Real-time device for power equipment safety state based on meteorological analysis

Abstract

The application provides a kind of real-time device of power equipment safety state based on meteorological analysis, belongs to the technical field of power equipment safety analysis, including sensing terminal layer, edge intelligence layer and cloud decision layer, sensing terminal layer is connected with edge intelligence layer, edge intelligence layer is connected with cloud decision layer, sensing terminal layer is used for all-round, stereoscopic monitoring of power operation site environment risk, equipment state and personnel health, edge intelligence layer is used for embedded intelligent gateway, sets up risk assessment model, realizes local data preprocessing, anomaly detection and preliminary warning, reduces network load and improves response speed. Breakthrough traditional monitoring limitations, realize accurate risk prediction, through meteorological-electricity-personnel multidimensional data deep fusion and coupling modeling, break data island, accurately quantify the influence of environmental mutation on equipment safety. Dynamic risk index reflects the health status of equipment in real time, the false alarm rate is reduced to ≤5%, which is 40% lower than the traditional system, effectively avoiding invalid operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of power equipment safety analysis technology, and in particular to a real-time power equipment safety status device based on meteorological analysis. Specifically designed for the maintenance and operation of ultra-high voltage power facilities, it can be safely and efficiently applied to various high-altitude operations in the ultra-high voltage field, ensuring the safety of both personnel and equipment. Background Technology

[0002] Power operation safety monitoring is a crucial link in ensuring the stable operation of the power system and the safety of personnel and property. Traditional power safety monitoring systems have significant limitations: First, they rely on single monitoring methods, often depending on independent wind speed, temperature, humidity, or leakage current sensors, lacking the ability to fuse multi-source data and failing to comprehensively reflect the integrated risks in complex operating environments. Second, their early warning mechanisms are lagging, triggering alarms only based on fixed thresholds, unable to dynamically correlate environmental changes with equipment status changes, resulting in high false alarm rates or untimely responses. Third, their systems have poor adaptability, especially under conditions of strong electromagnetic interference and extreme weather, where sensor accuracy decreases and data transmission stability is insufficient, making it difficult to meet the stringent requirements of scenarios such as high-altitude operations and substation inspections.

[0003] While some existing systems integrate environmental parameter monitoring, they fail to establish a quantitative correlation model between meteorological factors (such as wind speed, wind direction, temperature, and humidity) and electrical safety (such as leakage risk). This makes it impossible to accurately predict equipment failures or personnel safety risks induced by environmental factors. For example, key coupling mechanisms such as the relationship between decreased insulation performance and changes in leakage conductivity under high humidity conditions, and the impact of conductor galloping caused by strong winds on equipment electrical parameters, have not been effectively analyzed. Furthermore, existing solutions lack sufficient collaborative monitoring of workers' physiological states (such as heart rate and blood oxygen) and environmental risks, making it difficult to achieve coordinated early warning systems involving humans, machines, and the environment, thus leaving safety hazards unresolved.

[0004] Therefore, there is a need for an intelligent monitoring system that can integrate multi-dimensional monitoring data, dynamically analyze the coupling relationship between the environment and electrical risks, and achieve accurate early warning, in order to break through existing technological bottlenecks and improve the safety and efficiency of power operations. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time power equipment safety status device based on meteorological analysis, solving the technical problem that existing technologies cannot accurately predict power equipment failures and personnel safety risks induced by complex weather conditions. Overcoming the shortcomings of existing power safety monitoring technologies, such as limited monitoring dimensions, lagging early warning mechanisms, and lack of correlation analysis between environmental and electrical parameters, the core objective of this invention is to break down the barriers between environmental and electrical data in traditional monitoring.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A real-time power equipment safety status device based on meteorological analysis includes a sensing terminal layer, an edge intelligence layer, and a cloud decision layer. The sensing terminal layer is connected to the edge intelligence layer, and the edge intelligence layer is connected to the cloud decision layer. The sensing terminal layer is used for comprehensive and three-dimensional monitoring of environmental risks, equipment status, and personnel health at power operation sites. The edge intelligence layer is used as an embedded intelligent gateway to set up a risk assessment model, realize localized data preprocessing, anomaly detection, and preliminary early warning, reduce network load, and improve response speed. The cloud decision layer is used for data analysis engine and deep learning model training, LSTM+Attention time series prediction, risk evolution path deduction, and model self-learning optimization.

[0007] Furthermore, the sensing terminal layer includes meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors. The meteorological sensors are used to detect humidity, wind direction, and wind speed; the electrical sensors are used to detect the current, voltage, and temperature of electrical equipment; the personnel monitoring sensors are used to detect the physiological state of workers inside the electrical equipment; and the environmental sensors are used to detect and monitor salt deposits.

[0008] Furthermore, the meteorological sensors include ultrasonic anemometers and capacitive temperature and humidity sensors for real-time detection of wind speed and temperature and humidity. The electrical sensors include leakage current sensors, voltage sensors, equipment current sensors, and patch temperature sensors. The voltage sensors and equipment current sensors are used to detect the voltage and current of electrical equipment, the leakage current sensors are used to detect whether there is leakage in the equipment, and the patch temperature sensors are used to detect the temperature of the electrical equipment. The personnel monitoring sensor is designed to be worn by personnel entering the electrical equipment to monitor physiological characteristics. The personnel monitoring sensor is designed to be worn as a physiological monitoring module. The environmental sensor is a salt density sensor, which is placed on the surface of the insulator to monitor salt deposits.

[0009] Furthermore, meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors all achieve millisecond-level time alignment through the IEEE 1588v2 precision time synchronization protocol, ensuring spatiotemporal consistency of multi-source data. A unified data middleware is constructed, and the OPC UA protocol is used to enable plug-and-play functionality for sensors.

[0010] Furthermore, the nonlinear coupled response function of the risk assessment model is: Among them, R( t The real-time risk index is 0–100. Vw For wind speed, H For humidity, H 0 is the humidity safety threshold, Δ I l (t) Let T be the change in current and T be the temperature.α, β, and γ are The weighting coefficients are dynamically calibrated using historical fault data and online learning.

[0011] Furthermore, the edge intelligence layer includes a data preprocessing module, a risk assessment model module, an ontology decision-making module, a model compression module, and a time synchronization module. The data preprocessing module is used for data purification and optimal state estimation. The risk assessment model module is used to run the risk assessment model and calculate the risk score of the risk index in real time. The ontology decision-making module is used to execute preset control logic based on the results of the risk assessment model to achieve preliminary early warning and rapid response. The model compression module is used to optimize, prune, and quantize complex models trained in the cloud, enabling them to run efficiently on edge gateways with limited computing power and storage resources. The time synchronization module is used to achieve millisecond-level time alignment between sensors through the IEEE 1588 v2 precision time protocol, maintain the consistency of the processing timing within the entire edge computing node, and ensure that the received multi-channel sensor data has a unified and accurate timestamp, providing a reliable time reference for subsequent data fusion and correlation analysis.

[0012] Furthermore, the cloud-based decision-making layer includes a risk evolution model module, a deep learning engine module, a model self-learning module, a big data analysis module, and a visualization platform module. The risk evolution model module is used for medium- to long-term time-series prediction and risk extrapolation. Utilizing network-wide data collected from the edge layer, it constructs more complex predictive models to predict the probability of flashover in the next 24 hours. This not only assesses current risks but also predicts future risk trends, providing a basis for preventative maintenance decisions. The deep learning engine module provides model training and prediction capabilities, processes time-series data, and uncovers deep, non-linear correlations between environmental parameters and equipment status. The model self-learning module enables continuous optimization and dynamic evolution of the system risk assessment model. Through historical fault data and online learning for dynamic calibration, it updates coupling coefficients using the reinforcement learning DQN algorithm, collects new operational data and fault cases reported from the edge layer, and automatically adjusts and optimizes the weight parameters α, β, and α in the risk assessment model using reinforcement learning and other algorithms. γ enables the model to adapt to new situations such as equipment aging and environmental changes, becoming more accurate with use. The big data analysis module is used to store, manage, mine, and perform root cause analysis on all data. It is responsible for storing historical and real-time data from all sensing terminals and provides data query, statistical analysis, and in-depth mining capabilities. The visualization platform module provides a human-computer interaction interface to intuitively display system status, early warning information, and decision support.

[0013] Furthermore, when the device collects data on the degree of contamination on the insulator surface and the ambient humidity, during the edge processing stage, it does not judge whether the humidity is too high or the salt density exceeds the standard in isolation. Instead, it makes a combined judgment. If the humidity is >90% and the salt density is >0.08mg / cm², a first-level warning is triggered, and the dehumidification device is activated. The humidity threshold for triggering the warning is tied to the current specific salt density value. If the salt density value changes and decreases after cleaning, the critical humidity value for triggering the warning is also adjusted accordingly. The conditions for identifying risks change dynamically with the contamination status, rather than remaining fixed. The combination of salt density and temperature and humidity is a key precursor to the formation of a conductive water film on the insulator surface, which can lead to flashover accidents. By quantifying the combined effect of these two parameters, it is possible to identify a state of rapidly increasing risk before an electrical flashover actually occurs, thereby providing corresponding early warning time for taking preventive measures.

[0014] Furthermore, the edge intelligence layer processes data directly without uploading it to the cloud to await instructions. When the edge gateway determines a risk based on the dynamic threshold model, it directly activates the dehumidification device. By making decisions and executing near the data generation location, the end-to-end latency is reduced to less than 500ms, meeting the power system's need for rapid handling of some emergency faults. The cloud-based decision-making layer receives data from edge nodes across the entire network, constructs a risk evolution model, performs long-term and macro-level analysis to predict the probability of flashover in the next 24 hours, and, based on a global perspective, makes better scheduling decisions. It then pushes emergency isolation instructions to the SCADA system. The cloud-based decision-making layer uses the collected big data and reinforcement learning algorithms to continuously optimize and update the parameters α, β, and γ of the risk assessment model set at the edge, and then distributes the optimized model to the edge, so that the risk assessment capability can continuously evolve with the accumulation of operational experience.

[0015] The present invention, by adopting the above-described technical solution, has the following beneficial effects: (1) This invention breaks through the limitations of traditional monitoring and achieves accurate risk prediction. By deeply integrating and coupling multi-dimensional data from meteorology, electrical systems and personnel into a model, it breaks down data silos and accurately quantifies the impact of sudden environmental changes on equipment safety. The dynamic risk index (0-100) reflects the health status of equipment in real time, reducing the false alarm rate to ≤5%, which is 40% lower than that of traditional systems, effectively avoiding ineffective operation and maintenance.

[0016] (2) Significantly improves fault response timeliness and protection capabilities: The edge intelligent layer achieves millisecond-level local decision-making with an end-to-end response latency of <500ms. In emergencies, it can automatically trigger hard-contact tripping to prevent fault propagation. The hierarchical early warning mechanism (observation / enhanced monitoring / emergency isolation) combined with intelligent linkage control reduces costs and maintenance expenses. It is suitable for complex scenarios such as UHV transmission and coastal high-humidity substations, effectively preventing accidents such as flashover and galloping caused by extreme weather.

[0017] (3) Ensure operational safety and achieve collaborative protection between personnel and machines: Integrate physiological monitoring data of workers to assess the impact of environmental risks on the human body in real time, and avoid secondary accidents such as heatstroke and falls from heights caused by severe weather. Provide scientific decision support for live-line work and ensure the dual safety of personnel and equipment.

[0018] (4) Technological leadership and promotion value: The innovative nonlinear coupling response function and adaptive model optimization mechanism are protected by invention patents. The system complies with the IEC 61850 / 61970 international standards, supports protocol extension, and has the potential for promotion across regions and voltage levels of power grids, driving the transformation of power safety monitoring towards intelligence and initiative. Attached Figure Description

[0019] Figure 1 This is a block diagram of the real-time safety status device module for power equipment according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and preferred embodiments. However, it should be noted that many details listed in the specification are merely to provide the reader with a thorough understanding of one or more aspects of the present invention, and these aspects of the invention can be implemented even without these specific details.

[0021] like Figure 1 As shown, a real-time power equipment safety status device based on meteorological analysis includes a sensing terminal layer, an edge intelligence layer, and a cloud decision layer. The sensing terminal layer is connected to the edge intelligence layer, and the edge intelligence layer is connected to the cloud decision layer. The sensing terminal layer is used for comprehensive and three-dimensional monitoring of environmental risks, equipment status, and personnel health at power operation sites. The edge intelligence layer is used for embedded intelligent gateways to set up risk assessment models, realize localized data preprocessing, anomaly detection, and preliminary early warning, reduce network load, and improve response speed. The cloud decision layer is used for data analysis engine and deep learning model training, LSTM+Attention time series prediction, risk evolution path deduction, and model self-learning optimization.

[0022] This device solves the problem of low multi-source monitoring data fusion and realizes the panoramic perception of human-machine-environment. Existing power monitoring systems usually deploy meteorological sensors (wind speed, temperature, humidity) and electrical sensors (leakage, voltage) independently, resulting in data islands. The purpose of this invention is to construct a multi-source information fusion perception system that not only synchronously collects micro meteorological parameters such as wind speed, wind direction, temperature, and humidity, as well as electrical parameters such as leakage conductance and insulation resistance, but also innovatively introduces physiological sign data of operators (such as heart rate, blood oxygen). By establishing a unified data acquisition and synchronization mechanism, this invention aims to achieve all-round and three-dimensional monitoring of "environmental risks", "equipment status", and "personnel health" at the power operation site, filling the gap in the field of collaborative monitoring in the existing technology.

[0023] Solve the problem of unclear coupling mechanism between meteorological and electrical risks and establish a dynamic quantitative evaluation model. Traditional technologies rely only on fixed thresholds of single parameters for alarm, ignoring the dynamic impact of meteorological environment on the performance of electrical equipment. For example, in a high humidity environment, the decline of equipment insulation performance and leakage risk is not a linear relationship, and simply monitoring the leakage value often leads to a lag in response. The purpose of this invention is to deeply analyze the internal coupling mechanism between meteorological factors (such as wind speed, humidity) and electrical parameters (such as leakage conductance), and establish a quantitative model that can calculate the "meteorological-electrical coupling risk index" in real time. Through this model, the system can dynamically adjust the risk assessment criteria according to real-time environmental changes, so as to accurately identify potential equipment failures caused by sudden environmental changes.

[0024] Solve the problems of high false alarm rate and response delay of early warning, and achieve hierarchical and accurate early warning and active defense. Aiming at the pain points of high false alarm rate (such as still alarming when the wind is strong but the operator is in good condition) and long fault response time in existing systems, this invention aims to propose a hierarchical early warning mechanism based on multi-dimensional data fusion. The purpose of this invention is not only to "give an alarm", but more importantly, to "make a decision". By analyzing the superposition effect of environmental risks and human body status through algorithms, the system can intelligently divide risk levels (such as observation, enhanced monitoring, emergency isolation), and automatically trigger corresponding linkage control strategies (such as adjusting equipment load or emergency power-off). Ultimately, this invention aims to transform power safety monitoring from traditional "passive response" to "active prediction and defense", significantly reducing the unplanned downtime and the incidence of operation safety accidents, and improving the safety, intelligence level and operation and maintenance efficiency of power operations.

[0025] In this embodiment of the invention, the sensing terminal layer includes meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors. The meteorological sensors detect humidity, wind direction, and wind speed; the electrical sensors detect the current, voltage, and temperature of electrical equipment; the personnel monitoring sensors detect the physiological state of workers inside the electrical equipment; and the environmental sensors detect and monitor salt deposits. The meteorological sensors include an ultrasonic anemometer and a capacitive temperature and humidity sensor for real-time detection of wind speed and temperature / humidity. The electrical sensors include a leakage current sensor, a voltage sensor, an equipment current sensor, and a patch-type temperature sensor. The voltage sensor and equipment current sensor detect the voltage and current of the electrical equipment; the leakage current sensor detects whether leakage occurs in the equipment; the patch-type temperature sensor detects the temperature of the electrical equipment; the personnel monitoring sensors are worn by workers entering the electrical equipment to monitor physiological characteristics; the personnel monitoring sensors are wearable physiological monitoring modules; and the environmental sensors are salt density sensors installed on the surface of insulators to monitor salt deposits. Meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors all achieve millisecond-level time alignment through the IEEE 1588v2 precision time synchronization protocol, ensuring spatiotemporal consistency of multi-source data. A unified data middleware is constructed, and the OPC UA protocol is used to enable plug-and-play functionality for sensors.

[0026] In this embodiment of the invention, the edge intelligence layer includes a data preprocessing module, a risk assessment model module, an ontology decision module, a model compression module, and a time synchronization module. The data preprocessing module is used for data purification and optimal state estimation. The risk assessment model module is used to run the risk assessment model and calculate the risk score of the risk index in real time. The ontology decision module is used to execute preset control logic based on the results of the risk assessment model to achieve preliminary early warning and rapid response. The model compression module is used to optimize, prune, and quantize complex models trained in the cloud, enabling them to run efficiently on edge gateways with limited computing power and storage resources. The time synchronization module is used to achieve millisecond-level time alignment between sensors through the IEEE 1588 v2 precision time protocol, maintain the consistency of the processing timing within the entire edge computing node, and ensure that the received multi-channel sensor data has a unified and accurate timestamp, providing a reliable time reference for subsequent data fusion and correlation analysis.

[0027] The nonlinear coupled response function of the risk assessment model is: Among them, R( t The real-time risk index is 0–100. Vw For wind speed, H For humidity, H 0 is the humidity safety threshold, Δ I l(t) Let T be the change in current and T be the temperature. α, β, and γ are The weighting coefficients are dynamically calibrated using historical fault data and online learning.

[0028] In this embodiment of the invention, the cloud-based decision-making layer includes a risk evolution model module, a deep learning engine module, a model self-learning module, a big data analysis module, and a visualization platform module. The risk evolution model module is used for medium- to long-term time-series prediction and risk extrapolation. Utilizing network-wide data collected from the edge layer, it constructs more complex prediction models to predict the probability of flashover in the next 24 hours. This not only assesses current risks but also predicts future risk trends, providing a basis for preventative maintenance decisions. The deep learning engine module provides model training and prediction capabilities, processes time-series data, and uncovers deep, non-linear correlations between environmental parameters and equipment status. The model self-learning module enables continuous optimization and dynamic evolution of the system risk assessment model. It dynamically calibrates the model using historical fault data and online learning, updates coupling coefficients using the reinforcement learning DQN algorithm, collects new operational data and fault cases reported from the edge layer, and automatically adjusts and optimizes the weight parameters α, β, and α in the risk assessment model using reinforcement learning and other algorithms. γ enables the model to adapt to new situations such as equipment aging and environmental changes, becoming more accurate with use. The big data analysis module is used to store, manage, mine, and perform root cause analysis on all data. It is responsible for storing historical and real-time data from all sensing terminals and provides data query, statistical analysis, and in-depth mining capabilities. The visualization platform module provides a human-computer interaction interface to intuitively display system status, early warning information, and decision support.

[0029] In this embodiment of the invention, when the device collects data on the degree of contamination on the insulator surface and the ambient humidity, during the edge processing stage, it does not judge whether the humidity is too high or the salt density exceeds the standard in isolation. Instead, it makes a combined judgment. If the humidity is >90% and the salt density is >0.08mg / cm², a first-level warning is triggered, and the dehumidification device is activated. The humidity threshold for triggering the warning is tied to the current specific salt density value. If the salt density value changes and decreases after cleaning, the critical humidity value for triggering the warning is also adjusted accordingly. The conditions for identifying risks change dynamically with the contamination status, rather than remaining fixed. The combination of salt density and temperature and humidity is a key precursor to the formation of a conductive water film on the insulator surface and the occurrence of flashover accidents. By quantifying the combined effect of these two parameters, it is possible to identify a state of rapidly increasing risk before an electrical flashover actually occurs, thereby providing corresponding early warning time for taking preventive measures.

[0030] In this embodiment of the invention, the edge intelligence layer processes the data directly without uploading the data to the cloud and waiting for instructions. When the edge gateway determines the risk based on the dynamic threshold model, it directly activates the dehumidification device. By making decisions and executing near the data generation location, the end-to-end latency is reduced to <500ms, which meets the power system's need for rapid handling of some emergency faults. The cloud-based decision-making layer receives data from edge nodes across the entire network, constructs a risk evolution model, performs long-term and macro-level analysis to predict the probability of flashover in the next 24 hours, and, based on a global perspective, makes better scheduling decisions. It then pushes emergency isolation instructions to the SCADA system. The cloud-based decision-making layer uses the collected big data and reinforcement learning algorithms to continuously optimize and update the parameters α, β, and γ of the risk assessment model set at the edge, and then distributes the optimized model to the edge, so that the risk assessment capability can continuously evolve with the accumulation of operational experience.

[0031] Sensing terminal layer: Deploys a multimodal sensor network, including an ultrasonic anemometer (0–60 m / s, accuracy ±0.3 m / s), a capacitive temperature and humidity sensor (±0.5℃, ±3%RH), a μS-level leakage current sensor, and a wearable physiological monitoring module. All sensors achieve millisecond-level time alignment via the IEEE 1588v2 precision time synchronization protocol, ensuring spatiotemporal consistency of multi-source data.

[0032] Edge intelligence layer: Based on an embedded AI gateway (equipped with ARM Cortex-A78 + NPU), a lightweight risk assessment model is deployed to achieve localized data preprocessing, anomaly detection and preliminary early warning, thereby reducing network load and improving response speed.

[0033] Cloud-based decision-making layer: Deployed on a private cloud platform using a microservice architecture, it integrates a big data analysis engine and a deep learning model training module, supporting LSTM+Attention time series prediction, risk evolution path deduction, and model self-learning optimization.

[0034] A multi-source heterogeneous data fusion mechanism constructs a unified data middleware, and the OPC UA protocol enables plug-and-play sensor functionality. A time series alignment algorithm (DTW+Kalman filtering) addresses the issue of sampling frequency differences, generating standardized feature vectors that are input into the risk assessment model.

[0035] Tiered early warning and intelligent linkage control strategy; Level 1 early warning (medium risk): Push alarm to the operation and maintenance platform and initiate special equipment inspection. Level 2 early warning (high risk): Automatically issue control commands to the SCADA system to adjust the load or isolate the section. Emergency response: When the wind speed change rate is >5 m / s² and the leakage current increase is >20%, trigger the hard contact trip command to achieve millisecond-level power outage protection.

[0036] Real-time situational awareness and visualization, multi-dimensional risk assessment and early warning push, intelligent decision support and control linkage, historical data tracing and root cause analysis.

[0037] The multi-dimensional coupled evaluation model of meteorology, electrical engineering, and personnel breaks through the limitations of traditional single-parameter monitoring. The dynamic adaptive early warning mechanism has a false alarm rate of ≤5%, which is 40% lower than that of traditional systems. The edge-cloud collaborative architecture ensures high real-time performance under strong electromagnetic interference, with an end-to-end latency of <500ms. It supports IEC standard protocols and has good compatibility and scalability.

[0038] Application scenarios include monitoring for galloping of UHV transmission lines, preventing flashover at coastal high-humidity substations, and assisting in decision-making for personal safety during live-line work.

[0039] 1. Data Acquisition: The sensor synchronously collects environmental parameters (humidity, salinity) and electrical parameters (partial discharge amplitude, frequency) every 5 minutes.

[0040] 2. Edge Processing: Data Preprocessing: Outliers are removed, and the signal is smoothed using Kalman filtering. Real-time Assessment: If humidity > 90% and salinity > 0.08 mg / cm², a Level 1 warning is triggered, and the dehumidification device is activated.

[0041] 3. Cloud-based decision-making: Receive data uploaded from the edge, construct a risk evolution model, and predict the probability of flashover in the next 24 hours. If the risk index R ≥ 70 (high risk), push an "emergency isolation" command to the SCADA system.

[0042] 4. Interlocking control: The execution unit receives the instruction to shut down the circuit breaker in the faulty section and start the standby equipment.

[0043] 5. Closed-loop feedback: Monitor the dehumidification effect: If the humidity drops below 80%, stop heating and record the operation log.

[0044] By employing a salt density-humidity dynamic threshold model, flashover precursors are accurately identified. Edge-cloud collaboration enables "rapid local response + global optimized decision-making." After application in a pilot substation, the flashover early warning accuracy increased to 92%, and the fault response time was shortened to 300ms.

[0045] Matters not covered in this invention are common knowledge.

[0046] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A real-time device for the safety status of power equipment based on meteorological analysis, characterized in that: It includes a sensing terminal layer, an edge intelligence layer, and a cloud decision layer. The sensing terminal layer is connected to the edge intelligence layer, and the edge intelligence layer is connected to the cloud decision layer. The sensing terminal layer is used for comprehensive and three-dimensional monitoring of environmental risks, equipment status, and personnel health at power operation sites. The edge intelligence layer is used for embedded smart gateways to set up risk assessment models, realize localized data preprocessing, anomaly detection and preliminary early warning, reduce network load and improve response speed. The cloud decision layer is used for data analysis engine and deep learning model training, LSTM+Attention time series prediction, risk evolution path inference and model self-learning optimization.

2. The real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: The sensing terminal layer includes meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors. Meteorological sensors are used to detect humidity, wind direction, and wind speed; electrical sensors are used to detect the current, voltage, and temperature of electrical equipment; personnel monitoring sensors are used to detect the physiological state of workers inside electrical equipment; and environmental sensors are used to detect and monitor salt deposits.

3. The real-time power equipment safety status device based on meteorological analysis according to claim 2, characterized in that: The meteorological sensors include ultrasonic anemometers and capacitive temperature and humidity sensors for real-time detection of wind speed and temperature and humidity. The electrical sensors include leakage current sensors, voltage sensors, equipment current sensors, and patch temperature sensors. The voltage and equipment current sensors are used to detect the voltage and current of electrical equipment, the leakage current sensor is used to detect whether there is leakage in the equipment, and the patch temperature sensor is used to detect the temperature of the electrical equipment. The personnel monitoring sensor is designed to be worn by personnel entering the electrical equipment to monitor physiological characteristics. The personnel monitoring sensor is a wearable physiological monitoring module. The environmental sensor is a salt density sensor, which is placed on the surface of the insulator to monitor salt deposits.

4. A real-time power equipment safety status device based on meteorological analysis according to claim 2, characterized in that: Meteorological sensors, electrical sensors, personnel monitoring sensors, and environmental sensors all achieve millisecond-level time alignment through the IEEE 1588v2 precision time synchronization protocol, ensuring spatiotemporal consistency of multi-source data. A unified data middleware is constructed, and the OPCUA protocol is used to enable plug-and-play functionality for sensors.

5. A real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: The nonlinear coupled response function of the risk assessment model is: Among them, R( t The real-time risk index is 0–100. Vw For wind speed, H For humidity, H 0 is the humidity safety threshold, Δ I l (t) Let T be the change in current and T be the temperature. α,β,γ are The weighting coefficients are dynamically calibrated using historical fault data and online learning.

6. A real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: The edge intelligence layer includes a data preprocessing module, a risk assessment model module, an ontology decision-making module, a model compression module, and a time synchronization module. The data preprocessing module is used for data purification and optimal state estimation. The risk assessment model module is used to run the risk assessment model and calculate the risk score of the risk index in real time. The ontology decision-making module is used to execute preset control logic based on the results of the risk assessment model to achieve preliminary early warning and rapid response. The model compression module is used to optimize, prune, and quantize complex models trained in the cloud, enabling them to run efficiently on edge gateways with limited computing power and storage resources. The time synchronization module is used to achieve millisecond-level time alignment between sensors through the IEEE 1588 v2 precision time protocol, maintain the consistency of the processing timing within the entire edge computing node, and ensure that the received multi-channel sensor data has a unified and accurate timestamp, providing a reliable time reference for subsequent data fusion and correlation analysis.

7. A real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: The cloud-based decision-making layer comprises a risk evolution model module, a deep learning engine module, a model self-learning module, a big data analysis module, and a visualization platform module. The risk evolution model module is used for medium- to long-term time-series prediction and risk extrapolation. Utilizing network-wide data collected from the edge layer, it constructs more complex predictive models to predict the probability of flashover in the next 24 hours. This not only assesses current risks but also predicts future risk trends, providing a basis for preventative maintenance decisions. The deep learning engine module provides model training and prediction capabilities, processes time-series data, and uncovers deep, non-linear correlations between environmental parameters and equipment status. The model self-learning module enables continuous optimization and dynamic evolution of the system risk assessment model. Through historical fault data and online learning, it dynamically calibrates the model, updates coupling coefficients using the reinforcement learning DQN algorithm, collects new operational data and fault cases reported from the edge layer, and automatically adjusts and optimizes the weight parameters α, β, and α in the risk assessment model using reinforcement learning and other algorithms. γ enables the model to adapt to new situations such as equipment aging and environmental changes, becoming more accurate with use. The big data analysis module is used to store, manage, mine, and perform root cause analysis on all data. It is responsible for storing historical and real-time data from all sensing terminals and provides data query, statistical analysis, and in-depth mining capabilities. The visualization platform module provides a human-computer interaction interface to intuitively display system status, early warning information, and decision support.

8. A real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: When the device collects data on the degree of contamination on the insulator surface and the ambient humidity, during the edge processing stage, it does not judge whether the humidity is too high or the salt density exceeds the standard in isolation. Instead, it makes a combined judgment. If the humidity is >90% and the salt density is >0.08mg / cm², a level one warning is triggered, and the dehumidification device is activated. The humidity threshold for triggering the warning is tied to the current specific salt density value. If the salt density value changes and decreases after cleaning, the critical humidity value for triggering the warning is also adjusted accordingly. The conditions for identifying risks change dynamically with the contamination status, rather than remaining fixed. The combination of salt density and temperature and humidity is a key precursor to the formation of a conductive water film on the insulator surface, which can lead to flashover accidents. By quantifying the combined effect of these two parameters, it is possible to identify a state of rapidly increasing risk before an electrical flashover actually occurs, thus providing corresponding early warning time for taking preventive measures.

9. A real-time power equipment safety status device based on meteorological analysis according to claim 1, characterized in that: The edge intelligence layer processes data directly without uploading it to the cloud to await instructions. When the edge gateway determines a risk based on the dynamic threshold model, it directly activates the dehumidification device. By making decisions and executing near the data generation location, the end-to-end latency is reduced to less than 500ms, meeting the power system's need for rapid handling of some emergency faults. The cloud-based decision-making layer receives data from edge nodes across the entire network, constructs a risk evolution model, performs long-term and macro-level analysis to predict the probability of flashover in the next 24 hours, and, based on a global perspective, makes better scheduling decisions. It then pushes emergency isolation instructions to the SCADA system. The cloud-based decision-making layer uses the collected big data and reinforcement learning algorithms to continuously optimize and update the parameters α, β, and γ of the risk assessment model set at the edge, and then distributes the optimized model to the edge, so that the risk assessment capability can continuously evolve with the accumulation of operational experience.

Images