Typhoon disaster under power grid cascading failure early warning method and related equipment

Abstract

This application discloses a method and related equipment for early warning of cascading power grid failures under typhoon disasters, relating to the field of power system disaster early warning technology. The method includes: in response to identifying a typhoon disaster scenario, acquiring a dynamic spatiotemporal graph sequence of the power grid; identifying power grid equipment failures and predicting the failure probability of each power grid line based on the dynamic spatiotemporal graph sequence, and integrating them to obtain a set of failure events; performing power flow calculations on the power grid corresponding to the failure event set to calculate a candidate set of risky equipment; selecting the next failure equipment from the candidate set that meets preset simulation conditions and performing simulated failure deduction, iteratively updating the candidate set of risky equipment until a preset termination condition is met to obtain a cascading failure evolution path, and outputting cascading failure early warning information based on the cascading failure evolution path. This application solves the technical problem of insufficient real-time performance in predicting and warning of cascading power grid failures under typhoon disasters.

Description

Technical Field

[0001] This application relates to the field of power system disaster early warning technology, and in particular to a method and related equipment for early warning of power grid cascading failures under typhoon disasters. Background Technology

[0002] With the increasing frequency of extreme weather events, typhoons and their associated heavy rainfall pose a threat to the safe and stable operation of the power grid. Typhoon disasters may cause power grid equipment to trip and trigger cascading failures, resulting in widespread power outages. Therefore, it is necessary to predict and provide early warning of power grid tripping anomalies and cascading failures under typhoon disasters, providing emergency auxiliary decision-making support for power grid dispatching and operation personnel.

[0003] In related technologies, the prediction and early warning of power grid risks under typhoon disasters typically includes two approaches. One approach is based on typhoon wind field models and equipment failure probability models to calculate the failure probability of lines under typhoon disasters, and generate a set of anticipated failures accordingly. Then, power flow calculations are used to verify the risks of power flow exceeding limits and voltage exceeding limits, thereby conducting a static risk assessment. The other approach is based on training machine learning models using historical failure data to predict the failure probability of specific areas or equipment. The above-mentioned related technologies have the following shortcomings: the identification of tripping anomalies and the subsequent cascading failure simulation are disconnected, and the identification results are difficult to directly trigger the simulation; the failure simulation relies heavily on exhaustive verification of the anticipated failure set, which is inefficient in the dynamic environment of typhoons; furthermore, the failure simulation is difficult to effectively converge the failure propagation path, the simulation process is blind, and the scale of the failure simulation is prone to expansion, resulting in insufficient real-time performance of power grid cascading failure prediction and early warning under typhoon disasters. Summary of the Invention

[0004] The main purpose of this application is to provide a method and related equipment for early warning of cascading faults in power grids under typhoon disasters, aiming to solve the technical problem of insufficient real-time prediction and early warning of cascading faults in power grids under typhoon disasters.

[0005] To achieve the above objectives, this application proposes a method for early warning of cascading power grid failures under typhoon disasters, the method comprising: In response to the identification of typhoon disaster scenarios, a dynamic spatiotemporal map sequence of the power grid is obtained; Based on the dynamic spatiotemporal graph sequence, power grid equipment faults are identified and the fault probability of each power grid line is predicted, and the fault event set is obtained by integration. Perform power flow calculations on the power grid corresponding to the fault event set to obtain a candidate set of risky devices; Select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to the step of selecting the next faulty device that meets the preset simulation conditions, until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

[0006] In one embodiment, the step of obtaining the dynamic spatiotemporal graph sequence of the power grid includes: The system integrates grid operation data, typhoon disaster data, and micro-meteorological data, and performs time alignment and object mapping on the operation data, typhoon disaster data, and micro-meteorological data. The midpoint of the geographical location of the connection nodes at both ends of each power grid line is determined as the representative point of each power grid line. Based on the distance between the typhoon center and the representative point of each power grid line, the power grid lines within the typhoon's influence range are determined. For each power grid line within the typhoon's influence range, the wind speed and rainfall intensity of each meteorological observation station are inversely weighted according to the distance between each power grid line and each meteorological observation station to obtain the wind speed and rainfall intensity of each power grid line. The wind speed, rainfall intensity, and operational data of each power grid line are serialized in chronological order to obtain the dynamic spatiotemporal map sequence.

[0007] In one embodiment, the step of predicting the fault probability of each power grid line and integrating it to obtain a fault event set includes: The dynamic spatiotemporal graph sequence is input into a preset spatiotemporal graph neural network; Based on the topological connection relationship of each power grid line, the electrical quantity characteristics and wind and rain characteristics of each power grid line are spatially aggregated, and time modeling is performed based on the changes in characteristics at each moment within the sliding time window. Based on the characteristics of the nodes at both ends of each power grid line, the operating characteristics and wind and rain characteristics of each power grid line, the failure probability of each power grid line at each moment within the prediction time window is obtained. When the maximum failure probability of each power grid line within the predicted time window exceeds a preset threshold, the corresponding power grid line is identified as a high failure probability power grid line, and the failure events corresponding to the high failure probability power grid line are integrated into the failure event set.

[0008] In one embodiment, the step of performing power flow calculation on the power grid corresponding to the fault event set to obtain a candidate set of risky devices includes: By correlating active power with voltage phase angle and reactive power with voltage amplitude, the power grid corresponding to the fault event set is solved iteratively in an alternating manner to obtain the voltage amplitude and voltage phase angle of each bus. The current of each power grid line is calculated based on the voltage amplitude and voltage phase angle of the busbars at both ends of each power grid line. The load rate of each power grid line is obtained based on the ratio of the current of each power grid line to the thermal stability limit of each power grid line. The voltage exceedance limit of each busbar is obtained based on the deviation between the voltage amplitude of each busbar and the rated voltage. Grid lines with load rates exceeding a preset threshold and busbars with voltage exceeding a preset threshold are added to the candidate set of risky devices.

[0009] In one embodiment, the step of selecting the next faulty device that meets the preset simulation conditions includes: The perception state is determined by a preset reinforcement learning agent based on the load rate of each device in the candidate set of risk devices, the voltage limit exceedance of each bus, the typhoon intensity of the area where each device is located, and the current number of simulation steps. The reinforcement learning agent predicts the load loss ratio, average overload level, and voltage limit exceedance level of each device in the candidate set of risk devices as the next faulty device based on the perception state. Based on the load loss ratio, the average overload level, and the voltage limit exceedance level, assess the risk of system collapse caused by each device being the next failure device. From the candidate set of risky devices, the device with the highest risk is selected as the next faulty device.

[0010] In one embodiment, the step of outputting cascading failure early warning information based on the cascading failure evolution path includes: Based on the load rate and duration of each device in the cascading failure evolution path, determine the device-level risk indicators; Based on the probability of grid line reclosing failure, the probability of substation outage, and the risk of cascading failure propagation in the cascading failure evolution path, system-level risk indicators are determined. Based on the expected power shortage and average recovery time corresponding to the cascading failure evolution path, determine the global risk indicators; Based on the failure time sequence of each faulty device in the cascading failure evolution path, dynamic weights are determined for the device-level risk indicator, the system-level risk indicator, and the global risk indicator, respectively. A comprehensive risk value is obtained by weighting each risk indicator with the dynamic weights, and the cascading failure early warning information is generated based on the comprehensive risk value.

[0011] Furthermore, to achieve the above objectives, this application also proposes a power grid cascading failure early warning device under typhoon disasters, the power grid cascading failure early warning device under typhoon disasters comprising: The acquisition module is used to acquire the dynamic spatiotemporal sequence of the power grid in response to the identification of typhoon disaster scenarios; The integration module is used to identify power grid equipment faults and predict the fault probability of each power grid line based on the dynamic spatiotemporal graph sequence, and integrate them to obtain a fault event set. The calculation module is used to perform power flow calculation on the power grid corresponding to the fault event set and calculate the candidate set of risky equipment. The output module is used to select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to execute the step of selecting the next faulty device that meets the preset simulation conditions until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

[0012] In addition, to achieve the above objectives, this application also proposes a power grid cascading failure early warning device under typhoon disasters. The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the power grid cascading failure early warning method under typhoon disasters as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the power grid cascading fault early warning method under typhoon disasters as described above.

[0014] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the typhoon disaster cascading fault early warning method described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: Compared with the approach in related technologies that relies on exhaustive verification of a set of anticipated faults for fault inference, this application obtains a dynamic spatiotemporal graph sequence of the power grid in response to the identification of a typhoon disaster scenario; based on the dynamic spatiotemporal graph sequence, it identifies power grid equipment faults and predicts the fault probability of each power grid line, integrating them to obtain a fault event set; it performs power flow calculation on the power grid corresponding to the fault event set to calculate a candidate set of risky equipment; it selects the next faulty equipment that meets the preset simulation conditions from the candidate set of risky equipment, performs simulated fault inference on the next faulty equipment, updates the candidate set of risky equipment, and returns to the step of selecting the next faulty equipment that meets the preset simulation conditions until the preset termination condition is met, thereby obtaining a cascading fault evolution path, and outputs cascading fault early warning information based on the cascading fault evolution path. It is understandable that this application connects tripping anomaly identification with cascading fault simulation, and iteratively selects the next faulty device based on the intermediate power flow calculation results to converge the fault propagation path, so that the fault simulation is transformed from exhaustive verification to adaptive rolling simulation. Therefore, based on the above simulation, real-time or near-real-time cascading fault prediction and early warning can be achieved during typhoons, thereby solving the technical problem of insufficient real-time prediction and early warning of power grid cascading faults under typhoon disasters. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

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

[0018] Figure 1 This is a flowchart illustrating an embodiment of the power grid cascading fault early warning method under typhoon disasters provided in this application. Figure 2 This is a flowchart illustrating the method in an embodiment of this application; Figure 3 This is a basic framework diagram of STGNN prediction learning in the embodiments of this application; Figure 4 This is a schematic diagram of the module structure of the power grid cascading fault early warning device under typhoon disasters, as described in an embodiment of this application. Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the power grid cascading fault early warning method under typhoon disasters in the embodiments of this application.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution in this application's embodiments is: In response to the identification of typhoon disaster scenarios, a dynamic spatiotemporal map sequence of the power grid is obtained; Based on the dynamic spatiotemporal graph sequence, power grid equipment faults are identified and the fault probability of each power grid line is predicted, and the fault event set is obtained by integration. Perform power flow calculations on the power grid corresponding to the fault event set to obtain a candidate set of risky devices; Select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to the step of selecting the next faulty device that meets the preset simulation conditions, until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

[0023] In this embodiment, the application uses a power grid cascading fault early warning device under typhoon disaster as the main execution subject. For ease of description, it will be referred to as "device" for detailed description.

[0024] In related technologies, the identification of tripping anomalies and the subsequent cascading fault prediction are disconnected. Furthermore, fault prediction often relies on exhaustive verification of the expected fault set, resulting in low prediction efficiency. Moreover, fault prediction is difficult to effectively converge the fault propagation path, and the prediction process is somewhat blind, making it difficult to achieve real-time or near-real-time prediction and early warning of cascading faults.

[0025] This application provides a solution that enables power grid equipment fault identification to directly trigger cascading fault simulation, and adaptively selects the next faulty device based on intermediate power flow calculation results. Through iterative rolling simulation, the fault propagation path is converged, thereby achieving real-time or near-real-time cascading fault prediction and early warning during typhoons.

[0026] Based on this, this application provides a method for early warning of cascading power grid failures during typhoon disasters, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the power grid cascading failure early warning method under typhoon disasters in this application.

[0027] In this embodiment, the power grid cascading fault early warning method under typhoon disaster includes steps S10~S40: Step S10: In response to the identification of a typhoon disaster scenario, obtain the dynamic spatiotemporal map sequence of the power grid.

[0028] It should be noted that the typhoon disaster scenario refers to an operational scenario where the typhoon center enters or is about to enter the area where the power grid is located, and the power grid equipment faces the risk of tripping due to the impact of typhoon winds and rain. This scenario can be identified through elements such as wind speed and rainfall intensity in typhoon path forecast information, typhoon warning information, or meteorological observation data. The dynamic spatiotemporal graph sequence refers to a data sequence obtained by abstracting the power grid into a graph structure that changes over time, and then serializing the power grid operation characteristics and meteorological characteristics of consecutive moments in chronological order according to a sliding time window. In this sequence, the nodes and lines of the power grid correspond to the nodes and edges in the graph structure, respectively, and the node and line characteristics are dynamically updated over time.

[0029] Understandably, this step obtains the dynamic spatiotemporal graph sequence of the power grid under typhoon disasters, and organizes typhoon disaster data, micro-meteorological data and power grid operation data into a graph structure data that evolves over time. This provides a unified and high-quality input for subsequent power grid equipment fault identification, fault probability prediction and cascading fault inference, thus laying a data foundation for the connection between identification and inference.

[0030] Step S20: Based on the dynamic spatiotemporal graph sequence, identify power grid equipment faults and predict the fault probability of each power grid line, and integrate them to obtain a fault event set.

[0031] It should be noted that power grid equipment failures include tripping failures of power grid lines caused by typhoons and strong winds. Fault probability refers to the likelihood that each power grid line will experience a tripping failure within the predicted time window. The fault event set is a collection consisting of identified power grid equipment failure events and failure events corresponding to power grid lines predicted to have a high failure probability. This fault event set is used to trigger subsequent cascading failure simulations.

[0032] Understandably, this step identifies the fault probability of each power grid line while identifying the faults that have occurred in the power grid equipment, and integrates the two into a fault event set. This allows subsequent cascading fault simulations to be triggered by both the faults that have occurred in the power grid equipment and by the predicted results of high fault probabilities. This makes up for the shortcomings of the separation between power grid equipment fault identification and cascading fault simulation in related technologies, and improves the timeliness of triggering cascading fault simulations.

[0033] Step S30: Perform power flow calculation on the power grid corresponding to the fault event set to obtain a candidate set of risky devices.

[0034] It should be noted that power flow calculation refers to the process of solving for the voltage magnitude and phase angle of each bus and the power flow of each power grid line based on the power grid topology and injected power after the fault occurs, corresponding to the fault event set. The risk device candidate set refers to the set of power grid lines and buses whose load rate exceeds a preset threshold or whose voltage exceeds a preset threshold after power flow calculation. It is used to characterize devices in the power grid that have the risk of cascading faults after a fault occurs.

[0035] Understandably, this step, by performing power flow calculations on the power grid after a fault occurs, can accurately characterize the power flow transfer and voltage changes caused by the fault, and thereby screen out a candidate set of risky equipment with overload or over-limit risks, providing a range of candidate equipment for subsequent fault path deduction, thus supporting cascading fault deduction on the physical basis of accurate power flow calculation.

[0036] Step S40: Select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to execute the step of selecting the next faulty device that meets the preset simulation conditions until the preset termination condition is met, obtain the chain fault evolution path, and output the chain fault early warning information according to the chain fault evolution path.

[0037] It should be noted that the preset simulation conditions refer to the preset selection criteria used to select the next failing device from the candidate set of risky devices. Simulated failure deduction refers to the process of simulating the failure of the selected next failing device and re-performing power flow calculations to update the candidate set of risky devices. The cascading failure evolution path refers to the sequence of failing devices formed by the next failing devices selected in each iteration, used to characterize the propagation process of a cascading failure. The cascading failure early warning information refers to information generated based on the cascading failure evolution path to indicate the risk of cascading failures.

[0038] Understandably, this step selects the next faulty device from the candidate set of risky devices and performs simulated fault deduction. The candidate set of risky devices is continuously updated and the above selection is iteratively executed. This allows the fault deduction to converge the fault propagation path based on the intermediate power flow calculation results, avoiding the computational scale expansion caused by exhaustive verification of the expected fault set. Thus, when the preset termination conditions are met, the cascading fault evolution path is obtained and cascading fault early warning information is output. This achieves a closed loop from identification and deduction to early warning, improving the real-time performance of power grid cascading fault prediction and early warning under typhoon disasters.

[0039] This embodiment provides a method for early warning of cascading faults in the power grid under typhoon disasters. By acquiring a dynamic spatiotemporal graph sequence and identifying power grid equipment faults and predicting fault probabilities accordingly, performing power flow calculations to obtain a candidate set of risky equipment, iteratively selecting the next faulty equipment to simulate fault deduction, and outputting cascading fault early warning information, the method achieves real-time prediction and early warning of cascading faults in the power grid under typhoon disasters.

[0040] In one feasible implementation, the step of obtaining the dynamic spatiotemporal sequence of the power grid includes: The system integrates grid operation data, typhoon disaster data, and micro-meteorological data, and performs time alignment and object mapping on the operation data, typhoon disaster data, and micro-meteorological data. The midpoint of the geographical location of the connection nodes at both ends of each power grid line is determined as the representative point of each power grid line. Based on the distance between the typhoon center and the representative point of each power grid line, the power grid lines within the typhoon's influence range are determined. For each power grid line within the typhoon's influence range, the wind speed and rainfall intensity of each meteorological observation station are inversely weighted according to the distance between each power grid line and each meteorological observation station to obtain the wind speed and rainfall intensity of each power grid line. The wind speed, rainfall intensity, and operational data of each power grid line are serialized in chronological order to obtain the dynamic spatiotemporal map sequence.

[0041] It should be noted that the power grid operation data includes telemetry and teleindication data of buses and lines. Telemetry data includes at least active power, reactive power, voltage, and current, while teleindication data includes at least switch open / closed status. Typhoon disaster data includes typhoon paths and the radius of maximum wind speed. Micro-meteorological data includes wind speed and rainfall intensity observed by micro-meteorological stations. For each data source, key fields such as unique line identifiers, unique node identifiers, reference voltage, switch open / closed status, equipment path information, wind speed, and rainfall intensity are uniformly retained and standardized, and missing value processing, outlier removal, and timestamp standardization are completed. Object mapping refers to establishing the correspondence between unique node identifiers, unique line identifiers, and bus and line objects, while retaining the association between the two ends of the line and the unique node identifiers. Here, "line" is a general term, including line objects and line segment objects. When organizing by line segment, the line is refined into line segment objects. A segment-level unique identifier is set for each line segment, while retaining the unique identifier of its respective line and equipment path information as association fields, maintaining consistency with the mapping relationship with its connected nodes. Distance-inverse weighting refers to using the reciprocal of the distance between each power grid line and each meteorological observation station as the weight to weight the observed values ​​of each meteorological observation station.

[0042] For example, after determining the midpoint of the geographical locations of the connection nodes at both ends of each power grid line as the representative point of each power grid line, the distance between the typhoon center and the representative point of each power grid line can be used to determine whether each power grid line is within the typhoon's influence range. Indicatively, the typhoon's influence range can be determined using the following influence indicators: ; in, This represents the influence of line e at time t. A value of 1 indicates that line e is within the typhoon's influence range at time t, while a value of 0 indicates that line e is not affected by the typhoon at time t. This indicates the position of the typhoon center at time t; This represents the point on line e; This represents the radius of the typhoon's maximum wind speed at time t; This represents the geographic distance function.

[0043] Furthermore, for each power grid line within the typhoon's impact area, the wind speed and rainfall intensity at each meteorological observation station are estimated using a distance-inverse weighted average method, yielding the wind speed and rainfall intensity for each power grid line. Schematic, the wind speed of line o at time t can be expressed as: ; in, This represents the wind speed at time t along line o. Let represent the wind speed observed at time t by the j-th meteorological station; This represents the distance between the representative point of line o and the j-th meteorological observation station; Indicates the preset smoothing term; Indicates the preset power; This indicates the number of meteorological observation stations participating in the weighted estimate.

[0044] Furthermore, the distance-inverse weighted estimation method for rainfall intensity is consistent with that for wind speed, and the rainfall intensity at time t is calculated for line o. It can be represented as: ; in, This indicates the rainfall intensity at time t along line o; This represents the rainfall intensity observed at time t by the j-th meteorological station.

[0045] Furthermore, to reduce interference from non-typhoon-affected areas on the subsequent identification model, the wind speed and rainfall intensity on the line side are effectively processed, retaining wind and rain characteristics only when within the typhoon's influence range. The effectively processed wind speed and rainfall intensity can be expressed as follows: ; ; in, This represents the wind speed of line e at time t after the optimization process. This represents the rainfall intensity at time t for line e after the effective processing. This represents the influence of line e at time t.

[0046] Furthermore, the power grid can be abstracted as a time-varying graph structure, which can be represented as: ; in, A diagram showing the structure of a power grid as it changes over time; Represents the set of power grid nodes; Represents the set of power grid lines, any power grid line Connecting its two ends, the switching state of the power grid line determines its position at any given time. Whether it is put into operation; Indicates at time The adjacency matrix, which represents the topological connectivity of the power grid, is dynamically updated over time. At each moment, electrical quantities such as voltage and power of nodes are used as node features, and power flow, on / off status, associated wind speed and rainfall intensity of power lines are used as line features. The features of consecutive moments are serialized in chronological order according to a sliding time window to form a dynamic graph sequence.

[0047] It is understandable that this implementation method integrates typhoon disaster data, micro-meteorological data, and power grid operation data into a dynamic spatiotemporal map sequence that evolves over time by accessing multi-source data, time alignment, and object mapping, as well as by determining the typhoon impact range and inversely weighting the distance of wind and rain elements. This not only reduces the interference of non-typhoon-affected areas on subsequent identification models, but also provides a unified spatiotemporally aligned input for power grid equipment fault identification and fault inference.

[0048] In one feasible implementation, the step of predicting the fault probability of each power grid line and integrating it to obtain a fault event set includes: The dynamic spatiotemporal graph sequence is input into a preset spatiotemporal graph neural network; Based on the topological connection relationship of each power grid line, the electrical quantity characteristics and wind and rain characteristics of each power grid line are spatially aggregated, and time modeling is performed based on the changes in characteristics at each moment within the sliding time window. Based on the characteristics of the nodes at both ends of each power grid line, the operating characteristics and wind and rain characteristics of each power grid line, the failure probability of each power grid line at each moment within the prediction time window is obtained. When the maximum failure probability of each power grid line within the predicted time window exceeds a preset threshold, the corresponding power grid line is identified as a high failure probability power grid line, and the failure events corresponding to the high failure probability power grid line are integrated into the failure event set.

[0049] It should be noted that the identification of power grid equipment faults can be achieved based on the switch opening and closing status change information of the monitoring system. When the switch corresponding to the power grid line is detected to change from the closed position to the open position and fails to recover within a preset confirmation time, it is determined that the power grid line has actually tripped. The tripping time, the line's unique identifier, the equipment path information, and the switch's state before and after the change are recorded to form an actual tripping signal entry, which is then converted into an initial fault event set. Spatiotemporal graph neural networks refer to neural network models that can simultaneously model the spatial dependencies and temporal evolution patterns in graph-structured data. They improve fixed adjacency and fixed aggregation into a fusion topology of dynamic physical adjacency and learnable adaptive adjacency, and employ attention-weighted spatial aggregation that introduces edge features. Simultaneously, causal dilated temporal convolution replaces single recursive temporal modeling. For example, a spatiotemporal graph neural network may include a feature embedding layer, a spatial topology coding layer, a temporal dynamic coding layer, and a line-level decoding layer. The feature embedding layer provides a unified representation of the operational and disaster characteristics of nodes and lines. The spatial topology coding layer performs neighborhood aggregation of the electrical and disaster characteristics of nodes or lines based on the topological connectivity at time t. The temporal dynamic coding layer models the evolution of features over time within a sliding time window. The line-level decoding layer fuses the features of the nodes at both ends of the line, the line's own operational characteristics, and their associated wind and rain characteristics, mapping them to the fault probability of the line at each time within the prediction window. Spatial aggregation refers to performing neighborhood aggregation of the electrical and wind and rain characteristics of nodes or lines based on the topological connectivity of each power grid line to extract the spatial dependence inherent in the power grid topology. Temporal modeling refers to modeling the evolution of features at each time within the sliding time window over time.

[0050] For example, by inputting the dynamic spatiotemporal graph sequence into a preset spatiotemporal graph neural network, the fault probability of each power grid line at each moment within the prediction time window can be obtained. Schematic, the fault probability of line e at time t can be expressed as: ; in, This represents the probability of line e failing at time t; Represents a sequence of dynamic spatiotemporal graphs within a sliding time window of length L; This represents the mapping function learned by the spatiotemporal graph neural network; This represents the model parameters of the spatiotemporal graph neural network.

[0051] Furthermore, for power grid lines that meet the threshold conditions, a probability sequence is obtained within the future prediction window, and the expected occurrence time is determined accordingly. Using the maximum probability rule, the time with the highest fault probability within the prediction window is taken as the expected occurrence time. If the maximum value corresponds to multiple times, the earliest time is taken. Schematic, the expected occurrence time of line e can be expressed as: , Indicates the expected time of occurrence for line e; Indicates the traversal sequence number within the prediction window; Indicates the length of the prediction time window; Indicates time offset; Indicates that line e is at time The probability of failure.

[0052] Furthermore, a corresponding fault warning signal is generated for power grid lines that meet the threshold conditions. The fault warning signal includes at least a unique line identifier, equipment path information, fault probability, and expected occurrence time. Multiple fault warning signals generated within the same prediction window are arranged in order of expected occurrence time to form a fault warning list. When the maximum fault probability of any power grid line within the prediction window exceeds a preset threshold, subsequent cascading fault simulation is triggered, and the warning entries in the fault warning list are converted into initial fault event entries.

[0053] It is understood that this implementation method can improve the accuracy of fault probability prediction by simultaneously modeling the topological spatial dependence and characteristic temporal evolution of the power grid through a spatiotemporal graph neural network. When the maximum fault probability exceeds a preset threshold, the corresponding power grid line is integrated into the fault event set, thereby providing timely and accurate trigger input for cascading fault deduction.

[0054] In one feasible implementation, the step of performing power flow calculation on the power grid corresponding to the fault event set to obtain a candidate set of risky devices includes: By correlating active power with voltage phase angle and reactive power with voltage amplitude, the power grid corresponding to the fault event set is solved iteratively in an alternating manner to obtain the voltage amplitude and voltage phase angle of each bus. The current of each power grid line is calculated based on the voltage amplitude and voltage phase angle of the busbars at both ends of each power grid line. The load rate of each power grid line is obtained based on the ratio of the current of each power grid line to the thermal stability limit of each power grid line. The voltage exceedance limit of each busbar is obtained based on the deviation between the voltage amplitude of each busbar and the rated voltage. Grid lines with load rates exceeding a preset threshold and busbars with voltage exceeding a preset threshold are added to the candidate set of risky devices.

[0055] It should be noted that the power flow calculation employs a fast decoupled power flow algorithm. Based on the physical characteristics of power systems where active power is mainly related to voltage phase angle and reactive power is mainly related to voltage amplitude, this algorithm decouples and iteratively solves the relationship between active power and voltage phase angle, and reactive power and voltage amplitude. Before performing the power flow calculation, faulty equipment is removed from the grid connection model. Mathematically, this is represented by setting the elements of the corresponding faulty equipment branches in the grid admittance matrix to zero. Then, the power flow calculation is performed on the corrected grid. Load factor refers to the ratio of grid line current to the thermal stability limit of that grid line. Voltage exceedance refers to the degree of deviation of the bus voltage amplitude from the rated voltage.

[0056] For example, the correction equations for active power and voltage phase angle, and the correction equations for reactive power and voltage amplitude in power flow calculations can be expressed as follows: ; ; in, This represents the deviation vector of the active power injected into each bus. This represents the deviation vector of reactive power injected into each bus. Represents the bus voltage magnitude vector; This represents a constant matrix that reflects the sensitivity of active power flow to voltage phase angle; This represents a constant matrix reflecting the sensitivity of reactive power flow to voltage amplitude. This represents the phase angle correction vector for each bus voltage; This represents the vector of voltage magnitude correction for each bus. The two equations are solved iteratively, typically requiring 3-5 iterations to obtain high-precision values ​​for the voltage magnitude and phase angle of the entire network's buses.

[0057] Furthermore, for the power grid line connecting node i and node j, its current can be expressed as: ; in, This represents the current in the power grid line connecting node i and node j; , Let i and j represent the voltages at nodes i and j, respectively. , Voltage amplitude, , This refers to the voltage phase angle; This indicates the impedance of the power grid line.

[0058] The load factor of this power grid line can be expressed as: ; in, This indicates the load rate of the power grid line; This indicates the current in the power grid line; This indicates the thermal stability limit of the power grid line.

[0059] For the transformer connecting node i and node j, the apparent power flowing through it can be expressed as: ; in, This represents the apparent power of the transformer connecting node i and node j; Let represent the voltage at node i, where For voltage amplitude, This refers to the voltage phase angle; This indicates the conjugate of the transformer current.

[0060] The load factor of this transformer can be expressed as: ; in, This indicates the load rate of the transformer; This indicates the apparent power of the transformer; This indicates the rated capacity of the transformer.

[0061] The voltage over-limit of each bus can be expressed as: ; in, This indicates that the bus voltage has exceeded the limit; Indicates the voltage amplitude of the busbar; This indicates the rated voltage of the busbar.

[0062] It is understood that this implementation method accurately solves the voltage and power flow distribution of the power grid after a fault by decoupling iterative power flow calculation, and calculates the load rate and voltage limit exceedance accordingly. It can accurately screen out the candidate set of risky equipment with overload or limit exceedance risk, providing an accurate physical basis for subsequent fault path inference, thereby ensuring the accuracy of the inference results.

[0063] In one feasible implementation, the step of selecting the next faulty device that meets the preset simulation conditions includes: The perception state is determined by a preset reinforcement learning agent based on the load rate of each device in the candidate set of risk devices, the voltage limit exceedance of each bus, the typhoon intensity of the area where each device is located, and the current number of simulation steps. The reinforcement learning agent predicts the load loss ratio, average overload level, and voltage limit exceedance level of each device in the candidate set of risk devices as the next faulty device based on the perception state. Based on the load loss ratio, the average overload level, and the voltage limit exceedance level, assess the risk of system collapse caused by each device being the next failure device. From the candidate set of risky devices, the device with the highest risk is selected as the next faulty device.

[0064] It should be noted that reinforcement learning agents refer to pre-trained agents used to make selection decisions based on the perceived state. The perceived state is a state representation composed of the load rate of each device in the candidate set of risk devices, the degree of voltage exceedance on each bus, the typhoon intensity in the area where each device is located, and the current number of simulation steps. The load loss ratio refers to the proportion of load lost by the system after the simulated next failed device fails, relative to the total system load. The average overload level refers to the average overload level of the system after the simulated next failed device fails. The voltage exceedance level refers to the severity of voltage exceedance in the system after the simulated next failed device fails.

[0065] For example, the power flow calculation result is encoded as the perception state of the reinforcement learning agent, which can be represented as: ; in, This represents the perceptual state of a reinforcement learning agent; Indicates the preceding Normalized load rate of the most severely overloaded equipment; Indicates the degree to which the voltage of the critical busbar exceeds the limit; This indicates the quantitative values ​​of wind speed and rainfall intensity in the area where the equipment is located; This indicates the number of steps that have been taken in the current simulation.

[0066] For example, a reinforcement learning agent can be guided during the offline training phase using an immediate reward signal, which can be represented as: ; in, Indicates an immediate reward signal; This indicates the proportion of load loss caused by this simulation step; This indicates the average overload level of the system after this simulation step; This indicates the severity of the system's voltage exceeding the limit after this simulation step; , , This represents the weighting coefficient used to balance the importance of various risk factors.

[0067] Furthermore, the reinforcement learning agent is internally a deep neural network. Based on the experience learned during offline training, it evaluates and predicts the risk of system collapse after selecting each candidate device. The learning objective of the reinforcement learning agent is to maximize the cumulative reward of a series of actions performed in a deduction chain, thereby enabling it to actively identify and prioritize the failure path that will lead to the most serious consequences. After receiving the perception state, the reinforcement learning agent outputs an action, specifying which device to select from the risk device candidate set as the object for the next simulated failure.

[0068] It is understandable that this implementation method uses a reinforcement learning agent to predict the risks caused by each device as the next faulty device based on the perception state, and selects the device with the greatest risk as the next faulty device. This enables the fault simulation to prioritize exploring fault paths that lead to more serious consequences, thereby intelligently converging the fault propagation path and improving the pertinence of cascading fault simulation.

[0069] In one feasible implementation, when the simulation is triggered by a high-probability fault warning, since the warning is usually generated based on extreme typhoon weather, and the probability of permanent faults in power grid lines is significantly higher than that of transient faults in such scenarios, the simulation process for permanent faults is executed by default, skipping the initial fault nature judgment. When the simulation is triggered by an identified power grid equipment fault, the initial fault nature judgment logic for reclosing or alarm is executed. If reclosing is successful within 150-300 milliseconds after tripping, and the wind speed and rainfall intensity in the area where the equipment is located do not exceed the preset physical damage threshold, it is judged as a transient fault, the simulation is terminated and only recorded. If reclosing fails, the protection device issues a trip holding signal, or meteorological data indicates that there is a possibility of physical damage, it is judged as a permanent fault, and the simulation process is activated.

[0070] In one feasible implementation, multiple simulation threads are initiated, each independently performing the following loop: starting based on the initial fault or the result of the previous round; performing power flow calculation; the reinforcement learning agent selects the next faulty device based on the new perception state; simulating the fault of that device in the simulation; if the termination condition is met, the thread ends; otherwise, it returns to performing power flow calculation. Multiple simulation threads form multiple possible risk evolution paths.

[0071] In one feasible implementation, after all simulation threads have finished, the simulated fault chains are sorted according to their probability of occurrence and severity, and a critical risk scenario report is generated. For high-risk scenarios in the report, a pre-set handling plan is called from the strategy library, and the handling measure is input into the system state before the simulation chain occurs. The power flow calculation is re-executed, and by comparing the power flow results before and after the measure, it is quantitatively verified whether the measure can eliminate overload and block the fault chain, thus forming a decision-making closed loop.

[0072] In one feasible implementation, the step of outputting cascading failure early warning information based on the cascading failure evolution path includes: Based on the load rate and duration of each device in the cascading failure evolution path, determine the device-level risk indicators; Based on the probability of grid line reclosing failure, the probability of substation outage, and the risk of cascading failure propagation in the cascading failure evolution path, system-level risk indicators are determined. Based on the expected power shortage and average recovery time corresponding to the cascading failure evolution path, determine the global risk indicators; Based on the failure time sequence of each faulty device in the cascading failure evolution path, dynamic weights are determined for the device-level risk indicator, the system-level risk indicator, and the global risk indicator, respectively. A comprehensive risk value is obtained by weighting each risk indicator with the dynamic weights, and the cascading failure early warning information is generated based on the comprehensive risk value.

[0073] It should be noted that equipment-level risk indicators are used to characterize risks at the single equipment level, including equipment overload risk and voltage instability risk. Equipment overload risk is calculated based on the load rate and duration from the current or apparent power in the power flow results, while voltage instability risk is based on the deviation of the bus voltage from the limit. System-level risk indicators are used to characterize risks at the system level, including reclosing failure risk, substation outage risk, and cascading failure propagation risk. Reclosing failure risk assesses the probability of a permanent fault caused by automatic reclosing failure for high-failure-probability grid lines, taking into account the real-time wind and rain intensity and voltage level after power flow shift in the area. Substation outage risk analyzes the probability of the bus voltage dropping to 0 due to all incoming line trips or the complete shutdown of the main transformer in the power flow simulation results. Cascading failure propagation risk is measured by the product of the comprehensive probability of key risk scenarios and the final expected load loss. Global risk indicators are used to characterize risks at the entire network level, calculating the expected power shortage and average recovery time by comprehensively considering the probability and consequences of key risk scenarios. Dynamic weights refer to the weights of each risk indicator that are dynamically adjusted according to the failure time sequence of each faulty device in the cascading failure evolution path.

[0074] For example, the composite risk value can be expressed as: ; in, This represents the overall risk value; Indicates equipment-level risk indicators; This indicates a system-level risk indicator; This represents a global risk indicator; , , These represent the dynamic weights corresponding to equipment-level risk indicators, system-level risk indicators, and global risk indicators, respectively. Each dynamic weight is determined based on the failure time sequence of each faulty device in the cascading failure evolution path.

[0075] Furthermore, the calculated risk indicators at all levels, especially specific risks such as reclosing failure, heavy overload, and total shutdown, are combined with the geographic information system and mapped onto the actual geographical wiring diagram and single-line diagram of the power grid to generate a dynamic risk heat map. The dynamic risk heat map is refreshed at a frequency of minutes, and different levels of risk areas and equipment are marked with different colors and flashing. It also supports clicking to view the detailed risk value of any equipment and the key simulation scenario to which it belongs.

[0076] It is understood that this implementation method determines risk indicators at three levels: equipment level, system level, and global level. It also assigns dynamic weights to each risk indicator based on the fault time sequence to obtain a comprehensive risk value. This allows the cascading fault warning information to reflect the differentiated risk emphasis at different stages of fault evolution, thereby improving the pertinence and accuracy of cascading fault warnings.

[0077] For example, please refer to Figure 2 First, power grid operation data, typhoon disaster data, and micro-meteorological station data during typhoons are uniformly accessed, cleaned, and spatiotemporally aligned to form a joint power grid and typhoon dataset. This dataset is then used to construct a dynamic spatiotemporal map sequence as a unified input for power grid equipment fault identification and simulation calculations. Second, based on real-time switch position information from the monitoring system, actual tripping signals (a type of power grid equipment fault signal) are identified, forming an initial fault event set. The probability of line faults is predicted within a sliding time window, and when a threshold is exceeded, a fault warning signal is generated and summarized into a fault warning list. Third, simulations are triggered by identified power grid equipment faults or high-probability warnings to initialize scenarios. Subsequently, risky equipment is identified by combining power flow calculations and security boundary analysis, and adaptive fault path selection is performed using a DDQN intelligent agent. Finally, based on key risk scenarios and their accurate power flow results, three levels of risk indicators—equipment-level, system-level, and global resilience—are quantified. A dynamically updated risk heatmap is generated using a Geographic Information System (GIS) to achieve risk visualization and early warning output.

[0078] It is understood that this embodiment connects the identification of power grid equipment faults with the inference of cascading faults through the above process, and iteratively selects the next faulty device based on the intermediate power flow calculation results to converge the fault propagation path, so that the fault inference is transformed from exhaustive verification to adaptive rolling inference, thereby realizing real-time or near real-time cascading fault prediction and early warning during typhoons.

[0079] For example, please refer to Figure 3This embodiment provides a schematic diagram of the basic predictive learning framework of the Spatiotemporal Graph Neural Network (STGNN). Specifically, STGNN includes a data processing module, a spatiotemporal graph learning module, and a task-aware prediction module. Input data, after being formatted and feature-encoded by the data processing module, is input to the spatiotemporal graph learning module. The spatiotemporal graph learning module consists of a spatial learning network, a spatiotemporal fusion neural architecture, and a temporal learning network. The spatial learning network uses spectral domain graph convolutional networks (GCN), spatial domain GCN, and graph attention networks (GAT) to model the spatial dependencies of the power grid topology and node features. The spatiotemporal fusion neural architecture fuses spatial and temporal features. The temporal learning network uses recurrent neural networks (RNN), temporal convolutional networks (TCN), and time-series attention networks (TSAN) to model the temporal evolution of features within a sliding time window. The fused spatiotemporal features are then processed by the task-aware prediction module to output the final fault probability prediction result.

[0080] It is understood that this embodiment improves the accuracy of fault probability prediction by modeling the spatial dependence and characteristic temporal evolution of the power grid topology using spatial learning networks and temporal learning networks respectively, and then fuses them through a spatiotemporal fusion neural architecture, thus providing timely and accurate trigger input for subsequent cascading fault deduction.

[0081] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the power grid cascading failure early warning method under typhoon disasters. Any simple modifications based on this technical concept are within the protection scope of this application.

[0082] This application also provides a power grid cascading failure early warning device under typhoon disasters. Please refer to [reference needed]. Figure 4 The typhoon disaster power grid cascading fault early warning device includes: The acquisition module 10 is used to acquire the dynamic spatiotemporal map sequence of the power grid in response to the identification of a typhoon disaster scenario; The integration module 20 is used to identify power grid equipment faults and predict the fault probability of each power grid line based on the dynamic spatiotemporal diagram sequence, and integrate them to obtain a fault event set. Calculation module 30 is used to perform power flow calculation on the power grid corresponding to the fault event set and calculate the candidate set of risky equipment; Output module 40 is used to select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to execute the step of selecting the next faulty device that meets the preset simulation conditions until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

[0083] The typhoon-induced cascading fault early warning device for power grids provided in this application employs the typhoon-induced cascading fault early warning method described in the above embodiments, which can solve the technical problem of insufficient real-time performance in predicting and warning of cascading faults in power grids during typhoon disasters. Compared with the prior art, the beneficial effects of the typhoon-induced cascading fault early warning device for power grids provided in this application are the same as those of the typhoon-induced cascading fault early warning method provided in the above embodiments, and other technical features of the typhoon-induced cascading fault early warning device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0084] This application provides a power grid cascading failure early warning device under typhoon disasters. The power grid cascading failure early warning device under typhoon disasters includes: at least one processor; and a memory communicatively connected to at least one processor; wherein the memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor to enable at least one processor to execute the power grid cascading failure early warning method under typhoon disasters in the above embodiment 1.

[0085] The following is for reference. Figure 5 This document illustrates a structural schematic diagram of a power grid cascading fault early warning device suitable for implementing embodiments of this application. The power grid cascading fault early warning device in this application embodiment may include, but is not limited to, mobile terminals such as mobile phones, tablets, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital televisions and desktop computers. Figure 5 The power grid cascading failure early warning device shown under typhoon disasters is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0086] like Figure 5As shown, a typhoon disaster power grid cascading fault early warning device may include a processing unit 1001, which can perform various appropriate actions and processes according to a program stored in ROM 1002 or a program loaded from storage device 1003 into RAM 1004. RAM 1004 also stores various programs and data required for the operation of the typhoon disaster power grid cascading fault early warning device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. I / O interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touch screens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the power grid cascading failure early warning device under typhoon disasters to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows power grid cascading failure early warning devices under typhoon disasters with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0087] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0088] The typhoon-induced cascading fault early warning device for power grids provided in this application employs the typhoon-induced cascading fault early warning method described in the above embodiments, which can solve the technical problem of insufficient real-time performance in predicting and warning of cascading faults in power grids during typhoon disasters. Compared with the prior art, the beneficial effects of the typhoon-induced cascading fault early warning device for power grids provided in this application are the same as those of the typhoon-induced cascading fault early warning method provided in the above embodiments, and other technical features of the typhoon-induced cascading fault early warning device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0089] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0090] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0091] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the power grid cascading fault early warning method under typhoon disasters in the above embodiments.

[0092] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0093] The aforementioned computer-readable storage medium may be included in the typhoon disaster power grid cascading failure early warning device; or it may exist independently and not be assembled into the typhoon disaster power grid cascading failure early warning device.

[0094] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the power grid cascading fault early warning device under typhoon disaster, cause the power grid cascading fault early warning device under typhoon disaster to: in response to identifying the typhoon disaster scenario, acquire the dynamic spatiotemporal map sequence of the power grid; Based on the dynamic spatiotemporal graph sequence, power grid equipment faults are identified and the fault probability of each power grid line is predicted, and the fault event set is obtained by integration. Perform power flow calculations on the power grid corresponding to the fault event set to obtain a candidate set of risky devices; Select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to the step of selecting the next faulty device that meets the preset simulation conditions, until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

[0095] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0096] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0097] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0098] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described method for early warning of cascading power grid failures under typhoon disasters. This solves the technical problem of insufficient real-time performance in predicting and warning of cascading power grid failures under typhoon disasters. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the cascading power grid failure early warning method for typhoon disasters provided in the above embodiments, and will not be repeated here.

[0099] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for early warning of cascading power grid failures during typhoon disasters.

[0100] The computer program product provided in this application can solve the technical problem of insufficient real-time performance in predicting and warning of cascading power grid failures under typhoon disasters. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the cascading power grid failure warning method under typhoon disasters provided in the above embodiments, and will not be repeated here.

[0101] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A method for early warning of cascading power grid failures under typhoon disasters, characterized in that, include: In response to the identification of typhoon disaster scenarios, a dynamic spatiotemporal map sequence of the power grid is obtained; Based on the dynamic spatiotemporal graph sequence, power grid equipment faults are identified and the fault probability of each power grid line is predicted, and a fault event set is obtained by integration. Perform power flow calculations on the power grid corresponding to the fault event set to obtain a candidate set of risky devices; Select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to the step of selecting the next faulty device that meets the preset simulation conditions, until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

2. The method as described in claim 1, characterized in that, The steps for obtaining the dynamic spatiotemporal graph sequence of the power grid include: The system integrates grid operation data, typhoon disaster data, and micro-meteorological data, and performs time alignment and object mapping on the operation data, typhoon disaster data, and micro-meteorological data. The midpoint of the geographical location of the connection nodes at both ends of each power grid line is determined as the representative point of each power grid line. Based on the distance between the typhoon center and the representative point of each power grid line, the power grid lines within the typhoon's influence range are determined. For each power grid line within the typhoon's influence range, the wind speed and rainfall intensity of each meteorological observation station are inversely weighted according to the distance between each power grid line and each meteorological observation station to obtain the wind speed and rainfall intensity of each power grid line. The wind speed, rainfall intensity, and operational data of each power grid line are serialized in chronological order to obtain the dynamic spatiotemporal map sequence.

3. The method as described in claim 1, characterized in that, The step of predicting the fault probability of each power grid line and integrating it to obtain a fault event set includes: The dynamic spatiotemporal graph sequence is input into a preset spatiotemporal graph neural network; Based on the topological connection relationship of each power grid line, the electrical quantity characteristics and wind and rain characteristics of each power grid line are spatially aggregated, and time modeling is performed based on the changes in characteristics at each moment within the sliding time window. Based on the characteristics of the nodes at both ends of each power grid line, the operating characteristics and wind and rain characteristics of each power grid line, the failure probability of each power grid line at each moment within the prediction time window is obtained. When the maximum failure probability of each power grid line within the predicted time window exceeds a preset threshold, the corresponding power grid line is identified as a high failure probability power grid line, and the failure events corresponding to the high failure probability power grid line are integrated into the failure event set.

4. The method as described in claim 1, characterized in that, The step of performing power flow calculation on the power grid corresponding to the fault event set to obtain the candidate set of risky equipment includes: By correlating active power with voltage phase angle and reactive power with voltage amplitude, the power grid corresponding to the fault event set is solved iteratively in an alternating manner to obtain the voltage amplitude and voltage phase angle of each bus. The current of each power grid line is calculated based on the voltage amplitude and voltage phase angle of the busbars at both ends of each power grid line. The load rate of each power grid line is obtained based on the ratio of the current of each power grid line to the thermal stability limit of each power grid line. The voltage exceedance limit of each busbar is obtained based on the deviation between the voltage amplitude of each busbar and the rated voltage. Grid lines with load rates exceeding a preset threshold and busbars with voltage exceeding a preset threshold are added to the candidate set of risky devices.

5. The method as described in claim 1, characterized in that, The step of selecting the next faulty device that meets the preset simulation conditions includes: The perception state is determined by a preset reinforcement learning agent based on the load rate of each device in the candidate set of risk devices, the voltage limit exceedance of each bus, the typhoon intensity of the area where each device is located, and the current number of simulation steps. The reinforcement learning agent predicts the load loss ratio, average overload level, and voltage limit exceedance level of each device in the candidate set of risk devices as the next faulty device based on the perception state. Based on the load loss ratio, the average overload level, and the voltage limit exceedance level, assess the risk of system collapse caused by each device being the next failure device. From the candidate set of risky devices, the device with the highest risk is selected as the next faulty device.

6. The method as described in claim 1, characterized in that, The step of outputting cascading failure early warning information based on the cascading failure evolution path includes: Based on the load rate and duration of each device in the cascading failure evolution path, determine the device-level risk indicators; Based on the probability of grid line reclosing failure, the probability of substation outage, and the risk of cascading failure propagation in the cascading failure evolution path, system-level risk indicators are determined. Based on the expected power shortage and average recovery time corresponding to the cascading failure evolution path, global risk indicators are determined; Based on the failure time sequence of each faulty device in the cascading failure evolution path, dynamic weights are determined for the device-level risk indicator, the system-level risk indicator, and the global risk indicator, respectively. A comprehensive risk value is obtained by weighting each risk indicator with the dynamic weights, and the cascading failure early warning information is generated based on the comprehensive risk value.

7. A power grid cascading fault early warning device under typhoon disaster, characterized in that, The device includes: The acquisition module is used to acquire the dynamic spatiotemporal sequence of the power grid in response to the identification of typhoon disaster scenarios; The integration module is used to identify power grid equipment faults and predict the fault probability of each power grid line based on the dynamic spatiotemporal graph sequence, and integrate them to obtain a fault event set. The calculation module is used to perform power flow calculation on the power grid corresponding to the fault event set and calculate the candidate set of risky equipment. The output module is used to select the next faulty device that meets the preset simulation conditions from the risky device candidate set, perform simulated fault deduction on the next faulty device, update the risky device candidate set and return to execute the step of selecting the next faulty device that meets the preset simulation conditions until the preset termination condition is met, obtain the chain fault evolution path, and output chain fault early warning information according to the chain fault evolution path.

8. A power grid cascading fault early warning device under typhoon disaster, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the method for early warning of cascading power grid failures under typhoon disasters as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the power grid cascading fault early warning method under typhoon disaster as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the method for early warning of cascading power grid failures under typhoon disasters as described in any one of claims 1 to 6.

Images