regulation method based on knowledge graph and random sequence optimization is provided, including the following S1 to S5. Wherein: S1: Construct a power grid topology knowledge graph based on the power grid physical equipment, corresponding topology connections, and operational constraints.

[0024] The physical equipment of a power grid includes buses, lines, transformers, circuit breakers, generators, and loads. Operational constraints include series constraints (IN_SERIES), interlocking constraints (INTERLOCKS), protection constraints (PROTECTS), and supply constraints (SUPPORTED_BY). A power grid topology knowledge graph, or simply a knowledge graph, is constructed using these physical equipment as nodes and topology connections and operational constraints as edges. This knowledge graph is stored in a graph database (such as Neo4j). Furthermore, the static attributes of the power grid topology knowledge graph include the rated capacity, installation location, and maximum ramp rate of generators / PV panels.

[0025] Bus nodes include attributes such as voltage level, rated voltage, and priority; line nodes include attributes such as start point, end point, impedance, capacity, and protection zone; circuit breaker nodes include attributes such as location, status (open / closed), and operating time; generator nodes include attributes such as installed capacity, output range, and ramp rate; and load nodes include attributes such as capacity, adjustable range, and priority weight. Topology links are series connections between buses, circuit breakers, and equipment, i.e., physical conductor connections. An actual connection is only considered complete when all intermediate circuit breakers are closed. Series constraints prohibit certain operational combinations; interlocking constraints require at least one of two mutually exclusive switches to be open; protection constraints ensure automatic tripping of the protected area and protected equipment during a fault; power supply constraints quickly locate backup power paths during the search process, guiding the generation of recovery schemes and tending towards a "logical guidance relationship" rather than a simple "restrictive constraint"; power supply constraints refer to the power supply path between the load and the main power source, playing a heuristic role in the search algorithm and accelerating the recovery of power supply paths.

[0026] Power Grid Topology Knowledge Graph It is a directed graph used to describe the physical structure and logical constraints of the power grid, and its expression is as follows: ; in, For the set of vertices, , Let be the set of edges. , It is a set of physical device nodes in the power grid, representing physical entities in the power grid. A set of constraint nodes, representing logical objects such as protected areas, etc. This is a set of topological relationship edges, mainly containing electrical connection relationships (CONNECTS_TO), used to describe the physical connectivity between devices. This is a set of operational constraint edges, including series constraints, interlocking constraints, protection constraints, and power supply constraints, used to describe operational rules and safety constraints.

[0027] The detailed definitions of power grid physical equipment are shown in Table 1: Table 1. Detailed Definitions of Power Grid Physical Equipment

[0028] Detailed definition of topology connection: set up Indicates busbar Through the circuit breaker Connect to the line This edge means that only when... When in a closed state, and Only then was it electrically connected.

[0029] Detailed definition of the constraint edges: Set interlocking constraint edges This indicates that there is an interlocking relationship between the two switches; when the two switches cannot be closed simultaneously, [the interlocking relationship will occur]. It means that when When the switch is closed, ,when When closed, When the two switches are in a sequential relationship, that is... It must be closed first. Only then can it be closed, with Expression.

[0030] S2: Based on the output data, the power output of renewable energy is fitted with a probability distribution to obtain the output of renewable energy; and based on the output of renewable energy, several typical output scenarios are obtained.

[0031] S2 specifically includes: S21: Determine the output of renewable energy.

[0032] Power output data includes historical wind power data or weather forecast data. Historical wind power data includes historical active power output time series of wind farms, historical wind speed records, and solar irradiance series. Weather forecast information includes predicted wind speed, predicted wind direction, and predicted solar irradiance. Let... for Renewable energy output at any given time, based on historical data (Usually K≥30 days), it can be fitted to a probability distribution: ; in, It can be a normal distribution, a Beta distribution, or a Weibull distribution. The mean of the distribution reflects the average output level. , For historical data days, For the first The renewable energy output of the day, The standard deviation reflects the magnitude of force fluctuation. .

[0033] S22: Identify typical power output scenarios.

[0034] Based on renewable energy output, several typical output scenarios are obtained using methods such as percentile method, Gaussian quadrature method, or empirical method. Each scenario is assigned a corresponding probability weight, a renewable energy probability model for that period is constructed, and the generated scenario set is stored.

[0035] like Figure 3 As shown, this application uses the percentile method (selecting specific percentiles from historical data, such as the output values ​​corresponding to 10%, 30%, 50%, 70%, and 90% as scene nodes) to generate multiple typical output scenes. ,in, This represents the total number of typical power output scenarios. Each scenario is assigned a probability weight. ,satisfy To ensure the statistical representativeness of the scenarios, Numbered as typical power output scenarios, the first A typical power output scenario The expression is as follows: ; in, It is the inverse function of the distribution (quantile function). Percentile.

[0036] The five generated scenarios represent typical levels of output: very low, lower, medium, higher, and very high.

[0037] For typical output scenarios obtained using the percentile method, corresponding probability weights are derived, where... In this application, the probability weight is the area of ​​the scene interval. This allocation results in the highest weight for medium-level scenes (40%) and the lowest weight for extreme scenes (10% each), which is consistent with the characteristics of actual distribution.

[0038] S3: Based on the power grid topology knowledge graph and renewable energy output, construct a stochastic Markov decision process with the initial state of the power grid physical equipment after fault isolation.

[0039] Specifically, the state space in a stochastic Markov decision process This includes circuit breaker status, bus voltage, generator output, load consumption, and renewable energy output. Renewable energy output is a time-varying variable, determined by real-time monitoring data (Supervisory Control and Data Acquisition, SCADA), i.e., current historical wind power data or weather forecast information, within the state space. The calculation formula is as follows: ; in, This is the circuit breaker state matrix. , The bus voltage vector. , To provide power to the generator, , For load consumption, , Contribute to renewable energy , For the first The state matrix of each circuit breaker For the number of circuit breakers, For the first Voltage magnitude vector of each bus, Number of busbars For the first The active power output vector of each generator This represents the number of generators. For the number of loads, For the first The active power consumption vector of each load, For the first The output vector of a renewable energy source, This refers to the number of renewable energy power plants.

[0040] Action space in Markov decision-making process This includes topology operations, load dispatching, and generator regulation. The calculation formulas are as follows: ; in, For topological actions, , For load dispatching actions, , The adjustable load range is reflected in The upper and lower limit settings are used to constrain the true adjustable range of different loads. Loads can be divided into three categories according to their adjustment flexibility: important loads, ordinary loads, and flexible loads. The adjustable range of important loads is 5%-20%, such as important facility loads, which are not prone to large changes. The adjustable range of ordinary loads is 15%-50%, such as general commercial or industrial loads. The adjustable range of flexible loads is 50%-100%, and these are mostly movable or interruptible loads. The engine adjustment action is constrained by the climbing rate. , , For the first The change in the active power output of the generator. For time intervals, For the first The maximum climbing rate of the generator. For the first Topology action vector of each circuit breaker For the first The active power dispatching action vector for each load, For the first The active power regulation action vector of each generator. When When, it indicates the circuit breaker Turn on, When, it indicates the circuit breaker Remain unchanged, when When, it indicates the circuit breaker. closure.

[0041] The state transition function in a Markov decision process includes topology updates, power adjustments, and power flow calculations after actions are performed. The calculation formula is as follows: ; in, This refers to the physical status of the power grid equipment after fault isolation. This is the state transition function. for The state at any given moment, for Momentary actions for Renewable energy output at all times.

[0042] Topology update: According to topological actions Update the circuit breaker status, recalculate the bus connectivity matrix, and determine the set of de-energized buses. The calculation formula is as follows: ; in, Based on the updated circuit breaker state matrix, recalculate the bus connectivity matrix. ,in Indicates busbar and Connected; otherwise, 0.

[0043] Power adjustment: The power values ​​of each node are updated by applying load dispatching and generator regulation actions. The three-level load dispatching strategy is as follows: when the deficit is less than 10% of the total load, priority is given to adjusting the transferable load deficit; when the deficit is between 10% and 30%, adjustment plus partial reduction is applied; when the deficit is greater than 30%, emergency reduction plus tiered control is applied. The calculation formula is as follows: ; ; in, for The generator output at any given time for The generator output at any given time for Load consumption at any time for Load consumption at any time This represents the active power output adjustment of the generator over the time interval t to t+1. This variable is typically limited by the generator's maximum ramp rate and reflects the generator's responsiveness to system power balance. This variable represents the power adjustment of the load within the time interval t to t+1. It reflects the dynamic adjustment capability of the adjustable load to the system load demand and is limited by the adjustable range and type of the load.

[0044] Power flow calculation: The new bus voltage and line power are calculated using either DC or AC power flow algorithms. DC power flow is used for rapid estimation, while AC power flow is used for accurate verification. The DC power flow model is as follows: ; in For the first The phase angle of the busbars of the two busbars For line impedance, Power is injected into the bus.

[0045] Based on the new topology, loads, generators, and renewable energy output, run the power flow algorithm: ; in, for The system network loss at any time, for The bus voltage vector at time t.

[0046] S4: Starting from the initial state, with the goal of minimizing the cumulative cost function, using hard constraints as constraints, and topological distance as the heuristic function, an improved heuristic search algorithm is used to search in the state space to obtain the optimal operation sequence.

[0047] The heuristic search process based on the power grid topology knowledge graph is as follows: A best-first search is employed, starting with the system state after fault isolation. Topological distance is used as a heuristic function to guide the search, while the power grid topology knowledge graph constrains the range of legal actions. This allows for a rapid search in the state space for the optimal sequence of operations that satisfies all hard constraints. This is the standard usage of heuristic search. The state after fault isolation, i.e., the initial state, is the "root node" of the search tree. The goal is to find a path (sequence of operations) leading to the "goal state" (i.e., a state with power balance and satisfied constraints).

[0048] Specifically, the optimization objective is set as minimizing the cumulative cost function, which includes weighted indicators such as power imbalance, load loss, operational complexity, and recovery time. The calculation formula is as follows: ; in, The target for power imbalance is to reduce it to <1%. , For weighted load loss, important loads have higher weights. , For load Priority weights are assigned, with larger values ​​for critical loads. To account for operational complexity, including the number of switch operations and the number of interlock checks, , This refers to the number of switch operands, i.e., the number of circuit breaker operations in the action sequence. Interlock check count refers to the number of interlock rules that need to be verified when an action is executed. The number of load adjustments refers to the number of times the load power is adjusted during the dispatching process. Recovery time includes operation time and settling time. , This represents the total time consumed by the switching operation. For the first The time consumed by each switch operation For load response time, For generator adjustment time, This is the weighting coefficient for power imbalance. The weighting coefficient for the weighted load loss is... This is a weighting coefficient for operational complexity. To restore the time weighting coefficient, The standard configuration is .

[0049] Hard constraints include power balance constraints, bus voltage constraints, line capacity constraints, generator constraints, ramping constraints, interlocking constraints, and the N-1 safety criterion.

[0050] The formula for calculating the power balance constraint is as follows: ; in, For the first The active power output of the generator, For the first The active power consumption of a load, For the first The actual output of each renewable energy site For power grid system losses, This refers to the total load requirement of the power grid system.

[0051] The formula for calculating bus voltage constraints is as follows: ; in For the first Rated voltage of each busbar For the first Voltage amplitude of each bus.

[0052] The formula for calculating line capacity constraints is as follows: ; in, For the first The transmission power of each line, For the first The thermal stability limit of the line.

[0053] The formula for calculating generator constraints is as follows: ; in, For the first The maximum active power output of the generator. For the first The minimum active power output of the generator.

[0054] The formula for calculating the ramp constraint is as follows: ; in, For time intervals, For the first The maximum ramp rate of the generator.

[0055] Interlocking constraints: The logical constraints corresponding to the INTERLOCKS relationships in the power grid topology knowledge graph must be satisfied. The N-1 criterion states that the system can still operate stably after the failure of any single device.

[0056] This application uses topological distance as a heuristic function to guide the search. The formula for calculating the heuristic function is as follows: ; in, The topological distance weighting coefficient. , The topological distance from the currently de-energized bus to the nearest energized bus is calculated using BFS. These are the weighting coefficients for the operands. , Operands that have already been executed. This refers to the state space of the power grid physical equipment.

[0057] Improved heuristic search algorithms include Best-First Search (BFS) or The algorithm employed in this application is a best-first search algorithm. The initial state is the system state after fault isolation, and the target state is a state of power balance with no load loss. The search is performed in the state space, and a three-layer pruning strategy is implemented during the search process, such as... Figure 4 As shown, it includes hard constraint filtering, power grid topology knowledge graph constraint filtering, heuristic scoring and Top-K retention to ensure decision time <500ms.

[0058] The pruning strategy employs a three-tiered approach: first, hard constraint filtering, where any state violating hard constraints (power balance, voltage, capacity, etc.) is immediately discarded; second, grid topology knowledge graph constraint filtering, which, during action generation, queries the INTERLOCKS relationships in the grid topology knowledge graph to generate only legal actions that satisfy interlocking conditions, effectively eliminating illegal actions through relationship checks; and third, heuristic scoring and Top-K retention, which limit the search queue length and search depth, and control search time to ensure a total computation time of <500ms. The calculation formula is as follows: ; in, The time required for constructing and constraining the power grid topology knowledge graph includes querying the grid equipment connection relationships from the graph database and performing preliminary pruning operations such as interlocking constraint filtering. The generation and loading time for typical power output scenarios involves calculating and generating multiple renewable energy power output scenarios and their probability weights based on probability distributions. The execution time of the heuristic search algorithm is determined by the core state-space search and pruning evaluation process, which determines the optimal sequence of operations. To verify the operation sequence in multiple scenarios, the feasibility of the optimal solution is verified under different typical scenarios, including steps such as power flow calculation.

[0059] S5: Verify the optimal operation sequence based on multiple typical power output scenarios to obtain the final operation sequence; and use the final operation sequence to regulate the power grid.

[0060] S51: The optimal operation sequence is applied to multiple typical power output scenarios to obtain the corresponding topological relationships and load states.

[0061] The optimal operation sequence obtained from the search is then applied to several typical power output scenarios, and power flow calculations are performed for each: the output of renewable energy is set to... Determine the topology and load status, run the power flow algorithm to calculate and check whether all hard constraints are met, and record the feasibility indicators of typical output scenarios.

[0062] S52: Based on the topology and load conditions, determine the number of typical output scenarios that satisfy the hard constraints; and determine the success rate.

[0063] Success rate The ratio of the number of typical output scenarios that satisfy hard constraints to the total number of typical output scenarios is calculated using the following formula: ; To meet the number of typical output scenarios with hard constraints, This represents the total number of typical power output scenarios.

[0064] S53: Based on the success rate, obtain the final operation sequence.

[0065] When the success rate is greater than 90%, the plan is accepted, and its expected load loss is calculated to determine the ranking of the plans. The calculation formula is as follows: ; in, For the expectation, For load loss, For the first The probability weights of a typical output scenario For the first Load loss in a typical power output scenario.

[0066] Among the candidate schemes that meet the robustness requirements, they are ranked according to their comprehensive scores to determine the final operation sequence, expected recovery time, power balance, and scheme confidence.

[0067] In summary, this application employs heuristic search and multi-layer pruning, keeping the decision-making time below 500ms, thus improving response speed and ensuring rapid emergency decision-making. By combining a power grid topology knowledge graph with hard constraints, the final operation sequence satisfies 100% of the hard constraints, enhancing the application's security. Verification using multiple typical output scenarios ensures the feasibility of the final operation sequence in over 90% of scenarios, improving the application's robustness and adaptability. Furthermore, the application achieves a power balance accuracy greater than 99% (error <1%) and a load dispatch optimization degree greater than 85%, representing a significant improvement compared to equal load shedding. Simultaneously, the clear rules, standard interfaces, and lightweight algorithm design make the decision-making process traceable and easily integrated into existing systems, improving the application's interpretability and implementability. This application is suitable for dispatching renewable energy sources experiencing severe fluctuations due to extreme weather (ice storms, strong winds), rapid power balance restoration after line fault isolation, emergency control of high-proportion renewable energy grid-connected systems, and emergency recovery from regional power outages.

[0068] In one exemplary embodiment, such as Figure 5 As shown, if a power imbalance is caused by a line fault, assuming a three-phase short circuit occurs in LINE_1-2 in an IEEE 14-bus system, the protection device quickly isolates the circuit, resulting in: bus BUS_2 losing power; load LOAD_2 (50MW) losing power; and generator GEN_3 losing part of its load.

[0069] When using traditional methods, dispatchers need 2-3 minutes to determine the operation sequence by consulting the procedures: restore power to BUS_2 via LINE_2-3 or LINE_2-4, while also disconnecting 50MW of load to balance power and adjusting the output of GEN_1 and GEN_2.

[0070] When using this application for processing, the method of this application can provide the optimal solution within <500ms: close CB_2_3 → restore BUS_2 power supply; adjust LOAD_3 to 80% of the original (reduced by 20MW); adjust LOAD_4 to 60% of the original (reduced by 20MW); adjust LOAD_5 to 85% of the original (reduced by 10MW); increase GEN_1 output by 15MW, and increase GEN_2 output by 15MW; multi-scenario verification: the solution has a 100% success rate in 5 wind power output scenarios.

[0071] In summary, the efficiency of this application is significantly better than that of traditional methods, and the regulation quality is higher: it adopts distributed load regulation instead of centralized disconnection, reducing the outage area; at the same time, it has good robustness: it is feasible under various wind power output conditions.

[0072] In one exemplary embodiment, the issue is the fluctuation of renewable energy caused by extreme weather. Assuming that wind power accounts for 40% (500MW) of the system's installed capacity, and at a certain time the wind speed suddenly drops from 10m / s to 2m / s, the wind power output instantly drops from 400MW to 80MW (a decrease of 320MW).

[0073] Traditional fixed plans cannot adapt to such significant changes in power output. A plan formulated for "moderate wind" may result in power imbalance when executed for "weak wind". This application, however, can identify the current power output scenario (weak wind) in real time and perform rapid planning to search for the optimal load dispatch scheme: identifying a total of 300MW of adjustable load; prioritizing and adjusting 80MW of load; adjusting 240MW of generators based on ramping capability; robust verification: verified under three scenarios: weak wind, moderate wind, and strong wind; weak wind: feasible (successful); moderate wind: feasible (successful); strong wind: feasible (successful); overall success rate: 100%.

[0074] In summary, this application can maintain stable operation even under extreme conditions of drastic power fluctuations, with frequency deviation controlled within ±0.5Hz.

[0075] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores load power balancing control data based on knowledge graphs and random sequence optimization. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a load power balancing control method based on knowledge graphs and random sequence optimization.

[0076] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0077] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0078] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0079] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0080] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0081] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A load power balance regulation method based on knowledge graph and random sequence optimization, characterized in that, The load power balance regulation method based on knowledge graphs and random sequence optimization includes: A power grid topology knowledge graph is constructed based on the power grid physical devices and their corresponding topology connections and operational constraints; the power grid topology knowledge graph uses power grid physical devices as nodes and topology connections and operational constraints as edges; The power output of renewable energy is obtained by fitting a probability distribution to the power output data; and several typical power output scenarios are obtained based on the power output data; the power output data includes: historical wind power data or weather forecast data; Based on the power grid topology knowledge graph and the renewable energy output, a stochastic Markov decision process is constructed with the state of the power grid physical equipment after fault isolation as the initial state. The state space of the stochastic Markov decision process includes: circuit breaker state, bus voltage, generator output, load consumption, and renewable energy output; the action space includes topology operation, load scheduling, and generator regulation; and the state transition function includes topology update, power adjustment, and power flow calculation after the action is executed. Starting from the initial state, aiming to minimize the cumulative cost function, using hard constraints as conditions, and topological distance as a heuristic function, an improved heuristic search algorithm is employed to search the state space and obtain the optimal operation sequence. The hard constraints include power balance constraints, bus voltage constraints, line capacity constraints, generator constraints, ramping constraints, interlocking constraints, and the N-1 safety criterion. The improved heuristic search algorithm includes: a best-first search method or... algorithm; The optimal operation sequence is verified based on the multiple typical power output scenarios to obtain the final operation sequence; and the power grid is regulated using the final operation sequence.

2. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, The construction of a power grid topology knowledge graph based on power grid physical equipment, corresponding topology connections, and operational constraints specifically includes: Using formula Constructing a power grid topology knowledge graph ; in, For the set of vertices, , Let be the set of edges. , It is a set of physical equipment nodes of the power grid. For the set of constraint nodes, It is a set of topologically related edges. To run the set of constraint edges.

3. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, The step of fitting a probability distribution of the power output of renewable energy based on the output data to obtain the renewable energy output specifically includes: Using formula Probability distribution fitting is performed on the power output of renewable energy sources; in, for Renewable energy output at all times It can be a normal distribution, a Beta distribution, or a Weibull distribution. The mean of the distribution reflects the average output level. The standard deviation reflects the magnitude of force fluctuation.

4. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, The above describes several typical power output scenarios based on renewable energy output, specifically including: Based on renewable energy output, multiple typical output scenarios are obtained using the percentile method, Gaussian quadrature method, or empirical method; and each typical output scenario is assigned a corresponding probability weight.

5. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, The step of constructing a stochastic Markov decision process based on the power grid topology knowledge graph and the renewable energy output, with the initial state of the power grid physical equipment after fault isolation as the initial state, specifically includes: Using formula Determine the state space of power grid physical devices ; Using formula Determine the operating space of power grid physical equipment ; Using formula Determine the state transition function; in, This is the circuit breaker state matrix. The bus voltage vector. To provide power to the generator, For load consumption, Contribute to renewable energy For topological actions, For load dispatching actions, For engine adjustment actions, This refers to the state of the power grid physical equipment after the state transition. This is the state transition function. for The state at any given moment, for Momentary actions for Renewable energy output at all times.

6. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, Starting from the initial state, aiming to minimize the cumulative cost function, using hard constraints as conditions, and employing topological distance as a heuristic function, an improved heuristic search algorithm is used to search the state space to obtain the optimal operation sequence, specifically including: Using formula Determine the cumulative cost function ;in, This is the weighting coefficient for power imbalance. For power imbalance, The weighting coefficient for the weighted load loss is... For weighted load loss, This is a weighting coefficient for operational complexity. For operational complexity, To restore the time weighting coefficient, This refers to the recovery time. Using formula Determine the power balance constraints; where, For the first The active power output of the generator, For the first The active power consumption of a load, For the first The actual output of each renewable energy site For power grid system losses, The total load requirement of the power grid system; Using formula Determine the bus voltage constraints; where For the first Rated voltage of each busbar For the first Voltage amplitude of each busbar; Using formula Determine the line capacity constraints; among which, For the first The transmission power of each line, For the first The thermal stability limit of the line; Using formula Determine generator constraints; where, For the first The maximum active power output of the generator. For the first The minimum active power output of the generator; Using formula Determine the climbing constraints; among which, For the first The change in the active power output of the generator. For time intervals, For the first The maximum ramp rate of the generator; Using formula Determine the heuristic function ;in, The topological distance weighting coefficient. This represents the topological distance from the currently de-energized busbar to the nearest energized busbar. These are the weighting coefficients for the operands. Operands that have already been executed. This refers to the state space of the power grid physical equipment.

7. The load power balance regulation method based on knowledge graph and random sequence optimization according to claim 1, characterized in that, The step of verifying the optimal operation sequence based on the multiple typical power output scenarios to obtain the final operation sequence specifically includes: The optimal operation sequence is then applied to multiple typical power output scenarios to obtain the corresponding topological relationships and load states. Based on the topology and load conditions, determine the number of typical output scenarios that satisfy the hard constraints; and determine the success rate; the success rate is the ratio of the number of typical output scenarios that satisfy the hard constraints to the total number of typical output scenarios. Based on the success rate, the final operation sequence is obtained.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the load power balance regulation method based on knowledge graph and random sequence optimization as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the load power balance control method based on knowledge graph and random sequence optimization as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the load power balance control method based on knowledge graph and random sequence optimization as described in any one of claims 1-7.

Images