An intelligent scheduling and protection control method, system, device and medium for an AC-DC power grid

Abstract

The application discloses an AC-DC power grid intelligent scheduling and protection control method, system, device and medium, and belongs to the technical field of intelligent power grid operation control, which comprises the following steps: constructing an AC-DC collaborative scheduling model, dynamically predicting new energy output and load demand and generating a coordination instruction; based on the scheduling result, establishing a multi-level linkage protection architecture, defining AC and DC system protection collaborative logic; integrating heterogeneous operation data of AC and DC systems to construct a unified analysis model; adopting a deep reinforcement learning algorithm to rollingly optimize a dynamic power flow distribution strategy; dynamically correcting protection parameters and triggering edge side fault rapid response according to real-time operation state; forming a closed-loop technology link, with previous results driving subsequent actions, so that the adaptive collaboration of scheduling, protection and control is realized. Through the AC-DC collaborative scheduling and multi-level protection architecture, combined with the deep reinforcement learning algorithm and the dynamic protection mechanism, the application constructs a "scheduling-protection-control" three-dimensional decision system, and improves the stability of the power grid.

Description

Technical Field

[0001] This invention relates to the field of smart grid operation control technology, and in particular to a method, system, device and medium for intelligent dispatching and protection control of AC / DC power grids. Background Technology

[0002] The AC / DC hybrid power grid exhibits complex dynamic characteristics due to fluctuations in new energy sources and the risk of DC commutation. Existing dispatching and protection systems operate independently: dispatching strategies prioritize economic efficiency, while protection devices rely on fixed thresholds, resulting in a lack of real-time coordination between the two. When new energy output suddenly changes, protection settings cannot match dispatching changes, leading to a mismatch in the timing of AC protection and DC blocking actions, which can easily trigger cascading failures. Traditional methods struggle to balance energy absorption and safety.

[0003] The existing system suffers from three major problems: a disconnect between protection settings and scheduling strategies, insufficient cross-regional coordination, and lagging data interoperability. Protection thresholds cannot be dynamically adjusted according to operating conditions; during faults, the action logic of AC and DC systems is inconsistent; and differences in data formats cause delays in control commands and protection responses. Manual intervention is inefficient and struggles to meet the rapid control demands of high-proportion renewable energy grids. Therefore, a breakthrough in integrated scheduling and protection coordination technology is urgently needed. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent dispatching and protection control method, system, equipment, and medium for AC / DC power grids to address the problems that existing dispatching and protection methods for AC / DC hybrid power grids are unable to cope with new energy fluctuations, are prone to cascading failures, and cannot simultaneously meet safety and consumption needs.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent dispatching and protection control method for AC / DC power grids, comprising:

[0008] Acquire historical solar power output data and weather data, construct an AC / DC coordinated scheduling model, dynamically predict renewable energy output and load demand, and generate coordination instructions that include renewable energy output limits and DC power setpoints.

[0009] A multi-level linkage protection architecture is established based on the aforementioned coordination instructions, and the AC and DC system protection coordination logic is defined to obtain a multi-level linkage protection action sequence.

[0010] By integrating heterogeneous operating data of AC / DC systems and combining the coordination commands and the protection action sequence, a unified analysis model is constructed to obtain dynamic stability margin evaluation indicators.

[0011] Based on the dynamic stability margin evaluation index, a deep reinforcement learning algorithm is used to continuously optimize the dynamic power flow allocation strategy to obtain the optimized unit output command and DC control parameters.

[0012] Based on the real-time operating status and the optimized unit output command and DC control parameters, the protection parameters are dynamically corrected and a rapid response to edge-side faults is triggered, executing protection actions including cross-regional circuit breaker tripping and DC system blocking commands.

[0013] The data on the changes in the power grid state after the protection action is performed are fed back to the AC / DC coordinated dispatch model. Based on the feedback data, the limit boundary of the new energy output is corrected, forming an adaptive coordinated closed loop of dispatch, protection and control.

[0014] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0015] The aforementioned construction of an AC / DC coordinated dispatch model dynamically predicts renewable energy output and load demand, and generates coordination instructions that include renewable energy output limits and DC power setpoints, including:

[0016] By performing sliding window feature analysis on historical power output curves, the values ​​of weather-sensitive factors and the corresponding fluctuation values ​​of power output fluctuation characteristics are extracted to obtain a power output fluctuation probability distribution model.

[0017] The horizontal axis of the historical landscape output curve represents weather-sensitive factors.

[0018] The vertical axis of the historical wind power output curve represents the fluctuation value of the load fluctuation characteristics.

[0019] The weather-sensitive factors include wind speed, irradiance, temperature, humidity, and precipitation;

[0020] Based on AC load data and flexible DC power transmission plan, extract new energy output data from the flexible DC power transmission plan, pair AC load data with corresponding new energy output data, and establish a spatiotemporal correlation function between load demand and new energy output.

[0021] The power output fluctuation probability distribution model is combined with the spatiotemporal correlation function to construct an AC / DC coordinated scheduling model;

[0022] Capture the actual fluctuation values ​​of current weather-sensitive factors, input the actual fluctuation values ​​into the power output fluctuation probability distribution model, and obtain the actual load fluctuation demand;

[0023] The actual load fluctuation demand is input into the spatiotemporal correlation function to obtain the actual output fluctuation demand;

[0024] Based on the actual fluctuations in power output, a coordination command is generated that includes limits on new energy power output and DC power setting values.

[0025] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0026] The multi-level linkage protection architecture established based on the coordination instructions defines the AC and DC system protection coordination logic to obtain the multi-level linkage protection action sequence, including:

[0027] Based on the new energy output limit and DC power setting value in the coordination instruction, extract the AC section power fluctuation threshold and the DC converter station overload risk index;

[0028] The AC / DC power grid is uniformly divided into at least one local area. The ratio of the AC bus voltage deviation to the DC line current mutation characteristic is used as the priority weight of the local area fault action. The DC line current mutation characteristic is the maximum value of the DC line current mutation in the local area.

[0029] Based on the priority weight of fault actions in local areas, the priority of protection actions in different local areas is determined from large to small.

[0030] For each priority local area, define the AC and DC system protection coordination logic;

[0031] The protection coordination logic includes, in the protection action sequence, the ratio of the call duration of the AC circuit breaker tripping command to the DC blocking command is equal to the ratio of the magnitude of the AC current and the DC current during overcurrent.

[0032] Based on the protection coordination logic, an alternating trigger sequence containing AC circuit breaker tripping and DC blocking commands is generated to obtain a multi-level linkage protection action sequence.

[0033] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0034] The integrated AC / DC system heterogeneous operating data, combined with the coordinated commands and the protection action sequence, constructs a unified analysis model to obtain dynamic stability margin evaluation indicators, including:

[0035] Based on the current operating mode determined by the coordination instructions, historical data of AC section power fluctuation threshold and DC converter station overload risk index are extracted from the AC / DC coordinated scheduling model.

[0036] The actual measured values ​​of AC bus voltage deviation, AC system power angle change rate, and DC line current abrupt change characteristics in the protection architecture were collected.

[0037] Historical data and real-time measurements are synchronized and aligned in time and normalized in terms of feature dimensions, respectively.

[0038] Based on the normalized overload risk index of DC converter station and the characteristics of sudden change in DC line current, the probability of commutation failure of DC system is calculated by a preset probability model.

[0039] Establish a correlation mapping function between the power angle change rate of the AC system and the calculated commutation failure probability of the DC system, and obtain the output result of the correlation mapping function;

[0040] The output of the correlation mapping function is weighted and fused with the priority weight of local area fault actions in the protection action sequence to obtain the dynamic stability margin evaluation index.

[0041] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0042] Based on the dynamic stability margin evaluation index, a deep reinforcement learning algorithm is used to continuously optimize the dynamic power flow allocation strategy, resulting in optimized unit output commands and DC control parameters, including:

[0043] The dynamic stability margin evaluation index, together with real-time grid frequency, voltage, and tie-line power data, constitutes the input state vector for deep reinforcement learning.

[0044] Define the input state vector as the power grid operating state, and set thermal power output adjustment, new energy limit and DC power as output actions that can be executed by the intelligent agent;

[0045] The new energy output limit and DC power setting value determined by the generated coordination command are used as constraints in the deep reinforcement learning (DRL) process.

[0046] The initial strategy network is trained based on historical operating data, and during operation, the changes in the power grid state after the execution of output actions are used as actual feedback to update the action parameters online, thereby obtaining an optimized strategy that adapts to the current operating state.

[0047] Based on the optimization strategy, the optimized unit output command and DC control parameters are obtained.

[0048] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0049] The dynamic correction of protection parameters and triggering of rapid response to edge-side faults execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands, including:

[0050] Real-time acquisition of grid current, voltage, and frequency parameters;

[0051] Based on the optimized unit output command and the new energy output limit and DC power setting value in the DC control parameters, the overcurrent protection threshold adjustment is dynamically calculated to obtain the corrected protection action threshold.

[0052] The revised protection action threshold is sent to the area protection device, and the transient waveform change characteristics are monitored in real time;

[0053] When the detected transient waveform change characteristics exceed the corrected protection action threshold, it is determined to be a fault, triggering the cross-regional circuit breaker trip command and the DC system lockout command.

[0054] As a preferred embodiment of the intelligent dispatching and protection control method for AC / DC power grids described in this invention, wherein:

[0055] The step of feeding back the power grid state change data after the protection action is performed to the AC / DC coordinated dispatch model, and correcting the renewable energy output limit boundary based on the feedback data, includes:

[0056] The new energy limit and DC power setpoint output in the AC / DC coordinated dispatch model are converted into overcurrent protection threshold correction coefficients, and the protection action delay parameters are dynamically calculated in combination with the real-time frequency deviation of the power grid.

[0057] The revised protection action threshold and protection action delay parameters are synchronized to the area protection device through the communication channel;

[0058] The data on the change in grid status after the protection device performs protection actions is fed back to the AC / DC coordinated dispatch model, and the limit boundary of new energy output in the AC / DC coordinated dispatch model is corrected based on the feedback data.

[0059] Secondly, the present invention provides an intelligent dispatching and protection control system for AC / DC power grids, comprising:

[0060] The AC / DC coordinated dispatch prediction module is used to acquire historical wind and solar power output data and weather data, construct an AC / DC coordinated dispatch model, dynamically predict new energy power output and load demand, and generate coordination instructions that include new energy power output limits and DC power setpoints.

[0061] The multi-level linkage protection coordination module is used to establish a multi-level linkage protection architecture based on the coordination instructions, define the AC and DC system protection coordination logic, and obtain the multi-level linkage protection action sequence.

[0062] The dynamic stability margin assessment module is used to integrate heterogeneous operating data of AC / DC systems and combine the coordination commands and the protection action sequence to construct a unified analysis model and obtain dynamic stability margin assessment indicators.

[0063] The deep reinforcement learning optimization control module is used to continuously optimize the dynamic power flow allocation strategy based on the dynamic stability margin evaluation index and the deep reinforcement learning algorithm to obtain the optimized unit output command and DC control parameters.

[0064] The dynamic protection parameter correction module is used to dynamically correct protection parameters and trigger rapid response to edge-side faults based on the real-time operating status and the optimized unit output command and DC control parameters, and to execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands.

[0065] The closed-loop feedback module is used to feed back the power grid state change data after the protection action is performed to the AC / DC coordinated dispatch model. Based on the feedback data, the power output limit boundary of new energy sources is corrected to form an adaptive coordinated closed loop of dispatch, protection and control.

[0066] Thirdly, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the aforementioned intelligent dispatching and protection control method for AC / DC power grids.

[0067] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the aforementioned intelligent scheduling and protection control method for AC / DC power grids.

[0068] The beneficial effects of this invention are as follows: This invention dynamically predicts the output of new energy sources and load demand through an AC / DC coordinated scheduling model and generates coordinated instructions to avoid strategy conflicts caused by traditional hierarchical optimization; through a multi-level linkage protection architecture, it defines the protection coordination logic of AC and DC systems to prevent cascading faults caused by unilateral protection malfunctions; it uses a deep reinforcement learning algorithm to continuously optimize the dynamic power flow allocation strategy; it dynamically corrects protection parameters based on real-time operating status and triggers rapid response to edge-side faults; it forms a closed-loop technical link, with preceding results driving subsequent actions, achieving adaptive coordination of scheduling, protection, and control; through AC / DC coordinated scheduling and a multi-level protection architecture, combined with deep reinforcement learning algorithms and dynamic protection mechanisms, it constructs a three-dimensional decision-making system of "scheduling-protection-control" to improve power grid stability. Attached Figure Description

[0069] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1This is a basic flowchart illustrating an intelligent dispatching and protection control method for AC / DC power grids, provided as an embodiment of the present invention. Detailed Implementation

[0071] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0072] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for intelligent dispatching and protection control of AC / DC power grids is provided, such as... Figure 1 As shown, it includes:

[0073] S100: Acquire historical wind and solar power output data and weather data, construct an AC / DC coordinated scheduling model, dynamically predict new energy power output and load demand, and generate coordination instructions that include new energy power output limits and DC power setpoints.

[0074] In this embodiment of the invention, constructing an AC / DC coordinated dispatch model to dynamically predict renewable energy output and load demand and generate coordination instructions specifically includes:

[0075] A sliding window feature analysis was performed on the historical wind and solar power output curves to extract the values ​​of weather-sensitive factors and the corresponding fluctuation values ​​of power output fluctuation characteristics, thus obtaining a power output fluctuation probability distribution model. The horizontal axis of the historical wind and solar power output curves represents weather-sensitive factors, and the vertical axis represents the fluctuation values ​​of load fluctuation characteristics. Weather-sensitive factors include wind speed, irradiance, temperature, humidity, and precipitation.

[0076] In this embodiment of the invention, the sliding window feature analysis is specifically implemented by using a sliding window with a length of 1 hour and a step size of 5 minutes, and calculating the root mean square rate of change and fluctuation amplitude of the output data in each window as output fluctuation features.

[0077] Based on AC load data and flexible DC power transmission plan, extract new energy output data from the flexible DC power transmission plan, pair AC load data with corresponding new energy output data, and establish a spatiotemporal correlation function between load demand and new energy output.

[0078] By combining the power output fluctuation probability distribution model with the spatiotemporal correlation function, an AC / DC coordinated scheduling model is constructed.

[0079] Capture the actual fluctuation values ​​of current weather-sensitive factors, input the actual fluctuation values ​​into the power output fluctuation probability distribution model, and obtain the actual load fluctuation demand;

[0080] Input the actual load fluctuation demand into the spatiotemporal correlation function to obtain the actual output fluctuation demand;

[0081] Based on the actual fluctuations in power output, a coordination command is generated that includes limits on new energy power output and DC power setting values.

[0082] In this embodiment of the invention, the specific form of establishing the spatiotemporal correlation function between load demand and renewable energy output is to analyze the dynamic matching relationship between the spatiotemporal variation of AC load and renewable energy output through data mining and statistical modeling methods. Specifically, based on time series data, "AC load data" from the same or related regions and "renewable energy output data" extracted from the "flexible DC power transmission plan" are time-aligned and spatially paired. Vector autoregression (VAR), cointegration analysis, or machine learning models (such as LSTM neural networks) are used to construct correlation models between the two at different time scales (such as day-ahead and intraday) and spatial ranges. This quantifies the impact of load fluctuations on renewable energy absorption demand and the contribution of renewable energy output changes to grid power balance. Ultimately, a dynamic function is formed that can predict "the output responsibility or regulation target that renewable energy in the corresponding region should undertake under specific load demand," providing a decision-making basis for coordinated scheduling.

[0083] In this embodiment of the invention, the AC / DC coordinated scheduling model is not a simple assembly of modules, but a mathematical framework that organically integrates and collaboratively solves the "output fluctuation probability distribution model" and the "spatiotemporal correlation function." Specifically, the model is based on the output fluctuation probability distribution model to quantify the probabilistic impact of weather factors on the uncertainty of renewable energy output; at the same time, it introduces a spatiotemporal correlation function to characterize the dynamic coupling relationship between load changes and renewable energy output in geographical distribution and time series. During optimization, the model takes the minimum system operating cost or the maximum renewable energy absorption as the objective function, uses the renewable energy output limit and DC power setpoint as key decision variables, and considers constraints such as AC cross-sectional power flow, DC transmission capacity, and unit ramp rate. It dynamically generates coordination commands through rolling optimization algorithms (such as model predictive control). This model realizes the unified decision-making of probabilistic prediction of renewable energy fluctuations and cross-spatiotemporal matching of source-load-grid, which is the foundation for supporting subsequent protection and control coordination.

[0084] In this embodiment of the invention, after obtaining the "actual power output fluctuation demand", the dispatching system does not directly issue the demand value, but performs a decision decomposition and safety verification process: based on the current operating status, maximum available capacity and ramping capability of the renewable energy power station, it calculates the responsive power output adjustment range, and determines the upper limit of this range as the new "renewable energy power output limit" as a constraint instruction on the power station's output to prevent over-generation or under-generation; for the power difference that the renewable energy's own adjustment capability cannot meet, combined with the power flow distribution of the power grid and safety constraints, it calculates the amount of power required for cross-regional support or external transmission, and sets this value as the "DC power setpoint" to accurately control the transmission power of the flexible DC transmission system; the two safety-verified parameters, "renewable energy power output limit" and "DC power setpoint", are encapsulated into a comprehensive "coordination instruction", which is issued to the relevant power stations and DC control terminals through the dispatching communication network to achieve source-grid coordinated regulation.

[0085] It should be noted that the AC / DC coordinated dispatch model, by quantifying the impact of weather fluctuations on renewable energy output and coupling the spatiotemporal matching relationship between load and output, can significantly improve the dispatch accuracy and response speed of a high-proportion renewable energy power grid, reduce the risk of wind and solar curtailment or power shortage caused by extreme weather, and enhance the adaptive support capability of flexible DC transmission to renewable energy fluctuations, providing a dynamic decision-making basis for source-grid-load-storage coordinated optimization.

[0086] It should be noted that constructing a weather-sensitive AC / DC coordinated dispatch model to dynamically coordinate the output of new energy sources with load demand can reduce the impact of weather uncertainty on the power system and improve the absorption capacity of new energy sources and the stability of grid operation.

[0087] S200: Based on coordination instructions, a multi-level linkage protection architecture is established, defining the AC and DC system protection coordination logic to obtain a multi-level linkage protection action sequence;

[0088] In this embodiment of the invention, a multi-level linkage protection architecture is established, and the defined AC and DC system protection coordination logic specifically includes:

[0089] Based on the new energy output limit and DC power setpoint output by the scheduling model, the AC section power fluctuation threshold and DC converter station overload risk index are extracted.

[0090] The AC / DC power grid is uniformly divided into at least one local area. The ratio of the AC bus voltage deviation to the DC line current mutation characteristic is used as the priority weight of the local area fault action. The DC line current mutation characteristic is the maximum value of the DC line current mutation in the local area.

[0091] Based on the priority weight of local area fault actions, protection action sequences for different local areas are generated from large to small. The protection action sequence is an alternating trigger sequence of AC circuit breaker tripping and DC blocking commands.

[0092] The AC and DC system protection coordination logic is that the ratio of the call duration of the AC circuit breaker tripping and DC blocking commands in the protection action sequence is equal to the ratio of the magnitude of the AC current and the DC current during overcurrent.

[0093] It should be noted that this invention significantly improves the selectivity of the protection system under complex fault scenarios by dynamically quantifying the regional fault risk level and optimizing the AC / DC protection action sequence, reducing the probability of false tripping or failure to trip caused by traditional setting methods. At the same time, through the coordinated triggering mechanism of AC tripping and DC blocking, it effectively suppresses the fault propagation speed, reduces overload damage to key equipment, and ensures the power balance and voltage stability of AC / DC hybrid power grids during transient processes.

[0094] It should be noted that by constructing a collaborative protection mechanism for AC / DC systems, the priority of protection actions is dynamically generated based on regional fault risks. Through the time-sequential collaborative control of AC tripping and DC blocking, faults can be quickly isolated and the risk of grid cascading overload can be reduced, thus ensuring the transient stability of AC / DC hybrid power grids.

[0095] S300: Integrates heterogeneous operating data of AC / DC systems and combines coordinated commands and protection action sequences to construct a unified analysis model and obtain dynamic stability margin evaluation indicators;

[0096] In this embodiment of the invention, constructing a unified analysis model specifically includes:

[0097] Historical data on AC section power fluctuation thresholds and DC converter station overload risk indicators were extracted from the scheduling model. The "AC section power fluctuation threshold" refers to the maximum allowable power fluctuation range for each important transmission section, calculated by the scheduling model based on the N-1 safety criterion and transient stability analysis, used to assess the AC system's carrying capacity under disturbances. The "DC converter station overload risk indicator" refers to the degree of overload or risk level (usually expressed as a percentage or risk score) that the converter station equipment can withstand per unit time, assessed by the scheduling model based on the DC system's heat capacity, cooling conditions, and operating conditions. The extracted historical data includes time-series records of the above indicators under different operating modes, used for subsequent fusion analysis with real-time monitoring data to support the construction of a unified model.

[0098] Collect actual measured values ​​of AC bus voltage deviation and DC line current abrupt change characteristics in the protection architecture;

[0099] Time synchronization alignment and feature dimension normalization are performed on the two types of data respectively;

[0100] Establish a correlation mapping function between the power angle change rate of the AC system and the commutation failure probability of the DC system;

[0101] The mapping function is coupled with the priority weight of local area fault actions to generate a dynamic stability margin evaluation index.

[0102] In this embodiment of the invention, the correlation mapping function between the AC system power angle change rate and the DC system commutation failure probability can be constructed using statistical learning methods based on historical operating data. Specifically, a training dataset is constructed by collecting power angle change rate sequences and corresponding DC commutation failure event records under historical disturbance events (such as short-circuit faults and high-power fluctuations) in the power grid. A logistic regression model or a lightweight neural network is used, with the power angle change rate as the input feature and the commutation failure probability as the output label, for model training. During training, the model hyperparameters are optimized through cross-validation to ensure its generalization ability under different operating conditions. The trained mapping function can output an estimated probability of commutation failure in the DC system under the current conditions based on the real-time monitored power angle change rate, providing a key input for dynamic stability margin assessment.

[0103] In this embodiment of the invention, historical data refers to the analytical indicators output by the scheduling model during historical operating cycles, such as power fluctuation threshold and overload risk level.

[0104] The time synchronization alignment and feature dimension normalization processes for the two types of data are performed separately as follows:

[0105] Analyze the timestamps of the two types of data;

[0106] Find a common start time (For example, the earliest time of the largest of the two types of data) and end time (For example, the latest time with the smallest of the two types of data).

[0107] exist[ , Within the interval, based on the timestamp shared by both parties, the two types of data values ​​at the corresponding time are directly extracted;

[0108] If a certain category is missing at a certain point in time, interpolation or removal of that point can be performed;

[0109] In this embodiment of the invention, to ensure precise alignment of historical scheduling data and real-time protection measurements in the time dimension, the system employs a synchronous clock system based on the IEEE 1588 Precise Time Protocol (PTP). All data acquisition devices (such as SCADA RTU, PMU, and protection devices) are connected to the same PTP master clock, achieving microsecond-level time synchronization across the entire network. During the data processing phase, timestamp information in various data packets is parsed to ensure consistency in their time base. Data matching is performed within a defined common time interval, where the start time is the later of the earliest times of the two types of data, and the end time is the earlier of the latest times of the two types of data. For minor timestamp deviations caused by communication delays or differences in sampling periods, linear interpolation is used for correction, ensuring strict alignment of multi-source heterogeneous data on the time axis, providing a high-precision data foundation for subsequent feature fusion and analysis.

[0110] It should be noted that this invention, through the deep integration of historical pattern mining and real-time state perception, breaks through the limitations of isolated analysis of AC and DC systems in traditional stability assessment. It can accurately capture the coupling risks of AC power angle instability and DC commutation failure during the dynamic evolution of the power grid, identify weak links in advance and dynamically optimize protection strategies, significantly improve the defense capability of power grids with a high proportion of new energy to cope with complex disturbances, and at the same time reduce unnecessary power outage losses caused by malfunctions of the protection system.

[0111] It should be noted that establishing a real-time assessment mechanism for the dynamic stability of AC / DC systems, by integrating historical data with real-time monitoring of cross-system characteristics, quantifies the grid stability margin, provides accurate predictive basis for protection actions, prevents cascading failures, and enhances the resilience of hybrid power grids.

[0112] S400: Based on the dynamic stability margin evaluation index, a deep reinforcement learning algorithm is used to continuously optimize the dynamic power flow allocation strategy to obtain the optimized unit output command and DC control parameters.

[0113] In this embodiment of the invention, the strategy of using a deep reinforcement learning algorithm to continuously optimize the dynamic power flow allocation specifically includes:

[0114] The dynamic stability margin assessment index and real-time grid frequency, voltage and tie-line power data are input into the reinforcement learning model;

[0115] Define the power grid operating status as the input condition, and the thermal power output adjustment, new energy limit and DC power setting as the output actions;

[0116] The initial policy network is trained based on historical operational data, and the action parameters are updated online.

[0117] Generate unit output commands and DC control parameters based on the optimization strategy;

[0118] The deviation in the power grid state after the command is executed is fed back to the protection parameter self-correction module, and the overcurrent protection threshold is adjusted accordingly.

[0119] In this embodiment of the invention, the dynamic stability margin assessment index is quantified by performing a weighted product operation on the estimated probability of DC commutation failure and the priority weight of local area fault actions to obtain the dynamic stability margin assessment index of the area. The lower the index value, the higher the risk.

[0120] The architecture of the deep reinforcement learning algorithm is as follows;

[0121] Agent: The entity that makes decisions;

[0122] Environment: The external world in which intelligent agents interact;

[0123] State (s): A description of the environment at a given moment;

[0124] Action (a): The behavior that an intelligent agent can perform;

[0125] Reward (r): The feedback from the environment to the agent's actions is a scalar signal. The agent's ultimate goal is to maximize the cumulative reward.

[0126] Policy (π): The agent's behavior function, which defines the probability of choosing action a in state s (stochastic policy) or directly outputting an action (deterministic policy).

[0127] Value function (V(s)): measures the expected cumulative reward that can be obtained starting from state s and following policy π, representing the quality of the state;

[0128] Q-function (Q(s, a)): measures the expected cumulative reward that can be obtained by performing action a in state s and following policy π, representing the quality of the state-action pair;

[0129] Using the above architecture, a deep reinforcement learning algorithm can be trained and generated.

[0130] The initial policy network here adopts the architecture of a deep reinforcement learning algorithm, which is used to optimize policy generation after training.

[0131] The overcurrent protection threshold is obtained as follows:

[0132] Identify the overcurrent type and determine the protected object;

[0133] Under different overcurrent types, obtain the combination of current and voltage that the protected object can safely withstand. At this time, the maximum junction temperature is usually 150°C or 175°C. Calculate the minimum threshold based on the safe operating area (SOA).

[0134] Based on the protection speed (delay time), find the corresponding maximum allowable current on the safe operating area curve, and use it as the overcurrent protection threshold.

[0135] In this embodiment of the invention, when using a deep reinforcement learning algorithm to continuously optimize the dynamic power flow allocation strategy, to ensure the stability and convergence of the strategy, the Proximal Policy Optimization (PPO) algorithm is preferably used as the core optimizer. This algorithm, by introducing importance sampling and pruning mechanisms, effectively controls the magnitude of policy updates while ensuring learning efficiency, avoiding drastic oscillations during training. The policy network adopts a fully connected neural network structure containing an input layer, two hidden layers, and an output layer. The ReLU function is used as the activation function in the hidden layers to enhance nonlinear fitting capabilities. The input state vector is normalized and then fed into the network. The output layer uses either a linear or Softmax activation function depending on the continuity (e.g., thermal power output adjustment) or discreteness (e.g., unit start-up and shutdown) of the action space, thereby generating executable control actions.

[0136] In this embodiment of the invention, historical operating data refers to the actual operating records obtained from power grid SCADA, PMU and other systems;

[0137] It should be noted that this invention achieves global dynamic optimization of source-grid-load-storage control through reinforcement learning, breaking through the response speed bottleneck of traditional model prediction. It completes the coordinated decision-making of new energy limits, thermal power regulation and DC control on a millisecond time scale, which greatly improves the grid's ability to autonomously suppress transient problems such as frequency oscillations and voltage over-limits. At the same time, by adaptively adjusting the protection threshold according to the operating state, it effectively avoids the false triggering of protection devices under rapid fluctuations in new energy, and significantly improves the economic efficiency of grid safe operation and the utilization rate of renewable energy.

[0138] It should be noted that by constructing a power grid adaptive control closed loop based on reinforcement learning, the coordinated regulation of thermal power, new energy and DC systems is optimized through dynamic strategies, and operational status deviations are corrected in real time and protection thresholds are adjusted in conjunction with the system, thereby achieving autonomous optimization of the power grid's safety and stability and the coordination between power sources and the grid.

[0139] S500: Based on the real-time operating status and optimized unit output commands and DC control parameters, dynamically correct protection parameters and trigger rapid response to edge-side faults, and execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands.

[0140] In this embodiment of the invention, dynamically correcting protection parameters and triggering a rapid response to edge-side faults specifically includes:

[0141] Real-time acquisition of grid current, voltage, and frequency parameters;

[0142] Based on the renewable energy output limits and DC power setpoints in the current dispatch strategy, the overcurrent protection threshold adjustment is dynamically calculated. "Dynamic calculation" refers to the quantitative process of adaptively adjusting the overcurrent protection setpoints according to the current real-time grid operating conditions. Specifically, using the "renewable energy output limits" and "DC power setpoints" issued in the current dispatch instructions as inputs, their impact on the system's short-circuit current level is analyzed: when renewable energy output is high, the fault current may increase, requiring an appropriate increase in the overcurrent protection threshold for AC lines or buses to prevent maloperation due to current approaching the setpoint; conversely, the threshold can be lowered to enhance sensitivity. Simultaneously, combined with the DC system's control mode (e.g., constant power, constant current), its current response characteristics during faults are evaluated, and the thresholds for DC-side protection and converter station AC-side protection are adjusted accordingly. This adjustment is calculated in real-time using a preset mapping relationship or empirical formula (e.g., proportional adjustment based on the short-circuit capacity change rate), ensuring that the protection setpoints always match the actual operating mode.

[0143] The revised protection action threshold is sent to the area protection device;

[0144] Real-time monitoring of transient waveform changes at AC / DC nodes is achieved through edge nodes;

[0145] After identifying the fault, the cross-regional circuit breaker trips and DC blocking commands are triggered.

[0146] Feedback the power grid state parameters after the action to the scheduling model;

[0147] The correction and response process is executed on a second-level cycle, forming a closed-loop linkage between parameter adjustment and fault response.

[0148] In this embodiment of the invention, rapid fault response on the edge side is achieved through edge computing devices deployed at key nodes. Specifically, edge nodes with real-time computing capabilities are configured at locations such as flexible DC converter stations and hub substations. These nodes are connected to the local PMU (phasor measurement unit), protection devices, and control systems via high-speed communication interfaces (such as optical fibers) to collect voltage, current, and transient waveform data of AC and DC nodes in real time. The edge nodes have built-in fault identification algorithms (such as transient feature extraction based on wavelet transform). Once a fault feature is detected that exceeds the dynamically corrected protection threshold, a preset local control strategy is immediately triggered, such as sending a power reduction command to the flexible DC converter or initiating a DC blocking process. The entire process is completed locally, without relying on master station communication, ensuring millisecond-level fault response and effectively suppressing fault propagation.

[0149] It should be noted that this closed-loop system breaks through the bottlenecks of traditional fixed and rigid protection settings and slow dispatch response by using a real-time perception-decision-execution cycle at the second level. It achieves dynamic adaptation of protection thresholds and rapid rebalancing of power between the source and the grid under fault disturbances, effectively suppressing the risk of protection maloperation caused by random fluctuations in renewable energy. At the same time, through rapid isolation of cross-regional faults and linkage correction of dispatch strategies, it significantly shortens the grid recovery time, reduces power outage losses, and improves the survivability of grids with a high proportion of renewable energy under extreme conditions.

[0150] It should be noted that establishing a second-level closed-loop power grid protection and dispatch collaborative control system, through real-time parameter sensing, dynamic threshold setting, and rapid fault isolation linkage, enables instantaneous self-healing of AC / DC hybrid power grid faults and proactive enhancement of operational resilience, thereby reducing the risk of cascading collapse.

[0151] S600: Feeds back the grid state change data after the protection action to the AC / DC coordinated dispatch model, and corrects the limit boundary of new energy output based on the feedback data to form an adaptive coordinated closed loop of dispatch, protection and control.

[0152] In this embodiment of the invention, the data on the change in grid state after the execution of protection actions is fed back to the AC / DC coordinated dispatch model, and the boundary of the new energy output limit is corrected based on the feedback data, including:

[0153] The renewable energy limits and DC power setpoints output from the AC / DC coordinated dispatch model are converted into overcurrent protection threshold correction coefficients. These are then combined with real-time grid frequency deviation to dynamically calculate protection action delay parameters. "Conversion" refers to establishing a quantitative mapping relationship between renewable energy output limits, DC power setpoints, and protection settings. Through historical simulations or online calculations, a mapping model is pre-built, using the current dispatch output of renewable energy and DC transmission power as inputs to assess their impact on the system's short-circuit current level. A proportional coefficient is output as the overcurrent protection threshold correction coefficient, used to adjust the protection settings and avoid maloperation under high renewable energy output or insufficient sensitivity under low output. "Calculation" refers to dynamically adjusting the protection action delay using real-time monitored grid frequency deviations. When the frequency deviates significantly from the rated value (e.g., a rapid drop), the system is judged to be in an emergency state of high power shortage. In this case, the overcurrent protection action delay is appropriately extended to avoid erroneous tripping during transient processes, allowing time for frequency recovery. This delay parameter is calculated in real-time based on the magnitude of the frequency deviation using piecewise functions or fuzzy logic rules, achieving adaptive optimization of protection action characteristics.

[0154] The revised protection action threshold and protection action delay parameters are synchronized to the area protection device through the communication channel;

[0155] The power grid state change data after the protection device performs its protection action is fed back to the AC / DC coordinated dispatch model. Based on the feedback data, the renewable energy output limit boundary in the AC / DC coordinated dispatch model is corrected. "Correction" refers to the process of adaptively updating the preset renewable energy operation constraints in the dispatch model using the actual system response after the protection action. Specifically, when the protection device trips or blocks due to a fault or disturbance, the power grid state data after the fault is cleared (such as voltage, frequency, and tie-line power) is collected. If the system remains stable or recovers well, it indicates that the "renewable energy output limit boundary" set by the original dispatch model may be too conservative and can be appropriately relaxed (i.e., the upper limit is increased) within the allowable safety margin to improve the renewable energy absorption capacity. Conversely, if the system experiences voltage instability or frequency exceeding the limit, it indicates that the current limit boundary is too high and needs to be tightened (i.e., the upper limit is reduced) to enhance the safety margin. This correction process compares the feedback data with the model prediction results through parameter identification or online learning algorithms, dynamically adjusting the constraint parameters within the model to achieve closed-loop optimization and adaptive evolution of the dispatch strategy.

[0156] In this embodiment of the invention, the communication channel transmits the corrected protection settings to the area protection device in real time via the GOOSE (Generic ObjectOriented Substation Event) message of the IEC 61850 standard.

[0157] In this embodiment of the invention, the power grid state change data after the protection device operates is uploaded to the dispatch master station system via a power dispatch data network (such as a dedicated dispatch data network). Data transmission follows IEC 61850 or CIM / E (Common Information Model / Power Grid Model Exchange Format) standards to ensure semantic consistency and interoperability. After receiving the feedback data, the dispatch master station extracts key state information (such as voltage, frequency, and power distribution after a fault) from the data parsing module and inputs it into the parameter correction module of the AC / DC collaborative dispatch model. This module utilizes the deviation between historical data and current feedback, employing online learning or parameter identification algorithms to dynamically adjust key parameters in the model (such as the limit boundary of new energy output), completing the adaptive update of the dispatch model and thus forming a complete "execution-feedback-correction" closed-loop control link.

[0158] It should be noted that this invention, through real-time linkage of scheduling and protection parameters, effectively releases the potential for renewable energy consumption while ensuring the safe and stable operation of the power grid and reducing the risks of protection malfunctions and equipment overload. By achieving dynamic coordination between scheduling commands and protection settings, combined with a two-way coupling mechanism of threshold and delay parameters, and a feedback loop of data after protection actions, the invention improves the speed and accuracy of fault defense in scenarios with a high proportion of renewable energy access. By converting the operating limits output by scheduling into protection threshold correction coefficients, the safety boundary is dynamically optimized, overcoming the conservative operation problem caused by traditional fixed thresholds. Simultaneously, by dynamically adjusting protection delay parameters using real-time frequency deviation, the protection system's ability to identify transient overloads and permanent faults is enhanced. By establishing a closed-loop feedback mechanism for reverse correction of the scheduling model using state data after protection actions, the coordinated optimization of "scheduling-protection-power grid state" is achieved, further improving the overall safety level and operating efficiency of the system.

[0159] Example 2 is an embodiment of the present invention. This embodiment differs from the first embodiment in that it provides an intelligent dispatching and protection control system for AC / DC power grids.

[0160] It should be noted that the technical solution of the AC / DC power grid intelligent dispatch and protection control system is based on the same concept as the technical solution of the AC / DC power grid intelligent dispatch and protection control method described above. For details not described in detail in the technical solution of the AC / DC power grid intelligent dispatch and protection control system in this embodiment, please refer to the description of the technical solution of the AC / DC power grid intelligent dispatch and protection control method described above.

[0161] This embodiment of an intelligent dispatching and protection control system for AC / DC power grids includes:

[0162] The AC / DC coordinated dispatch prediction module is used to acquire historical wind and solar power output data and weather data, construct an AC / DC coordinated dispatch model, dynamically predict new energy power output and load demand, and generate coordination instructions that include new energy power output limits and DC power setpoints.

[0163] The multi-level linkage protection coordination module is used to establish a multi-level linkage protection architecture based on coordination commands, define the AC and DC system protection coordination logic, and obtain the multi-level linkage protection action sequence.

[0164] The dynamic stability margin assessment module is used to integrate heterogeneous operating data of AC / DC systems and combine them with coordinated commands and protection action sequences to build a unified analysis model and obtain dynamic stability margin assessment indicators.

[0165] The deep reinforcement learning optimization control module is used to continuously optimize the dynamic power flow allocation strategy based on the dynamic stability margin evaluation index and the deep reinforcement learning algorithm to obtain the optimized unit output command and DC control parameters.

[0166] The dynamic protection parameter correction module is used to dynamically correct protection parameters and trigger rapid response to edge-side faults based on real-time operating status, optimized unit output commands, and DC control parameters, and to execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands.

[0167] The closed-loop feedback module is used to feed back the power grid state change data after the protection action is performed to the AC / DC collaborative dispatch model. Based on the feedback data, the limit boundary of new energy output is corrected to form an adaptive collaborative closed loop of dispatch, protection and control.

[0168] This embodiment also provides an electronic device applicable to an AC / DC power grid intelligent dispatching and protection control method, including:

[0169] The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement a smart dispatching and protection control method for AC / DC power grids as described in the above embodiments.

[0170] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the intelligent dispatching and protection control method for AC / DC power grids as proposed in the above embodiments.

[0171] The storage medium proposed in this embodiment and the method for implementing intelligent dispatching and protection control of AC / DC power grid proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0172] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

Claims

1. A method for intelligent dispatching and protection control of AC / DC power grids, characterized in that, include: Acquire historical solar power output data and weather data, construct an AC / DC coordinated scheduling model, dynamically predict renewable energy output and load demand, and generate coordination instructions that include renewable energy output limits and DC power setpoints. A multi-level linkage protection architecture is established based on the aforementioned coordination instructions, and the AC and DC system protection coordination logic is defined to obtain a multi-level linkage protection action sequence. By integrating heterogeneous operating data of AC / DC systems and combining the coordination commands and the protection action sequence, a unified analysis model is constructed to obtain dynamic stability margin evaluation indicators. Based on the dynamic stability margin evaluation index, a deep reinforcement learning algorithm is used to continuously optimize the dynamic power flow allocation strategy to obtain the optimized unit output command and DC control parameters. Based on the real-time operating status and the optimized unit output command and DC control parameters, the protection parameters are dynamically corrected and a rapid response to edge-side faults is triggered, executing protection actions including cross-regional circuit breaker tripping and DC system blocking commands. The data on the change in grid status after the protection action is performed is fed back to the AC / DC coordinated dispatch model. Based on the feedback data, the limit boundary of new energy output is corrected to form an adaptive coordinated closed loop of dispatch, protection and control. The multi-level linkage protection architecture established based on the coordination instructions defines the AC and DC system protection coordination logic to obtain the multi-level linkage protection action sequence, including: Based on the new energy output limit and DC power setting value in the coordination instruction, extract the AC section power fluctuation threshold and the DC converter station overload risk index; The AC / DC power grid is uniformly divided into at least one local area. The ratio of the AC bus voltage deviation to the DC line current mutation characteristic is used as the priority weight of the local area fault action. The DC line current mutation characteristic is the maximum value of the DC line current mutation in the local area. Based on the priority weight of fault actions in local areas, the priority of protection actions in different local areas is determined from large to small. For each priority local area, define the AC and DC system protection coordination logic; The protection coordination logic includes, in the protection action sequence, the ratio of the call duration of the AC circuit breaker tripping command to the DC blocking command is equal to the ratio of the magnitude of the AC current and the DC current during overcurrent. Based on the protection coordination logic, an alternating trigger sequence containing AC circuit breaker tripping and DC blocking commands is generated to obtain a multi-level linkage protection action sequence.

2. The intelligent dispatching and protection control method for AC / DC power grids as described in claim 1, characterized in that: The aforementioned construction of an AC / DC coordinated dispatch model dynamically predicts renewable energy output and load demand, and generates coordination instructions that include renewable energy output limits and DC power setpoints, including: By performing sliding window feature analysis on historical power output curves, the values ​​of weather-sensitive factors and the corresponding fluctuation values ​​of power output fluctuation characteristics are extracted to obtain a power output fluctuation probability distribution model. The horizontal axis of the historical landscape output curve represents weather-sensitive factors. The vertical axis of the historical wind power output curve represents the fluctuation value of the load fluctuation characteristics. The weather-sensitive factors include wind speed, irradiance, temperature, humidity, and precipitation; Based on AC load data and flexible DC power transmission plan, extract new energy output data from the flexible DC power transmission plan, pair AC load data with corresponding new energy output data, and establish a spatiotemporal correlation function between load demand and new energy output. The power output fluctuation probability distribution model is combined with the spatiotemporal correlation function to construct an AC / DC coordinated scheduling model; Capture the actual fluctuation values ​​of current weather-sensitive factors, input the actual fluctuation values ​​into the power output fluctuation probability distribution model, and obtain the actual load fluctuation demand; The actual load fluctuation demand is input into the spatiotemporal correlation function to obtain the actual output fluctuation demand; Based on the actual fluctuations in power output, a coordination command is generated that includes limits on new energy power output and DC power setting values.

3. The intelligent dispatching and protection control method for AC / DC power grids as described in claim 2, characterized in that: The integrated AC / DC system heterogeneous operating data, combined with the coordinated commands and the protection action sequence, constructs a unified analysis model to obtain dynamic stability margin evaluation indicators, including: Based on the current operating mode determined by the coordination instructions, historical data of AC section power fluctuation threshold and DC converter station overload risk index are extracted from the AC / DC coordinated scheduling model. The actual measured values ​​of AC bus voltage deviation, AC system power angle change rate, and DC line current abrupt change characteristics in the protection architecture were collected. Historical data and real-time measurements are synchronized and aligned in time and normalized in terms of feature dimensions, respectively. Based on the normalized overload risk index of DC converter station and the characteristics of sudden change in DC line current, the probability of commutation failure of DC system is calculated by a preset probability model. Establish a correlation mapping function between the power angle change rate of the AC system and the calculated commutation failure probability of the DC system, and obtain the output result of the correlation mapping function; The output of the correlation mapping function is weighted and fused with the priority weight of local area fault actions in the protection action sequence to obtain the dynamic stability margin evaluation index.

4. The intelligent dispatching and protection control method for AC / DC power grids as described in claim 3, characterized in that: Based on the dynamic stability margin evaluation index, a deep reinforcement learning algorithm is used to continuously optimize the dynamic power flow allocation strategy, resulting in optimized unit output commands and DC control parameters, including: The dynamic stability margin evaluation index, together with real-time grid frequency, voltage, and tie-line power data, constitutes the input state vector for deep reinforcement learning. Define the input state vector as the power grid operating state, and set thermal power output adjustment, new energy limit and DC power as output actions that can be executed by the intelligent agent; The new energy output limit and DC power setting value determined by the generated coordination command are used as constraints in the deep reinforcement learning process. The initial strategy network is trained based on historical operating data, and during operation, the changes in the power grid state after the execution of output actions are used as actual feedback to update the action parameters online, thereby obtaining an optimized strategy that adapts to the current operating state. Based on the optimization strategy, the optimized unit output command and DC control parameters are obtained.

5. The intelligent dispatching and protection control method for AC / DC power grids as described in claim 4, characterized in that: The dynamic correction of protection parameters and triggering of rapid response to edge-side faults execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands, including: Real-time acquisition of grid current, voltage, and frequency parameters; Based on the optimized unit output command and the new energy output limit and DC power setting value in the DC control parameters, the overcurrent protection threshold adjustment is dynamically calculated to obtain the corrected protection action threshold. The revised protection action threshold is sent to the area protection device, and the transient waveform change characteristics are monitored in real time; When the detected transient waveform change characteristics exceed the corrected protection action threshold, it is determined to be a fault, triggering the cross-regional circuit breaker trip command and the DC system lockout command.

6. The intelligent dispatching and protection control method for AC / DC power grids as described in claim 5, characterized in that: The step of feeding back the power grid state change data after the protection action is performed to the AC / DC coordinated dispatch model, and correcting the renewable energy output limit boundary based on the feedback data, includes: The new energy limit and DC power setpoint output in the AC / DC coordinated dispatch model are converted into overcurrent protection threshold correction coefficients, and the protection action delay parameters are dynamically calculated in combination with the real-time frequency deviation of the power grid. The revised protection action threshold and protection action delay parameters are synchronized to the area protection device through the communication channel; The data on the change in grid status after the protection device performs protection actions is fed back to the AC / DC coordinated dispatch model, and the limit boundary of new energy output in the AC / DC coordinated dispatch model is corrected based on the feedback data.

7. An intelligent dispatching and protection control system for AC / DC power grids, employing the intelligent dispatching and protection control method for AC / DC power grids as described in any one of claims 1-6, characterized in that, include: The AC / DC coordinated dispatch prediction module is used to acquire historical wind and solar power output data and weather data, construct an AC / DC coordinated dispatch model, dynamically predict new energy power output and load demand, and generate coordination instructions that include new energy power output limits and DC power setpoints. The multi-level linkage protection coordination module is used to establish a multi-level linkage protection architecture based on the coordination instructions, define the AC and DC system protection coordination logic, and obtain the multi-level linkage protection action sequence. The dynamic stability margin assessment module is used to integrate heterogeneous operating data of AC / DC systems and combine the coordination commands and the protection action sequence to construct a unified analysis model and obtain dynamic stability margin assessment indicators. The deep reinforcement learning optimization control module is used to continuously optimize the dynamic power flow allocation strategy based on the dynamic stability margin evaluation index and the deep reinforcement learning algorithm to obtain the optimized unit output command and DC control parameters. The dynamic protection parameter correction module is used to dynamically correct protection parameters and trigger rapid response to edge-side faults based on the real-time operating status and the optimized unit output command and DC control parameters, and to execute protection actions including cross-regional circuit breaker tripping and DC system blocking commands. The closed-loop feedback module is used to feed back the power grid state change data after the protection action is performed to the AC / DC coordinated dispatch model. Based on the feedback data, the power output limit boundary of new energy sources is corrected to form an adaptive coordinated closed loop of dispatch, protection and control.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent dispatching and protection control method for AC / DC power grids as described in any one of claims 1 to 6.

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 steps of the intelligent dispatching and protection control method for AC / DC power grids as described in any one of claims 1 to 6.

Images