Power grid node health index assessment method suitable for remote operation and maintenance

Abstract

The invention relates to the technical field of intelligent power grid operation and maintenance, and discloses a power grid node health index assessment method suitable for remote operation and maintenance, and the method comprises the steps: S1, carrying out the fusion collection of multi-source data; s2, dynamic weight self-adaptive distribution is carried out; s3, dynamically correcting the life reference; step S4, layered health index calculation; s5, health state decision making and visualization are carried out; the data weight is automatically adjusted according to the timeliness difference of high and low frequency data through a dynamic weight adaptive distribution mechanism, so that the problem of health score long-term distortion possibly caused by low frequency data updating lag is solved, key parameters can be monitored in real time through a real-time anomaly detection and response mechanism, a parameter overrun early warning threshold value is set, and the early warning accuracy is improved. When the key parameters continuously exceed the limit, an online diagnosis mode is triggered, the data sampling frequency is increased, it is ensured that abnormal conditions can be quickly found and processed, the ability of the system to deal with sudden faults is improved, and the fault risk is reduced.

Description

Technical Field

[0001] The present invention relates to the field of smart grid operation and maintenance technology, and in particular to a grid node health index assessment method suitable for remote operation and maintenance. Background Art

[0002] Power distribution stations and the distribution network switchgear within them are the most fundamental electrical equipment in the power system. They have a large market application and are managed by end users from various industries. Typically, each end-user's power distribution station has multiple distribution network switchgear, which, in different combinations, form one or more power distribution systems, providing distribution, carrying, and control for the user's power supply.

[0003] After searching, the Chinese patent number CN113112037A discloses that the invention provides a remote operation and maintenance health management system and health assessment method for power distribution stations. The system includes: a data management unit, a permission management unit, a health assessment unit and a fault prediction unit. The data management unit includes a database interface module, an evaluation strategy module, and a data sorting module. The database interface module is used to obtain the operating data of the distribution network switch equipment in the power distribution station from the public cloud database; the method divides the operating data of the distribution network switch equipment into five layers of indicator data, and adopts a bottom-up recursive calculation method to obtain the health score data of the power distribution station, and forms a digital archive of the equipment and an industrial knowledge base to provide data-based precise operation and maintenance services.

[0004] However, the above evaluation method also has some problems. First, in the above scheme, most of the operating condition data need to be collected through a mixed method of sensors (seconds), manual inspections (months) and professional tests (years / several years). For low-frequency data such as annual test results and high-frequency data such as temperature, the lack of dynamic weighting of low-frequency data update lag may lead to long-term distortion of the health score. For example, when the annual insulation test is not updated, the system cannot reflect the recent degradation of the equipment. In addition, the reference value in the above scheme can be based on the equipment life reduction, the average value of similar equipment or experience assignment, but the applicable scenarios and priorities are not clear. For example, the equipment life reduction coefficient is fixed at 0.05 / year, and there is a lack of differentiated adjustment for the actual operating environment of the equipment, such as high temperature and high humidity, which may lead to a large deviation between the reference value and the actual life. Based on this, the present invention designs a power grid node health index evaluation method suitable for remote operation and maintenance to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a grid node health index assessment method suitable for remote operation and maintenance, which solves the problems of lack of dynamic weighting and unclear scenarios and priorities in the background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: The grid node health index assessment method applicable to remote operation and maintenance includes: Step S1: Multi-source data fusion acquisition. High-frequency sensor data and low-frequency test data are collected in real time through the Internet of Things interface. High-frequency data includes temperature, current, and mechanical parameters at the second to minute level, and low-frequency data includes insulation performance, mechanical wear, and chemical test results at the month to year level. Sliding window filtering is performed on the high-frequency data with a window length of 5-10 sampling periods, and outliers are removed based on the 3σ criterion. Edge preprocessing is performed on the high-frequency data, including filtering, denoising, and outlier removal. Timestamps and credibility tags are added to the low-frequency data to form a structured data set. The collected data is checked for integrity, missing data is interpolated and supplemented based on data from adjacent time points, and the processed data is stored in a multi-dimensional data pool according to a preset data structure. Step S2: Dynamic weight adaptive allocation. Based on the timeliness difference between high-frequency and low-frequency data, a dynamic attenuation factor is used to adjust the data weight. The weight of high-frequency data increases as the timeliness of low-frequency data decreases. A weight allocation threshold is set. When the timeliness of low-frequency data exceeds the preset half-life, a data update request is triggered and its confidence level is lowered. Step S3: Dynamically correct the lifespan benchmark. Based on the equipment's operating years and environmental parameters, the environmental accelerated degradation coefficient is calculated and the equipment design lifespan is dynamically corrected. The initial lifespan benchmark is differentiated by equipment type, and a dynamic lifespan reference curve is generated in combination with real-time environmental data. Step S4, stratified health index calculation, constructing a four-layer health indicator system, including physical layer, functional layer, environmental layer and historical layer indicators, and assigning weights to each layer through the hierarchical analysis method; using the fuzzy comprehensive evaluation method to aggregate the indicator deviation layer by layer, output the health index HI on a scale of 0-100, and calculate the confidence interval; Step S5, health status decision and visualization, divides the health status into four categories according to the HI value, including normal, warning, maintenance and emergency shutdown status, and generates an operation and maintenance strategy based on the degradation trend prediction; outputs a multi-dimensional visualization report, including a health heat map, key parameter trend analysis and life loss distribution map.

[0007] Preferably, the high-frequency sensor data includes: temperature parameters: busbar joint, switch contact temperature; electrical parameters: three-phase current effective value, partial discharge pulse frequency; mechanical parameters: opening and closing time, energy storage mechanism pressure; the data is subjected to sliding window mean filtering through the edge computing node, and abnormal data is eliminated using the 3σ criterion.

[0008] Preferably, the low-frequency test data includes: insulation performance test results: polarization index, dielectric loss factor; mechanical property test data: contact wear, operating mechanism lubrication status; chemical test data: SF6 gas purity, dissolved gas content in oil; the data is entered manually or uploaded to a mobile terminal, and compared with historical test data for consistency verification.

[0009] Preferably, the dynamic weight adaptive allocation includes the following steps: Step s2.1, define the benchmark sampling frequency of high-frequency data as f_h, and the benchmark sampling frequency of low-frequency data as f_l; Step s2.2, when the time interval of low-frequency data exceeds the set threshold, start the weight redistribution strategy and gradually increase the weight of high-frequency data to the upper limit value; Step s2.3, introduce the half-life model to dynamically downgrade the calculation contribution of expired data; Step s2.4, according to the historical reliability records of the data, set basic weights for high-frequency and low-frequency data respectively, and dynamically adjust the weight distribution based on the real-time fluctuation characteristics of the data. When the data fluctuation is abnormal, reduce the weight of the corresponding data; Step s2.5, regularly evaluate and optimize the weight distribution strategy, and update the weight distribution model according to the evaluation results.

[0010] Preferably, the dynamic correction of the life benchmark includes the following steps: In step s3.1, the environmental accelerated degradation coefficient is jointly calibrated according to the temperature gradient, humidity saturation and pollution level of the environment in which the equipment is located; in step s3.2, for equipment in highly corrosive environments, an additional salt spray corrosion correction factor is added to shorten the life benchmark update cycle; in step s3.3, an equipment aging model is established, and the model input includes environmental parameters, operating conditions and equipment type. The remaining life of the equipment is predicted based on the equipment aging model and compared with the dynamic life reference curve; in step s3.4, the environmental accelerated degradation coefficient is adjusted according to the comparison results, and the dynamic correction process of the life benchmark is logged for traceability and analysis.

[0011] Preferably, the specific steps of calculating the hierarchical health index include: step s4.1, normalization of the bottom-level indicators, converting each monitoring parameter into a deviation in the interval [0,1]; step s4.2, middle-layer weight allocation, calibrating the indicator weights through the entropy weight method combined with expert experience; step s4.3, top-level aggregation calculation, using weighted summation and fuzzy rule base fusion to output the HI value, and additional trend smoothing processing.

[0012] Preferably, the health status decision and visualization also includes the following steps: step s5.1, setting corresponding thresholds and warning levels according to the health status divided by HI value; step s5.2, when the HI value exceeds the preset threshold, automatically generating an operation and maintenance work order and pushing it to the corresponding operation and maintenance personnel; step s5.3, dynamically updating the health status according to the execution results of the operation and maintenance strategy, and recording and analyzing the health status decision process for optimizing the operation and maintenance strategy.

[0013] Preferably, a real-time anomaly detection and response mechanism is also included, including the following steps: step s6.1, real-time monitoring of key parameters and setting parameter out-of-limit warning thresholds; step s6.2, when key parameters exceed the limit continuously, triggering the online diagnostic mode and increasing the data sampling frequency; step s6.3, starting an emergency assessment process for sudden changes in environmental parameters, quickly responding to abnormal situations, and recording the anomaly detection and response process.

[0014] Preferably, the multi-dimensional visualization report further includes the following modules: The equipment health matrix diagram, with HI values ​​and degradation directions marked by distribution bay unit, provides an intuitive overview of equipment health. The life loss prediction map displays the contribution of mechanical, electrical, and environmental factors to the remaining life, assisting decision-making. The operation and maintenance strategy decision tree recommends inspection, testing, or replacement operations based on HI levels. Report generation templates and custom configuration functions are used to meet the needs of different users.

[0015] Preferably, system performance monitoring and optimization is also included, including the following steps: Step s7.1: Real-time monitoring of the performance indicators of the edge computing device and the cloud operation and maintenance management platform, including CPU usage, memory usage, and network latency. Step s7.2: Dynamically adjust system resource allocation based on performance monitoring data to ensure stable system operation. Step s7.3: Regularly evaluate and optimize the system's performance to improve data processing efficiency and response speed. Step s7.4: Record the system performance monitoring and optimization process.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. In the present invention, the data weight is automatically adjusted according to the timeliness difference of high-frequency and low-frequency data through a dynamic weight adaptive allocation mechanism, which solves the problem that the long-term distortion of health scores may be caused by the lag in low-frequency data updates. The real-time anomaly detection and response mechanism can monitor key parameters in real time and set parameter over-limit warning thresholds. When key parameters exceed the limit continuously, the online diagnosis mode is triggered and the data sampling frequency is increased to ensure that abnormal situations can be quickly discovered and handled, thereby improving the system's ability to cope with sudden failures and reducing the risk of failures.

[0017] 2. In this invention, multi-source data fusion is used to acquire high-frequency sensor data and low-frequency test data in real time using the Internet of Things interface. For high-frequency data, sliding window filtering and the 3σ criterion are used for preprocessing, which can effectively remove noise and outliers and improve the purity and reliability of the data. For low-frequency data, timestamps and credibility tags are added to form a structured data set, which facilitates tracking the temporal characteristics and credibility of the data.

[0018] 3. In the present invention, the environmental accelerated degradation coefficient is calculated and the equipment life is corrected based on the equipment's years of operation and environmental parameters through dynamic correction of the life benchmark to generate a dynamic life reference curve. This process takes into account the impact of the equipment's actual operating environment on its life, such as high temperature, high humidity and other harsh environmental factors, and avoids the problem of large deviation between the reference value and the actual life caused by simple methods such as fixed reduction coefficients, making the prediction of the remaining life of the equipment more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is the overall flow chart of the present invention; Figure 2 This is a flow chart of data collection and preprocessing of the present invention; Figure 3 This is a flow chart of the dynamic weight adaptive allocation of the present invention; Figure 4 This is a flowchart of the stratified health index calculation of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Example 1; please refer to Figure 1-Figure 4 , a grid node health index evaluation method suitable for remote operation and maintenance, including: Step S1: Multi-source data fusion acquisition, real-time acquisition of high-frequency sensor data and low-frequency test data through the Internet of Things interface. High-frequency data includes temperature, current, and mechanical parameters at the second to minute level, and low-frequency data includes insulation performance, mechanical wear, and chemical test results at the month to year level. Sliding window filtering is performed on high-frequency data with a window length of 5-10 sampling periods, and outliers are eliminated based on the 3σ criterion; Implement edge preprocessing on high-frequency data, including filtering, denoising, and outlier removal, and attach timestamps and credibility labels to low-frequency data to form a structured data set. Perform integrity checks on the collected data, interpolate missing data based on data from adjacent time points, and store the processed data in a multi-dimensional data pool according to the preset data structure; Step S2: Dynamic weight adaptive allocation. According to the timeliness difference between high-frequency and low-frequency data, the data weight is adjusted using a dynamic attenuation factor. The weight of high-frequency data increases as the timeliness of low-frequency data decreases. Set a weight allocation threshold. When the low-frequency data exceeds the preset half-life, a data update request is triggered and its confidence level is lowered. Step S3: Dynamically correct the lifespan benchmark. Based on the equipment's operating years and environmental parameters, the environmental accelerated degradation coefficient is calculated and the equipment design lifespan is dynamically corrected. The initial lifespan benchmark is differentiated by equipment type, and a dynamic lifespan reference curve is generated in combination with real-time environmental data. Step S4, stratified health index calculation, constructing a four-layer health indicator system, including physical layer, functional layer, environmental layer and historical layer indicators, and assigning weights to each layer through the hierarchical analysis method; using the fuzzy comprehensive evaluation method to aggregate the indicator deviation layer by layer, output the health index HI on a scale of 0-100, and calculate the confidence interval; Step S5, health status decision and visualization, divides the health status into four categories according to the HI value, including normal, warning, maintenance and emergency shutdown status, and generates an operation and maintenance strategy based on the degradation trend prediction; outputs a multi-dimensional visualization report, including a health heat map, key parameter trend analysis and life loss distribution map.

[0022] High-frequency sensor data includes: temperature parameters: busbar joint and switch contact temperature; electrical parameters: three-phase current effective value, partial discharge pulse frequency; mechanical parameters: opening and closing time, energy storage mechanism pressure; data is filtered using a sliding window mean filter through the edge computing node, and abnormal data is eliminated using the 3σ criterion.

[0023] Low-frequency test data includes: insulation performance test results: polarization index, dielectric loss factor; mechanical property test data: contact wear, operating mechanism lubrication status; chemical test data: SF6 gas purity, dissolved gas content in oil; data is entered manually or uploaded via mobile terminals and compared with historical test data for consistency verification.

[0024] Dynamic weight adaptive allocation includes the following steps: step s2.1, defining the benchmark sampling frequency of high-frequency data as f_h, and the benchmark sampling frequency of low-frequency data as f_l; step s2.2, when the time interval of low-frequency data exceeds the set threshold, starting the weight redistribution strategy, and gradually increasing the weight of high-frequency data to the upper limit value; step s2.3, introducing the half-life model to dynamically downgrade the calculation contribution of expired data; step s2.4, according to the historical reliability records of the data, setting basic weights for high-frequency and low-frequency data respectively, based on the real-time fluctuation characteristics of the data, dynamically adjusting the weight distribution, and when the data fluctuation is abnormal, reducing the weight of the corresponding data; step s2.5, regularly evaluating and optimizing the weight distribution strategy, and updating the weight distribution model according to the evaluation results.

[0025] The dynamic correction of the life benchmark includes the following steps: Step s3.1, the environmental accelerated degradation coefficient is jointly calibrated according to the temperature gradient, humidity saturation and pollution level of the environment in which the equipment is located; Step s3.2, for equipment in highly corrosive environments, an additional salt spray corrosion correction factor is added to shorten the life benchmark update cycle; Step s3.3, an equipment aging model is established, and the model input includes environmental parameters, operating conditions and equipment type, and the remaining life of the equipment is predicted according to the equipment aging model and compared with the dynamic life reference curve; Step s3.4, the environmental accelerated degradation coefficient is adjusted according to the comparison results, and the dynamic correction process of the life benchmark is logged for traceability and analysis.

[0026] The specific steps for calculating the hierarchical health index include: step s4.1, normalization of the bottom-level indicators, converting each monitoring parameter into a deviation in the interval [0,1]; step s4.2, middle-level weight allocation, calibrating the indicator weights through the entropy weight method combined with expert experience; step s4.3, top-level aggregation calculation, using weighted summation and fuzzy rule base fusion to output the HI value, and additional trend smoothing processing.

[0027] Health status decision and visualization also includes the following steps: Step s5.1, setting corresponding thresholds and warning levels based on the health status divided by HI value; Step s5.2, when the HI value exceeds the preset threshold, automatically generating an operation and maintenance work order and pushing it to the corresponding operation and maintenance personnel; Step s5.3, dynamically updating the health status based on the execution results of the operation and maintenance strategy, and recording and analyzing the health status decision process to optimize the operation and maintenance strategy.

[0028] The working principle of the embodiment of the present invention is as follows: in the multi-source data fusion and collection link, the Internet of Things interface is like a sensitive antenna, capturing high-frequency sensor data and low-frequency test data in real time. The high-frequency data covers various key operating parameters at the second to minute level, and the low-frequency data is an important detection indicator at the month to grade level. In addition, differentiated processing strategies are adopted for data of different frequencies. Sliding window filtering is combined with the 3σ criterion to accurately purify high-frequency data, laying a solid data foundation for subsequent processing. Edge preprocessing, integrity checking, and storage operations further improve data quality and availability.

[0029] Dynamic weight adaptive allocation is based on the timeliness difference between high-frequency and low-frequency data, and cleverly uses dynamic attenuation factors to achieve dynamic balance of weights. The dynamic correction of life benchmark is like a precise clock calibration mechanism. According to the equipment's years of operation and environmental parameters, the environmental accelerated degradation coefficient is calculated, and the equipment design life is dynamically corrected. At the same time, the initial life benchmark is set based on the differentiation of equipment types, and a dynamic life reference curve is generated to provide an accurate reference for equipment life assessment and assist in the subsequent health index calculation process to fully consider the equipment life factors.

[0030] The hierarchical health index calculation constructs a four-tier health indicator system. The combined application of the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method achieves layer-by-layer aggregation calculation, from the normalization of bottom-level indicators to the top-level health index. The system outputs a health index (HI) on a scale of 0-100 and its confidence interval, providing strong support for the quantitative assessment of grid node health. Health status decision-making and visualization classifies health status categories based on HI values. Deterioration trend predictions facilitate the generation of precise operation and maintenance strategies. Multi-dimensional visualization reports present complex data in intuitive charts, helping operation and maintenance personnel quickly understand equipment conditions and make informed decisions.

[0031] Example 2: Please refer to Figure 1-Figure 4 In an embodiment of the present invention, a real-time anomaly detection and response mechanism is also included, including the following steps: step s6.1, real-time monitoring of key parameters and setting parameter over-limit warning thresholds; step s6.2, when key parameters exceed the limit continuously, triggering the online diagnosis mode and increasing the data sampling frequency; step s6.3, starting the emergency assessment process for sudden changes in environmental parameters, quickly responding to abnormal situations, and recording the anomaly detection and response process.

[0032] The multi-dimensional visual report also includes the following modules: an equipment health status matrix diagram, which annotates the HI value and degradation direction by distribution bay unit, providing an intuitive overview of equipment health; a life loss prediction map, which is used to display the contribution ratio of mechanical, electrical, and environmental factors to the remaining life, to assist in decision-making; an operation and maintenance strategy decision tree, which recommends inspection, testing, or replacement operations based on the HI level; and report generation templates and custom configuration functions to meet the needs of different users.

[0033] It also includes system performance monitoring and optimization, including the following steps: Step s7.1, real-time monitoring of the performance indicators of edge computing devices and cloud operation and maintenance management platforms, including CPU usage, memory usage and network latency; Step s7.2, dynamically adjusting system resource allocation based on performance monitoring data to ensure stable system operation; Step s7.3, regularly evaluating and optimizing the system performance to improve data processing efficiency and response speed; Step s7.4, recording the system performance monitoring and optimization process.

[0034] The working principle of the embodiment of the present invention is as follows: through multi-source data fusion collection, the Internet of Things interface is used to obtain high-frequency sensor data and low-frequency test data in real time, providing a basis for subsequent evaluation. For high-frequency data, sliding window filtering and the 3σ criterion are used for preprocessing to eliminate outliers; timestamps and credibility tags are added to low-frequency data to form a structured data set, and integrity checks and storage are performed.

[0035] The dynamic correction of the lifespan benchmark calculates the environmental accelerated degradation coefficient based on the equipment's operating age and environmental parameters, dynamically corrects the equipment's design life, and generates a dynamic lifespan reference curve based on real-time environmental data. The stratified health index calculation constructs a four-tier health indicator system, using the analytic hierarchy process and fuzzy comprehensive evaluation method to aggregate indicator deviations layer by layer, outputting a health index (HI) and confidence interval. Health status decision-making and visualization classifies health states based on HI values, generates operation and maintenance strategies, and outputs multidimensional visualization reports, including health heat maps, key parameter trend analysis, and lifespan loss distribution diagrams.

[0036] The real-time anomaly detection and response mechanism monitors key parameters in real time, sets parameter over-limit warning thresholds, and triggers online diagnostic mode and increases data sampling frequency when key parameters exceed the limit continuously. It also initiates an emergency assessment process for sudden changes in environmental parameters, quickly responds to abnormal situations, and records the anomaly detection and response process.

[0037] System performance monitoring and optimization monitors the performance indicators of edge computing devices and cloud operation and maintenance management platforms in real time, dynamically adjusts system resource allocation based on monitoring data, regularly evaluates and optimizes system performance, and improves data processing efficiency and response speed.

[0038] Example 3; please refer to Figure 1-Figure 4 In an embodiment of the present invention, a method for evaluating the health index of a power grid node based on multi-source data fusion and dynamic weight optimization is provided. The specific implementation process is as follows: Distributed fiber-optic temperature measurement devices, wireless current sensors, and vibration sensors collect key operating parameters of power grid nodes in real time, including busbar joint temperature, three-phase current RMS value, partial discharge pulse frequency, opening and closing time, and energy storage mechanism pressure. The temperature threshold is set at 90 degrees Celsius, the three-phase current baseline is 400 amperes with a tolerance of ±5%, and the partial discharge frequency threshold is limited to 50 times per minute. High-frequency data is filtered using a sliding window mean filter at the edge computing node, with a window length of 8 sampling periods. Anomalous data is effectively eliminated based on the 3σ criterion.

[0039] The dynamic weight allocation module adjusts the calculated weights based on the timeliness of high- and low-frequency data. The baseline sampling frequency for high-frequency data is set at 10 Hz, and the low-frequency data is updated monthly. When the low-frequency data's timeliness exceeds the preset six-month half-life, the system automatically initiates a weight redistribution strategy, gradually increasing the weight of high-frequency data from an initial value of 0.7 to an upper limit of 0.9. The weight model is also dynamically adjusted based on the salt spray corrosion level of the equipment's environment, with a corrosion correction factor of 1.2 for coastal areas and 1.0 for inland areas, ensuring accurate weight allocation and environmental adaptability.

[0040] Each indicator is normalized and converted to a deviation between 0 and 1. A fuzzy comprehensive evaluation method is used for aggregation and calculation, resulting in a health index on a scale of 0 to 100. This index is then smoothed with a sliding average window to eliminate short-term fluctuations. The health status decision module categorizes the health index into four levels: normal (a health index above 80), warning (70 to 80), maintenance (60 to 70), and emergency shutdown (below 60).

[0041] The multi-dimensional visualization reporting module generates equipment health heat maps, life loss prediction maps, and operation and maintenance decision trees, intuitively displaying the health status and degradation direction of each distribution bay unit. Mechanical wear contributes 45% to the life loss of substation busbar nodes, electrical aging accounts for 35%, and environmental corrosion affects 20%. Based on this, the system recommends priority strategies such as contact replacement or strengthening insulation monitoring.

[0042] The real-time anomaly detection module sets out-of-limit warning thresholds for key parameters. If the temperature exceeds 90°C three times consecutively or the frequency of partial discharges exceeds 50 times per minute, it triggers online diagnostic mode, increases the sampling frequency to 20 Hz, and simultaneously initiates an emergency assessment process. The system performance monitoring module tracks resource utilization on edge computing nodes and the cloud platform in real time, dynamically optimizing memory allocation and network bandwidth to ensure CPU utilization is below 85% and network latency is less than 50 milliseconds. Through regular evaluation and log analysis, system response time has been reduced from 2.5 seconds to 1.8 seconds.

[0043] The embodiment of the present invention uses multi-source data fusion and dynamic weight optimization technology to achieve a health index assessment error rate of less than 3%, an equipment life prediction accuracy of more than 92%, and an operation and maintenance response efficiency improved by 40%. It is suitable for power grid node health management in complex environments and provides high-precision decision support for remote operation and maintenance.

[0044] Working Principle: The grid node health index assessment method integrates multi-source data collection and utilizes IoT interfaces to acquire high-frequency sensor data and low-frequency test data in real time. High-frequency data is pre-processed using sliding window filtering and the 3σ criterion, while low-frequency data is timestamped and labeled with credibility to form a structured dataset. Dynamic weighting adaptively adjusts weights based on data timeliness to ensure assessment accuracy. Dynamic life benchmark correction is based on the equipment's operating age and environmental parameters. The environmental accelerated degradation factor is calculated and corrected for equipment life to generate a dynamic life reference curve.

[0045] The hierarchical health index calculation constructs a four-tier health indicator system, combining the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation methods to output a health index (HI) and confidence interval. Health status decision-making and visualization classifies status categories based on HI values, generates operation and maintenance strategies, and outputs multidimensional visualization reports. The real-time anomaly detection and response mechanism monitors key parameters, sets warning thresholds, and triggers online diagnosis and emergency assessment processes. System performance monitoring and optimization monitors performance indicators in real time, dynamically adjusts resource allocation, and improves data processing efficiency and response speed. The entire system achieves a complete closed-loop from data collection to health assessment, anomaly response, and operation and maintenance decision-making, ensuring the safe and stable operation of the power grid.

[0046] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A power grid node health index assessment method suitable for remote operation and maintenance, characterized in that: include: Step S1: Multi-source data fusion acquisition, real-time acquisition of high-frequency sensor data and low-frequency test data through the Internet of Things interface. High-frequency data includes temperature, current, and mechanical parameters at the second to minute level, and low-frequency data includes insulation performance, mechanical wear, and chemical test results at the month to year level. Sliding window filtering is performed on high-frequency data with a window length of 5-10 sampling periods, and outliers are eliminated based on the 3σ criterion; Implement edge preprocessing on high-frequency data, including filtering, denoising, and outlier removal, and attach timestamps and credibility labels to low-frequency data to form a structured data set. Perform integrity checks on the collected data, interpolate missing data based on data from adjacent time points, and store the processed data in a multi-dimensional data pool according to the preset data structure; Step S2: Dynamic weight adaptive allocation. According to the timeliness difference between high-frequency and low-frequency data, the data weight is adjusted using a dynamic attenuation factor. The weight of high-frequency data increases as the timeliness of low-frequency data decreases. Set a weight allocation threshold. When the low-frequency data exceeds the preset half-life, a data update request is triggered and its confidence level is lowered. Step S3: Dynamically correct the lifespan benchmark. Based on the equipment's operating years and environmental parameters, the environmental accelerated degradation coefficient is calculated and the equipment design lifespan is dynamically corrected. The initial lifespan benchmark is differentiated by equipment type, and a dynamic lifespan reference curve is generated in combination with real-time environmental data. Step S4, calculating the hierarchical health index, constructing a four-layer health indicator system, including physical layer, functional layer, environmental layer and historical layer indicators, and assigning weights to each layer through the hierarchical analysis method; The fuzzy comprehensive evaluation method is used to aggregate the index deviation layer by layer, output the health index HI on a scale of 0-100, and calculate the confidence interval; Step S5: Health status decision and visualization: the health status is divided into four categories according to the HI value, including normal, warning, maintenance and emergency shutdown status, and the operation and maintenance strategy is generated in combination with the degradation trend prediction; Output multi-dimensional visual reports, including health heat maps, key parameter trend analysis and life loss distribution maps.

2. The method for evaluating the health index of a power grid node suitable for remote operation and maintenance according to claim 1, characterized in that: The high-frequency sensor data includes: Temperature parameters: busbar connector and switch contact temperature; Electrical parameters: three-phase current effective value, partial discharge pulse frequency; Mechanical parameters: opening and closing time, energy storage mechanism pressure; The data is filtered using a sliding window mean filter at the edge computing node, and the 3σ criterion is used to eliminate abnormal data.

3. The power grid node health index evaluation method suitable for remote operation and maintenance according to claim 1 is characterized in that: The low-frequency test data includes: Insulation performance test results: polarization index, dielectric loss factor; Mechanical characteristics test data: contact wear, operating mechanism lubrication status; Chemical test data: SF6 gas purity, dissolved gas content in oil; The data is manually entered or uploaded via a mobile terminal and compared with historical test data for consistency verification.

4. The method for evaluating the health index of a power grid node suitable for remote operation and maintenance according to claim 1, characterized in that: The dynamic weight adaptive allocation comprises the following steps: Step s2.1, define the high-frequency data reference sampling frequency as f_h, and the low-frequency data reference sampling frequency as f_l; Step s2.2: When the low-frequency data time interval exceeds the set threshold, the weight redistribution strategy is activated to gradually increase the high-frequency data weight to the upper limit; Step s2.3, introduce the half-life model to dynamically downgrade the calculated contribution of expired data; Step s2.4: Based on the historical reliability records of the data, basic weights are set for high-frequency and low-frequency data respectively. Based on the real-time fluctuation characteristics of the data, the weight distribution is dynamically adjusted. When the data fluctuation is abnormal, the weight of the corresponding data is reduced; In step s2.5, the weight allocation strategy is evaluated and optimized regularly, and the weight allocation model is updated according to the evaluation results.

5. The power grid node health index evaluation method suitable for remote operation and maintenance according to claim 1 is characterized in that: The dynamic correction of the life benchmark includes the following steps: Step s3.1: The environmental accelerated degradation coefficient is calibrated based on the temperature gradient, humidity saturation, and pollution level of the environment in which the equipment is located. Step s3.2: For equipment in highly corrosive environments, an additional salt spray corrosion correction factor is added to shorten the life benchmark update cycle; Step s3.3: Establish an equipment aging model. The model inputs include environmental parameters, operating conditions, and equipment type. The remaining life of the equipment is predicted based on the equipment aging model and compared with the dynamic life reference curve. Step s3.4, adjust the environmental accelerated degradation coefficient according to the comparison results, and log the dynamic correction process of the life benchmark for traceability and analysis.

6. The power grid node health index evaluation method suitable for remote operation and maintenance according to claim 1 is characterized in that: The specific steps of calculating the stratified health index include: Step s4.1, normalize the underlying indicators and convert each monitoring parameter into a deviation in the [0,1] interval; Step s4.2, middle layer weight allocation, calibrating indicator weights by combining entropy weight method with expert experience; Step s4.3, top-level aggregation calculation, uses weighted summation and fuzzy rule base fusion to output HI value, and additional trend smoothing processing.

7. The method for evaluating the health index of power grid nodes suitable for remote operation and maintenance according to claim 1, characterized in that: The health status decision and visualization further includes the following steps: Step s5.1, setting corresponding thresholds and warning levels according to the health status divided by HI value; Step s5.2: When the HI value exceeds the preset threshold, an operation and maintenance work order is automatically generated and pushed to the corresponding operation and maintenance personnel; In step s5.3, the health status is dynamically updated according to the execution results of the operation and maintenance strategy, and the health status decision-making process is recorded and analyzed for optimizing the operation and maintenance strategy.

8. The method for evaluating the health index of power grid nodes suitable for remote operation and maintenance according to claim 1, characterized in that: It also includes a real-time anomaly detection and response mechanism, including the following steps: Step s6.1: Monitor key parameters in real time and set parameter over-limit warning thresholds; Step s6.2: When the key parameters exceed the limit continuously, the online diagnostic mode is triggered and the data sampling frequency is increased; In step s6.3, an emergency assessment process is initiated for sudden changes in environmental parameters, an abnormal situation is quickly responded to, and the abnormality detection and response process is recorded.

9. The method for evaluating the health index of power grid nodes suitable for remote operation and maintenance according to claim 1, characterized in that: The multi-dimensional visualization report also Includes the following modules: The equipment health matrix diagram, with HI values ​​and degradation directions marked by distribution bay unit, provides an intuitive overview of equipment health. The life loss prediction map displays the contribution of mechanical, electrical, and environmental factors to the remaining life, assisting decision-making. The operation and maintenance strategy decision tree recommends inspection, testing, or replacement operations based on HI levels. Report generation templates and custom configuration functions are used to meet the needs of different users.

10. The power grid node health index evaluation method suitable for remote operation and maintenance according to claim 1, characterized in that: It also includes system performance monitoring and optimization, including the following steps: Step s7.1: Real-time monitoring of performance indicators of the edge computing device and the cloud operation and maintenance management platform, including CPU usage, memory usage, and network latency; Step s7.2: Dynamically adjust system resource allocation based on performance monitoring data to ensure stable system operation; Step s7.3: Regularly evaluate and optimize the system's performance to improve data processing efficiency and response speed; Step s7.4, record the system performance monitoring and optimization process.

Images