Power transmission line waterlogging disaster risk assessment method and system based on time event driving

Abstract

The invention discloses a power transmission line waterlogging disaster risk assessment method and system, and the method comprises the steps: generating time sequence flow data of environment, equipment state and weather prediction through the classification and fusion of multi-modal time sequence data, and setting an initial triggering threshold value according to the data; and through a depth time sequence feature extraction network, a depth feature relationship is extracted from the time sequence flow data, and a power transmission line waterlogging disaster hidden danger map is constructed. The map displays a risk path of waterlogging to the transmission tower through a node connection relationship. And the system adjusts a trigger threshold according to the depth characteristic relationship, preliminarily evaluates the waterlogging risk, calculates a risk value based on the node attribute value, divides the risk level, and achieves the evaluation of the waterlogging disaster risk of the power transmission line. According to the method, through multi-modal data fusion and depth feature extraction, the limitation of a single data source is overcome, multi-dimensional correlation analysis of the waterlogging risk of the power transmission line is realized, and through time event driving and map construction, the efficiency and accuracy of disaster early warning are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system transmission line safety protection, and particularly relates to a transmission line waterlogging disaster risk assessment method and system based on time event driving. Background Art

[0002] As a core component of the power system, transmission lines are responsible for efficiently transporting electricity from power plants to substations and load centers. They are crucial for optimizing the allocation of regional power resources and rationalizing the energy structure. Their safe and stable operation directly determines the safety and reliability of the entire power system.

[0003] Flooding poses numerous threats to the safe operation of power transmission lines. Waterlogging can cause short circuits or reduced insulation performance in substation equipment, leading to equipment burnout, line tripping, and even fires. Underground or low-lying substation buildings are particularly susceptible to flooding due to overloaded drainage systems, resulting in regional power outages. Furthermore, floodwaters and loose soil can undermine the stability of tower foundations, causing them to tilt, fall, or break transmission lines. Flooded objects carried by strong currents can also directly impact towers and lines, exacerbating physical damage. Furthermore, flooding significantly increases the difficulty of operation and maintenance. Waterlogging hinders repair personnel and equipment from entering the site, prolonging troubleshooting and repair time. Furthermore, damp cable joints, distribution boxes, and other facilities pose a risk of electrical leakage, threatening personnel safety.

[0004] Currently, the industry primarily relies on static weather forecasts and historical rainfall data to predict the risk of waterlogging along power transmission lines. This approach relies on single-source analysis, processing only rainfall or water level data independently and lacking dynamic correlation across multiple sources. Furthermore, manual inspections are used to assess the impact of waterlogging on transmission facilities, resulting in significant response delays. Summary of the Invention

[0005] In view of this, the present invention provides a transmission line waterlogging disaster risk assessment method and system based on time event driving, aiming to calculate the comprehensive risk value by multi-source data fusion, while reducing manual intervention, and realizing accurate prediction of waterlogging disaster risk.

[0006] In order to achieve the above object, the technical solution provided by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for assessing the risk of waterlogging disasters in power transmission lines based on time events, comprising the following steps:

[0008] Classify and fuse the multimodal time series data of the assessment area to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data, and generate time events based on the initial trigger thresholds of different types of data;

[0009] Based on the deep time series feature extraction network, the deep feature relationship of different types of data triggering time events is extracted from the environmental state time series stream data, equipment state time series stream data and weather forecast time series stream data;

[0010] Based on the time series data of environmental status, equipment status, and meteorological forecast, the attribute values ​​of different types of nodes in the graph are calculated, and a transmission line waterlogging disaster risk map is constructed based on different types of nodes. The transmission line waterlogging disaster risk map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes.

[0011] The initial trigger threshold of the time event is adjusted according to the deep feature relationship, and the risk of urban flooding disaster is preliminarily assessed based on the adjusted trigger threshold; based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby realizing the risk assessment of urban flooding disaster in the transmission line of the assessed area.

[0012] Furthermore, the multimodal time series data of the area to be evaluated is classified and fused, including:

[0013] Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data, and meteorological time series data;

[0014] For different types of runtime series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the Hampel filter is used to remove outliers to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data.

[0015] Furthermore, the deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer;

[0016] The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting deep feature relationships, as follows:

[0017]

[0018] Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, is the interval between current time and historical time;

[0019] The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

[0020] Furthermore, a map of potential waterlogging hazards along transmission lines was constructed, including:

[0021] Determine the nodes in the graph based on predefined entity node types and relationship node types; entity node types include rainfall events, water depth, drainage load, terrain elevation, and transmission towers; relationship nodes include cause, aggravate, and associate;

[0022] The connection relationship between each node in the graph is determined according to the predefined edge types; the edge types include heavy rainfall-causing edge, drainage-aggravating edge and terrain-associated edge.

[0023] Furthermore, the risk value is calculated using the attribute values ​​of each node in the risk path, including:

[0024] Calculate the value of each assessment factor based on the attribute value of each node in the risk path and determine the weight of each assessment factor; the assessment factors include cumulative rainfall, drainage load rate, terrain elevation and water depth;

[0025] Normalize the values ​​of each evaluation factor;

[0026] The risk value is obtained by multiplying the normalized values ​​of each evaluation factor by the corresponding weight and summing them up.

[0027] In a second aspect, the present invention provides a transmission line waterlogging disaster risk assessment system based on time event driving, comprising:

[0028] The time series data processing module is used to classify and fuse the multimodal time series data of the assessment area, obtain the environmental status time series stream data, equipment status time series stream data and weather forecast time series stream data, and generate time events according to the initial trigger thresholds of different types of data;

[0029] A feature extraction module is used to extract deep feature relationships of different types of data triggering time events from environmental state time series stream data, equipment state time series stream data, and weather forecast time series stream data based on a deep time series feature extraction network;

[0030] The graph construction module is used to calculate the attribute values ​​of different types of nodes in the graph nodes based on the environmental status time series flow data, equipment status time series flow data, and meteorological forecast time series flow data, and to construct a transmission line waterlogging disaster hazard map based on different types of nodes. The transmission line waterlogging disaster hazard map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes;

[0031] The risk assessment module is used to adjust the initial trigger threshold of the time event according to the deep feature relationship, and preliminarily assess the risk of urban flooding disasters based on the adjusted trigger threshold; based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby realizing the risk assessment of urban flooding disasters in the transmission line of the assessed area.

[0032] Furthermore, in the time series data processing module, the multimodal time series data of the evaluation area is classified and fused, including:

[0033] Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data, and meteorological time series data;

[0034] For different types of runtime series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the Hampel filter is used to remove outliers to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data.

[0035] Furthermore, in the feature extraction module, the deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer;

[0036] The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting deep feature relationships, as follows:

[0037]

[0038] Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, is the interval between current time and historical time;

[0039] The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

[0040] Furthermore, in the map construction module, a map of hidden dangers of waterlogging disasters in transmission lines is constructed, including:

[0041] Determine the nodes in the graph based on predefined entity node types and relationship node types; entity node types include rainfall events, water depth, drainage load, terrain elevation, and transmission towers; relationship nodes include cause, aggravate, and associate;

[0042] The connection relationship between each node in the graph is determined according to the predefined edge types; the edge types include heavy rainfall-causing edge, drainage-aggravating edge and terrain-associated edge.

[0043] Furthermore, in the risk assessment module, the risk value is calculated using the attribute values ​​of each node in the risk path, including:

[0044] Calculate the value of each assessment factor based on the attribute value of each node in the risk path and determine the weight of each assessment factor; the assessment factors include cumulative rainfall, drainage load rate, terrain elevation and water depth;

[0045] Normalize the values ​​of each evaluation factor;

[0046] The risk value is obtained by multiplying the normalized values ​​of each evaluation factor by the corresponding weight and summing them up.

[0047] In summary, the present invention provides a method and system for assessing the risk of waterlogging disasters in transmission lines driven by time events. By classifying and fusing the multimodal time series data of the area to be assessed, the environmental state time series stream data, the equipment state time series stream data and the weather forecast time series stream data are obtained, and time events are generated according to the initial trigger thresholds of different types of data; the deep feature relationship between the time events triggered by different types of data is extracted from the environmental state time series stream data, the equipment state time series stream data and the weather forecast time series stream data based on the deep time series feature extraction network; the deep feature relationship between the time events triggered by different types of data is extracted from the environmental state time series stream data, the equipment state time series stream data and the weather forecast time series stream data based on the deep time series feature extraction network; the deep feature relationship between the time events triggered by different types of data is extracted from the environmental state time series stream data, the equipment state time series stream data and the weather forecast time series stream data based on the deep time series feature extraction network. The attribute values ​​of different types of nodes in the sequence data calculation graph nodes are used to construct a transmission line waterlogging disaster hidden danger map based on different types of nodes; the transmission line waterlogging disaster hidden danger map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes; the initial trigger threshold of the time event is adjusted according to the deep feature relationship, and the risk of waterlogging disasters is preliminarily evaluated based on the adjusted trigger threshold; based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby realizing the transmission line waterlogging disaster risk assessment of the area to be assessed. The present invention breaks the limitation of a single data source through multimodal time series data fusion and deep feature extraction, and realizes multi-dimensional dynamic correlation analysis of transmission line waterlogging risks; at the same time, based on time event drive and graph construction, the risk level is assessed to solve the problem of slow response of manual inspections, and significantly improve the efficiency and accuracy of disaster warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 A flow chart of a method for assessing transmission line waterlogging risk based on time event driving provided by an embodiment of the present invention;

[0050] Figure 2 A diagram illustrating the implementation process of a time event-driven transmission line waterlogging disaster risk assessment and analysis method provided by an embodiment of the present invention;

[0051] Figure 3 A flowchart of multimodal time series data acquisition and fusion provided by an embodiment of the present invention;

[0052] Figure 4 A flowchart of the deep temporal feature extraction network and adaptive edge computing provided by an embodiment of the present invention;

[0053] Figure 5 A flowchart for constructing a waterlogging disaster risk map for power transmission lines according to an embodiment of the present invention;

[0054] Figure 6 A diagram of a time-driven risk assessment and real-time warning engine provided by an embodiment of the present invention;

[0055] Figure 7 A block diagram of a transmission line waterlogging risk assessment system based on time event driving provided by an embodiment of the present invention;

[0056] Figure 8 A block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purposes, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0058] See also Figure 1 The embodiment of the present invention provides a method for assessing the risk of waterlogging disasters in transmission lines based on time events, comprising the following steps:

[0059] S11: Classify and fuse the multimodal time series data of the area to be assessed to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data, and generate time events based on the initial trigger thresholds of different types of data.

[0060] It should be noted that multimodal time series data covers various types of data such as environmental status, equipment status, weather forecasts, etc., and is arranged in chronological order; the initial trigger threshold is pre-set and is used to determine whether the data reaches the numerical value that triggers a specific time event.

[0061] This step classifies and fuses multimodal time series data, integrating scattered heterogeneous data into logically related environmental, equipment, and meteorological time series stream data. Time events are generated based on the initial trigger threshold, converting continuously changing data into discrete, meaningful events to facilitate subsequent system analysis and processing, providing a structured data foundation for risk assessment.

[0062] S12: Based on the deep time series feature extraction network, deep feature relationships of different types of data triggering time events are extracted from environmental state time series stream data, equipment state time series stream data, and weather forecast time series stream data.

[0063] It should be noted that the deep time series feature extraction network is a network model based on deep learning that can automatically mine complex and deep feature relationships in time series data.

[0064] This step uses a deep time series feature extraction network to process three types of time series stream data: environment, equipment, and meteorology. Through the nonlinear transformation and feature learning of multi-layer neural networks, the deep feature relationships hidden when different types of data trigger time events are extracted from massive data, capturing the inherent complex connections between data and providing key feature basis for subsequent risk assessment.

[0065] S13: Calculate the attribute values ​​of different types of nodes in the graph nodes based on the environmental status time series flow data, the equipment status time series flow data, and the meteorological forecast time series flow data, and construct a transmission line waterlogging disaster hazard map based on different types of nodes; the transmission line waterlogging disaster hazard map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes.

[0066] It should be noted that the graph node is the basic unit in the transmission line waterlogging disaster hazard map, representing various influencing factors; the attribute value is a quantitative indicator that describes the characteristics and status of the graph node; the transmission line waterlogging disaster hazard map is a visual graph structure constructed with nodes and edges, which is used to show the risk path of waterlogging disasters affecting transmission towers.

[0067] This step calculates the attribute values ​​of the graph nodes based on three types of time-series stream data, quantitatively characterizes the status of each influencing factor, and then constructs a hidden danger map based on the correlation between different types of nodes, transforming complex multi-source data into an intuitive graph structure, clearly presenting the risk transmission path of urban flooding disasters to transmission towers.

[0068] S14: Adjust the initial trigger threshold of the time event according to the deep feature relationship, and preliminarily evaluate the risk of waterlogging disaster based on the adjusted trigger threshold; based on the preliminary risk assessment results, calculate the risk value using the attribute value of each node in the risk path, and divide the risk level according to the risk value, so as to realize the waterlogging disaster risk assessment of the transmission line in the area to be assessed.

[0069] It should be noted that the preliminary risk assessment is a preliminary judgment on the possibility of waterlogging disasters occurring in transmission lines based on the adjusted trigger threshold; the risk value is a quantitative indicator calculated by comprehensively considering the attribute values ​​of each node in the risk path to measure the degree of waterlogging disaster risk; the risk level classification is the process of dividing the waterlogging disaster risk into different severity levels based on the risk value.

[0070] This step adjusts the initial trigger threshold of the time event based on the extracted deep feature relationship to make the threshold more consistent with the actual data characteristics. Based on this, a preliminary risk assessment is performed, and the risk value is calculated based on the node attribute values ​​in the risk path of the hidden danger map. Finally, the risk level is divided according to the set standards to achieve a comprehensive and accurate assessment of the risk of waterlogging disasters on the transmission line.

[0071] This embodiment provides a time-event-driven transmission line waterlogging risk assessment method. This method breaks through the limitations of a single data source, integrates multimodal time series data, and comprehensively reflects information related to waterlogging disasters. It introduces a deep time series feature extraction network to automatically obtain deep feature relationships of data, improving analysis accuracy. It constructs a transmission line waterlogging disaster hazard map to intuitively present risk paths and facilitate decision-making. It dynamically adjusts the trigger threshold based on the deep feature relationship to achieve dynamic and precise risk assessment, effectively solving the problems of the existing technology lacking multi-source dynamic association and response delay.

[0072] See also Figure 2 , Figure 2 This paper demonstrates the implementation process of a time-event-driven transmission line waterlogging risk assessment and analysis method, including multimodal time series data acquisition and fusion, a deep temporal feature extraction network (DTFEN) and adaptive edge computing, the construction of a transmission line waterlogging hazard map, and a time-driven risk assessment and real-time warning engine. This process is described below in conjunction with some embodiments of the present invention.

[0073] See also Figure 3 , Figure 3 This paper presents a multimodal time series data collection and fusion process. This process first collects multimodal time series data, covering transmission line operation, environmental dynamics, meteorological conditions, and equipment status. The collected data is then classified into multiple sources, resulting in environmental time series streams, equipment status time series streams, and meteorological forecast time series streams. Finally, through time alignment strategies and outlier robustness processing, data fusion is achieved, providing a foundation for subsequent analysis.

[0074] based on Figure 3 The process shown, in one embodiment, classifies and fuses multimodal time series data of the area to be evaluated, including:

[0075] S21: Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data and meteorological time series data.

[0076] It should be noted that in specific implementation, four-dimensional time series data can be collected in real time through IoT devices, meteorological platforms, and power monitoring systems, covering transmission line operation, environmental dynamics, meteorological characteristics, and equipment status, including:

[0077] Transmission line operation sequence data, including equipment commissioning time, historical maintenance timestamps, fault event time series, etc.;

[0078] Environmental dynamic time series data, including rainfall time series (minute level), water level change rate (second level), and temperature and humidity time series records;

[0079] Meteorological time series data, including wind speed fluctuation time curve, rainfall intensity time distribution model, etc.;

[0080] Equipment status time series data, including the time trend of insulator leakage current and the time series characteristics of conductor galloping frequency.

[0081] For meteorological data, real-time access to meteorological bureau data is provided through an API, collecting parameters including rainfall intensity (mm / h, updated every 5 minutes) and cumulative rainfall duration. For drainage system data, the system integrates the status of drainage pump stations (operating / faulty / standby) and pipeline flow (e.g., a pipeline with an instantaneous flow of 8 m³ / s, a designed capacity of 10 m³ / s, and a load factor of 0.8). Alarms are triggered when the load factor exceeds 1.0. For terrain data, a 5-meter resolution Digital Elevation Model (DEM) can be used to analyze grid slopes (e.g., a grid with an 8° slope is marked as an "easily drainable area") and catchment areas (using ArcGIS (Arc Geographic Information System) to identify a basin as a "high-risk area for waterlogging").

[0082] S22: For different types of runtime time series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the Hampel filter is used to remove outliers to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data.

[0083] It should be noted that the collected data is divided into three types of time series flows:

[0084] Environmental time series stream data includes minute-level rainfall, second-level soil moisture change curves, and temperature fluctuation time series; equipment status time series stream data includes conductor tension change rate, insulator leakage current time series, tower vibration frequency, etc.; weather forecast time series stream data is connected to the Meteorological Bureau's forecast of rainfall intensity in the next two hours, as well as the wind speed time trend predicted by the LSTM model.

[0085] The time alignment operation can use the sliding window interpolation method to unify the time axes of data with different sampling frequencies.

[0086] In addition, the data is processed for outliers. Specifically, the Hampel filter is used to remove outliers. The formula is:

[0087]

[0088] Among them, x filtered is the data value processed by the Hampel filter, that is, the data after outliers are removed and corrected, midian(W) is the median of the data in the sliding window W, x is the original data value to be processed, W is the sliding window, and MAD (Median Absolute Deviation) is the median absolute deviation.

[0089] After classification, alignment and denoising, standardized environmental status time series stream data, equipment status time series stream data and weather forecast time series stream data are generated.

[0090] See also Figure 4 , Figure 4 This paper illustrates a deep temporal feature extraction network (DTFEN) and adaptive edge computing process. The network architecture utilizes an improved form, incorporating a multi-head temporal attention mechanism and a lightweight feature compression layer. Training and optimization are then performed, involving loss function design and edge device-adaptive training. Finally, a load-aware scheduling algorithm dynamically allocates computing resources, enabling efficient network operation and accurate extraction of deep feature relationships.

[0091] based on Figure 4 In the process shown, in one embodiment, the deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer;

[0092] The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting deep feature relationships, as follows:

[0093]

[0094] Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, The interval between the current time and the historical time.

[0095] It's important to note that the multi-head temporal attention mechanism adjusts the attention paid to data from different periods when extracting deep feature relationships using time-decay weights, increasing the focus on recent data to better capture long-term dependencies in the time series. Time-decay weights reduce the weight of older data in the attention calculation, prioritizing the focus on recent data. For example, when analyzing the risk of flooding along power transmission lines, recent data such as rainfall and water level changes are more critical to the current risk assessment. This mechanism can highlight the role of this recent data.

[0096] The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

[0097] It should be noted that the lightweight feature compression layer uses a 1D convolution kernel with a width of 5 to compress high-dimensional features. This significantly reduces the model's computational complexity without losing critical information, improving efficiency and making the model more adaptable to the limited computing resources of edge devices. Specifically, a 1D convolution kernel (with a width of 5, meaning each convolution operation considers data features from five consecutive time steps) is used to perform a convolution operation on the input high-dimensional time series feature data. This convolution operation extracts local feature correlations and maps high-dimensional features into a low-dimensional space. For example, for high-dimensional time series data on the status of power line equipment (including the time-varying changes in multiple parameters collected by multiple sensors), processing with this lightweight feature compression layer can reduce feature dimensionality and computational complexity by approximately 80%, effectively improving model speed and practicality on edge devices.

[0098] To ensure that the Deep Temporal Feature Extraction Network (DTFEN) can not only accurately predict the risk of waterlogging on transmission lines but also provide good interpretability of model features, a loss function that jointly optimizes prediction error and feature interpretability is designed as follows:

[0099]

[0100] Given the limited computing and storage resources of edge devices (such as monitoring terminals deployed near power transmission lines, which typically use the ARM architecture), knowledge distillation technology is employed to migrate the knowledge of large models trained on massive amounts of data in the cloud into a smaller model. Specifically, the complex prediction logic and rich knowledge learned from the large cloud model are distilled and transferred to a lightweight model (<10MB) suitable for the ARM architecture.

[0101] In actual application scenarios, multiple devices (such as power line monitoring sensor nodes at different locations) participate in data collection and model inference tasks. To rationally allocate computing resources, a load-aware scheduling algorithm is used to dynamically select computing nodes based on the remaining power (Eremaining) and data traffic (Drate) of the devices. The expression is as follows:

[0102]

[0103] Nodes with high priority perform model inference, and the others are only responsible for data collection.

[0104] See also Figure 5 , Figure 5 This article illustrates the process for constructing a transmission line flooding hazard map. The process begins by defining graph nodes, including entity nodes and relationship nodes. Next, graph edges are defined to clarify node connections. The graph is then stored and queried, with typical query logic configured. Finally, dynamic graph updates are implemented to ensure that the map reflects flooding hazard risk paths and other information in a timely manner as data changes.

[0105] based on Figure 5 The process shown, in one embodiment, constructs a transmission line waterlogging disaster risk map, including:

[0106] S31: Determine the nodes in the graph according to predefined entity node types and relationship node types; entity node types include rainfall events, water depth, drainage load, terrain elevation and transmission towers; relationship nodes include cause, aggravate and associate.

[0107] It should be noted that among the entity nodes, the rainfall event node is used to represent rainfall-related information. Its attributes include time, location, rainfall intensity, and duration. These attributes can be used to analyze the impact of rainfall on subsequent waterlogging risks; the water accumulation depth node records the location, current value, historical peak value and other information of the water accumulation. These data provide a basis for assessing the flooding risk faced by transmission facilities; the drainage load node contains drainage pipe ID, load rate, coverage and other attributes. Through these attributes, the operating status of the drainage system and its impact on the waterlogging risk of the transmission line can be understood; the terrain elevation node has grid coordinates, elevation value, slope and other attributes. This information can be used to determine whether the area is an easily drained area or a high-risk area for waterlogging; the transmission tower node records tower ID, location, foundation depth, voltage level and other information.

[0108] Among the relationship nodes, the CAUSES node is used to represent the causal relationship between heavy rainfall events and increased water depth, that is, heavy rainfall events will directly lead to increased water depth; the AGGRAVATES node is used to represent the relationship between drainage overload and the risk of flooding of power transmission facilities. When the drainage system is overloaded, the flooding risk faced by power transmission facilities will be aggravated. The RELATES node is used to describe the connection between low-lying terrain and historical waterlogging events, indicating that there is a correlation between low-lying terrain areas and historical waterlogging events. The definitions of the above nodes are shown in the following table:

[0109]

[0110] S32: Determine the connection relationship between each node in the graph according to predefined edge types; the edge types include heavy rainfall-causing edge, drainage-aggravating edge and terrain-associated edge.

[0111] It should be noted that the starting point of the heavy rainfall-causing edge is the rainfall event, and the end point is the depth of waterlogging. The weight calculation rule is weight = rainfall intensity × duration. The weight reflects the degree of influence of heavy rainfall on the increase in the depth of waterlogging. The larger the weight, the more significant the effect of heavy rainfall on the increase in the depth of waterlogging. The starting point of the drainage-aggravating edge is drainage overload (that is, the case where the drainage load rate is greater than a certain threshold), and the end point is the transmission flooding risk. The weight calculation rule is weight = drainage load rate × the inverse of the tower foundation depth. The weight reflects the degree to which drainage overload aggravates the flooding risk of transmission facilities. Terrain-associated edge: The starting point is a low-lying area (determined by terrain elevation and other data), and the end point is a historical waterlogging event. The weight calculation rule is weight = elevation value × historical waterlogging frequency. The weight indicates the close correlation between the low-lying terrain and the historical waterlogging event. The definitions of each edge are shown in the following table:

[0112]

[0113] Furthermore, the atlas uses a graph database (such as Neo4j) to store nodes and edges, using geographic grids as index units. Each grid contains meteorological, topographic, and facility data, facilitating complex temporal queries. For example, based on the query criteria "areas with rainfall > 50 mm and water depth > 0.5 meters," all heavy rainfall event nodes can be retrieved, matched with associated water depth nodes, and grid areas that meet the threshold conditions can be selected. A risk heat map is then generated based on the atlas, visually displaying the risk distribution using different colors (red for high risk, yellow for medium risk, and green for low risk). New rainfall events trigger an update to the atlas, automatically linking data such as the load rate of surrounding drainage facilities and tower locations to calculate flooding risk weights. Furthermore, the atlas optimizes the model based on flooding cases from the past five years. Using training data such as historical rainfall, drainage load, water depth, and tower failure records, the weight calculation formula for the "drainage load → flooding risk" edge is optimized to improve prediction accuracy.

[0114] See also Figure 6 , Figure 6 This paper presents a time-driven risk assessment and real-time early warning engine. It first constructs a dynamic risk assessment model. It then calculates the risk value by assigning assessment factors and weights, normalizing data, and calculating a comprehensive risk value. Finally, it classifies risk levels based on the risk score (RiskScore) and triggers corresponding response strategies, completing the waterlogging disaster risk assessment.

[0115] based on Figure 6 The process shown, in one embodiment, calculates the risk value using the attribute value of each node in the risk path, including:

[0116] S41: Calculate the value of each evaluation factor based on the attribute value of each node in the risk path, and determine the weight of each evaluation factor; the evaluation factors include cumulative rainfall, drainage load rate, terrain elevation and water depth.

[0117] It should be noted that accumulated rainfall (R) is calculated by counting the total rainfall over a unit time period (e.g., 24 hours) and is expressed in millimeters (mm). For example, if the assessed area receives 60 mm of accumulated rainfall over a 24-hour period, this value reflects the degree of rainfall accumulation and is a key factor in the development of waterlogging. The drainage load factor (D) can be calculated by comparing the current discharge capacity of the drainage system to its designed maximum discharge capacity (D = actual discharge capacity / designed discharge capacity). For example, if the designed discharge capacity of a drainage system is 10 cubic meters per second and the actual discharge capacity is 8 cubic meters per second, the drainage load factor D = 8 ÷ 10 = 0.8, reflecting the operating load of the drainage system. Terrain elevation (T) is obtained by taking the elevation (in meters) of the transmission line's location. For example, a transmission tower is located at an elevation of 5 meters. Generally, low-lying areas have relatively low elevation values, approaching 0, which influences the accumulation and drainage of accumulated water. Water depth (W) is obtained by monitoring the real-time water depth (in centimeters) at monitoring points around the transmission facility. For example, the real-time water depth around a certain transmission tower is 30 centimeters, which directly reflects the degree of water threat currently faced by the transmission facilities.

[0118] The weight coefficient of each factor is determined by the expert scoring method. For example, the weight distribution is as follows:

[0119]

[0120] S42: Normalize the values ​​of each evaluation factor.

[0121] It should be noted that the Min-Max normalization method is used to map each indicator to the interval [0,1]. For each evaluation factor X (X can be R, D, T, W), its normalization formula is:

[0122]

[0123] Among them, X i is the original index value, X min is the minimum value of the indicator in historical data, X max It is the maximum value of this indicator in historical data.

[0124] S43: Multiply the normalized values ​​of each assessment factor by the corresponding weight and sum them up to obtain the risk value.

[0125] It should be noted that before calculating the risk value, the trigger threshold for waterlogging events is dynamically adjusted based on time series data (such as historical rainfall, water depth, etc.) to avoid misjudgment or omission of fixed thresholds due to environmental changes. The formula is defined as follows:

[0126]

[0127] The time decay function attenuates the impact of historical events, ensuring that recent events have a higher weight:

[0128]

[0129] λ=0.05 is the attenuation coefficient.

[0130] The normalized values ​​of each evaluation factor are multiplied by the corresponding weights and then summed to obtain the risk score (RiskScore). The calculation formula is:

[0131]

[0132] in,

[0133]

[0134] After obtaining the risk value, the risk level is divided according to different risk value ranges and corresponding response strategies are adopted: when 0≤RiskScore<0.3, it is low risk, and inspection tasks are pushed to operation and maintenance personnel to check drainage facilities; when 0.3≤RiskScore<0.6, it is medium risk, and the backup drainage pump is started and the changes in accumulated water are monitored in real time, and water level reports are generated every hour; when 0.6≤RiskScore≤1, it is high risk, and the underground cable power supply is automatically isolated, a mobile drainage vehicle is dispatched to the site, and the emergency power supply guarantee plan is activated. The details are as follows:

[0135]

[0136] Based on the same inventive concept, the present application also provides a time-event-driven transmission line waterlogging risk assessment system for implementing the aforementioned time-event-driven transmission line waterlogging risk assessment method. The solution provided by this system is similar to the solution described in the aforementioned method. Therefore, the specific limitations of the following embodiments of the time-event-driven transmission line waterlogging risk assessment system can be found in the limitations of the time-event-driven transmission line waterlogging risk assessment method described above and will not be further elaborated here.

[0137] See also Figure 7 The embodiment of the present invention further provides a transmission line waterlogging disaster risk assessment system based on time event driving, comprising:

[0138] The time series data processing module is used to classify and fuse the multimodal time series data of the assessment area, obtain the environmental status time series stream data, equipment status time series stream data and weather forecast time series stream data, and generate time events according to the initial trigger thresholds of different types of data;

[0139] A feature extraction module is used to extract deep feature relationships of different types of data triggering time events from environmental state time series stream data, equipment state time series stream data, and weather forecast time series stream data based on a deep time series feature extraction network;

[0140] The graph construction module is used to calculate the attribute values ​​of different types of nodes in the graph nodes based on the environmental status time series flow data, equipment status time series flow data, and meteorological forecast time series flow data, and to construct a transmission line waterlogging disaster hazard map based on different types of nodes. The transmission line waterlogging disaster hazard map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes;

[0141] The risk assessment module is used to adjust the initial trigger threshold of the time event according to the deep feature relationship, and preliminarily assess the risk of urban flooding disasters based on the adjusted trigger threshold; based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby realizing the risk assessment of urban flooding disasters in the transmission line of the assessed area.

[0142] Furthermore, in the time series data processing module, the multimodal time series data of the evaluation area is classified and fused, including:

[0143] Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data, and meteorological time series data;

[0144] For different types of runtime series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the Hampel filter is used to remove outliers to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data.

[0145] Furthermore, in the feature extraction module, the deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer;

[0146] The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting deep feature relationships, as follows:

[0147]

[0148] Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, is the interval between current time and historical time;

[0149] The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

[0150] Furthermore, in the map construction module, a map of hidden dangers of waterlogging disasters in transmission lines is constructed, including:

[0151] Determine the nodes in the graph based on predefined entity node types and relationship node types; entity node types include rainfall events, water depth, drainage load, terrain elevation, and transmission towers; relationship nodes include cause, aggravate, and associate;

[0152] The connection relationship between each node in the graph is determined according to the predefined edge types; the edge types include heavy rainfall-causing edge, drainage-aggravating edge and terrain-associated edge.

[0153] Furthermore, in the risk assessment module, the risk value is calculated using the attribute values ​​of each node in the risk path, including:

[0154] Calculate the value of each assessment factor based on the attribute value of each node in the risk path and determine the weight of each assessment factor; the assessment factors include cumulative rainfall, drainage load rate, terrain elevation and water depth;

[0155] Normalize the values ​​of each evaluation factor;

[0156] The risk value is obtained by multiplying the normalized values ​​of each evaluation factor by the corresponding weight and summing them up.

[0157] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0158] Reference Figure 8 An embodiment of the present invention further provides a computer device, comprising: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, it implements the time event-driven transmission line waterlogging disaster risk assessment method as described in any one of the above methods.

[0159] The computer device may be a desktop computer, notebook computer, PDA, cloud server or other computing device. The computer device may include, but is not limited to, a processor and a memory. It will be understood by those skilled in the art that Figure 8 The computer device is merely an example and does not constitute a limitation on the computer device. The computer device may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, etc.

[0160] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0161] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped with the computer device. Furthermore, the memory may include both an internal storage unit of the computer device and an external storage device. The memory is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is about to be output.

[0162] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for assessing the risk of waterlogging disasters in a transmission line based on time event driving as described in any one of the above methods is implemented.

[0163] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a camera / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, removable hard drives, magnetic disks, or optical disks. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals or telecommunication signals.

[0164] An embodiment of the present invention provides a computer program product, including a computer program, which implements any one of the above methods when executed by a processor.

[0165] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0166] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0167] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely schematic. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0168] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for assessing the risk of waterlogging in transmission lines based on time event driving, characterized in that: The steps include: Classify and fuse the multimodal time series data of the assessment area to obtain environmental status time series stream data, equipment status time series stream data, and weather forecast time series stream data, and generate time events based on the initial trigger thresholds of different types of data; Extracting deep feature relationships between different types of data triggering the time events from the environmental state time series stream data, the device state time series stream data, and the weather forecast time series stream data based on a deep time series feature extraction network; Calculating attribute values ​​of different types of nodes in a graph node based on the environmental state time series flow data, the equipment state time series flow data, and the meteorological forecast time series flow data, and constructing a transmission line waterlogging disaster hazard map based on the different types of nodes; the transmission line waterlogging disaster hazard map is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes; Adjusting the initial trigger threshold of the time event according to the depth feature relationship, and preliminarily assessing the risk of waterlogging disaster based on the adjusted trigger threshold; Based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby achieving the transmission line waterlogging disaster risk assessment in the area to be assessed.

2. The method for assessing transmission line waterlogging disaster risk based on time event driving according to claim 1 is characterized in that: Classify and fuse the multimodal time series data of the assessment area, including: Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data, and meteorological time series data; For different types of runtime time series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the outliers are removed through the Hampel filter to obtain the environmental status time series flow data, the equipment status time series flow data and the weather forecast time series flow data.

3. The method for assessing transmission line waterlogging disaster risk based on time event driving according to claim 1, characterized in that: The deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer; The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting the deep feature relationship, as follows: Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, is the interval between current time and historical time; The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

4. The method for assessing transmission line waterlogging disaster risk based on time event driving according to claim 1, characterized in that: Construct a map of potential waterlogging hazards along transmission lines, including: Determine nodes in the graph based on predefined entity node types and relationship node types; the entity node types include rainfall events, water depth, drainage load, terrain elevation, and transmission towers; the relationship nodes include cause, exacerbate, and associate; The connection relationship between each node in the graph is determined according to predefined edge types; the edge types include heavy rainfall-causing edges, drainage-aggravating edges and terrain-associated edges.

5. The method for assessing transmission line waterlogging disaster risk based on time event driving according to claim 1, characterized in that: Calculating a risk value using the attribute value of each node in the risk path includes: Calculating the value of each evaluation factor based on the attribute value of each node in the risk path and determining the weight of each evaluation factor; the evaluation factors include cumulative rainfall, drainage load rate, terrain elevation and water depth; Normalizing the values ​​of each evaluation factor; The risk value is obtained by multiplying the normalized value of each assessment factor by the corresponding weight and then summing the results.

6. A transmission line flooding disaster risk assessment system based on time event drive, characterized in that: include: The time series data processing module is used to classify and fuse the multimodal time series data of the assessment area, obtain the environmental status time series stream data, equipment status time series stream data and weather forecast time series stream data, and generate time events according to the initial trigger thresholds of different types of data; A feature extraction module is used to extract deep feature relationships between different types of data triggering the time events from the environmental state time series stream data, the device state time series stream data, and the weather forecast time series stream data based on a deep time series feature extraction network; a graph construction module for calculating attribute values ​​of different types of nodes in a graph node based on the environmental state time series flow data, the equipment state time series flow data, and the meteorological forecast time series flow data, and constructing a transmission line waterlogging disaster hazard graph based on the different types of nodes; the transmission line waterlogging disaster hazard graph is used to represent the risk path of waterlogging disasters affecting transmission towers through the connection relationship between different nodes; a risk assessment module, configured to adjust the initial trigger threshold of the time event according to the depth feature relationship, and preliminarily assess the risk of waterlogging disaster based on the adjusted trigger threshold; Based on the preliminary risk assessment results, the risk value is calculated using the attribute value of each node in the risk path, and the risk level is divided according to the risk value, thereby achieving the transmission line waterlogging disaster risk assessment in the area to be assessed.

7. The time event driven transmission line waterlogging disaster risk assessment system according to claim 6, characterized in that: In the time series data processing module, the multimodal time series data of the area to be evaluated is classified and fused, including: Real-time collection of transmission line equipment operation time series data, environmental dynamic time series data, and meteorological time series data; For different types of runtime time series data, the sliding window interpolation method is used to align the data time axes with different sampling frequencies, and the outliers are removed through the Hampel filter to obtain the environmental status time series flow data, the equipment status time series flow data and the weather forecast time series flow data.

8. The time event driven transmission line waterlogging disaster risk assessment system according to claim 6, characterized in that: In the feature extraction module, the deep temporal feature extraction network adopts an improved Transformer architecture, which includes a multi-head temporal attention mechanism and a lightweight feature compression layer; The multi-head temporal attention mechanism is used to adjust the attention of data in different periods by time decay weight when extracting the deep feature relationship, as follows: Where, Represents the calculation process of the multi-head temporal attention mechanism, 、 and are query vector, key vector and value vector respectively, Indicates actual data information. represents the scaling factor, represents element-wise multiplication, is the time decay weight, is the attenuation coefficient, is the interval between current time and historical time; The lightweight feature compression layer is used to compress high-dimensional features using a 1D convolution kernel with a width of 5.

9. The time event driven transmission line waterlogging disaster risk assessment system according to claim 6, characterized in that: In the map construction module, a map of hidden dangers of waterlogging disasters in power transmission lines is constructed, including: Determine nodes in the graph based on predefined entity node types and relationship node types; the entity node types include rainfall events, water depth, drainage load, terrain elevation, and transmission towers; the relationship nodes include cause, exacerbate, and associate; The connection relationship between each node in the graph is determined according to predefined edge types; the edge types include heavy rainfall-causing edges, drainage-aggravating edges and terrain-associated edges.

10. The transmission line waterlogging disaster risk assessment system based on time event drive according to claim 6, characterized in that: In the risk assessment module, the risk value is calculated using the attribute value of each node in the risk path, including: Calculating the value of each evaluation factor based on the attribute value of each node in the risk path and determining the weight of each evaluation factor; the evaluation factors include cumulative rainfall, drainage load rate, terrain elevation and water depth; Normalizing the values ​​of each evaluation factor; The risk value is obtained by multiplying the normalized value of each assessment factor by the corresponding weight and then summing the results.

Images