Intelligent electric power perimeter early warning and diagnosis method and equipment based on microwave radar shooting control

Abstract

The invention discloses an intelligent electric power perimeter early warning and diagnosis method and equipment based on microwave radar shooting control, and the method comprises the steps: firstly collecting multi-dimensional data through a plurality of sensors, including microwave radar, video images, environment monitoring, electric power system operation, facility static information and the like; the microwave radar and video image data are preprocessed and subjected to space-time fusion, and target sensing information is obtained. Thirdly, constructing a three-dimensional digital twinborn model based on the static information of the facility, mapping the fused target sensing, environment monitoring and operation data into the model in real time, and generating a real-time twinborn situation; meanwhile, an electric power knowledge graph library is constructed, real-time twinning situations are associated, perimeter safety risks are recognized through semantic reasoning, grades are determined, and root causes are positioned through causal reasoning during faults. And based on the risk level and the real-time situation, dynamically evaluating the risk, generating adaptive early warning information, predicting the potential risk, giving a maintenance suggestion, and intelligently matching an emergency plan. The objective of the invention is to solve the defects of the prior art in the aspect of electric power perimeter safety early warning, realize the conversion from passive response to active prediction and intelligent decision, and improve the safety protection level and operation and maintenance efficiency of electric power facilities.

Description

Technical Field

[0001] This invention relates to the field of power system safety early warning and diagnosis technology, specifically to an intelligent power perimeter early warning and diagnosis method and device based on microwave radar monitoring. Background Technology

[0002] High-voltage power transmission and distribution facilities are critical infrastructure, and their safe and stable operation is of paramount importance. However, the perimeter of power facilities faces various security threats, including but not limited to human sabotage, illegal intrusion, natural disasters, and animal disturbances. These factors can all damage power facilities and thus affect the stable and secure supply of electricity.

[0003] Traditional methods for perimeter security of power facilities primarily rely on physical isolation measures, such as fences and warning signs, as well as simple infrared or video surveillance systems. However, these methods have many limitations in practical applications. Specifically, physical isolation measures are costly, occupy a large amount of space, and are difficult to fully cover in complex terrain conditions. They also lack remote evidence collection capabilities, making it impossible to effectively collect evidence and analyze accident liability for actions taken after a warning. While electronic warning signs can provide some warning, they typically lack network communication capabilities, and their infrared sensors have limited detection range and a high false alarm rate. They can only provide offline voice broadcasts and cannot achieve remote recording and evidence collection of illegal activities.

[0004] With the continuous development of artificial intelligence and sensor technology, AI-based perimeter early warning devices have gradually emerged, improving early warning capabilities to some extent. However, these technologies still have shortcomings. First, in high-voltage electromagnetic environments, noise and target signals highly overlap, and traditional noise reduction methods easily lead to the loss of effective signals, affecting the accuracy of target identification. Second, single sensors or simple multi-sensor fusion cannot guarantee robustness of identification under adverse weather or complex lighting conditions, resulting in decreased identification accuracy. Furthermore, existing early warning devices mainly focus on detection and response after an event occurs, with relatively fixed early warning modes. They lack in-depth correlation analysis of power system operating status, environmental changes, and potential risks, making it difficult to achieve proactive prediction and intelligent diagnosis. Finally, existing early warning systems typically operate as independent units, failing to deeply integrate with the operation, maintenance, fault diagnosis, and emergency management of the entire power system, resulting in information silos and hindering comprehensive decision-making. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an intelligent power perimeter early warning and diagnosis method and device based on microwave radar monitoring. Its purpose is to solve the shortcomings of the existing technology in power perimeter security early warning, realize the transformation from passive response to active prediction and intelligent decision-making, and improve the safety protection level and operation and maintenance efficiency of power facilities.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: According to a first aspect of the present invention, a method for intelligent power perimeter early warning and diagnosis based on microwave radar surveillance is provided, comprising: Multidimensional data is collected by various sensors deployed around the power grid perimeter. The multidimensional data includes at least microwave radar data and video image data for perimeter target perception, environmental monitoring data, and operational data and facility static information from the power information system. The microwave radar data and video image data are preprocessed and spatiotemporally fused to obtain fused target perception information. Based on the static information of the facilities, a three-dimensional digital twin model of the power facilities and their surrounding environment is constructed; the fused target perception information, environmental monitoring data and operation data are mapped into the three-dimensional digital twin model in real time to synchronize the physical world and the digital world, and generate a real-time twin potential that includes the dynamic behavior of the perimeter targets and the operating status of the equipment. A power knowledge graph database containing power equipment, environmental factors, behavior types, risk knowledge, and emergency plans is constructed; the real-time digital twin potential is associated with the power knowledge graph database, and perimeter security risks are identified and risk levels are determined through semantic reasoning; and when a fault occurs, causal reasoning is performed based on historical data stored in the three-dimensional digital twin model and the power knowledge graph database to locate the root cause of the fault. Based on the risk level and the real-time twin potential, the system dynamically assesses risks and generates adaptive early warning information; predicts potential risks of power facilities based on historical and real-time data and generates predictive maintenance recommendations; and intelligently matches emergency plans for high-risk events or faults based on the power knowledge graph; finally, it outputs the early warning information, maintenance recommendations, and emergency plans.

[0007] In one possible implementation of the first aspect, the preprocessing of the microwave radar data includes: The microwave radar data is denoised using an adaptive threshold denoising method based on wavelet transform. This method determines the threshold of each decomposition layer by estimating the noise standard deviation of the wavelet coefficients and uses a nonlinear function to process the wavelet coefficients to preserve the signal abrupt change characteristics.

[0008] In one possible implementation of the first aspect, in the adaptive threshold denoising method, the first... Threshold of decomposition layer Determined according to the following formula:

[0009] in, For the first Estimates of the noise standard deviation of the layer wavelet coefficients. For the first The number of layer wavelet coefficients.

[0010] In one possible implementation of the first aspect, the preprocessing of the video image data includes: An image enhancement algorithm based on illuminance-reflectance component separation is used to enhance the video image data. By estimating and removing the illuminance component in the image, the reflectance component of the target is restored to achieve image enhancement.

[0011] In one possible implementation of the first aspect, the spatiotemporal fusion includes: establishing a mapping relationship between the radar coordinate system and the video image coordinate system through calibration, so as to achieve spatiotemporal alignment between the radar-detected target and the target in the video image, and obtain fused target perception information.

[0012] In one possible implementation of the first aspect, the process of mapping the fused target perception information, environmental monitoring data, and operational data to the three-dimensional digital twin model in real time employs a Kalman filter algorithm for data fusion.

[0013] In one possible implementation of the first aspect, the semantic reasoning employs a rule-based reasoning method, wherein the rule is a conditional rule of the form IF-THEN. When the event information in the real-time twin potential satisfies the IF condition of the rule, the THEN conclusion is triggered to identify the risk.

[0014] In one possible implementation of the first aspect, the dynamic risk assessment includes: calculating a comprehensive risk score using a multi-factor weighted model, wherein the calculation formula for the multi-factor weighted model is:

[0015] in, For comprehensive risk scoring, The basic risk score is derived from the knowledge graph. Risk score derived from weather conditions; Risk score introduced by line load; Risk score introduced by the health status of the equipment; The risk score is determined by the time factor. These are the weighting coefficients corresponding to the risk scores.

[0016] According to a second aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the aforementioned intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance.

[0017] According to a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance.

[0018] According to a fourth aspect of the present invention, a computer program product is provided, which, when executed by a processor, implements the aforementioned intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance.

[0019] Compared with the prior art, the present invention has at least the following beneficial effects: This invention provides an intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring. By fusing multi-source sensor data from microwave radar and video images, combined with spatiotemporal synchronous processing, it effectively overcomes noise interference in high-voltage electromagnetic environments, avoids the loss of effective signals, and improves the accuracy of target recognition. Simultaneously, under adverse weather or complex lighting conditions, the collaborative operation of multiple sensors can complement perception blind spots, enhance recognition stability, and reduce false alarm rates. Based on a three-dimensional digital twin model, it maps the physical world state in real time and integrates environmental monitoring data, operational data, and facility static information to generate a dynamic twin potential. Combined with a power knowledge graph for semantic reasoning, it can deeply correlate equipment operation, environmental factors, and behavioral types, enabling early identification and level assessment of perimeter security risks, and achieving risk prediction and proactive intervention. In the event of a fault, historical data stored in the three-dimensional digital twin model and the power knowledge graph are used for causal reasoning to quickly locate the root cause of the fault, improve fault handling efficiency, and reduce power outage time. Adaptive early warning information is dynamically generated based on risk level and real-time twin dynamics. Potential risks are predicted based on historical and real-time data, generating predictive maintenance recommendations. Simultaneously, emergency plans are intelligently matched for high-risk events, providing targeted early warnings and achieving intelligent operation and maintenance, reducing the cost of manual intervention. By deeply integrating perimeter early warning with the power information system (including operation and maintenance, fault diagnosis, and emergency management), data sharing and business linkage are achieved, avoiding information fragmentation caused by traditional independent systems and improving overall operation and maintenance efficiency. Utilizing multi-sensor data fusion and digital twin technology, remote real-time monitoring and historical backtracking of perimeter target behavior are achieved, enabling data retention and evidence analysis of illegal activities, compensating for the shortcomings of traditional physical isolation or electronic warning signs in evidence collection. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the specific embodiments of the present invention, the drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart of an intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring according to the present invention; Figure 2 This is a block diagram of an intelligent power perimeter early warning and diagnosis system based on microwave radar monitoring according to the present invention; Figure 3 This is a schematic diagram of the multidimensional data acquisition and preprocessing module of the present invention; Figure 4 This is a schematic diagram of the components of the digital twin modeling module for power facilities in this invention; Figure 5 This is a schematic diagram of the power graph library construction and reasoning module of the present invention; Figure 6 This is a schematic diagram of the intelligent early warning and decision-making module of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Combination Figure 1 and Figure 2 As shown in the figure, the present invention provides an intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring. The overall architecture of the system implemented by this method is as follows: Figure 2 As shown, it mainly includes: a multi-dimensional data acquisition and preprocessing module, a power facility digital twin modeling module, a power map library construction and reasoning module, and an intelligent early warning and decision-making module. These modules work together to achieve proactive prediction, intelligent early warning, and fault diagnosis of high-voltage transmission and distribution perimeter risks. The early warning and diagnosis methods specifically include the following steps: S1. Collect multi-dimensional data by deploying various sensors around the power grid perimeter. The multi-dimensional data includes at least microwave radar data and video image data for perimeter target perception, environmental monitoring data, and operation data and facility static information from the power information system. Preprocess the microwave radar data and video image data and perform spatiotemporal fusion to obtain fused target perception information.

[0024] like Figure 3 As shown, in specific implementation, the multi-dimensional data acquisition and preprocessing module includes a microwave data acquisition and noise reduction unit, a video image data acquisition and enhancement unit, a multi-sensor spatiotemporal calibration unit, an environmental and operational data acquisition unit, and a geographic information and equipment information integration unit, as detailed below: The microwave data acquisition and noise reduction unit employs a millimeter-wave radar module to monitor target motion information within the perimeter area in real time. To address strong noise interference in the high-voltage electromagnetic environment, an adaptive threshold noise reduction method based on wavelet transform is used to denoise the microwave radar data. Specifically, the original radar signal is first subjected to adaptive wavelet decomposition. By analyzing the signal's spectral characteristics, the optimal wavelet basis function (e.g., db4 or sym8) and decomposition level (typically 4-6 levels) are determined. The wavelet decomposition formula is:

[0025] in, These are wavelet coefficients. For observing signals, It is a family of wavelet functions.

[0026] During the noise reduction process, a modified Stein unbiased risk estimation (SURE) thresholding criterion is used for hierarchical thresholding. j Threshold of decomposition layer Determined according to the following formula:

[0027] in, For the first j The estimated noise standard deviation of the layer wavelet coefficients is calculated using the median method:

[0028] For the first j The number of layer wavelet coefficients. The threshold function uses either soft or hard thresholding and introduces a nonlinear compression function. f( , ) This preserves the signal's abrupt change characteristics. Finally, the denoised signal is reconstructed using inverse wavelet transform. To address the pulse noise unique to high-voltage electromagnetic environments, this unit also incorporates robust statistical methods (such as median filtering and Kalman filtering) to further suppress sudden, strong interference.

[0029] The video image data acquisition and enhancement unit employs a 2-megapixel infrared camera module to capture video images of the perimeter area. To address visual recognition challenges under adverse weather conditions such as nighttime, rain, and fog, an image enhancement algorithm based on illuminance-reflectance component separation is used to enhance the video image data. The core algorithm is Multi-Scale Retinex (MSR), whose theory is based on image... It can be decomposed into reflection components. and illuminance component The product of:

[0030] The MSR algorithm estimates the illuminance component by convolving Gaussian kernels of different scales, thereby recovering the reflectance component. The specific steps are as follows: First, regarding the image's... Single-scale Retinex calculation is performed on each color channel:

[0031] in, The scale is Gaussian kernel function, This indicates that the original image is convolved using a Gaussian kernel function.

[0032] The final output of MSR is a weighted sum of reflection components at multiple scales:

[0033] in, These are weighting coefficients. This refers to the number of scales. Furthermore, this unit utilizes the YOLOv5 model combined with optical flow to extract the spatiotemporal features of the target from the video stream in real time, including the target's position, velocity, orientation, and behavioral posture.

[0034] The multi-sensor spatiotemporal calibration unit addresses the spatiotemporal discrepancy between radar and video by establishing a spatiotemporal correlation through a calibration target. The calibration target integrates three infrared reflection points (radar calibration reference) and three ARUCO visual feature points (video calibration reference), with the spatial coordinates of the reflection points and feature points known. .

[0035] Spatiotemporal deviation detection includes: Time deviation, compared with radar point cloud timestamps With video frame timestamps Calculate synchronization error Corrected to using a cross-correlation algorithm .

[0036] Spatial deviation, based on the principle of perspective transformation, establishes radar polar coordinates. With video pixel coordinates Mapping relationship:

[0037] in, Given a 3×4 perspective transformation matrix, solve using 3 sets of known coordinate points.

[0038] Dynamic calibration and trajectory completion are achieved through the following methods: Time calibration is achieved by generating a synchronization pulse signal through an FPGA to control the simultaneous acquisition of data by the radar and camera.

[0039] Spatial calibration converts the polar coordinates of the radar target into pixel coordinates, achieving spatial alignment of the data.

[0040] Blind spot filling is achieved by introducing Kalman filtering to predict the target trajectory. The state equation and observation equation are as follows: The state equation is

[0041] The observation equation is

[0042] The above method achieves spatiotemporal alignment between radar-detected targets and targets in video images, resulting in fused target perception information.

[0043] The environmental and operational data acquisition unit is equipped with various environmental sensors, including icing sensors, temperature and humidity sensors, anemometers, and tilt sensors, to monitor the physical environmental parameters around the power lines in real time (such as icing thickness, temperature and humidity, wind speed, and tower tilt status). Simultaneously, it interfaces with existing power systems such as the power SCADA system and EMS system to acquire real-time operational data of the lines, such as load, voltage, current, and equipment health status (such as transformer oil temperature and partial discharge).

[0044] The Geographic Information and Equipment Information Integration Unit is responsible for integrating the static geographic and structural information of power facilities. It acquires precise geographic coordinates, surrounding topography, and vegetation distribution information for power facilities (such as poles, lines, and substation locations) through GIS data. It integrates BIM data to obtain detailed 3D structures, material properties, installation dates, and maintenance records for equipment such as poles, conductors, insulators, and transformers. Furthermore, it utilizes laser point cloud scanning technology to create high-precision 3D models of power facilities and their surrounding environment, providing foundational data for the construction of digital twin models.

[0045] S2. Construct a three-dimensional digital twin model of the power facility and its surrounding environment based on the static information of the facility; map the fused target perception information, environmental monitoring data and operation data into the three-dimensional digital twin model in real time to synchronize the physical world and the digital world, and generate a real-time twin potential that includes the dynamic behavior of the perimeter target and the operating status of the equipment.

[0046] like Figure 4 As shown, in specific implementation, the digital twin modeling module for power facilities includes a 3D high-fidelity modeling unit, a real-time state mapping and synchronization unit, and a behavior and event visualization unit, as detailed below: The 3D high-fidelity modeling unit, based on data provided by the geographic information and equipment information integration unit, constructs a detailed 3D digital twin model of high-voltage transmission and distribution lines, towers, substations, and their surrounding environment. Utilizing the large-scale geographic background provided by GIS data, combined with equipment details from BIM data and precise geometric information from laser point cloud data, a highly realistic virtual power system is generated. This model accurately reflects the geometric structure, topological relationships, and spatial distribution of power facilities in the physical world, and through techniques such as texture mapping and material rendering, it achieves a high degree of visual consistency with the real scene. For example, it can accurately display the direction of each conductor, the type of tower, the layout of substations, and surrounding mountains, rivers, and buildings.

[0047] The real-time status mapping and synchronization unit is responsible for accurately mapping the real-time dynamic data acquired by the multi-dimensional data acquisition and preprocessing module onto the three-dimensional digital twin model, and maintaining real-time synchronization between the physical world and the digital twin model. The mapped data includes: the position, speed, and trajectory of targets (such as people and vehicles) detected by microwave radar; the behavior type (such as climbing, fishing, construction) and target identity identified by video images; ice thickness, temperature and humidity, wind speed, and tower tilt monitored by environmental sensors; and line load, voltage, current, and equipment health status provided by the power SCADA / EMS system.

[0048] The data fusion process is optimized using the Kalman filter algorithm. For nonlinear systems, the Extended Kalman Filter (EKF) is used, and its steps are as follows: The prediction steps include: State prediction

[0049] Covariance prediction

[0050] The update steps include: Kalman gain,

[0051] Status update,

[0052] Covariance update

[0053] in, It is a nonlinear observation function. yes Jacobian matrix, It refers to the observation noise covariance. This algorithm eliminates data noise and uncertainty, ensuring accurate and consistent information in the digital twin model.

[0054] The behavior and event visualization unit within the digital twin model visually and dynamically displays perimeter security-related behaviors and events. For example, an intruder's movement trajectory is depicted in real-time as colored lines on the 3D model; identified behavior types (such as "climbing" or "fishing") are displayed above the target via labels or animations; warning areas are highlighted with transparent blocks; and warning levels are differentiated using color coding (such as green - normal, yellow - caution, and red - danger). Simultaneously, the unit visualizes the distribution of environmental parameters (such as wind speed and temperature) and the real-time load status of power lines, enhancing the situational awareness capabilities of maintenance personnel.

[0055] S3. Construct a power knowledge graph library containing power equipment, environmental factors, behavior types, risk knowledge, and emergency plans; associate the real-time digital twin potential with the power knowledge graph library, identify perimeter security risks and determine risk levels through semantic reasoning; and when a fault occurs, perform causal reasoning based on historical data stored in the three-dimensional digital twin model and the power knowledge graph library to locate the root cause of the fault.

[0056] like Figure 5 As shown, in specific implementation, the power graph library construction and reasoning module includes a graph library construction unit, a semantic risk reasoning unit, and a fault root cause diagnosis unit. The details of each unit are as follows: The graph construction unit builds a comprehensive and structured power knowledge graph, represented in the form of triples (entity 1, relation, entity 2), and stored in a graph database (such as Neo4j). Nodes (entities) include: Electrical equipment, such as poles, conductors, transformers, switches, and insulators, includes attributes such as their model, parameters, location, and health status.

[0057] Environmental factors, such as weather (wind, rain, snow, hail, lightning), topography (mountains, plains, water), and vegetation (tree type, height).

[0058] Behavioral types, such as intrusion, climbing, fishing, construction, hanging of foreign objects, and drone flights, include their characteristic descriptions and potential hazards.

[0059] Risk levels, such as low risk, medium risk, high risk, and extremely high risk, along with the corresponding triggering conditions and consequences.

[0060] Safety regulations, such as the "Electric Power Work Regulations" and the "Regulations on the Protection of Electric Power Facilities," contain specific clauses and operating procedures.

[0061] Historical accident cases include the time, location, cause, process, consequences, and handling measures of the accidents.

[0062] Laws and regulations, specifically legal provisions related to the protection of power facilities.

[0063] Emergency response plans are emergency procedures and measures tailored to different accident types and risk levels.

[0064] The edges (relationships) in the power knowledge graph represent semantic associations between entities. For example, "tower" is located in "mountainous area," "strong wind" affects "conductor galloping," "climbing" leads to "electric shock accident," and "construction" violates "safety regulations." The graph is constructed using a semi-automatic approach, combining expert experience (acquiring tacit knowledge through expert interviews, questionnaires, etc.) and natural language processing (NLP) technology to automatically extract entities and relationships from textual materials such as power industry standards, regulations, and accident reports, and then perform knowledge fusion and verification.

[0065] The semantic risk reasoning unit associates the real-time mapped perceptual data and event information in the twin model with a knowledge graph, and uses a rule-based reasoning method for semantic reasoning. The rules are IF-THEN conditional rules. When the event information in the real-time twin potential meets the IF condition of the rule, the THEN conclusion is triggered to identify the risk. For example: IF(Target Type = "Crane") AND (Target Location = "Under High Voltage Line") AND (Target Behavior = "Lifting Operation") THEN (Event Type = "High-Risk Violation") AND (Associated Risk = "Electrocution Accident") AND (Recommended Procedure = "Maintain Safe Distance") Through this reasoning, the system can identify potential violations, safety hazards and their associated risks, and determine the risk level.

[0066] The fault root cause diagnosis unit, when a power system fault occurs, performs causal reasoning based on historical data stored in a 3D digital twin model and a power knowledge graph to locate the root cause of the fault. It achieves causal chain reasoning by constructing an event sequence graph and combining it with causal relationships in the knowledge graph for path searching and verification. Cause→Event1→Event2→ →Effect This expression is a logical illustration, not a strict mathematical formula. It describes the basic structure of a causal chain, where one or more initial **causes** lead to a certain **effect** through a series of intermediate events. In a fault diagnosis scenario: Cause refers to the root cause of a fault. Examples include "trees growing too tall," "insulators aging," and "illegal construction."

[0067] Event1, Event2, ... refer to a series of intermediate phenomena or events caused by a fundamental cause. For example, "trees growing too tall" may lead to "trees touching power lines in strong winds" (Event1), which in turn leads to "short circuits" (Event2), and then to "protective devices tripping" (Event3).

[0068] Effect refers to the final observed fault phenomenon. For example, "power outage".

[0069] The goal of root cause diagnosis is to reverse-engineer the causal chain when the final result (Effect) is observed, by tracing back historical data and utilizing causal relationships in a knowledge graph (e.g., a knowledge graph storing knowledge such as "a tree touching a conductor causes a short circuit"), thereby finding the initial root cause. This method helps to delve into the essence from the phenomenon, achieving precise fault location.

[0070] For example, by analyzing data changes and event sequences before and after a failure, the system can quickly pinpoint the root cause of the failure, such as external damage, equipment aging, severe weather, or operational errors, and generate a diagnostic report.

[0071] S4. Based on the risk level and the real-time twin potential, dynamically assess the risk and generate adaptive early warning information; predict the potential risks of power facilities based on historical and real-time data and generate predictive maintenance suggestions; and intelligently match emergency plans for high-risk events or faults based on the power knowledge graph; finally output the early warning information, maintenance suggestions and emergency plans.

[0072] like Figure 6 As shown, in specific implementation, the intelligent early warning and decision-making module includes an adaptive early warning strategy unit, a predictive maintenance suggestion unit, an emergency plan intelligent recommendation unit, and a multi-mode early warning output unit, as detailed below: The adaptive early warning strategy unit dynamically adjusts the early warning strategy based on the real-time situational awareness provided by the digital twin model and the risk reasoning results from the knowledge graph. This unit dynamically adjusts the early warning strategy according to the real-time situational information (such as target type, location, and behavior) provided by the digital twin model and the risk reasoning results (such as risk level, associated procedures, and historical cases) from the knowledge graph. Unlike the traditional fixed three-level response, the early warning level of this invention is dynamically generated after comprehensively evaluating multiple factors. These evaluation factors include: Risk levels are derived from knowledge graphs; for example, the risk level of "construction under high-voltage lines" is higher than that of "ordinary people approaching".

[0073] Severe weather conditions, such as strong winds, heavy rain, and icing, can increase the risk level.

[0074] When a line is under high load, any external interference may lead to more serious consequences, thus increasing the risk level.

[0075] The health status of the equipment is assessed; if there are potential health hazards in surrounding equipment, the risk level increases.

[0076] Due to time factors, the risk level may need to be upgraded during nighttime, holidays, or special sensitive periods.

[0077] Through a multi-factor weighted assessment model, the system can generate a dynamic and refined early warning level for each potential risk event and trigger corresponding early warning responses. For example, when the same approaching behavior is detected, if the system predicts upcoming strong winds or that the line is operating under high load, the system will issue a higher-level alarm in advance and suggest that maintenance personnel take preventative measures, such as remote announcements or initiating drone inspections. The comprehensive risk score is calculated using a multi-factor weighted model.

[0078] These are sub-scores representing different risk factors. Each sub-score is a normalized numerical value (e.g., between 0 and 1), explained in detail below: This is a comprehensive risk score; the higher the score, the more severe the risk. The basic risk score is inferred from the knowledge graph. For example, the risk score for illegal construction is higher than that for people approaching. The risk score is determined by weather conditions; for example, the risk score is higher for windy or rainy days than for sunny days. The risk score is determined by the line load; for example, the risk score is higher when the line is under high load than when it is under low load. The risk score is derived from the health condition of the equipment; for example, the risk score is higher when there is old equipment nearby. Risk scores are determined by time factors, such as higher risk scores at night or on holidays; As corresponding weighting coefficients to the risk scores, the weights determine the importance of each risk factor in the final total score. For example, if the behavioral risks inferred from the knowledge graph are considered the most important, then... The value can be set relatively high. The weights can be set by domain experts based on experience, or they can be trained from historical data using machine learning methods, with the sum of all weights being 1.

[0079] This formula enables the system to integrate qualitative or quantitative risk factors from different information sources into a unified, quantitative risk assessment framework, thereby achieving adaptive and dynamic adjustment of the warning level.

[0080] Warning Level Determined dynamically based on RiskScore:

[0081] in , , This is a preset risk threshold. Based on the warning level, a corresponding warning response is triggered, such as remote announcements or initiating drone inspections.

[0082] The warning level formula is a hierarchical function that maps the comprehensive risk score to discrete warning levels, transforms the quantified risk assessment results into specific operational instructions, and ultimately determines the warning level, such as four levels: "normal", "caution", "warning", and "danger".

[0083] If the risk score is below the threshold If the risk score is within a certain range, the situation is considered normal and no warning is triggered. and If the risk score falls between these values, a "Caution" level warning is triggered. and If the risk score falls between these thresholds, a "warning" level alert will be triggered. If the risk score exceeds the threshold... If this occurs, the highest level of "danger" warning will be triggered.

[0084] Thresholds can be set by operations and maintenance experts based on management requirements and experience, or optimized based on historical data to achieve a balance between alert sensitivity and false alarm rate. In this way, the system simplifies complex risk assessment results into clear and actionable alert levels, making it easier for operations and maintenance personnel to quickly understand and respond.

[0085] The predictive maintenance recommendation unit utilizes historical data and machine learning algorithms (such as Long Short-Term Memory networks, LSTM) to predict future risks. Combined with environmental data from digital twin models (such as icing thickness, wind speed, and temperature trends), it predicts potential risks to power lines under specific weather conditions, such as fallen trees, hanging objects, conductor galloping, and insulator flashover. Based on the predictions, the system proactively provides maintenance personnel with recommendations, such as pruning trees along the power line corridor, reinforcing towers, and conducting preventative maintenance on specific equipment. This shifts the focus from reactive emergency repairs to proactive prevention, reducing the probability of accidents.

[0086] LSTM networks are suitable for time series forecasting, and their unit calculation formula is as follows: Input gate,

[0087] The Gate of Oblivion

[0088] Output gate,

[0089] Candidate memory units,

[0090] Memory unit:

[0091] Hidden state

[0092] in, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. This represents element-wise multiplication. By training an LSTM model, future environmental parameters or risk indicators can be predicted, and maintenance recommendations (such as pruning trees and reinforcing towers) can be generated.

[0093] The intelligent emergency response plan recommendation unit intelligently recommends the most appropriate emergency response measures based on emergency plans and historical handling experience stored in the knowledge graph when the system issues a high-level warning or diagnoses a fault. For example, when a fire is detected under high-voltage lines, the system will immediately recommend a fire emergency plan and list detailed handling steps, including power outage, notifying the fire department, and personnel evacuation. Simultaneously, the system can dynamically adjust the recommended plan based on the real-time situation to ensure that maintenance personnel can respond quickly and effectively, minimizing losses.

[0094] A similarity-based contingency plan matching method is adopted, which calculates the cosine similarity between the current event and contingency plans in the knowledge graph:

[0095] in, It is a feature vector of the current event and contingency plan. This vector is converted into a multi-dimensional mathematical vector by natural language processing technology (such as BERT, Word2Vec, etc.) to transform the text description of the event (e.g., a fire occurred near tower No. 3, and the wind speed was 5 meters per second), capturing the key semantic information of the event. It is a feature vector of a certain emergency plan. Similarly, the text description of the emergency plan is also converted into a vector of the same dimension. The dot product of two vectors measures how close the two vectors are in terms of direction. The dot product is the largest if the directions are the same and the smallest if the directions are opposite. Let represent the Euclidean norm (or modulus) of the two vectors, i.e., the lengths of the vectors. The formula for calculating them is... ; The final similarity score ranges from -1 to 1; a score of 1 indicates that the two vectors are in the same direction and the event and the plan are highly matched; a score of 0 indicates that the two vectors are orthogonal and have no correlation; a score of -1 indicates that the two vectors are in completely opposite directions and are completely mismatched. By calculating the cosine similarity between each emergency response plan in the knowledge base and the current event, the system can find the plan with the highest score and recommend it as the most suitable solution. The advantage of this method is that it is unaffected by vector length (i.e., text length), focusing only on content and semantic similarity, recommending the emergency response plan with the highest similarity, and dynamically adjusting based on the real-time situation.

[0096] The multi-mode early warning output unit outputs early warning information in multiple modes, including: Local audio-visual warnings are provided through the device's integrated speaker for adaptive voice broadcasting (with customizable MP3 voice content and volume), and flashing warnings via high-brightness LEDs. The volume and flashing frequency are adaptively adjusted according to the risk level and ambient noise.

[0097] Remote alarms are sent to maintenance and management personnel via SMS, mobile app push notifications, and telephone notifications.

[0098] The system features coordinated control, automatically dispatching drones to verify the warning area in conjunction with the drone inspection system; automatically triggering fire-fighting equipment in conjunction with the fire protection system; and automatically switching to the video feed of the warning area in conjunction with the video surveillance system.

[0099] Data retention involves uploading video recordings and photographic evidence taken at the time of an early warning event to the cloud for storage in real time, serving as the basis for subsequent accident analysis and accountability.

[0100] Through the above specific implementation methods, the present invention realizes a complete technical solution from multi-dimensional data collection to intelligent decision-making, effectively improving the proactive and accurate perimeter security protection of power facilities, and also has good robustness.

[0101] This invention represents a leap from perception to cognition. Existing technologies primarily address the question of "what is seen," while this invention, by introducing digital twins and a graph library, achieves a deep understanding and semantic reasoning of perimeter security events, solving the questions of "what does it mean" and "what will happen," thus advancing from simple perception to deep cognition. The digital twin model, through multi-dimensional data fusion, overcomes the limitations of single sensors in harsh environments. Combined with semantic reasoning from the graph library, it can more accurately identify violations and safety hazards in complex scenarios, significantly reducing false alarm and false negative rates. Based on the real-time mapping of the physical world by the digital twin model and the risk reasoning capabilities of the graph library, the system can predict potential risk points and failure modes, transforming passive event response into proactive risk management, reducing accidents at their source, and ensuring the safe and stable operation of power facilities. This invention provides maintenance personnel with data-driven, knowledge-based decision support, including adaptive early warning strategies, predictive maintenance suggestions, and intelligent emergency plan recommendations, improving maintenance efficiency and decision-making accuracy. This invention integrates perimeter security from an isolated subsystem into the overall operation, maintenance, fault diagnosis, and emergency management system of the power system, achieving information sharing and collaborative decision-making, and improving the overall intelligent management level of the power system. Through pre-incident data backtracking using a digital twin model and causal chain reasoning from a graph library, the root cause of a fault can be quickly located, shortening troubleshooting time and reducing power outage losses.

[0102] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used in the operation of an intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring.

[0103] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be Random Access Memory (RAM) or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring in the above embodiments.

[0104] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0105] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] This invention also provides a computer program product, which is used to execute any of the above-described intelligent power perimeter early warning and diagnosis methods based on microwave radar monitoring. Since the computer program product provided by this invention and the above-described intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring belong to the same inventive concept, the computer program product provided by this invention possesses all the advantages of the above-described intelligent power perimeter early warning and diagnosis method based on microwave radar monitoring. Therefore, the beneficial effects of the computer program product provided by this invention will not be elaborated upon here.

[0109] In this invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0110] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent power perimeter early warning and diagnosis based on microwave radar surveillance, characterized in that, include: Multidimensional data is collected by various sensors deployed around the power perimeter. The multidimensional data includes at least microwave radar data and video image data for perimeter target perception, environmental monitoring data, and operational data and facility static information from the power information system. The microwave radar data and video image data are preprocessed and spatiotemporally fused to obtain fused target perception information. Based on the static information of the facilities, a three-dimensional digital twin model of the power facilities and their surrounding environment is constructed; the fused target perception information, environmental monitoring data and operation data are mapped into the three-dimensional digital twin model in real time to synchronize the physical world and the digital world, and generate a real-time twin potential that includes the dynamic behavior of the perimeter targets and the operating status of the equipment. Construct a power knowledge graph that includes power equipment, environmental factors, behavior types, risk knowledge, and emergency plans; associate the real-time twin potential with the power knowledge graph, and identify perimeter security risks and determine risk levels through semantic reasoning; In addition, when a fault occurs, causal reasoning is performed based on the historical data stored in the three-dimensional digital twin model and the power knowledge graph to locate the root cause of the fault. Based on the risk level and the real-time twin potential, the risk is dynamically assessed and adaptive early warning information is generated. Predict potential risks to power facilities based on historical and real-time data, and generate predictive maintenance recommendations; And based on the power knowledge graph, intelligent matching of emergency plans for high-risk events or faults; Finally, the system outputs the aforementioned warning information, maintenance suggestions, and emergency response plan.

2. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The preprocessing of the microwave radar data includes: The microwave radar data is denoised using an adaptive threshold denoising method based on wavelet transform. This method determines the threshold of each decomposition layer by estimating the noise standard deviation of the wavelet coefficients and uses a nonlinear function to process the wavelet coefficients to preserve the signal abrupt change characteristics.

3. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 2, characterized in that, In the adaptive threshold noise reduction method, the first... Threshold of decomposition layer Determined according to the following formula: in, For the first Estimates of the noise standard deviation of the layer wavelet coefficients. For the first The number of layer wavelet coefficients.

4. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The preprocessing of the video image data includes: An image enhancement algorithm based on illuminance-reflectance component separation is used to enhance the video image data. By estimating and removing the illuminance component in the image, the reflectance component of the target is restored to achieve image enhancement.

5. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The spatiotemporal fusion includes: establishing a mapping relationship between the radar coordinate system and the video image coordinate system through calibration, so as to achieve spatiotemporal alignment between the radar-detected target and the target in the video image, and obtain the fused target perception information.

6. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The process of mapping the fused target perception information, environmental monitoring data, and operational data to the three-dimensional digital twin model in real time employs a Kalman filter algorithm for data fusion.

7. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The semantic reasoning employs a rule-based reasoning method, where the rule is an IF-THEN conditional rule. When the event information in the real-time twin potential satisfies the IF condition of the rule, the THEN conclusion is triggered to identify the risk.

8. The intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance according to claim 1, characterized in that, The dynamic risk assessment includes: calculating a comprehensive risk score using a multi-factor weighted model, the calculation formula of which is: in, For comprehensive risk scoring, The basic risk score is inferred from the knowledge graph. Risk score derived from weather conditions; The risk score is determined by the line load. Risk score derived from equipment health status; The risk score is determined by the time factor. These are the weighting coefficients corresponding to the risk scores.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent power perimeter early warning and diagnosis method based on microwave radar surveillance as described in any one of claims 1 to 8.

Images