Line early warning judgment method and system based on multi-source data fusion

Abstract

The invention is suitable for the technical field of power transmission line maintenance, and provides a line early warning judgment method and system based on multi-source data fusion, and the method comprises the steps: constructing a multi-dimensional knowledge graph, and building a semantic association network; collecting meteorological and equipment state data in real time, fusing the data into a knowledge graph space through a mapping function, identifying an abnormal meteorological-equipment combination mode by using fuzzy reasoning, and calculating a thickness change rate in combination with an icing growth prediction model; the icing thickness and the galloping probability are calculated, and a composite early warning level is generated; and dynamically generating a risk thermodynamic diagram on the GIS map, and visually displaying a risk propagation path. Through semantic fusion of multi-source data, composite risk modeling and spatialization decision making, the system breaks through single parameter criterion limitation of traditional early warning, realizes intelligent identification and dynamic early warning of meteorological-equipment coupling risks, and significantly improves accuracy of power transmission line disaster early warning and operation and maintenance decision making efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of transmission line maintenance, and in particular relates to a line early warning determination method and system based on multi-source data fusion. Background Art

[0002] As transmission networks grow in size and complexity, transmission lines face increasingly severe risks from meteorological disasters (such as icing and swaying) and abnormal equipment conditions. Existing technologies for early warning and assessment of transmission lines have evolved from single-parameter monitoring to multi-source data fusion, involving the real-time collection and preliminary analysis of meteorological data (temperature, humidity, wind speed) and equipment status data (conductor tension, insulator resistance).

[0003] However, the industry's integration of multi-source data still remains at the level of simple superposition at the data level, lacking in-depth exploration of semantic correlations and causal relationships between data. It is difficult to effectively identify complex risk patterns under the coupling of meteorological and equipment status, resulting in insufficient accuracy and reliability of early warning judgments.

[0004] Current, comparable technologies primarily store meteorological and equipment data in separate databases, employing single threshold criteria or simple weighted models for risk warnings. Some systems incorporate physical models to predict ice growth, but fail to construct semantic association networks between the data. For example, existing solutions may set ice thickness thresholds based on historical data and trigger warnings based on real-time humidity and temperature data. However, these solutions ignore the impact of equipment characteristics (such as conductor material and surface coating) on ​​the icing process. They also fail to structure the association of historical failure cases with real-time monitoring data. Consequently, the development of warning rules relies on manual experience and lacks data-driven intelligent reasoning capabilities. Summary of the Invention

[0005] The purpose of the present invention is to provide a line warning determination method based on multi-source data fusion, aiming to solve the technical problems existing in the prior art identified in the background technology.

[0006] The present invention is implemented as follows: a line warning determination method based on multi-source data fusion, the method comprising: Construct a multi-dimensional knowledge graph of transmission lines and establish an association network; Real-time meteorological monitoring data and equipment status data are collected and mapped into the knowledge graph space. Abnormal meteorological-equipment status combination patterns are identified, and the rate of change of conductor ice thickness is calculated using an ice growth prediction model. The pre-installed expert rule base in the knowledge graph is called upon to calculate ice thickness and ice dancing probability based on the output of the ice growth prediction model, and then validated to generate a composite warning level. A risk heat map is dynamically generated on the GIS map, and color gradient coding is used to display the risk propagation paths in different sections.

[0007] As a further solution of the present invention, the construction of a multi-dimensional knowledge graph of power transmission lines specifically includes: The seven-tuple model is used to define the knowledge graph architecture, which includes entity sets and relationship sets: The entity set contains: Meteorological entity set , device entity set , icing mode entity set , Fault Case Entity Set ; The relationship set contains: ; Build entity embedding vectors based on the knowledge representation model and perform relationship modeling by minimizing the energy function; Establish semantic association relationships between entities to form an entity network of weather-equipment-icing-fault.

[0008] As a further solution of the present invention, the real-time data is mapped to the knowledge graph space, abnormal weather-equipment status combination patterns are identified, and the rate of change of conductor ice thickness is calculated using an ice growth prediction model, which specifically includes: Establish a mapping function from real-time data to knowledge graph , where the input vector is: ; Respectively represent real-time temperature, real-time humidity, real-time wind speed, real-time equipment data, and wire data; The pre-processed real-time data is mapped to the constructed multi-dimensional knowledge graph of power transmission lines through a mapping function, and the real-time data is semantically aligned with the meteorological entity set and the equipment entity set. The triangular membership function is used to construct a fuzzy inference rule base to define the meteorological humidity anomaly pattern: Based on fuzzy inference algorithms, the mapped real-time data is evaluated to identify abnormal weather and equipment status combination patterns; An ice growth prediction model is established to calculate and analyze the rate of change of conductor ice thickness.

[0009] As a further solution of the present invention, the definition of the abnormal meteorological humidity mode is as follows: The membership function of is: ; in, Indicates humidity The membership function is used to measure the humidity The degree of humidity anomaly mode, The lower limit of normal humidity, is the critical humidity threshold for icing.

[0010] As a further solution of the present invention, the ice growth prediction model is established to calculate and analyze the change rate of the conductor ice thickness, specifically: ; in, Indicates the rate of change of the ice thickness on the conductor, is the current ice thickness, is the surface temperature of the conductor, To balance the humidity, is the humidity influence coefficient, is the wind speed suppression coefficient, is the natural shedding coefficient.

[0011] As a further solution of the present invention, the output results of the ice growth prediction model are used to calculate ice thickness and ice dancing probability and perform validation to generate a composite warning level, specifically including: Calculate the predicted ice thickness and dancing probability every hour in the future; Calling the expert rule base stored in the multi-dimensional knowledge graph of transmission lines to identify preset rules related to ice thickness and dancing probability; Comparing the predicted ice thickness and the predicted dancing probability with the preset rules defined in the expert rules to determine whether each data item meets the specific disaster triggering conditions; Logical synthesis and weighted calculation are used to generate composite warning levels.

[0012] As a further solution of the present invention, the calculation of the predicted ice thickness and dancing probability every hour in the future is specifically as follows: ; ; Where, for The predicted ice thickness at the time, To predict the dancing probability, is the initial dancing probability.

[0013] As a further solution of the present invention, the generation of composite warning levels: ; ; ; in, Indicates the composite warning level, is the risk magnification factor, represent the icing risk factor and the dancing risk factor respectively, represents the critical threshold of ice thickness, represents the critical threshold of dancing probability.

[0014] As a further solution of the present invention, the risk heat map is dynamically generated on the GIS map, and the risk propagation path of different sections is displayed using color gradient coding, which specifically includes: Map the composite warning level, ice oscillation probability, and ice thickness to the GIS system based on the geographic coordinates of the transmission lines to form spatial basic data; Normalize the mapped data according to the preset risk indicators and generate a dynamic risk heat map using color gradient coding; Analyze spatial data distribution, simulate and mark risk transmission paths.

[0015] Another object of the present invention is to provide a line warning determination system based on multi-source data fusion, the system comprising: The association building module is used to construct a multi-dimensional knowledge graph of transmission lines and establish an association network; The icing status analysis module is used to collect meteorological monitoring data and equipment status data in real time, map the real-time data into the knowledge graph space, identify abnormal meteorological-equipment status combination patterns, and use the icing growth prediction model to calculate the rate of change of conductor ice thickness; The warning level analysis module is used to call the expert rule base preset in the knowledge graph, calculate the ice thickness and dancing probability based on the output of the ice growth prediction model, and perform validation to generate a composite warning level; The transmission path analysis module is used to dynamically generate risk heat maps on GIS maps, using color gradient coding to display the risk transmission paths of different sections.

[0016] The beneficial effects of the present invention are: This solution constructs a multi-dimensional knowledge graph of transmission lines, structures and models entities such as meteorology, equipment, icing, and faults, and their relationships, realizes semantic fusion and causal reasoning of multi-source data, and solves the problems of data isolation and lack of association analysis in traditional solutions.

[0017] Real-time data mapping and fuzzy inference mechanisms accurately identify complex anomaly patterns in weather and equipment status. Combined with an ice growth prediction model, they dynamically calculate the rate of change in thickness, enhancing real-time quantification of icing processes in complex environments. Composite warning levels generated based on a knowledge graph and expert rule base effectively quantify the synergistic risk of ice thickness and ice dancing probability by integrating physical model predictions with domain knowledge, avoiding the one-sidedness of single threshold criteria.

[0018] Through GIS risk heat maps and propagation path simulation, abstract risks are transformed into visual spatial decision-making information, realizing the full chain intelligence of "data collection-risk identification-early warning judgment-visualization application".

[0019] This solution significantly improves the accuracy, real-time performance and spatial decision-making capabilities of transmission line disaster warnings, provides a scientific basis for efficient scheduling of operation and maintenance resources and disaster prevention, and effectively ensures the safe and stable operation of transmission lines. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flowchart of a line warning determination method based on multi-source data fusion provided by an embodiment of the present invention; Figure 2 A flowchart of constructing a multi-dimensional knowledge graph of power transmission lines and establishing an associated network provided by an embodiment of the present invention; Figure 3 A flow chart for calculating the rate of change of conductor ice thickness provided by an embodiment of the present invention; Figure 4 A flowchart of generating a composite warning level provided by an embodiment of the present invention; Figure 5 A flowchart of dynamically generating a risk heat map according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a line warning determination system based on multi-source data fusion provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0022] Figure 1 The flowchart of the line warning determination method based on multi-source data fusion provided by the embodiment of the present invention is as follows: Figure 1 As shown, the method includes: S100, constructs a multi-dimensional knowledge graph of transmission lines and establishes an association network; This step defines a knowledge graph architecture using a seven-tuple model, systematically integrating multiple key entities and complex relationships. This entity set includes not only meteorological entities (including real-time meteorological elements such as temperature, humidity, and wind speed) and equipment entities (covering basic equipment parameters such as conductor model, tower structure, and insulator status), but also icing pattern entities (such as rime icing and fog icing, along with their corresponding temperature, humidity, and wind speed conditions) and fault case entities (recording the time, location, and causative parameter combinations of historical icing trips and conductor breakage).

[0023] The relationship set uses semantic associations such as "weather-icing, causes", "equipment-icing, affects", and "icing-failure, triggers" to construct a complete causal chain, from meteorological conditions triggering icing formation, equipment status affecting icing development, and ultimately leading to failure.

[0024] When constructing entity embedding vectors based on the knowledge representation model, classic models such as TransE and ComplEx can be used to convert entities and relationships into low-dimensional dense vectors. By minimizing the energy function (such as the marginal-based ranking loss function), the relative positions of entities in the vector space are optimized to ensure that implicit associations such as "high humidity + low temperature" and "icing onset", "conductor diameter + surface roughness" and "icing growth rate" are accurately expressed in the vector space, forming a semantic association network of meteorology-equipment-icing-fault.

[0025] This step builds the underlying knowledge framework for multi-source data fusion, transforming fragmented meteorological data, equipment status, and historical failure cases into a structured knowledge network, providing a semantically aligned benchmark space for subsequent real-time data mapping.

[0026] For example, when the real-time collected humidity value is higher than the critical humidity threshold for icing, the knowledge graph can automatically associate it with the corresponding icing mode entity through the "weather-icing" relationship, retrieve historical fault cases under this mode, and predict possible equipment risks.

[0027] Compared with traditional independent database storage, the semantic association characteristics of knowledge graphs can explore potential causal relationships between data. For example, it discovered the implicit rule that "under the combined conditions of wind speed 6m / s, humidity 85%, and temperature -3℃, the ice thickness growth rate of a certain type of wire is 15% faster than the predicted value of the theoretical model." This ability to associate historical data with real-time monitoring through an entity network gives the system an intelligent foundation for knowledge reasoning and pattern recognition.

[0028] In addition, the multi-dimensional knowledge graph provides a structured carrier for the expert rule base, so that rules such as "a red alert is triggered when the ice thickness exceeds the critical value and the dancing probability is greater than 60%" can accurately anchor the quantitative relationship between entities, avoiding the ambiguity of traditional rule bases that rely on manual experience classification, improving the logical rigor and data-driven capabilities of warning judgments, and laying the foundation for cross-domain knowledge integration for subsequent abnormal pattern recognition, ice growth prediction, and composite warning level generation.

[0029] like Figure 2 As shown, the construction of a multi-dimensional knowledge graph of transmission lines specifically includes: S110 uses a seven-tuple model to define the knowledge graph architecture, which includes entity sets and relationship sets: The entity set contains: Meteorological entity set , device entity set , icing mode entity set , Fault Case Entity Set ; The relationship set contains: ; S120, constructs entity embedding vectors based on the knowledge representation model and performs relationship modeling by minimizing the energy function; S130: Establish semantic association relationships between entities to form an entity network of weather-equipment-icing-fault.

[0030] S200 collects meteorological monitoring data and equipment status data in real time, maps the real-time data into a knowledge graph space, identifies abnormal meteorological-equipment status combination patterns, and uses an ice growth prediction model to calculate the rate of change of conductor ice thickness. This step acquires multidimensional data in real time through a sensor network deployed along the transmission line (such as temperature sensors, humidity sensors, anemometers, conductor tension sensors, insulator leakage current monitoring devices, etc.). After preprocessing such as noise reduction filtering and spatiotemporal alignment, the input vector X containing real-time temperature, humidity, wind speed, equipment operating parameters (such as conductor sag and insulator resistance), and conductor physical properties (such as diameter and surface coating material) is semantically matched with the meteorological entity set (such as "low temperature" and "high humidity" entities) and equipment entity set (such as "certain type of conductor" and "composite insulator" entities) in the knowledge graph through a mapping function, realizing the conversion of real-time data from physical space to knowledge space.

[0031] In the abnormal pattern recognition link, the fuzzy inference rule base constructed based on the triangular membership function can not only quantitatively evaluate the abnormal degree of a single meteorological parameter (such as humidity), but also identify multi-parameter coupling anomalies through fuzzy logic operations (such as fuzzy "and" and "or" operations). For example, when "temperature ≤ 0℃" and "humidity ≥ 80%" and "wind speed ≤ 3m / s", it is judged as a "high-risk combination mode of rime and ice cover".

[0032] The ice growth prediction model dynamically calculates the rate of change of ice thickness by integrating real-time meteorological factors (humidity, wind speed, and conductor surface temperature) with the current ice status (initial thickness). For example, when the humidity is continuously higher than the equilibrium humidity, the model automatically increases the humidity influence coefficient to reflect the accelerating effect of supersaturated water vapor on ice growth.

[0033] This step builds a closed-loop processing mechanism of "data collection-semantic mapping-pattern recognition-dynamic prediction", realizing the deep integration and intelligent analysis of multi-source heterogeneous data.

[0034] For example, when the conductor sensor in a certain section detects that the humidity suddenly rises to 85% (higher than the critical humidity threshold for icing), the temperature drops to -2°C, and the insulator resistance value fluctuates abnormally, the system quickly locates the associated path in the knowledge graph through the mapping function: "The hydrophobicity of the insulator surface decreases in high humidity and low temperature environment → the probability of ice adhesion increases". Combined with fuzzy reasoning, it identifies the compound risk pattern of "meteorological anomaly + equipment status degradation", and simultaneously starts the ice growth prediction model to calculate the thickness change rate, providing real-time dynamic physical parameter support for subsequent early warnings.

[0035] Compared with traditional monitoring systems that only analyze a single meteorological parameter or equipment indicator in isolation, this step uses the semantic association capability of the knowledge graph to accurately capture the "meteorological-equipment" coupling anomalies. For example, it can identify the "hidden icing risk when the humidity does not reach the absolute threshold but the surface contamination of the equipment is high", avoiding the one-sidedness of single indicator warnings.

[0036] At the same time, the combination of fuzzy reasoning and physical models enables the system to handle data uncertainties (such as sensor measurement errors and gradual changes in environmental parameters), thereby improving the robustness of abnormal pattern recognition. The dynamic prediction of ice growth based on real-time data can more accurately reflect the immediate changing trend of ice cover in the current environment (such as the thickness surge rate during a sudden cold wave) compared to static models that rely on historical average data. It provides high-precision basic data for the dynamic generation of subsequent warning levels, effectively shortens the response time from data collection to risk identification, and enhances the system's real-time warning capabilities for sudden icing disasters.

[0037] like Figure 3 As shown, the real-time data is mapped to the knowledge graph space, abnormal weather-equipment status combination patterns are identified, and the ice growth prediction model is used to calculate the change rate of the conductor ice thickness, specifically including: S210, establish a mapping function from real-time data to knowledge graph , where the input vector is: ; Respectively represent real-time temperature, real-time humidity, real-time wind speed, real-time equipment data, and wire data; S220, mapping the preprocessed real-time data to the constructed multi-dimensional knowledge graph of the power transmission line through a mapping function, and semantically aligning the real-time data with the meteorological entity set and the equipment entity set; S230 uses the triangular membership function to build a fuzzy inference rule base to define the meteorological humidity anomaly pattern: S240, based on fuzzy inference algorithms, evaluates the mapped real-time data and identifies abnormal weather and equipment status combination patterns; S250: Establish an ice growth prediction model to calculate and analyze the rate of change of ice thickness on the conductor.

[0038] In this step, the definition of the abnormal meteorological humidity mode is as follows: The membership function of is: ; in, Indicates humidity The membership function is used to measure the humidity The degree of humidity anomaly mode, The lower limit of normal humidity, is the critical humidity threshold for icing.

[0039] when , , indicating humidity It is below the lower limit of normal and does not belong to the abnormal humidity mode; when , , indicating humidity Between the lower limit of normal and the critical humidity threshold of icing, the degree of humidity anomaly is The value of is related; when , , indicating humidity Reaching or exceeding the critical humidity threshold for icing is considered an abnormal humidity mode.

[0040] The ice growth prediction model is established to calculate and analyze the change rate of the conductor ice thickness, specifically: ; in, Indicates the rate of change of the ice thickness on the conductor, is the current ice thickness, is the surface temperature of the conductor, To balance the humidity, is the humidity influence coefficient, is the wind speed suppression coefficient, is the natural shedding coefficient.

[0041] S300: Calling the pre-set expert rule base in the knowledge graph, based on the output of the ice growth prediction model, calculates the ice thickness and dancing probability, performs validation, and generates a composite warning level; In this step, the thickness change rate output by the ice growth prediction model is substituted into the time recursion formula to dynamically calculate the predicted ice thickness values ​​for multiple time scales such as 1 hour, 3 hours, and 6 hours in the future.

[0042] At the same time, a dancing probability prediction model is constructed by combining the dynamic characteristics of the conductor with meteorological parameters (such as wind speed and ice eccentricity), forming a quantitative relationship of dancing risk that is strongly correlated with the ice thickness.

[0043] The expert rule base pre-stores composite criteria based on historical fault data and industry standards, such as "an orange warning is triggered when the predicted ice thickness exceeds 1.2 times the conductor design tolerance threshold and the dancing probability is greater than 50%" and "encrypted monitoring is started when the ice thickness change rate is greater than 1.5 mm / h for three consecutive hours and the humidity is greater than 90%." These rules are precisely associated with the entities of the knowledge graph and correspond to the specific nodes of "equipment entity-icing mode-fault case" (such as the critical ice thickness corresponding to a certain type of conductor and the specific dancing wind speed range).

[0044] During the verification phase, the system not only compares whether a single parameter exceeds the limit, but also performs logical synthesis through the rule engine. For example, when "the predicted ice thickness does not reach the critical value but the equipment status data shows an abnormal increase in the conductor tension", it triggers a special assessment of uneven ice distribution or local heavy ice to avoid underestimation of risks caused by the assumption of uniform ice cover.

[0045] This step realizes the intelligent decision-making closed loop of "data prediction-rule reasoning-multi-source verification", breaking through the limitations of traditional early warning systems that rely on a single threshold or simple weighting.

[0046] For example, when the predicted ice thickness of a conductor in a certain section is 18 mm (close to the design threshold of 20 mm), the initial dancing probability is 30%. However, the historical cases associated with the expert rule base show that when the ice cover of this type of conductor is rime (corresponding to the ice cover pattern entity in the knowledge graph), the dancing threshold will be reduced by 15% due to the decrease in the smoothness of the surface ice layer. The system automatically corrects the dancing probability to 45% through rule matching, and combines the composite warning formula to generate a level that is closer to the actual risk.

[0047] This mechanism, which deeply integrates physical model predictions with domain knowledge, can not only use real-time data to capture the dynamic evolution of current risks, but also make up for the shortcomings of model assumptions (such as not considering the impact of conductor surface contamination on ice adhesion) through expert experience, effectively reducing missed reports and false alarms.

[0048] In addition, the generation of composite warning levels is not a simple superposition, but reflects the characteristics of the coordinated disaster-causing factors of icing disasters through logical thresholds and weighted functions. For example, when the ice thickness exceeds the critical value, the amplification effect of the dancing risk will significantly increase the overall warning level through the risk amplification coefficient γ, providing more accurate risk ranking for operation and maintenance decisions (such as giving priority to the "high ice cover + high dancing risk" section rather than a single high ice cover section).

[0049] This step leverages the rule-structured storage and rapid retrieval capabilities of the knowledge graph, enabling the system to dynamically call exclusive rule sets based on the equipment characteristics of different transmission lines (such as voltage level and conductor type) to achieve differentiated early warnings. For example, a lower warning threshold for the probability of galloping is set for lines with large spans across valleys. This improves the adaptability and engineering practicality of the early warning system, and provides a scientific hierarchical decision-making basis for subsequent risk heat map generation and operation and maintenance resource scheduling.

[0050] like Figure 4 As shown, based on the output of the ice growth prediction model, the ice thickness and dancing probability are calculated and verified to generate a composite warning level, specifically including: S310, calculating the predicted ice thickness and dancing probability every hour in the future; S320, calling an expert rule base stored in a multi-dimensional knowledge graph of transmission lines to identify preset rules related to ice thickness and dancing probability; S330, comparing the predicted ice thickness and the predicted dancing probability with preset rules defined in the expert rules to determine whether each data item meets a specific disaster triggering condition; S340 uses logical synthesis and weighted calculation to generate a composite warning level.

[0051] In this step, the predicted ice thickness and dancing probability for each hour in the future are calculated as follows: ; ; Where, for The predicted ice thickness at the time, To predict the dancing probability, is the initial dancing probability.

[0052] Dancing probability and ice thickness are two core parameters for early warning judgment of transmission lines. They reflect the operating risks of the lines from different dimensions, and the two together form the basis for early warning decisions.

[0053] Ice thickness is a direct physical risk indicator. Its increase will lead to increased mechanical loads on conductors, potentially causing accidents such as line breakage and tower collapse. When it exceeds a critical value (such as 10 mm), an early warning is triggered. As shown in Table 1, the risk level is divided into 1-5 levels (safe to extremely severe risk), providing a basic basis for risk assessment.

[0054] The dancing probability reflects the risk of dynamic instability caused by self-excited vibration of ice-covered conductors under specific wind speed and direction conditions. As shown in Table 2, it is divided into five risk levels according to the probability value. When the threshold is exceeded, the corresponding level warning is triggered. The final warning level is determined together with the ice thickness and equipment status as a weight factor. In the expert rule, when the ice thickness and wind speed meet specific conditions, the formula is corrected to reflect the synergistic effect of multiple factors.

[0055] Table 1. Ice cover status analysis table

[0056] Table 2. Dancing probability analysis table

[0057] The composite warning level is generated: ; ; ; in, Indicates the composite warning level, is the risk magnification factor, represent the icing risk factor and the dancing risk factor respectively, represents the critical threshold of ice thickness, represents the critical threshold of dancing probability.

[0058] S400 dynamically generates risk heat maps on GIS maps, using color gradient coding to display risk propagation paths in different sections.

[0059] In this step, an integrated “location-risk parameter” dataset is formed by spatially correlating the geographical coordinates of the transmission lines (latitude and longitude, tower location, and span distribution) with attribute data such as the composite warning level, real-time ice thickness, and ice dancing probability.

[0060] First, the multi-dimensional risk data (e.g., ice thickness of 15 mm corresponds to a risk value of 0.7, ice dancing probability of 60% corresponds to a risk value of 0.8, and composite warning level red corresponds to a risk value of 1.0) are standardized according to preset rules (e.g., normalized to the interval [0,1]). The inverse distance weighted interpolation or kriging interpolation algorithm is used to spread the risk values ​​of discrete monitoring points to the entire line corridor to generate a continuous risk thermal layer.

[0061] The color gradient coding strategy usually adopts a progressive color band (such as from blue to red corresponding to low risk to extremely high risk respectively), and displays multiple risk parameters through transparency adjustment (such as a semi-transparent heat map overlaid on a line vector map), so that operation and maintenance personnel can intuitively identify the superimposed risk areas of "high ice thickness + high dancing risk".

[0062] In the risk transmission path analysis, the system uses GIS spatial analysis functions (such as network analysis and buffer zone analysis) and combines meteorological factors (such as prevailing wind direction and terrain slope) with equipment layout (such as tower spacing and conductor tension distribution) to simulate the diffusion path of icing disasters from the "initial high-risk section" to adjacent sections (for example, propagation along the line direction along the prevailing wind direction, or a chain reaction path of load transfer to adjacent towers due to local heavy ice), and marks potential transmission trajectories with arrow symbols or highlighted lines.

[0063] The significant advantage of this step is that it builds a spatial visualization decision-making platform for risk situations, transforming abstract warning data into intuitive and interactive geographic information products.

[0064] For example, when a transmission line in a mountainous area issues an icing warning due to a cold wave, the risk heat map can display in real time the red high-temperature area (high composite warning level) in the section above 800 meters above sea level, and through path simulation, it can prompt the risk to spread along the valley to the southeast tower group, guiding the operation and maintenance team to prioritize ice melting operations on the tension towers in this direction.

[0065] Compared to traditional tabular or list-based warnings, GIS heat maps leverage human visual sensitivity to spatial distribution to help decision-makers quickly locate risk clusters and key transmission nodes (such as large-span sections across rivers and power towers at wind outlets), avoiding delays in analysis and judgment caused by data fragmentation. Dynamic generation enables the system to respond to data updates in real time (e.g., refreshing the heat map every 5 minutes), adapting to the dynamic nature of icing disasters. Color gradient coding not only enables a hierarchical display of risk levels but also enhances the effectiveness of warning information through visual metaphors (e.g., red represents emergency), reducing the likelihood of human misjudgment.

[0066] The simulation of risk propagation paths breaks through the limitations of single-point early warning and can reveal the spatial correlation effects of disasters (for example, ice overload on a certain tower may trigger a chain tripping of adjacent towers), providing a global perspective combining "points, lines and surfaces" for preventive maintenance.

[0067] For example, during a freezing rain disaster, the system identified in advance the path of "ice covering the mountain spreading from the pole tower on the top to the foot of the mountain", helping the operation and maintenance department to deploy drone inspections and ice melting equipment before the disaster spread, reducing the accident rate by 40%.

[0068] This spatialized and visual risk presentation method not only improves the practicality of the early warning system, but also, through deep integration with GIS, provides a scientific decision-making basis based on geographic space for the intelligent operation and maintenance of transmission lines, achieving a key leap from "passive response" to "active prevention."

[0069] like Figure 5As shown, the risk heat map is dynamically generated on the GIS map, and the risk propagation paths of different sections are displayed using color gradient coding, specifically including: S410, mapping the composite warning level, ice oscillation probability, and ice thickness to the GIS system based on the geographic coordinates of the transmission lines to form spatial basic data; S420, normalizing the mapped data according to the preset risk indicators, and generating a dynamic risk heat map using color gradient coding; S430, analyze spatial data distribution, simulate and mark risk transmission paths.

[0070] Figure 6 The structural block diagram of the line warning judgment system based on multi-source data fusion provided by the embodiment of the present invention is as follows: Figure 6 As shown, the system includes: The association building module 100 is used to construct a multi-dimensional knowledge graph of power transmission lines and establish an association network; Icing status analysis module 200 is used to collect meteorological monitoring data and equipment status data in real time, map the real-time data into the knowledge graph space, identify abnormal meteorological-equipment status combination patterns, and calculate the rate of change of conductor ice thickness using an ice growth prediction model; The warning level analysis module 300 is used to call the expert rule base preset in the knowledge graph, calculate the ice thickness and dancing probability based on the output of the ice growth prediction model, perform validation, and generate a composite warning level; The transmission path analysis module 400 is used to dynamically generate a risk heat map on a GIS map, and use color gradient coding to display the risk transmission paths of different sections.

[0071] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0073] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A line warning determination method based on multi-source data fusion is characterized by: The method comprises: Construct a multi-dimensional knowledge graph of transmission lines and establish an association network; Real-time meteorological monitoring data and equipment status data are collected and mapped into the knowledge graph space. Abnormal meteorological-equipment status combination patterns are identified, and the rate of change of conductor ice thickness is calculated using an ice growth prediction model. The pre-installed expert rule base in the knowledge graph is called upon to calculate ice thickness and ice dancing probability based on the output of the ice growth prediction model, and then validated to generate a composite warning level. A risk heat map is dynamically generated on the GIS map, and color gradient coding is used to display the risk propagation paths in different sections.

2. The method according to claim 1, characterized in that The construction of a multi-dimensional knowledge graph of transmission lines specifically includes: The seven-tuple model is used to define the knowledge graph architecture, which includes entity sets and relationship sets: The entity set contains: Meteorological entity set , device entity set , icing mode entity set , Fault Case Entity Set ; The relationship set contains: ； Build entity embedding vectors based on the knowledge representation model and perform relationship modeling by minimizing the energy function; Establish semantic association relationships between entities to form an entity network of weather-equipment-icing-fault.

3. The method according to claim 1, characterized in that The real-time data is mapped to the knowledge graph space, abnormal weather-equipment status combination patterns are identified, and the rate of change of conductor ice thickness is calculated using an ice growth prediction model. Specifically, the following steps are involved: Establish a mapping function from real-time data to knowledge graph , where the input vector is: ； Respectively represent real-time temperature, real-time humidity, real-time wind speed, real-time equipment data, and wire data; The pre-processed real-time data is mapped to the constructed multi-dimensional knowledge graph of power transmission lines through a mapping function, and the real-time data is semantically aligned with the meteorological entity set and the equipment entity set. The triangular membership function is used to construct a fuzzy inference rule base to define the meteorological humidity anomaly pattern: Based on fuzzy inference algorithms, the mapped real-time data is evaluated to identify abnormal weather and equipment status combination patterns; An ice growth prediction model is established to calculate and analyze the rate of change of conductor ice thickness.

4. The method according to claim 3, characterized in that The definition of meteorological humidity anomaly mode, humidity The membership function of is: ； in, Indicates humidity The membership function is used to measure the humidity The degree of humidity anomaly mode, The lower limit of normal humidity, is the critical humidity threshold for icing.

5. The method according to claim 3, characterized in that The ice growth prediction model is established to calculate and analyze the change rate of the conductor ice thickness, specifically: ； in, Indicates the rate of change of the ice thickness on the conductor, is the current ice thickness, is the surface temperature of the conductor, To balance the humidity, is the humidity influence coefficient, is the wind speed suppression coefficient, is the natural shedding coefficient.

6. The method according to claim 3, characterized in that The output results of the ice growth prediction model are used to calculate ice thickness and ice dancing probability and perform validation to generate a composite warning level, specifically including: Calculate the predicted ice thickness and dancing probability every hour in the future; Calling the expert rule base stored in the multi-dimensional knowledge graph of transmission lines to identify preset rules related to ice thickness and dancing probability; Comparing the predicted ice thickness and the predicted dancing probability with the preset rules defined in the expert rules to determine whether each data item meets the specific disaster triggering conditions; Logical synthesis and weighted calculation are used to generate composite warning levels.

7. The method according to claim 6, characterized in that The calculation of the predicted ice thickness value and dancing probability every hour in the future is specifically as follows: ； ； Where, for The predicted ice thickness at the time, To predict the dancing probability, is the initial dancing probability.

8. The method according to claim 7, characterized in that The composite warning level is generated: ； ； ； in, Indicates the composite warning level, is the risk magnification factor, represent the icing risk factor and the dancing risk factor respectively, represents the critical threshold of ice thickness, represents the critical threshold of dancing probability.

9. The method according to claim 1, characterized in that The risk heat map is dynamically generated on the GIS map, and the risk propagation paths of different sections are displayed using color gradient coding, specifically including: Map the composite warning level, ice oscillation probability, and ice thickness to the GIS system based on the geographic coordinates of the transmission lines to form spatial basic data; Normalize the mapped data according to the preset risk indicators and generate a dynamic risk heat map using color gradient coding; Analyze spatial data distribution, simulate and mark risk transmission paths.

10. The line warning judgment system based on multi-source data fusion is characterized by: The system comprises: The association building module is used to construct a multi-dimensional knowledge graph of transmission lines and establish an association network; The icing status analysis module is used to collect meteorological monitoring data and equipment status data in real time, map the real-time data into the knowledge graph space, identify abnormal meteorological-equipment status combination patterns, and use the icing growth prediction model to calculate the rate of change of conductor ice thickness; The warning level analysis module is used to call the expert rule base preset in the knowledge graph, calculate the ice thickness and dancing probability based on the output of the ice growth prediction model, and perform validation to generate a composite warning level; The transmission path analysis module is used to dynamically generate risk heat maps on GIS maps, using color gradient coding to display the risk transmission paths of different sections.

Images