Power distribution network simulation deduction and load regulation and control method based on Internet of Things data

Abstract

The invention belongs to the technical field of electrical engineering, and particularly relates to a power distribution network simulation deduction and load regulation and control method based on Internet of Things data, which comprises the steps of acquiring multi-source Internet of Things data and performing timestamp checking, topology consistency deduction and cross-source error reforming to form a standardized power grid state data set; constructing a cross-scale operation feature based on the data set, generating a panoramic operation situation through topology and time sequence alignment, inputting the panoramic operation situation into a reinforcement learning agent, executing exploration in a simulation environment to generate an extreme compound fault scene, extracting risk severity from the scene, constructing a risk evolution path, and obtaining a risk evolution result; the method comprises the steps of identifying key induction nodes and easy-to-diffuse links, constructing a defensive load regulation and control strategy oriented to a causal chain, performing multi-working-condition simulation verification on the strategy, constructing an optimized regulation and control strategy set based on simulation deviation, and driving a power distribution network control system to execute load regulation and control actions so as to improve system stability and risk suppression capability.

Description

Technical Field

[0001] This invention relates to the field of electrical engineering technology, specifically to a method for power distribution network simulation and load control based on Internet of Things (IoT) data. Background Technology

[0002] The distribution network simulation and load regulation method based on IoT data relies on the Internet of Things to collect real-time data of the distribution network (such as equipment status, electricity consumption behavior, etc.), build a model to simulate the operation of the power grid and predict load changes, and formulate dispatch strategies to dynamically adjust the power generation, energy storage and user loads, so as to realize intelligent technology for safe operation, economic dispatch and new energy consumption.

[0003] In existing technologies, IoT technology is used to comprehensively perceive real-time data of all aspects of the distribution network (such as equipment status, line parameters, and user electricity consumption behavior), and big data analysis and machine learning algorithms are used to build digital models to simulate and predict future load change trends in the power grid operation status. Based on the simulation results, optimized scheduling strategies are formulated to dynamically adjust the power generation, energy storage, and user-side load resources, thereby achieving safe and stable operation of the distribution network, economical and efficient scheduling, and optimization of new energy consumption.

[0004] The above-mentioned solutions still have some problems in practical application. Although the existing technology can complete the function of load regulation in the distribution network, in the operation of the distribution network, due to the long-term inconsistency between multi-source monitoring data in terms of time synchronization, structural correspondence and cross-regional measurement offset, the power grid status cannot form an accurate and unified overall picture in both spatial and temporal dimensions. When the system enters the stage of disturbance, load fluctuation or local fault propagation, this information misalignment will cause the overall trend of change to be perceived late or misinterpreted, making it impossible for operation and maintenance personnel to identify the real risk transmission chain in a timely manner, and making it difficult to judge the pressure level of key nodes. This further leads to the deviation in judgment of the timing, scope and intensity of load control, thereby increasing the possibility of chain instability of the system under high pressure conditions.

[0005] To this end, the present invention provides a method for power distribution network simulation and load control based on Internet of Things data. Summary of the Invention

[0006] This application provides a distribution network simulation and load control method based on IoT data, which makes the entire strategy formation process closer to the dynamic evolution of the real power grid, thereby significantly improving the ability to identify abnormal scenarios and the ability to predict complex faults.

[0007] To achieve the above objectives, this application adopts the following technical solution: This application provides a method for power distribution network simulation and load control based on Internet of Things (IoT) data. The method includes: Acquire multi-source IoT data from the distribution network, and perform timestamp gap verification, network topology consistency inference, and cross-source measurement error reshaping on the multi-source IoT data to form a standardized power grid status dataset; A cross-scale operation feature set is constructed based on the standardized power grid state dataset. The cross-scale operation feature set includes local dynamic sub-features, wide-area correlation sub-features, and time-series drift indicators. Based on the local dynamic sub-features, the wide-area correlation sub-features, and the temporal drift index, topological alignment in the physical structure dimension and temporal alignment in the time evolution dimension are performed, and a panoramic operational status feature set is generated through a dual-dimensional joint spatiotemporal alignment mechanism. The panoramic operational status feature set is input into the reinforcement learning agent, and exploratory actions are performed using a simulation environment built based on the power distribution network operation mechanism. The exploratory behavior is guided by a reward convergence mechanism built through system instability constraints, and an extreme complex fault scenario dataset is generated. Based on the aforementioned extreme complex failure scenario dataset, risk severity data is extracted, and risk evolution path data is constructed. Based on the aforementioned risk evolution path data, a defensive load regulation strategy oriented towards the causal chain is constructed by identifying key triggering nodes and easily spread links in the propagation chain. The defensive load control strategy is subjected to multi-condition simulation verification to obtain simulation results. The simulation deviation is obtained by comparing the simulation results with the expected control target of the defensive load control strategy, and an optimized control strategy set is constructed based on the simulation deviation. The optimized control strategy set drives the power distribution network control system to perform load control actions.

[0008] In some possible implementations, the step of performing topological alignment in the physical structure dimension and temporal alignment in the time evolution dimension based on the local dynamic sub-features, the wide-area correlation sub-features, and the temporal drift index, and generating a panoramic operational situation feature set through a dual-dimensional joint spatiotemporal alignment mechanism, includes: Based on the local dynamic sub-features, the topological consistency calibration of the operating state related to local nodes is performed in the physical structure dimension to obtain local topological calibration results. Based on the aforementioned wide-area correlation sub-features, topology matching processing of cross-regional correlation relationships is performed at the physical structure dimension to obtain a wide-area topology correlation mapping result consistent with the overall power grid structure; Based on the aforementioned time-series drift index, time-series alignment correction is performed on the time evolution dimension of cross-source measurement data to obtain time-series data under a unified time reference. The local topology calibration results, the wide-area topology mapping results, and the time-series data under the unified time reference are fused by a dual-dimensional joint spatiotemporal alignment mechanism to generate a panoramic operational status feature set.

[0009] In some possible implementations, the panoramic operational status feature set is input into a reinforcement learning agent, and exploratory actions are performed using a simulation environment constructed based on the power distribution network operation mechanism. The exploratory behavior is guided by a reward convergence mechanism constructed through system instability constraints, generating a dataset of extreme complex fault scenarios, including: The panoramic operational situation feature set is mapped to the state vector of the reinforcement learning agent to obtain state vector data. Based on the state vector data, exploratory actions of the reinforcement learning agent are executed in a simulation environment built based on the distribution network operation mechanism, including the simulation of operations on load nodes, switch states and line parameters, and the action execution result data is obtained. Based on the action execution result data, system instability constraint indicators, and key node risk indicators, calculate the reward value for each action to obtain reward value data. The key node risk indicators include at least one of node betweenness centrality, load importance level, and fault propagation impact range. Based on the reward value data, the action selection probability of the reinforcement learning agent is adjusted through the reward convergence mechanism to guide the reinforcement learning agent to prioritize actions that can cause high risk and cross-regional propagation of faults, and to obtain optimized exploration strategy data. Record the action sequence and corresponding system response of the optimized exploration strategy data in the simulation environment, and comprehensively generate an extreme complex fault scenario dataset.

[0010] In some possible implementations, the extraction of risk severity data and the construction of risk evolution path data based on the extreme complex failure scenario dataset include: The system instability quantification index is extracted from the extreme complex fault scenario dataset. The system instability quantification index includes at least one of voltage over-limit severity, frequency deviation degree and line cascading overload ratio. Based on the quantitative index of system instability, and combined with the action sequences and system responses centrally recorded in the extreme complex fault scenario dataset, the key path node sequence of risk propagation is identified. Based on the sequence of critical path nodes, risk evolution path data is constructed.

[0011] In some possible implementations, the defensive load regulation strategy oriented towards the causal chain is constructed based on the risk evolution path data by identifying key triggering nodes and easily spreadable links in the propagation chain, including: Based on the risk evolution path data, risk driving force analysis is performed on the node sequence in the risk evolution path data to obtain the key triggering node identification results; Based on the risk evolution path data, the expansion trend analysis of the propagation chain links in the risk evolution path data is performed to obtain the propagation chain diffusion link identification results. Based on the identification results of the key inducing nodes and the identification results of the propagation chain diffusion links, control actions and load regulation measures are designed to construct a defensive load regulation strategy oriented towards the causal chain.

[0012] In some possible implementations, performing multi-condition simulation verification on the defensive load control strategy to obtain simulation results includes: Under multiple different loads and operating scenarios, the defensive load control strategy is executed, and the simulation action sequence and corresponding system response data are recorded to obtain simulation action and response data; Based on the simulation action and response data, the load control effect, system stability index and deviation under each working condition are statistically analyzed to obtain simulation result data.

[0013] In some possible implementations, obtaining the simulation deviation by comparing the simulation results with the expected control objective of the defensive load regulation strategy includes: Based on the simulation results data, the load control effect, system stability index and deviation recorded in the simulation results data are structured and organized to obtain the simulation results feature set; Based on the feature set of the simulation results and the target feature set corresponding to the expected control target of the defensive load regulation strategy, a point-by-point comparison analysis is performed to quantify the degree of deviation of each control target in the simulation scenario and obtain the simulation deviation.

[0014] In some possible implementations, the construction of an optimized control strategy set based on the simulation deviation includes: Based on the simulation deviation data, feature analysis is performed on the deviation components in the simulation deviation data that reflect the degree to which the expected control target is not met, and the deviation feature analysis results are obtained. Based on the deviation characteristic analysis results, the load control actions are adjusted according to the risk suppression priority to obtain the optimized control action results; An optimized set of control strategies is constructed based on the results of the aforementioned control actions.

[0015] In some possible implementations, the process of driving the distribution network control system to perform load regulation actions based on the optimized control strategy set includes: Based on the optimized control strategy set, the target load node, control parameters and action timing corresponding to each optimized control strategy are analyzed to obtain the control execution instruction set. According to the control execution instruction set, the corresponding load control operation is executed to obtain system response data; Load regulation actions are performed based on the system response data.

[0016] The beneficial effects of this invention are as follows: 1. The distribution network simulation and load control method based on IoT data described in this invention utilizes a constructed panoramic operation status feature set to drive a reinforcement learning agent to explore actions. It can continuously output state change sequences consistent with the real power grid operation mechanism in the simulation environment, thereby systematically presenting the dynamic response of the power grid under complex conditions such as high-voltage load impact, cross-regional disturbance, and chain overload propagation. This continuous exploration process fully exposes the hidden potential fault accumulation paths, critical state jump points, and high-risk coupling trigger chains, forming a structured dataset of extreme complex fault scenarios. This dataset further enhances the ability to identify high-risk modes of the power grid, enabling the system to anticipate the evolution trend that may lead to serious instability in advance, discover the risk source that may trigger chain diffusion at an earlier stage, and provide a reliable foundation for subsequent risk extrapolation and defense strategy design. 2. The distribution network simulation and load control method based on IoT data described in this invention identifies key inducing nodes and propagation chain diffusion links through risk evolution path data, and constructs a causal chain-based defensive load control strategy on this basis. This enables precise intervention in the direction and pace of risk transmission, establishing a one-to-one correspondence between control actions and fault propagation characteristics. After performing multi-condition simulation verification and generating simulation deviations, the deviations are further analyzed and summarized to form a structured set of optimized control strategies. This makes the load control actions more precise in terms of time investment, node selection, and suppression depth, thereby enabling the control process to simultaneously suppress multiple potential diffusion branches, significantly improving the stability maintenance capability under complex operating conditions, and thus continuously strengthening the overall operational safety of the power grid. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the distribution network simulation and load control method based on IoT data according to the present invention; Detailed Implementation The terms "first," "second," and "third," etc., used in this application specification, claims, and drawings are used to distinguish different objects, not to limit a specific order.

[0019] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0020] Research has revealed that while existing technologies can perform load regulation in distribution networks, inconsistencies in time synchronization, structural correspondence, and cross-regional measurement offsets among multi-source monitoring data prevent the formation of an accurate and unified overall picture of the network's operation in both spatial and temporal dimensions. When the system enters a stage of disturbance, load fluctuation, or local fault propagation, this information misalignment can delay the perception or misinterpretation of the overall situation's changing trends. This makes it difficult for maintenance personnel to identify the true risk transmission chain in a timely manner and to judge the degree of pressure on key nodes. Furthermore, it can lead to deviations in the judgment of the timing, scope, and intensity of load control, thereby increasing the possibility of cascading instability of the system under high-pressure conditions.

[0021] Example 1 To solve the above problems, such as Figure 1 As shown, this application provides a method for power distribution network simulation and load control based on Internet of Things (IoT) data, the method comprising: Step 1: Acquire multi-source IoT data of the distribution network, and perform timestamp gap verification, network topology consistency inference, and cross-source measurement error reshaping on the multi-source IoT data to form a standardized power grid status dataset; In step one: multi-source IoT data: including information such as voltage, current, load, switch status and line parameters collected by various measurement terminals, sensors and monitoring devices in the power distribution network; Timestamp gap verification: This refers to comparing and supplementing the timestamps of data from different sources to ensure the continuity of data in time. Network topology consistency inference: refers to topology matching and correction of collected data based on the actual structure and connection relationships of the power grid; Cross-source measurement error reshaping: refers to the correction and unification of measurement errors present in data from different sources; For example: After collecting raw data from various measurement points, the system first verifies the timestamps and automatically fills in any missing or abnormal timestamps to ensure data continuity and reliability. Next, the system performs topology consistency inference on the data based on the relationship between power grid nodes and lines, correcting data that does not conform to the actual structure. Then, the system performs unified processing on the data deviations from different sensors to ensure that all data are consistent on the measurement scale, ultimately generating a complete and standardized power grid status dataset, ensuring that the data can be directly used for subsequent operation analysis and load regulation. It should be noted that this step ensures the consistency of the collected data in terms of time and structure, and eliminates cross-source errors, thereby forming a high-quality standardized dataset, providing a reliable foundation for generating panoramic operational status characteristics of the distribution network and load regulation.

[0022] Step 2: Construct a cross-scale operation feature set based on the standardized power grid state dataset. The cross-scale operation feature set includes local dynamic sub-features, wide-area correlation sub-features, and time-series drift indicators. In step two: Cross-scale operation feature set: refers to the feature set obtained by multi-dimensional and multi-level analysis of distribution network status data, which is used to describe the operation characteristics of the power grid at different spatial and temporal scales; Local dynamic sub-features: used to reflect the instantaneous state and dynamic changes of a single node or a small area; Wide-area correlation sub-features: used to describe the interrelationships and overall behavioral patterns between different regions or nodes; Time-series drift metric: used to quantify the offset trend and change pattern of cross-source measurement data over time; For example, the system takes a standardized power grid status dataset as input. First, it analyzes the operating data such as voltage, current, and load of each node and its neighboring areas, extracts the dynamic features of local nodes to form local dynamic sub-features, performs correlation analysis and pattern recognition between nodes in different areas to form wide-area correlation sub-features that reflect the mutual influence between regions, and calculates the time evolution offset by tracking historical measurement data to obtain the time series drift index. Finally, these three types of features are organically integrated to form a complete cross-scale operating feature set, providing a unified and coherent input for the subsequent generation of panoramic operating status features. It should be noted that this step can systematically integrate local dynamics, cross-regional correlations, and time evolution characteristics while ensuring data continuity. It not only clearly describes the current operating status of the power grid, but also provides comprehensive and reliable feature support for intelligent simulation, risk assessment, and load control strategies.

[0023] Step 3: Based on the local dynamic sub-features, the wide-area correlation sub-features, and the temporal drift index, perform topological alignment in the physical structure dimension and temporal alignment in the time evolution dimension, and generate a panoramic operational status feature set through a dual-dimensional joint spatiotemporal alignment mechanism; In step three: topology alignment in the physical structure dimension: refers to mapping local and wide-area features to a topology model consistent with the actual power grid structure based on the actual connection relationships of power grid nodes and lines; Temporal alignment in the time evolution dimension: refers to correcting the time series of cross-source measurement data according to the time drift index, so that all features are synchronized under a unified time reference; Dual-dimensional joint spatiotemporal alignment mechanism: refers to the fusion of the above topology alignment and time-series alignment results to form a comprehensive description of the power grid operation status; For example: The system takes local dynamic sub-features, wide-area correlation sub-features and time-series drift indicators as inputs. In a unified data processing flow, the features of each node and region are first mapped to the actual topology model of the power grid. At the same time, the feature data of each node and region are adjusted according to the time-series drift indicators to synchronize them under a unified time reference. Then, the system fuses the topology calibration results and time alignment results to form a panoramic operation status feature set that reflects the overall operation status of the power grid. Thus, local dynamics, cross-regional correlation and time evolution features are reflected in the same model at the same time. It should be noted that this step can unify and integrate local dynamics, cross-regional correlations, and temporal evolution characteristics in space and time, generating a complete and coherent panoramic operational status feature set, providing reliable input for subsequent generation of extreme complex fault scenarios, risk assessment, and defensive load control strategies.

[0024] Step 4: Input the panoramic operation status feature set into the reinforcement learning agent, and use the simulation environment built based on the power distribution network operation mechanism to perform exploratory actions; In step four: Reinforcement learning agent: refers to an algorithmic model that can autonomously select actions based on input state characteristics and optimize strategies through reward feedback in a simulation environment; Simulation environment: A simulation platform built based on the operation mechanism of the power distribution network, used to simulate the impact of load nodes, switch states and line parameter changes on system operation; Exploratory actions: refer to reinforcement learning agents trying different combinations of operations in a simulation environment in order to discover situations that may cause system instability or risk propagation; For example, the system uses the panoramic operational status feature set as the state input of the reinforcement learning agent. In the simulation environment, it simulates the operating status of each node and line of the power grid and allows the agent to perform exploratory actions, such as adjusting the power of load nodes, switching switch states, or modifying line parameters. At the same time, it records the system response caused by each action. Based on the system instability constraints and key node risk indicators, the agent evaluates the effect of the exploratory actions and adjusts the action selection strategy through a reward feedback mechanism, so that it gradually leans towards the operation path that can reveal potential high risks and cross-regional propagation, thereby generating a comprehensive exploratory data sequence in the simulation environment.

[0025] It should be noted that this step, under controlled simulation conditions, enables the autonomous exploration of reinforcement learning agents to map the panoramic operational status characteristics into action sequences and system response data. This provides reliable data support for the generation of extreme complex fault scenarios, risk evolution path analysis, and defensive load control strategies, while ensuring that the data source is traceable, continuous, and representative.

[0026] Step 5: Guide exploration behavior through a reward convergence mechanism built using system instability constraints to generate a dataset of extreme complex failure scenarios; In step five: System instability constraints: used to limit the exploration range of the reinforcement learning agent. By setting thresholds for key operating indicators such as voltage over-limit, line overload and node power fluctuation, the exploration process is guided to remain within a controllable risk range. Reward convergence mechanism: refers to rewarding or punishing each exploration action based on the system's instability constraints, so that the reinforcement learning agent's policy gradually stabilizes and converges to the action direction that can trigger typical high-risk patterns; Extreme complex fault scenario dataset: refers to a data set generated by reinforcement learning agents in a simulation environment, containing multi-node and multi-line coupled disturbances and operating state sequences that may lead to system instability; For example: The exploratory action sequence generated by the reinforcement learning agent in step four and the system response data of the simulation environment are used as inputs. Operating state variables such as voltage offset, line load rate changes, and node power oscillation amplitude are substituted into the system instability constraints for successive verification to determine whether the current exploratory action approaches or touches a potential high-risk boundary. Subsequently, the system inputs the above verification results into the reward convergence mechanism, updating the reward value according to the power grid operation mechanism. This makes the reinforcement learning agent more inclined to choose action paths that can trigger cross-regional associated risks, chain load disturbances, and multi-node synchronous instability in subsequent explorations. In multiple iterations, the system gradually accumulates high-risk operation sequences generated by the agent, corresponding power grid response states, and composite fault modes formed by the superposition of multi-dimensional disturbances. Through the converged reward strategy, typical representative scenarios are selected, ultimately forming a dataset of extreme composite fault scenarios with a clear structure, complete event chains, and the ability to realistically reflect the propagation path of extreme operating risks in the distribution network. It should be noted that this step, through the combined effect of system instability constraints and reward convergence mechanisms, ensures that the exploration of reinforcement learning agents is no longer disorderly diffusion, but rather gradually focuses on potentially high-risk operational spaces, achieving efficient capture of cross-regional risk coupling, chain disturbance propagation, and multi-factor collaborative instability mechanisms. The resulting extreme complex fault scenario dataset can not only characterize the complex instability modes that real power grids may face, but also provide a reliable data foundation for subsequent risk assessment model construction, defensive load control strategy design, and multi-scenario verification.

[0027] Step Six: Extract risk severity data based on the extreme complex failure scenario dataset and construct risk evolution path data; In step six: Risk severity data: used to quantify the risk intensity information presented by each node or region under extreme complex failure scenarios; Risk evolution path data: used to describe the time series evolution trajectory of extreme complex failures from initial triggering to chain propagation, reflecting the continuous process of the failure's impact range, propagation direction, and evolution stages; For example, using an extreme complex fault scenario dataset as input, the system first identifies the state change trends of each node before and after the fault disturbance in a unified data analysis process. Based on voltage offset, power flow over-limit amplitude, and component load changes, it calculates risk severity data that reflects the risk intensity. Subsequently, the system summarizes these state changes according to the time sequence from the fault triggering moment to the propagation stage, and further connects the continuous change process according to the influence relationship between nodes to form a complete evolution link from the initial disturbance to the chain propagation. On this basis, the system organizes these evolution links into structured risk evolution path data according to the time progression logic and the order of related nodes, so that the generation, expansion, and trend changes of extreme complex faults can be presented coherently in the same data structure, ensuring that subsequent analysis can be carried out on a consistent and continuous information basis. It should be noted that this step enables the precise quantification of the risk intensity caused by extreme compound failures and provides a structured expression of the risk propagation process in time and space, thereby ensuring that subsequent risk classification, risk trend inference, and load control strategy optimization have a clear and reliable data foundation.

[0028] Step 7: Based on the risk evolution path data, construct a defensive load regulation strategy oriented towards the causal chain by identifying key triggering nodes and easily spread links in the propagation chain; In step seven: key triggering node: refers to the node in the risk evolution path data that first triggers the spread of the fault or contributes the most to the subsequent risk evolution; Easily spread links in the propagation chain: These refer to critical links in the risk evolution path where faults can easily spread rapidly from upstream nodes to downstream nodes. Defensive load control strategies oriented towards causal chains: These are control strategies that proactively intervene in the load of key nodes to suppress the spread of risk based on the causal relationship of risk evolution. For example: Taking risk evolution path data as input, the system first analyzes the state changes and chain diffusion sequence of each node in the path to identify the key triggering node that has the greatest impact on the overall evolution trend in the initial stage of the fault. Then, the system compares the risk transmission relationship between nodes segment by segment along the risk evolution path to locate the link in the propagation chain where the fault is most likely to be amplified and spread. Combining the positional relationship of these nodes in the chain structure, the system determines the load node that is most sensitive to the system stability. At this time, the system constructs a defensive load control strategy consistent with the risk evolution sequence based on the causal triggering role of the key triggering node and the diffusion trend of the easily diffused link in the propagation chain. This allows the control command to act on the corresponding load node in time before the risk spreads, thereby achieving the pre-suppression and link weakening of potential diffusion paths and forming a complete control strategy structure oriented towards the causal chain. It should be noted that this step can transform risk evolution path data into structured causal chain information, thereby identifying key triggering nodes and easily spread links in the propagation chain. This enables load control strategies to intervene proactively at the source and main transmission path of fault propagation, effectively blocking the risk propagation chain and improving the defense capability and operational stability of the distribution network under extreme complex fault conditions.

[0029] Step 8: Perform multi-condition simulation verification on the defensive load control strategy and obtain simulation results; In step eight: Multi-condition simulation verification: refers to simulating and testing the defensive load control strategy under different load levels, line conditions and node disturbances to evaluate the effectiveness and stability of the strategy under various operating environments; Simulation results refer to the data recorded during the simulation process, such as the load response of each node, system stability indicators, and risk propagation status, which are used for subsequent analysis and strategy optimization. For example: The constructed defensive load control strategy is imported into the simulation platform, and the control operation is continuously executed under various operating conditions such as different load levels, key node disturbances, and line anomalies. During the simulation, the system records the load adjustment of each node, the risk changes of key nodes, and the overall stability indicators of the system in real time, and associates these data with the strategy input and execution actions. Secondly, the system analyzes the response effect of the strategy under various operating conditions based on the recorded data, identifies weaknesses or potential risk propagation paths, and makes necessary adjustments and optimizations to the defensive load control strategy to ensure that the strategy can be stably implemented under various extreme and complex fault conditions, and generates a complete and coherent simulation result dataset, providing a reliable basis for subsequent strategy verification and optimization. It should be noted that this step can verify the feasibility and effectiveness of the defensive load control strategy through multi-condition simulation and obtain detailed simulation results, providing reliable data support for strategy optimization, risk control effect evaluation and practical application deployment.

[0030] Step 9: Obtain the simulation deviation by comparing the simulation results with the expected control target of the defensive load regulation strategy, and construct an optimized regulation strategy set based on the simulation deviation; In step nine: Simulation deviation: refers to the difference between the actual system response and the expected control target of the defensive load regulation strategy in multi-condition simulation; Optimized control strategy set: refers to multiple strategy combinations generated based on simulation deviations, used to further improve the accuracy, stability and risk suppression effect of load control; For example: The simulation results of multiple operating conditions are compared item by item with the expected control objectives of the defensive load control strategy to identify the degree of deviation of load adjustment and key risk indicators at each node, thereby obtaining the simulation deviation. Secondly, the system will automatically adjust the load allocation strategy and key node control scheme based on the deviation analysis results, generate a set of optimized control strategies, and ensure that each strategy maintains a defensive constraint against the spread of potential risks during the generation process, and record the applicable conditions and corresponding effects of each strategy to provide data support for further iterative optimization. It should be noted that this step can perform a closed-loop comparison between the actual simulation feedback and the expected control target, forming a set of targeted optimized control strategies, thereby improving the accuracy and reliability of defensive load control strategies, and providing a set of feasible solutions for the safe operation of the system under extreme complex fault scenarios.

[0031] Step 10: Drive the distribution network control system to perform load regulation actions based on the optimized control strategy set; In step ten: Load regulation action: refers to the operation behavior of the distribution network control system in adjusting the load distribution, switch status or voltage control device of each node according to the optimized regulation strategy set, so as to improve system stability and suppress risks; For example: The optimized control strategy set is used as input, and instructions are issued by the distribution network control system to adjust the load distribution, switch status and operating parameters of related control equipment at each node in sequence. During the execution, the control system monitors the response of key nodes and the operation status of the power grid in real time, and automatically corrects the operation sequence or magnitude according to the strategy requirements to ensure that the load control action meets the safety constraints and minimizes potential risks. After the execution is completed, the system records all control actions and their corresponding power grid responses to provide data for subsequent effect evaluation and strategy optimization. It should be noted that this step can transform the optimized control strategy set into actual executable load control operations, realize proactive control of the power grid operation status, thereby reducing risk propagation, improving system stability, and ensuring that the control process is traceable and verifiable in actual operation.

[0032] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A power distribution network simulation deduction and load regulation method based on Internet of Things data, characterized in that, The method comprises: acquiring multi-source Internet of Things data of a power distribution network, and performing timestamp gap checking, network topology structure consistency inference, and cross-source measurement error reprocessing on the multi-source Internet of Things data to form a standardized power grid state data set; constructing a cross-scale operation feature set based on the standardized power grid state data set, the cross-scale operation feature set comprising local dynamic sub-features, wide-area correlation sub-features, and time series drift indicators; performing topology alignment in the physical structure dimension and time series alignment in the time evolution dimension based on the local dynamic sub-features, the wide-area correlation sub-features, and the time series drift indicators, and generating a panoramic operation situation feature set through a two-dimensional joint space-time alignment mechanism; inputting the panoramic operation situation feature set into a reinforcement learning agent, and executing exploratory actions using a simulation environment constructed based on the operation mechanism of the power distribution network, guiding the exploratory behavior through a reward convergence mechanism constructed based on system instability constraints to generate an extreme compound fault scenario data set; extracting risk severity data based on the extreme compound fault scenario data set, and constructing risk evolution path data; based on the risk evolution path data, constructing a defensive load regulation strategy oriented to causal chains by identifying key induced nodes and easily diffusing links in the propagation chain; performing multi-working-condition simulation verification on the defensive load regulation strategy to obtain simulation results, comparing the simulation results with expected control targets of the defensive load regulation strategy to obtain simulation deviations, and constructing an optimized regulation strategy set based on the simulation deviations; driving the power distribution network control system to execute load regulation actions based on the optimized regulation strategy set. 2.The power distribution network simulation deduction and load regulation method based on Internet of Things data according to claim 1, wherein, The method comprises: based on the local dynamic sub-features, performing topology consistency calibration of the operation state related to the local nodes in the physical structure dimension to obtain a local topology calibration result; based on the wide-area correlation sub-features, performing topology matching processing of the cross-regional correlation relationship in the physical structure dimension to obtain a wide-area topology correlation mapping result consistent with the overall structure of the power grid; based on the time series drift indicators, performing time series alignment correction of the time evolution offset of the cross-source measurement data in the time evolution dimension to obtain time series data under a unified time reference; fusing the local topology calibration result, the wide-area topology mapping result, and the time series data under the unified time reference through a two-dimensional joint space-time alignment mechanism to generate a panoramic operation situation feature set. 3.The power distribution network simulation deduction and load regulation method based on Internet of Things data according to claim 1, characterized in that, The method comprises: mapping the panoramic operation situation feature set into a state vector of the reinforcement learning agent to obtain state vector data; Based on the state vector data, exploratory actions of the reinforcement learning agent are performed in a simulation environment constructed based on the operation mechanism of the power distribution network, including operation simulation of load nodes, switch states and line parameters, and obtaining action execution result data; According to the action execution result data and system instability constraint indicators and key node risk indicators, the reward value of each action is calculated to obtain reward value data, and the key node risk indicators include at least one of node betweenness centrality, load importance level and fault propagation influence range; Based on the reward value data, the action selection probability of the reinforcement learning agent is adjusted through a reward convergence mechanism to guide the reinforcement learning agent to preferentially perform actions that can trigger high-risk, cross-regional fault propagation, and obtain optimized exploration strategy data; Record the action sequence and corresponding system response of the optimized exploration strategy data in the simulation environment to comprehensively generate extreme composite fault scenario data set. 4.The power distribution network simulation deduction and load regulation method based on Internet of Things data according to claim 3, characterized in that, The risk severity data is extracted based on the extreme composite fault scenario data set, and the risk evolution path data is constructed, including: Extracting system instability degree quantitative indicators from the extreme composite fault scenario data set, the system instability degree quantitative indicators including at least one of voltage out-of-limit severity, frequency deviation degree and line chain overload ratio; Based on the system instability degree quantitative indicators, the key path node sequence of risk propagation is identified by combining the action sequence and system response recorded in the extreme composite fault scenario data set; Based on the key path node sequence, the risk evolution path data is constructed.

5. The power distribution network simulation deduction and load regulation method based on Internet of Things data according to claim 4, characterized in that, Based on the risk evolution path data, a defensive load regulation strategy oriented to the causal chain is constructed by identifying key induced nodes and propagation chain diffusion links, including: Based on the risk evolution path data, risk driving force analysis is performed on the node sequence in the risk evolution path data to obtain key induced node identification results; Based on the risk evolution path data, expansion trend analysis is performed on the propagation chain links in the risk evolution path data to obtain propagation chain diffusion link identification results; Based on the key induced node identification results and the propagation chain diffusion link identification results, control actions and load regulation measures are designed to construct a defensive load regulation strategy oriented to the causal chain. 6.The power distribution network simulation deduction and load regulation method based on Internet of Things data according to claim 5, characterized in that, The defensive load regulation strategy is simulated and verified under multiple working conditions to obtain simulation results, including: In multiple different load and operation scenarios, the defensive load regulation strategy is executed, and simulation action sequence and corresponding system response data are recorded to obtain simulation action and response data; Based on the simulation action and response data, the load regulation effect, system stability indicators and deviation under each working condition are counted to obtain simulation result data. 7.The power distribution network simulation deduction and load regulation method based on IOT data according to claim 6, characterized in that, The simulation deviation is obtained by comparing the simulation results with the expected control target of the defensive load regulation strategy, including: Based on the simulation result data, the load regulation effect, system stability indicators and deviation recorded in the simulation result data are structured and arranged to obtain a simulation result feature set; Performing index-by-index comparison analysis based on the simulation result feature set and the target feature set corresponding to the expected control target of the defensive load regulation strategy, quantifying the deviation degree of each control target in the simulation scenario, and obtaining simulation deviation. 8.The power distribution network simulation deduction and load regulation method based on IOT data according to claim 7, wherein, Based on the simulation deviation, an optimized regulation strategy set is constructed, including: Based on the simulation deviation data, the deviation components reflecting the unmet degree of the expected control target in the simulation deviation data are analyzed, and the deviation feature analysis result is obtained; Based on the deviation feature analysis result, the load regulation action is adjusted according to the risk suppression priority, and the regulation action optimization result is obtained; Based on the regulation action optimization result, an optimized regulation strategy set is constructed. 9.The power distribution network simulation deduction and load regulation method based on IOT data according to claim 8, characterized in that, Based on the optimized regulation strategy set, the power distribution network control system is driven to perform load regulation action, including: Based on the optimized regulation strategy set, the target load node, control parameter and action time sequence corresponding to each optimized regulation strategy are analyzed, and the regulation execution instruction set is obtained; According to the regulation execution instruction set, the corresponding load regulation operation is performed, and the system response data is obtained; Based on the system response data, the load regulation action is completed.

Images