Real-time safety risk monitoring and emergency response device of new energy power system

Abstract

The invention relates to the technical field of new energy electric power construction and operation control, in particular to a real-time security risk monitoring and emergency response device of a new energy electric power system, which is characterized in that an edge layer completes time alignment, quality marking and local abnormity emergency disposal of multi-source heterogeneous data; the cloud control layer constructs an integrated three-dimensional map fusing electrical topology, asset relationship and geographic space information, carries out risk assessment in combination with time sequence characteristics and an electrical equipment mechanism model, and outputs a risk degree, credibility, a propagation range and a disposal time window result; the intelligent decision-making layer generates a disposal action sequence meeting system and equipment constraints through multi-objective optimization by utilizing collaborative learning strategy prior based on a risk assessment result, and maps the disposal action sequence into an executable instruction set; and the execution and auditing layer issues a control instruction and collects execution feedback data. All the layers are connected through preset interfaces to form a closed-loop cooperative control system, so that the safety operation and risk response capability of the station is improved.

Description

Technical Field

[0001] This invention relates to the field of new energy power construction and operation control technology, specifically to a real-time safety risk monitoring and emergency response device for new energy power systems. Background Technology

[0002] New energy power systems refer to power generation and grid-connected operation systems that primarily utilize wind and solar power and incorporate energy storage devices. These systems are characterized by rapidly changing output due to weather conditions, decreased inertia and short-circuit capacity, and weakened voltage and frequency support capabilities. Against this backdrop, operational safety monitoring and emergency response rely on multi-source monitoring methods, including monitoring and data acquisition systems, phasor measurement units, energy storage battery management systems, and sensor data such as temperature, vibration, partial discharge, and infrared imaging. Simultaneously, it is necessary to identify and trigger appropriate actions based on the system's status within a relatively short timescale.

[0003] Current engineering practices generally employ a combination of threshold alarms and rule bases for online monitoring and alarms, supplemented by manual handling based on the dispatch terminal's operational diagrams and emergency plans. Existing systems are often built in a layered and domain-specific manner along the data link, lacking unified spatiotemporal data representation and quality labeling standards among monitoring and data acquisition systems, wide-area measurement systems, protection and automation devices, and station control systems. Time synchronization accuracy and data integrity are significantly affected by network conditions. In the analysis phase, single-source or few-source data-driven methods struggle to characterize the mechanistic constraints of the coupling between new energy equipment and the power grid, resulting in limited lead time and reliability for risk identification. In the handling phase, pre-implementation operational constraint verification and post-implementation evidence recording are scattered across different systems, making it difficult to form a consistent closed-loop management system.

[0004] Therefore, the technical problem to be solved is: in scenarios with a high proportion of renewable energy grid connection, to provide a unified spatiotemporal expression and management method for multi-source heterogeneous monitoring data, and to form a reliable safety risk assessment result and disposal time window determination within the real-time constraint, so as to achieve consistent connection and traceable recording with the control execution process, thereby supporting the online safety risk monitoring and emergency response triggering judgment of the renewable energy power system. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a real-time safety risk monitoring and emergency response device for new energy power systems, which addresses the shortcomings of the prior art and solves the technical problems of SS.

[0006] The objective of this invention is achieved through the following technical solutions: In a first aspect, the present invention provides a real-time safety risk monitoring and emergency response device for a new energy power system, comprising: The edge layer is used to access multi-source heterogeneous data, perform time alignment and quality marking on the multi-source heterogeneous data, perform preliminary emergency handling on abnormal data in the quality marking, and output spatiotemporal data. The cloud control layer is used to receive the spatiotemporal data output by the edge layer, construct a three-dimensional map that integrates electrical topology, asset relationships and geospatial information based on the spatiotemporal data, and conduct risk assessment based on the three-dimensional map, combined with the temporal characteristics in the spatiotemporal data and the electrical equipment mechanism model, and output the risk assessment results. The risk assessment results include at least the risk level, credibility, propagation range and disposal time window. The intelligent decision-making layer is used to generate at least one candidate sequence of actions for handling the risk assessment results based on the risk assessment results output by the cloud control layer, and based on a preset multi-objective optimization model and constraints, and to map the sequence of actions for handling the risk assessment results into an executable instruction set through security projection. The execution and audit layer is used to receive the set of executable instructions output by the intelligent decision-making layer, issue control instructions to the automatic generation control, automatic voltage control and station control execution interfaces, and collect feedback data and system measurement results after instruction execution.

[0007] As a further improvement of the present invention, the edge layer is specifically configured as follows: Key electrical parameters of wind turbines and photovoltaic inverters in new energy power systems are sampled at intervals of seconds or half seconds. The state of charge and temperature data of the energy storage battery management subsystem in the new energy power system are sampled at a period of half a second. The phasor measurement unit in the new energy power system is sampled at a sampling rate of 30-120 points / second; the infrared and conductor temperature sensors in the new energy power system are sampled at a sampling rate of a first set frequency; and the acoustic vibration sensors in the new energy power system are sampled at a sampling rate of a second set frequency. The data sampled from wind turbines and photovoltaic inverters, energy storage battery management subsystems, phasor measurement units, infrared and wire temperature sensors and acoustic vibration sensors are preprocessed to form multi-source heterogeneous data. Add timestamps based on a unified time base and quality marker fields that characterize data quality to multi-source heterogeneous data frames.

[0008] As a further improvement of the present invention, the edge layer uses the IEEE 1588 time protocol for station clock synchronization, and controls the time synchronization error to be no higher than 1 millisecond.

[0009] As a further improvement of the present invention, the cloud control layer includes a spatiotemporal data carrying component. The spatiotemporal data carrying component is used to divide the spatiotemporal data into a raw data module, a feature data module, and a labeled data module according to different types of spatiotemporal data. The electrical topology, asset relationships, and geospatial information in the raw data module, feature data module, and labeled data module are fused as the dimensions of the three-dimensional map. The equipment, buses, and feeders in the raw data module, feature data module, and labeled data module are used as nodes, and the equivalent impedance, geographical distance, and control coupling coefficient in the raw data module, feature data module, and labeled data module are used as edges to construct the three-dimensional map. The feature data module includes current fluctuation rate, frequency deviation, reactive power margin, DC side ripple rate, temperature rise rate, and phase angle difference change rate calculated based on the original data; the annotation data module records the types of fault events and their start and end times that are annotated in history or in real time.

[0010] As a further improvement of the present invention, the cloud control layer also includes a risk assessment engine, which is used for: Based on the aforementioned three-dimensional map, a graphical model is used to infer the propagation of risk in the spatiotemporal dimension within one to three steps, resulting in data-driven results. Causal reasoning methods are used to determine the correlation between fault precursor signals and potential events in time series characteristics; An evidence fusion method is used to synthesize the data-driven results and the mechanism consistency results calculated based on the electrical equipment mechanism model to obtain a fusion risk value; The fusion risk value is classified based on a preset threshold, and the risk assessment result is output.

[0011] As a further improvement of the present invention, the cloud control layer further includes an online simulation verification component, which is used to perform operational constraint verification before the action sequence generated by the intelligent decision-making layer is executed, including: Power flow calculations are performed using the Newton-Raphson method, with a minimum of 10 -5 The convergence accuracy verification system power flow distribution; Eigenvalue analysis was used to perform small disturbance stability analysis, and a damping ratio of not less than 0.05 was used as the stability threshold. Based on the equipment thermal model, the temperature rise and power ratio of the transformer and energy storage equipment when executing candidate strategies are calculated and verified to see if they exceed the upper limit of the allowable threshold.

[0012] As a further improvement of the present invention, the multi-objective optimization model of the intelligent decision-making layer includes: Set optimization objectives, which include at least minimizing system residual risk, minimizing power loss, minimizing load shedding costs, and minimizing the number of operations. The decision variables include at least the active / reactive power setpoints of energy storage, the curtailment ratio of wind and solar power generation, the combination of switching actions for power flow reconfiguration, and the system segmentation and disconnection actions; The constraints include at least the system active / reactive power balance constraints, line thermal stability limit constraints, bus voltage upper and lower limit constraints, system frequency deviation limit constraints, energy storage state of charge dynamic change constraints, and converter current limit constraints. A collaborative solution algorithm combining mixed integer quadratic programming and heuristic neighborhood search is used to obtain a sequence of actions for at least one candidate disposal.

[0013] As a further improvement of the present invention, the intelligent decision-making layer further includes: The collaborative learning module is used to transfer and collaboratively train policy knowledge among multiple new energy power plants. It uses secure aggregation technology and differential privacy to process the gradient summary of the uploaded multi-objective optimization model. The formation strategy module operates under a centralized training and distributed execution architecture. Its state space includes at least the energy storage charge status of each station, the voltage of key bus, the system frequency, and the global risk level. Its action space includes at least the adjustment of energy storage charging and discharging power, the adjustment of reactive power support capacity, and the issuance of disconnection commands. The reward function is designed to penalize the increase in risk value, the increase in voltage deviation, and the increase in abandoned electricity.

[0014] As a further improvement of the present invention, the execution and audit layer includes: Based on the executable instruction set, continuous adjustment commands are issued through the automatic generation control and automatic voltage control channels; Based on the executable instruction set, discrete operation instructions are executed through the station control, and throttling control is implemented on the operation of the same switch to ensure that the number of times the same switch is operated does not exceed the threshold within any set time window; discrete operation instructions include switch opening and closing and segmented disconnection issued by the interface.

[0015] Secondly, the present invention provides a method for real-time safety risk monitoring and emergency response in a new energy power system, comprising: Acquire multi-source heterogeneous data, perform time alignment and quality labeling on the multi-source heterogeneous data, perform preliminary emergency handling on detected local anomalies, and output spatiotemporal data; Based on the aforementioned spatiotemporal data, a three-dimensional map integrating electrical topology, asset relationships, and geospatial information is constructed. Based on the three-dimensional map, combined with temporal characteristics and electrical equipment mechanism models, a risk assessment is conducted, and the risk assessment results are output. The risk assessment results include at least the risk level, credibility, propagation range, and disposal time window. Based on the risk assessment results, and the preset multi-objective optimization model and constraints, at least one candidate sequence of actions for handling the risk assessment results is generated, and the sequence of actions is mapped into an executable instruction set through safety projection. The executable instruction set sends control commands to the automatic generation control, automatic voltage control and station control execution interfaces, and collects feedback data and system measurement results after the command execution.

[0016] The beneficial effects of this invention are as follows: This invention provides a real-time safety risk monitoring and emergency response device for a new energy power system. In this device, the edge layer accesses multi-source heterogeneous data and performs time alignment, quality marking, and local anomaly emergency handling, outputting spatiotemporal data to achieve accurate and efficient data preprocessing and local anomaly handling, improving data consistency and anomaly response timeliness. The cloud control layer constructs a three-dimensional map integrating electrical topology, asset relationships, and geospatial information based on spatiotemporal data, and performs risk assessment by combining time-series characteristics and electrical equipment mechanism models, outputting multi-dimensional assessment results such as risk level, credibility, propagation range, and handling time window, achieving comprehensive quantification and precise characterization of risk assessment. The intelligent decision-making layer generates candidate action sequences based on risk assessment results and a preset multi-objective optimization model and constraints. These sequences are then mapped into an executable instruction set through safe projection, enabling decision optimization and safe instruction conversion under multiple constraints. The execution and audit layer receives the executable instruction set and sends control commands to the automatic generation control, automatic voltage control, and station control execution interfaces. It also collects execution feedback data and system measurement results to achieve accurate instruction execution and effective feedback on execution results. Each functional layer connects to the control channel through preset data interfaces to form a closed-loop collaborative control system, achieving efficient collaboration and closed-loop optimization of functions at each layer, and collaboratively improving the safe operation capability and risk response efficiency of new energy power plants. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a real-time safety risk monitoring and emergency response platform for a new energy power system according to the present invention; Figure 2 This is an internal structural diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives and technical solutions of this invention clearer and easier to understand, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Example 1 This embodiment provides a real-time safety risk monitoring and emergency response device for a new energy power system. The device includes an edge layer, a cloud control layer, an intelligent decision-making layer, and an execution and auditing layer. The edge layer is used to access multi-source heterogeneous data, perform time alignment and quality marking on the multi-source heterogeneous data, and perform preliminary emergency handling on abnormal data detected in the quality markings, outputting spatiotemporal data. The cloud control layer is used to receive spatiotemporal data output from the edge layer. Based on the spatiotemporal data, it constructs a three-dimensional map that integrates electrical topology, asset relationships, and geospatial information. Based on the three-dimensional map, combined with the temporal characteristics in the spatiotemporal data and the electrical equipment mechanism model, it conducts risk assessment and outputs risk assessment results. The risk assessment results include at least the risk level, credibility, propagation range, and disposal time window. The intelligent decision-making layer is used to generate at least one candidate sequence of actions based on the risk assessment results output by the cloud control layer, based on a preset multi-objective optimization model and constraints, and to map the sequence of actions into an executable instruction set through security projection. The execution and audit layer is used to receive the executable instruction set output by the intelligent decision layer, issue control instructions to the automatic generation control, automatic voltage control and station control execution interfaces, and collect feedback data and system measurement results after instruction execution.

[0022] The edge layer is specifically configured as follows: Key electrical parameters of wind turbines and photovoltaic inverters in the new energy power system are sampled at intervals of seconds or half-seconds; the state of charge and temperature data of the energy storage battery management subsystem in the new energy power system are sampled at intervals of half-seconds; phasor measurement units in the new energy power system are sampled at a sampling rate of 30-120 points / second; infrared and conductor temperature sensors in the new energy power system are sampled at a sampling rate of a first set frequency; acoustic vibration sensors in the new energy power system are sampled at a sampling rate of a second set frequency; the sampled data from wind turbines and photovoltaic inverters, energy storage battery management subsystems, phasor measurement units, infrared and conductor temperature sensors, and acoustic vibration sensors are preprocessed to form multi-source heterogeneous data; a timestamp based on a unified time base and a quality marker field characterizing data quality are added to the multi-source heterogeneous data frames. The edge layer uses the IEEE 1588 time protocol for station clock synchronization, controlling the time synchronization error to be no higher than 1 millisecond.

[0023] The cloud control layer includes a spatiotemporal data carrier component, which is divided into raw data modules, feature data modules, and labeled data modules according to different types of spatiotemporal data. Electrical topology, asset relationships, and geospatial information from the raw data modules, feature data modules, and labeled data modules are integrated as dimensions of a three-dimensional map. Equipment, buses, and feeders from the raw data modules, feature data modules, and labeled data modules are used as nodes, and equivalent impedance, geographical distance, and control coupling coefficients from the raw data modules, feature data modules, and labeled data modules are used as edges to construct the three-dimensional map. The feature data module includes at least current fluctuation rate, frequency deviation, reactive power margin, DC side ripple rate, temperature rise rate, and phase angle difference change rate calculated based on the raw data. The labeled data module records historical or real-time labeled fault event categories and their start and end times. By categorizing the spatiotemporal data carrier components into raw data modules, feature data modules, and labeled data modules, the system achieves classified management of different types of spatiotemporal data, improving data processing efficiency. The electrical topology, asset relationships, and geospatial information from each module are integrated as dimensions of the 3D map, forming a multi-dimensional information fusion-based three-dimensional data representation framework, enhancing the map's ability to represent complex spatiotemporal relationships. The 3D map is constructed using equipment, buses, and feeders as nodes, giving power network entities a clear topological location within the map. The 3D map is constructed using equivalent impedance, geographical distance, and control coupling coefficients as edges, transforming physical parameters, spatial distance, and control relationships into map edge attributes, achieving a quantitative mapping of electrical, spatial, and control characteristics. Through the collaborative definition of the above nodes and edges, a 3D map integrating electrical topology, asset relationships, and geospatial information is ultimately formed. Compared to traditional 2D maps, this map possesses a more three-dimensional spatiotemporal relationship representation capability, a more accurate entity association analysis capability, and a more comprehensive multi-dimensional information integration capability, thereby improving the accuracy and comprehensiveness of power system spatiotemporal data analysis.

[0024] The cloud control layer also includes a risk assessment engine, which is used to: infer the propagation of risk in one to three steps in the spatiotemporal dimension based on a three-dimensional graph and a graph model to obtain data-driven results; determine the correlation between fault precursor signals and potential events using causal reasoning methods; synthesize the data-driven results and the mechanism consistency results calculated based on the electrical equipment mechanism model using evidence fusion methods to obtain a fused risk value; classify the fused risk value based on a preset threshold and output the risk assessment results.

[0025] The cloud control layer also includes an online simulation verification component. This component is used to perform operational constraint verification before the action sequence generated by the intelligent decision-making layer is executed. This includes: performing power flow calculations using the Newton-Raphson method, and setting a minimum value of 10. -5 The convergence accuracy of the system power flow distribution is verified; the eigenvalue analysis method is used to perform small disturbance stability analysis, and a damping ratio of not less than 0.05 is used as the stability threshold; based on the equipment thermal model, the temperature rise and power ratio of the transformer and energy storage equipment when executing candidate strategies are calculated and verified to see if they exceed the upper limit of the allowable threshold.

[0026] The multi-objective optimization model of the intelligent decision-making layer includes: setting optimization objectives, which at least include minimizing system residual risk, minimizing power loss, minimizing load shedding costs, and minimizing the number of operations; decision variables at least include energy storage active / reactive power setpoints, wind and solar power generation curtailment ratios, switching action combinations for power flow reconfiguration, and system segmentation actions; constraints at least include system active / reactive power balance constraints, line thermal stability limit constraints, bus voltage upper and lower limit constraints, system frequency deviation limit constraints, energy storage state of charge dynamic change constraints, and converter current limiting constraints; and employing a collaborative solution algorithm combining mixed integer quadratic programming and heuristic neighborhood search to obtain a sequence of actions for at least one candidate disposal.

[0027] The intelligent decision-making layer also includes: a collaborative learning module, used for transferring and collaboratively training strategy knowledge among multiple new energy power stations, employing secure aggregation technology and differential privacy to process the gradient summary of the uploaded multi-objective optimization model; and a formation strategy module, which operates under a centralized training and distributed execution architecture. Its state space includes at least the energy storage charge state of each power station, key bus voltage, system frequency, and global risk level, while its action space includes at least the adjustment of energy storage charging and discharging power, adjustment of reactive power support capacity, and issuance of decoupling instructions. The reward function is designed to penalize increases in risk value, voltage deviation, and abandoned power simultaneously.

[0028] Example 2 This embodiment addresses the technical problems of delayed risk identification, scattered data sources, lack of mechanism constraint verification, and incomplete response chain in new energy power systems under high-proportion grid connection conditions. It provides a real-time safety risk monitoring and emergency response platform for new energy power systems. (See also...) Figure 1 The platform is divided into an edge layer, a cloud control layer, an intelligent decision-making layer, and an execution and audit layer, forming a technical chain of monitoring and data collection, risk assessment, strategy generation, secure execution, and closed-loop auditing. Through layered collaboration and unified data and control channels, the platform ensures that the entire process from risk discovery to handling and execution is verifiable, traceable, and reproducible.

[0029] The edge layer consists of a data acquisition gateway, edge intelligent units, and a local emergency interface. The data acquisition gateway interfaces with wind turbines, photovoltaic inverters, energy storage battery management systems, phasor measurement units, and sensors for temperature, vibration, partial discharge, and infrared detection. It supports IEC61850 MMS and GOOSE, IEC 60870-5-104, IEEE C37.118, and Modbus-TCP protocols, performing message parsing, unified timestamp encapsulation, and generating quality tag fields. The sampling period for key electrical parameters of wind turbines and inverters is one second or half a second; the sampling period for the state of charge and string temperature on the energy storage side is half a second; the phasor measurement unit has a sampling rate of 30 to 120 points per second; infrared and wire temperature measurement is at 10 Hz; and acoustic vibration is at 1 kHz. The edge intelligent unit performs isolated forest and cumulative sum detection, with a sliding window of 60 seconds, an update step size of 5 seconds, and a sensitivity threshold of twice the standard deviation; it also performs incremental training using self-supervised contrastive learning at 30-minute intervals. Time synchronization adopts the IEEE 1588 precision time protocol, with an intra-station synchronization error of no more than one millisecond. Under network outage conditions, drift compensation is performed by a real-time clock, with a drift rate of no more than one in a million. The local emergency interface is directly connected to the station control system and the energy storage battery management system, and the delay from triggering to execution of the action link is no more than three hundred milliseconds.

[0030] The cloud control layer consists of a spatiotemporal data bus and a feature warehouse (i.e., the spatiotemporal data carrier component in Example 1), a risk engine, and a digital twin simulation sandbox. The spatiotemporal data bus is divided into themes based on station, voltage level, and equipment type, with unified millisecond-level timestamps and a single-theme throughput of no less than 50,000 frames per second. The feature warehouse is divided into a raw data layer, a feature data layer, and a labeled data layer. Feature items include current fluctuation rate, frequency deviation, reactive power margin, DC side ripple rate, temperature rise rate, and phase angle difference change rate. The labeled layer records the fault type and start and end times. The risk engine conducts assessments based on a three-dimensional map integrating electrical topology, asset topology, and geospatial information. Nodes are equipment, buses, and feeders, and edge weights are determined by equivalent impedance, geographical distance, and control coupling coefficient. State mutations trigger incremental updates in the neighborhood. Risk propagation using a graph neural network and causal reasoning using a dynamic Bayesian network are executed on this map, and evidence fusion is used to calculate the fused risk value and credibility, propagation range, and remaining disposal time. The graph neural network uses two layers of spectral domain convolution and sixty-four-dimensional hidden units, with a propagation step size of one to three steps. Evidence fusion employs the Dempster-Shafer method, calculating conflict factors and credibility decay based on data-driven scoring and mechanism consistency scoring. A digital twin simulation sandbox maintains a bidirectional mapping between electrical topology, equipment parameters, and operating points. Power flow calculations utilize the Newton-Raphson method with a convergence accuracy of at least 10 to the power of -5. Small disturbance stability analysis employs the eigenvalue method with a damping ratio threshold of at least 0.05. The equipment thermal model provides allowable temperature rise and upper bounds for transformers and energy storage. The time delay from receiving candidate strategies to providing voltage, thermal limits, and stability verification results is no more than thirty seconds.

[0031] The intelligent decision-making layer consists of federated and formation strategies, a multi-objective optimizer, and a secure projection module. The federated and formation strategies perform policy migration and multi-agent collaboration across multiple sites, employing secure aggregation and differential privacy-based gradient summary uploads, with an aggregation cycle of fifteen minutes. The multi-objective optimizer performs joint optimization with the goals of integrating risk value, energy loss, load shedding cost, and number of operations. The decision vector includes active and reactive power settings for energy storage, wind and solar power curtailment ratios, power flow reconfiguration switching actions, and segmented disconnection actions. Constraints include active and reactive power balance, line thermal limits, bus voltage upper and lower bounds, frequency deviation limits, dynamic state of charge of energy storage, and converter current limits. The solution employs a hybrid integer quadratic programming and heuristic neighborhood search collaborative approach, with a computation time not exceeding five seconds. The secure projection module maps candidate actions to the constrained feasible region using the L2 minimum distance principle, eliminating mutually exclusive or limit-crossing actions.

[0032] The execution and audit layer consists of an automatic power generation control interface, an automatic voltage control and station control execution interface, and blockchain auditing and situation visualization. The end-to-end latency of the automatic power generation control and automatic voltage control channels is no more than one second. Station control execution sets throttling operations for switch actions, with a single switch not exceeding one action within fifteen minutes. Blockchain auditing uses a consortium blockchain and Raft consensus, with a block interval of ten seconds. On-chain content includes the strategy number, version number, timestamp, and parameter summary, supporting playback and verification of the time range of strategy sequences and measured values. Situation visualization displays a panoramic view of the situation, alarm heatmaps, strategy playback, and execution effects, and is linked to audit information.

[0033] In the perception phase, the edge intelligent unit completes quality labeling and initial screening. When voltage fluctuations exceed 5% and last for more than 10 seconds, or DC-side ripple rates exceed 10% and last for more than 3 minutes, candidate events are generated and uploaded via the spatiotemporal data bus. In the fusion and inference phase, the risk engine performs graph neural network propagation and dynamic Bayesian inference on the 3D graph, calculates data-driven scores and mechanism consistency scores, and outputs fused risk values, credibility, propagation range, and remaining handling time through evidence fusion. Values ​​above 0.7 are considered high-risk, and values ​​between 0.4 and 0.7 are considered medium-risk. Key cut sets are also identified. In the simulation verification phase, the digital twin sandbox loads the current operating point and performs power flow, thermal limit, and stability verification on candidate strategies. Strategies that violate bus voltage bandwidth, line thermal limit, and converter current limit are eliminated, and a feasible region constraint description is output. In the strategy generation phase, federation and formation strategies provide prior knowledge. The multi-objective optimizer solves for the optimal action sequence of energy storage output and reactive power support, generation limitation ratio, power flow reconfiguration, and segmented disconnection within the feasible region. The safety projection module completes the feasible region mapping and forms the final instruction set. In the execution and closed-loop phase, automatic generation control, automatic voltage control, and station control execution interfaces issue instructions and collect execution receipts. The blockchain completes on-chain evidence storage, and the situation visualization displays the execution effect. Receipts and measurements are fed back to the risk engine, simulation sandbox, and federation strategy module via the spatiotemporal data bus for review and parameter updates.

[0034] The time synchronization benchmark is the IEEE 1588 Precision Time Protocol, with an intra-station synchronization error of no more than one millisecond. Under network outage conditions, the time benchmark is maintained by a real-time clock, with a drift rate of no more than one part per million. Sampling frequency, channel throughput, and end-to-end latency are executed according to the aforementioned hierarchical definitions, with a single-topic throughput of no less than 50,000 frames per second and an edge-to-cloud end-to-end data latency of no more than one second. The fusion risk value classification thresholds are two levels: 0.7 and 0.4. The bus voltage bandwidth is ±5% of the nominal value. The bearing temperature rise threshold after ten minutes is 15 degrees Celsius, and the energy storage operating state of charge range is 20 to 85 degrees Celsius. The simulation convergence accuracy is no less than 10 to the power of -5, the stability damping ratio threshold is no less than 0.05, and the strategy verification latency is no more than 30 seconds. Safety execution limits the number of times a single switch can operate within 15 minutes to no more than once. The upper limit of the energy storage rate is given by the thermal model and does not exceed the equipment's rated continuous rate.

[0035] This embodiment and Figure 1 Consistent with the accompanying drawings, this embodiment establishes a closed-loop response path from monitoring to execution through multi-source data fusion via edge and cloud collaboration, risk propagation and causal reasoning based on 3D graphs, credibility assessment of evidence fusion, pre-verification using digital twins, strategy generation for multi-objective optimization and secure projection, and transparent auditing via consortium blockchains. Compared to conventional solutions relying on static thresholds and offline contingency plans, this embodiment achieves substantial improvements in risk identification lead time, false alarm suppression, response feasibility, and execution traceability, fully supporting the technical features described in the claims and meeting the requirements of clarity and support in patent texts.

[0036] Example 3 This embodiment provides a real-time safety risk monitoring and emergency response method for a new energy power system, including: acquiring multi-source heterogeneous data, performing time alignment and quality labeling on the multi-source heterogeneous data, and conducting preliminary emergency handling for detected local anomalies, outputting spatiotemporal data; constructing a three-dimensional map integrating electrical topology, asset relationships, and geospatial information based on the spatiotemporal data, and conducting risk assessment based on the three-dimensional map, combined with time-series characteristics and electrical equipment mechanism models, outputting risk assessment results, which at least include risk level, credibility, propagation range, and handling time window; generating at least one candidate sequence of handling actions based on the risk assessment results and a preset multi-objective optimization model and constraints, and mapping the sequence of handling actions into an executable instruction set through safety projection; issuing control commands to the automatic generation control, automatic voltage control, and station control execution interfaces of the executable instruction set, and collecting feedback data and system measurement results after command execution.

[0037] Example 4 In another embodiment of the present invention, a computer-readable storage medium is provided as a storage component within a terminal device, the function of which is to store programs and data. It should be noted that the computer-readable storage medium here encompasses not only the built-in storage components of the terminal device but also extended storage components supported by the device. Essentially, it is a tangible medium capable of containing or storing programs that can be invoked by or in conjunction with an instruction execution system, device, or apparatus. This storage medium provides storage areas for the terminal's operating system and stores one or more instructions suitable for processor loading and execution, which can constitute one or more computer programs containing program code.

[0038] Specifically, examples of computer-readable storage media (a non-exclusive list) include: electrical connections with one or more wires, portable disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable optical disc read-only memory, optical storage devices, magnetic storage devices, or any reasonable combination of the above types.

[0039] The storage medium may also include data signals propagated as part of a baseband portion or a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any reasonable combination of both. Furthermore, computer-readable storage medium may also refer to other readable media besides conventional readable storage media, capable of sending, propagating, or transmitting programs for use or operation by an instruction execution system, apparatus, or device. Program code on the storage medium can be transmitted via any suitable medium, including but not limited to wireless, wired, optical fiber, or any reasonable combination thereof.

[0040] The program code used to implement the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C. The execution modes of the program code include: running entirely on the user's computing device, running partially on the user's device as a standalone software package, running partially in a distributed manner on both the user's device and a remote computing device, or running entirely on a remote computing device or server. When a remote computing device is involved, the device can be connected to the user's computing device via any type of network such as a local area network (LAN) or a wide area network (WAN), or connected to an external computing device via the Internet through an Internet service provider.

[0041] The processor is capable of loading and executing one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the real-time safety risk monitoring and emergency response method for the new energy power system described in Example 1.

[0042] Example 5 Figure 2This is a schematic diagram of a computer device provided according to an embodiment of the present invention.

[0043] Please see Figure 2 The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When executed by the processor 61, the computer program 63 implements the real-time safety risk monitoring and emergency response method for the new energy power system in this embodiment. To avoid repetition, these details are not elaborated here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the real-time safety risk monitoring and emergency response component computing system of the new energy power system in this embodiment. To avoid repetition, these details are not elaborated here.

[0044] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 2 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0045] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, CPUs, graphics processing units (GPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, quantum computing-based data processing logic units, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0046] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the computer device 60.

[0047] Furthermore, the memory 62 may include both internal storage units and external storage devices of the computer device 60. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0048] Any references to memory, databases, or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (Read-Only Memory). Memory includes ROM, magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0049] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

Claims

1. A real-time security risk monitoring and emergency response device for new energy power systems, characterized in that, The method comprises the following steps: An edge layer is configured to access multi-source heterogeneous data, time-align and quality-label the multi-source heterogeneous data, and preliminarily handle abnormal data in the quality label, and output spatio-temporal data; A cloud control layer is configured to receive the spatio-temporal data output by the edge layer, construct a three-dimensional graph integrating electrical topology, asset relationship and geographic spatial information based on the spatio-temporal data, and perform risk assessment based on the three-dimensional graph, the time sequence characteristics in the spatio-temporal data and the electrical equipment mechanism model, and output risk assessment results, which at least include risk degree, credibility, propagation range and disposal time window; An intelligent decision layer is configured to generate at least one candidate disposal action sequence for the risk assessment results based on a preset multi-objective optimization model and constraint condition according to the risk assessment results output by the cloud control layer, and map the disposal action sequence to an executable instruction set through security projection; An execution and audit layer is configured to receive the executable instruction set output by the intelligent decision layer, issue control instructions to an automatic generation control, an automatic voltage control and a station control execution interface, and collect feedback data and system measurement results after the instructions are executed. 2.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The edge layer is specifically configured as follows: Key electrical parameters of wind turbines and photovoltaic inverters in a new energy power system are sampled at a period of seconds or half seconds; State of charge and temperature data of energy storage battery management subsystems in the new energy power system are sampled at a period of half seconds; Phasor measurement units in the new energy power system are sampled at a sampling rate of 30-120 points per second; Infrared and conductor temperature measurement sensors in the new energy power system are sampled at a first set sampling rate; Acoustic vibration sensors in the new energy power system are sampled at a second set sampling rate; Data sampled by the wind turbines and photovoltaic inverters, the energy storage battery management subsystems, the phasor measurement units, the infrared and conductor temperature measurement sensors and the acoustic vibration sensors are preprocessed to form multi-source heterogeneous data; A timestamp based on a unified time reference and a quality label field representing data quality are added to the multi-source heterogeneous data frame. 3.The real-time security risk monitoring and emergency response device of new energy power system according to claim 2, characterized in that, The edge layer uses IEEE 1588 time protocol for station clock synchronization, and controls the time synchronization error to be not higher than 1 millisecond. 4.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The cloud control layer comprises a spatio-temporal data bearing component, which is configured to divide the spatio-temporal data into an original data module, a feature data module and a labeled data module according to different types of the spatio-temporal data, integrate electrical topology, asset relationship and geographic spatial information in the original data module, the feature data module and the labeled data module as dimensions of a three-dimensional graph, and construct the three-dimensional graph by taking devices, busbars and feeders in the original data module, the feature data module and the labeled data module as nodes and taking equivalent impedance, geographic distance and control coupling coefficient in the original data module, the feature data module and the labeled data module as edges. The feature data module comprises current fluctuation rate, frequency deviation, reactive power margin, direct current side ripple rate, temperature rise rate and phase angle difference change rate calculated based on the original data; and the labeled data module records historical or real-time labeled fault event categories and their start and end time. 5.The real-time security risk monitoring and emergency response device of new energy power system according to claim 4, characterized in that, The cloud control layer further comprises a risk assessment engine, which is configured to: based on the three-dimensional atlas, using a graph model to infer the propagation of risks within one to three steps in the space-time dimension, and obtaining a data-driven result; using a causal reasoning method to judge the correlation between fault precursor signals and potential events in the time series characteristics; using an evidence fusion method, the data-driven result and the mechanism consistency result calculated based on the mechanism model of the electrical equipment are synthesized to obtain a fused risk value; based on a preset threshold, the fused risk value is graded, and a risk assessment result is output. 6.The real-time security risk monitoring and emergency response device of new energy power system according to claim 5, characterized in that, The cloud control layer further comprises an online simulation verification component, which is configured to perform running constraint checking before the treatment action sequence generated by the intelligent decision-making layer is executed, including: The Newton-Raphson method is used to calculate the power flow, and the power flow distribution of the system is checked with a convergence accuracy of not less than 10 -5 . using the eigenvalue analysis method to perform small perturbation stability analysis, and taking a damping ratio not less than 0.05 as the stability threshold; based on the device thermal model, calculate and check whether the temperature rise and power ratio of the transformer and energy storage device when executing the candidate strategy exceed the upper limit of the allowed threshold. 7.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The multi-objective optimization model of the intelligent decision-making layer comprises: setting optimization goals, the optimization goals at least including minimizing system residual risk, minimizing power loss, minimizing load shedding cost, and minimizing operation times; decision variables at least include energy storage active / reactive power set value, wind and solar power generation limit generation ratio, switch action combination for power flow reconstruction, and system segmentation splitting action; constraint conditions at least include system active / reactive power balance constraint, line thermal stability limit constraint, bus voltage upper and lower limit constraint, system frequency deviation limit constraint, energy storage state of charge dynamic change constraint, and converter current limiting constraint; using a collaborative solving algorithm combining mixed integer quadratic programming and heuristic neighborhood search, at least one candidate treatment action sequence is obtained. 8.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The intelligent decision-making layer further comprises: a collaborative learning module for migrating and collaboratively training strategy knowledge among multiple new energy sites, using a secure aggregation technique combined with differential privacy to process uploaded multi-objective optimization model gradient summaries; a formation strategy module that works under a centralized training and distributed execution architecture, with a state space including energy storage state of charge, key bus voltage, system frequency, and global risk level, and an action space including energy storage charging and discharging power adjustment, reactive power support capability adjustment, and splitting instruction issuance, and a reward function designed to punish risk value increase, voltage deviation increase, and power curtailment increase. 9.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The execution and audit layer comprises: according to the executable instruction set, issuing continuous adjustment instructions through automatic generation control and automatic voltage control channels; according to the executable instruction set, issuing discrete operation instructions through station control, and implementing throttle control on the operation of the same switch to ensure that the number of actions of the same switch within any specified time window does not exceed the threshold; discrete operation instructions include switch opening and closing and segment splitting. 10.The real-time security risk monitoring and emergency response device of new energy power system according to claim 1, characterized in that, The device process comprises: acquiring multi-source heterogeneous data, time aligning and quality labeling the multi-source heterogeneous data, and performing preliminary emergency treatment on detected local abnormalities, and outputting space-time data; Based on the aforementioned spatiotemporal data, a three-dimensional map integrating electrical topology, asset relationships, and geospatial information is constructed. Based on the three-dimensional map, combined with temporal characteristics and electrical equipment mechanism models, a risk assessment is conducted, and the risk assessment results are output. The risk assessment results include at least the risk level, credibility, propagation range, and disposal time window. Based on the risk assessment results, and the preset multi-objective optimization model and constraints, at least one candidate sequence of actions for handling the risk assessment results is generated, and the sequence of actions is mapped into an executable instruction set through safety projection. The executable instruction set sends control commands to the automatic generation control, automatic voltage control and station control execution interfaces, and collects feedback data and system measurement results after the command execution.

Images