A method and system for on-line monitoring and evaluation of the operating state of a surge arrester

Abstract

The application discloses a kind of lightning arrester operating state on-line monitoring and evaluation method and system, it is related to lightning arrester operating state monitoring technical field, the present application includes three steps of collaborative correction, health degree evaluation and local defect identification.Collaborative correction is through multi-sensor network acquisition data, establishes environmental compensation model and optimizes leakage current measurement accuracy, and dynamically adjusts filter to suppress interference signal.Health degree evaluation fuses multidimensional features, adopts dynamic entropy weight method and adaptive membership function to realize accurate state classification, and improves parameter coupling analysis ability by covariance correction.Local defect identification utilizes distributed optical fiber sensing network and digital twin technology, realizes real-time positioning and growth prediction of micro-defects.The present application can effectively improve the accuracy and reliability of lightning arrester operating state monitoring, timely discover potential failure, reduce maintenance cost, and ensure the safe and stable operation of power system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lightning arrester operation state monitoring, in particular to a lightning arrester operation state online monitoring and evaluation method and system. BACKGROUND

[0002] As a key infrastructure for the operation of modern society, the safe and stable operation of the power system is of great importance. As a core protection element in the power system, the lightning arrester bears the heavy responsibility of limiting overvoltage, protecting electrical equipment from lightning impact and operating overvoltage damage. With the continuous expansion of the power grid scale and the increasing complexity of the operating environment, the reliability and stability requirements of the lightning arrester are also increasing. Traditional manual inspection and periodic preventive test methods gradually expose many shortcomings in monitoring the operation state of the lightning arrester, such as untimely detection, inability to monitor in real time, difficulty in detecting internal defects, etc., which cannot meet the needs of modern power systems for efficient and accurate monitoring of lightning arresters.

[0003] The existing lightning arrester online monitoring systems on the market have a series of problems in terms of function and performance. On the one hand, the data acquisition accuracy is not ideal and is easily disturbed by environmental factors. For key electrical characteristic parameters such as leakage current, the measurement results often have large errors and cannot accurately reflect the real operating state of the lightning arrester. On the other hand, the multi-source data fusion analysis capability is lacking. Most systems can only process simple monitoring data of a single type, and cannot organically integrate electrical, mechanical, thermodynamic and environmental multi-dimensional data, so as to comprehensively and deeply evaluate the health status of the lightning arrester, and easily cause misjudgment or omission. In addition, in terms of fault early warning, the existing system's early warning mechanism is simple and rough, and cannot timely and accurately issue a graded early warning signal according to the different fault degrees of the lightning arrester, and cannot provide sufficient and effective decision support for operation and maintenance personnel.

[0004] In order to solve the above-mentioned defects, a technical solution is provided. SUMMARY

[0005] The purpose of the present application is to solve the problems of existing lightning arrester monitoring technology in data acquisition accuracy, multi-dimensional feature fusion, fault early warning mechanism and microscopic defect monitoring, and to propose a lightning arrester operation state online monitoring and evaluation method and system.

[0006] The purpose of the present application can be achieved by the following technical solution:

[0007] A lightning arrester operation state online monitoring and evaluation method, comprising the following steps:

[0008] S1, cooperative correction, collecting data through a multi-sensor network, establishing an environmental compensation model, optimizing the measurement accuracy of the leakage current, and dynamically adjusting the filter to suppress interference signals;

[0009] S2, health assessment, fusion of electrical, mechanical, thermodynamic, environmental multidimensional characteristics, using dynamic entropy weight method and adaptive membership function to realize accurate state classification, and improving parameter coupling analysis ability through covariance correction;

[0010] S3, local defect discrimination, using distributed optical fiber sensing network and digital twin technology to realize real-time positioning and growth prediction of micro defects, and combining edge-cloud collaborative computing to optimize defect classification efficiency.

[0011] Further, the specific process of S1 is as follows:

[0012] S101, distributed sensing network construction: 4 groups of wideband current sensors are arranged in a ring within 0.5m range at the top of the arrester body, forming an anti-interference differential measurement array;

[0013] 5 groups of thin film temperature sensors are embedded along the axis of the arrester umbrella skirt at equal intervals, forming an axial temperature gradient monitoring chain;

[0014] Three-axis acceleration sensors are installed on the base to collect real-time mechanical vibration spectrum of 0.1-2000Hz;

[0015] A miniature weather station is integrated at the top to monitor environmental temperature, humidity, wind speed and air pressure synchronously;

[0016] S102, establish leakage current correction model: , wherein is the corrected leakage current, is the original leakage current, is the temperature compensation coefficient, is the humidity correction factor, T is the real-time temperature, is the rated operating temperature of the device, is the real-time air pressure and standard air pressure respectively, is the air pressure standardization factor, RH is the real-time relative humidity;

[0017] S103, on the basis of the leakage current correction model, add a surface pollution compensation factor, and realize non-inductive pollution estimation through multi-sensor data fusion;

[0018] S104, intelligent filtering processing: using a parameter adaptive finite impulse response filter set, the cutoff frequency is dynamically adjusted according to the real-time environmental temperature, specifically:

[0019] When the temperature rises by 1℃, the cutoff frequency moves up by 0.5Hz;

[0020] When the humidity increases by 10%RH, the stopband attenuation increases by 3dB;

[0021] Used to suppress interference signals in the 20-50kHz frequency band caused by rain and fog.

[0022] Furthermore, the specific operation steps of S103 are as follows:

[0023] Establish a pollution rate prediction equation based on environmental parameters: ,in For surface contamination density, The temperature decay factor, Humidity influence coefficient This is the wind speed suppression coefficient, with an initial value of 0.08. The relative humidity over time τ Let τ be the wind speed. Let be the surface temperature over time τ, and e be the base of the natural logarithm. Let be the integral variable, representing each moment from the start time to the current time t. Represents the time variable Integrate the pollution rate from time 0 to time t;

[0024] Introduce a pollution compensation term into the leakage current correction model: ,in The corrected leakage current. The pollution sensitivity coefficient, Given the current surface contamination density, Critical pollution density, It is the hyperbolic tangent function;

[0025] A sliding window optimizer is deployed on the edge computing node to calculate the pollution compensation effect every 30 minutes. The calculation logic is as follows: ;in To represent the pollution compensation effect, N is the number of data points within the sliding window. Here, is the partial derivative, representing the corrected leakage current. The rate of change of pollution density p Let i be the time interval for the i-th data point;

[0026] when Time-triggered parameter self-tuning: ;in A new pollution sensitivity coefficient; This is the original pollution sensitivity coefficient; A symbolic function used to determine... The positive and negative, if If it is positive, then the value is 1; if If it is 0, then the value is 0. If the number is negative, its value is -1;

[0027] Acquire weather station and temperature sensor data, pollution model output as health degree assessment feature input, compensation result feedback to adaptive filter, when Automatic enhancement 50-100Hz band filter.

[0028] Further, the specific operation steps of S2 are as follows:

[0029] S201, multi-dimensional feature engineering construction, extract state indicators, including:

[0030] Electrical characteristics: resistive current third harmonic distortion rate, discharge pulse repetition rate;

[0031] Mechanical characteristics: energy ratio of base vibration signal in 100-400Hz frequency band;

[0032] Thermodynamic characteristics: product of maximum value and standard deviation of axial temperature gradient;

[0033] Environmental characteristics: weighted comprehensive index of temperature and humidity change rate and wind speed;

[0034] S202, dynamic weight distribution mechanism, calculate index weight in real time by using improved entropy weight method, calculate the dispersion of each characteristic parameter:

[0035] Generate dynamic weight: ; Wherein is the dispersion of the jth characteristic parameter, is the dynamic weight of the jth feature; is the standardized value of the jth feature of the ith sample, n and m are the total number of samples and the total number of characteristics respectively; Make the weight distribution automatically adjust with the device aging process;

[0036] S203, fuzzy decision mechanism, construct S-type membership function: , wherein is the membership of the jth feature, is the sensitivity adjustment factor, the initial value is 2.5, is the historical baseline value of each characteristic quantity; And introduce dynamic membership function parameter adaptive adjustment mechanism;

[0037] By calculating the comprehensive health degree , establish multi-level early warning, wherein S is the comprehensive health degree, is the weight of the jth feature, is the membership of the jth feature;

[0038] When , trigger device emergency isolation, when , notify manual for manual review and decision making, when​ At the same time, maintain the regular monitoring mode.

[0039] Furthermore, the specific steps for introducing the dynamic membership function parameter adaptive adjustment mechanism in S203 are as follows:

[0040] Deploy a sliding window statistics module on the edge computing node to perform statistics on each feature parameter. Calculate the data distribution characteristics within the calculation window: , Dynamically update the membership function threshold: ;in Features The sliding window means and standard deviation; N is the number of data points within the sliding window. Let x be the value of the feature x of the k-th data point. The updated membership function threshold. The threshold value is the original membership function threshold.

[0041] Dual-modal regulation strategy: Steady-state operating condition, i.e., health status change rate < 0.01 / h: ;in These are the new feature weight coefficients, used to adjust the feature weights under steady-state conditions. These are the original feature weight coefficients. A symbolic function used to determine... Positive and negative, For health score S, the characteristics The partial derivatives;

[0042] Transient operating conditions, i.e., health status change rate ≥ 0.01 / h: , The parameter is the increment over the most recent hour;

[0043] A covariance matrix correction mechanism is introduced, and the covariance matrix of the characteristic parameters is calculated every 6 hours. : Where N is the sample size, Let i be the i-th feature value of the κ-th sample. Let be the mean of the i-th feature;

[0044] Modify the membership function: ;in Let be the membership degree of the i-th feature after correction. As a sensitivity adjustment factor, Let i be the covariance between feature i and feature j. For all The features are multiplied together. are the standard deviations of the i-th and j-th features, respectively;

[0045] The entropy weight value calculated by the dynamic weight distribution module is obtained, and when the weight of a certain parameter corresponds to The adjustment step is expanded to 0.15;

[0046] The abnormal working condition automatically triggers the parameter freezing mechanism: when a certain parameter mutation exceeds 3σ, the membership parameter is locked to the previous period value;

[0047] The optimization result is fed back to the artificial early warning, and When the cumulative adjustment exceeds 15%, a parameter calibration work order is generated.

[0048] Further, the specific operation steps of the S3 are as follows:

[0049] S301, optical fiber sensing network deployment, embedding distributed optical fiber array between the valve piece layers in the lightning arrester:

[0050] Φ0.2mm OFDR optical fiber is adopted, which is arranged in axial spiral winding, covering all the valve piece contact surfaces; FBG grating sensors are set every 10cm to form a temperature-strain dual-parameter sensing chain;

[0051] A miniaturized edge computing node is integrated at the top fitting, with an optical fiber demodulation module built-in, and a sampling rate of 2kHz;

[0052] S302, multi-physical field coupling signal analysis, constructing a temperature-strain joint decoupling model: , wherein is the temperature sensitivity coefficient, is the strain-temperature cross-influence factor, is the strain sensitivity coefficient, is the current wavelength, is the reference wavelength; is the temperature change, is the wavelength change, and L is the optical fiber length; is the temperature coefficient;

[0053] S303, real-time diagnosis on the edge side, using a lightweight defect recognition model for diagnosis, specifically through an improved one-dimensional deep separable convolution, the convolution kernel width is 5, the channel number is 64, the axial abnormal signal is extracted, the Squeeze-and-Excitation module is introduced in the spatial dimension, and the defect sensitive area is weighted and focused. Random forest algorithm is used to realize defect classification;

[0054] S304, cooperative optimization mechanism, dynamic task allocation strategy: the edge node processes more than 80% of the regular data, and the time delay requirement is <50ms; the cloud platform centrally analyzes complex working condition data, including damage assessment after lightning impulse;

[0055] Synchronize edge node model parameters every week, incremental learning data volume ≥10000 groups, when new defect patterns are detected, automatically trigger federal learning process, participating nodes ≥50;

[0056] S305, three-dimensional defect visualization, build lightning arrester digital twin: map optical fiber monitoring data to three-dimensional model, use HSV color space to encode temperature-strain composite index:

[0057] Red channel: temperature anomaly degree, 0-120℃ is mapped to 0-255;

[0058] Blue channel: strain over-standard rate, 0-2000με is mapped to 0-255;

[0059] Defect growth simulation: based on phase field method to predict crack propagation path, when strain exceeds 80% of material yield strength, automatically generate maintenance coordinates.

[0060] A lightning arrester operation state online monitoring and evaluation system, comprising:

[0061] Multi-source data acquisition module, for integrating wideband current sensor, temperature gradient monitoring chain, three-axis acceleration sensor and micro weather station, real-time acquisition of lightning arrester electrical, mechanical, thermodynamic and environmental multidimensional data;

[0062] Environment compensation and filtering module, for dynamic interference suppression through leakage current correction model and pollution compensation algorithm combined with adaptive filter, to improve data reliability;

[0063] Dynamic feature evaluation module, for extracting five types of features, calculating dynamic weight by improved entropy weight method, and generating comprehensive health degree score through covariance correction and parameter adaptive adjustment;

[0064] Collaborative diagnosis module, through edge node processing of regular data, cloud platform analysis of complex working conditions; capture micro defects by distributed optical fiber array, combined with lightweight deep learning model and federal learning to realize high-precision defect classification and positioning;

[0065] Digital twin decision module, build lightning arrester three-dimensional digital twin, encode temperature-strain data through HSV color space, predict crack propagation path based on phase field method, automatically generate maintenance coordinates, and trigger multi-level early warning strategy.

[0066] Compared with the prior art, the beneficial effects of the present application are:

[0067] (1) The present application effectively improves the measurement accuracy and anti-interference ability of the operating state monitoring data of the lightning arrester by means of multi-sensor cooperative correction and dynamic filtering technology. Multi-dimensional data is collected by a multi-sensor network, an environmental compensation model is established to optimize the measurement accuracy of leakage current, and a parameter-adaptive filter set is used to dynamically adjust in real time according to the environment to suppress interference signals. The problems of low data acquisition accuracy and susceptibility to environmental interference of existing monitoring systems are solved, and high-quality data basis is provided for subsequent health state evaluation;

[0068] (2) The present application adopts dynamic entropy weight method and adaptive membership function, fuses multi-dimensional features to realize accurate grading evaluation of the health state of the lightning arrester, and improves the parameter coupling analysis ability through covariance correction; the dynamic weight is calculated by extracting the state index, so that the weight distribution is automatically adjusted with the aging of the equipment; the constructed membership function is dynamically optimized according to real-time data, and a multi-level early warning strategy is established; the operation and maintenance personnel can master the health state of the lightning arrester in real time, discover potential faults in advance, and effectively improve the reliability of the operation of the lightning arrester;

[0069] (3) The present application realizes real-time positioning and growth prediction of micro-defects in the lightning arrester by means of distributed optical fiber sensing network and digital twin technology, combines edge-cloud collaborative computing to optimize defect classification efficiency; a distributed optical fiber array is embedded in the lightning arrester to form a temperature-strain dual-parameter sensing chain; through a lightweight defect recognition model and a federated learning algorithm, the edge node processes regular data, the cloud platform analyzes complex working conditions, and high-precision defect classification and positioning are realized; at the same time, a three-dimensional digital twin is constructed, the crack propagation path is predicted based on the phase field method, and the maintenance coordinates are automatically generated. These technologies significantly improve the monitoring and prediction ability of micro-defects in the lightning arrester, and provide a scientific basis for equipment maintenance and repair. BRIEF DESCRIPTION OF DRAWINGS

[0070] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings;

[0071] Figure 1 The method flowchart of the present application. DETAILED DESCRIPTION

[0072] The technical solutions of the present application will be described below in conjunction with the embodiments, obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0073] It should be understood that the terms "comprises" and "comprising," when used in this specification and claims, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0074] It should also be understood that the terms used in the specification and claims herein are for the purpose of describing particular embodiments and are not intended to be limiting of the disclosure. As used in this specification and in the claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It should also be further understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0075] As shown in FIG. 1, a method for on-line monitoring and evaluation of the operating state of a lightning arrester includes the following steps: Figure 1

[0076] Step one, coordinated correction, collect data through a multi-sensor network, establish an environmental compensation model (temperature and humidity, air pressure, pollution compensation), optimize leakage current measurement accuracy, and dynamically adjust the filter to suppress interference signals.

[0077] Distributed sensing network construction: 4 groups of wideband current sensors (100 kHz-10 MHz frequency response) are arranged in a ring within 0.5 m of the top of the lightning arrester body, forming an anti-interference differential measurement array; 5 groups of thin film temperature sensors (resolution 0.1 °C) are embedded equidistantly along the axis of the lightning arrester skirt, forming an axial temperature gradient monitoring chain; a three-axis acceleration sensor (range ±20g) is installed at the base, collecting real-time mechanical vibration spectrum of 0.1-2000 Hz; a miniature weather station is integrated at the top, synchronously monitoring environmental temperature and humidity (accuracy ±1% RH), wind speed (range 0-60 m / s), and air pressure (300-1100 hPa);

[0078] Establish a leakage current correction model: wherein is the corrected leakage current, is the original leakage current, is the temperature compensation coefficient (0.0042 / °C for silicon rubber material), is the humidity correction factor (0.018), T is the real-time temperature, is the rated operating temperature of the device, and is set to 40 °C, is the air pressure standardization factor, is the real-time air pressure and standard air pressure, respectively; RH is the real-time relative humidity;

[0079] ​On the basis of the leakage current correction model, a surface contamination compensation factor is added, and non-inductive contamination estimation is realized through multi-sensor data fusion. The specific steps are as follows:

[0080] The pollution rate prediction equation based on environmental parameters is established: , wherein is the surface contamination density, is the temperature attenuation factor, and the initial value is 0.03 / ℃; is the humidity influence coefficient, and the initial value is 0.15, is the wind speed inhibition coefficient, and the initial value is 0.08, is the relative humidity of time τ, is the wind speed of time τ, is the surface temperature of time τ; e is the base of natural logarithm, is the integral variable, indicating each time from the start time to the current time t, used for cumulative calculation of contamination density; indicates the integral of the time variable from time 0 to time t, which is used for cumulative calculation of pollution rate;

[0081] A pollution compensation term is introduced in the leakage current correction model: , wherein is the corrected leakage current, is the pollution sensitivity coefficient, and the value is 0.12, is the current surface contamination density, is the critical contamination density, , is the hyperbolic tangent function, which realizes nonlinear compensation and avoids over-correction;

[0082] A sliding window optimizer is deployed on the edge computing node, and the pollution compensation effect is calculated every 30 minutes. The calculation logic is: ; wherein is the pollution compensation effect, which is used to measure the influence degree of pollution compensation on leakage current correction, N is the number of data points in the sliding window, is the partial derivative, indicating the rate of the corrected leakage current changing with the pollution density p, reflecting the influence degree of the change of pollution density on the leakage current; is the time interval of the i-th data point;

[0083] When , the parameter self-tuning is triggered: ; wherein is the new pollution sensitivity coefficient, which is the pollution sensitivity coefficient after self-tuning according to the pollution compensation effect , which is used to update the sensitivity parameter in the pollution compensation model; This is the original pollution sensitivity coefficient; A symbolic function used to determine... The positive and negative, if If it is positive, then the value is 1; if If it is 0, then the value is 0. If the number is negative, its value is -1;

[0084] Data from weather stations (temperature, humidity, wind speed) and temperature sensors is acquired. The pollution model output is used as the feature input for health assessment, and the compensation result is fed back to the adaptive filter. Automatically enhances filtering in the 50-100Hz frequency band.

[0085] Intelligent filtering: It adopts a parameter adaptive finite impulse response filter bank, whose cutoff frequency is dynamically adjusted according to the real-time ambient temperature: for every 1°C increase in temperature, the cutoff frequency moves up by 0.5Hz; for every 10%RH increase in humidity, the stopband attenuation increases by 3dB; and it suppresses interference signals in the 20-50kHz frequency band caused by rain and fog.

[0086] Step 2: Health assessment, integrating multi-dimensional features (electrical, mechanical, thermodynamic, environmental), using dynamic entropy weight method and adaptive membership function to achieve accurate state classification, and improving parameter coupling analysis capability through covariance correction;

[0087] Extract status indicators, including:

[0088] Electrical characteristics: resistive current third harmonic distortion rate, partial discharge pulse repetition rate; mechanical characteristics: energy proportion of base vibration signal in the 100-400Hz frequency band; thermodynamic characteristics: product of maximum axial temperature gradient and standard deviation; environmental characteristics: weighted composite index of temperature and humidity change rate and wind speed.

[0089] The dynamic weight allocation mechanism calculates the index weights in real time using an improved entropy weight method, and also calculates the dispersion of each feature parameter. Generate dynamic weights: ,in Let be the dispersion of the j-th feature parameter. The dynamic weight of the j-th feature; Let be the standardized value of the j-th feature of the i-th sample, and n and m be the total number of samples and the total number of features, respectively; the weight allocation is automatically adjusted as the equipment ages, and the weight coefficient is automatically increased by 35%-60% when the temperature gradient increases abnormally.

[0090] Fuzzy decision-making mechanism, constructing an S-shaped membership function: ,in Let j be the membership degree of the j-th feature. This is the sensitivity adjustment factor, initially set to 2.5. is the historical baseline value of each feature;

[0091] A dynamic membership function parameter adaptive adjustment mechanism is introduced, and the process is as follows:

[0092] A sliding window statistical module is deployed on the edge computing node (window size = 24 hours, step = 1 hour), and the data distribution characteristics in the window are calculated for each feature parameter , , , the membership function threshold is dynamically updated: ; wherein are the sliding window mean and sliding window standard deviation of the feature respectively; N is the number of data points in the sliding window, is the value of the feature x of the kth data point, is the updated membership function threshold, is the original membership function threshold;

[0093] Dual-mode adjustment strategy: steady state condition (health rate change <0.01 / h): ; wherein is the new feature weight coefficient, used to adjust the weight of the feature in the steady state condition, is the original feature weight coefficient, is a sign function, used to determine the positive and negative of , is the partial derivative of the health S to the feature ; transient state condition (health rate change ≥0.01 / h): , is the parameter increment in the last 1 hour;

[0094] A covariance matrix correction mechanism is introduced, and the feature parameter covariance matrix is calculated every 6 hours: ; wherein N is the number of samples, is the value of the ith feature of the κth sample, is the mean value of the ith feature;

[0095] Correct the membership function: ; wherein is the corrected membership of the ith feature, is the sensitivity adjustment factor, is the covariance between feature i and feature j, is the product operation of all features, are the standard deviations of the ith and jth features respectively;

[0096] The entropy weight value calculated by the dynamic weight allocation module is obtained When the weight of a certain parameter Adjust the step size to 0.15; Abnormal working condition automatically triggers parameter freezing mechanism: when a certain parameter mutation exceeds 3σ, its membership parameter is locked to the previous period value; The optimization result is fed back to the artificial early warning, and When the cumulative adjustment exceeds 15%, generate a parameter calibration work order.

[0097] By calculating the comprehensive health degree , a multi-level early warning is established, where S is the comprehensive health degree, is the weight of the jth feature, is the membership degree of the jth feature; When , trigger the emergency isolation of the device, when , notify the artificial to make artificial decision-making, when , keep the normal monitoring mode.

[0098] Step three, local defect identification, using distributed optical fiber sensing network and digital twin technology, realize real-time positioning and growth prediction of micro defects, combined with edge-cloud collaborative computing to optimize defect classification efficiency;

[0099] Embed a distributed optical fiber array between the valve plate layers inside the lightning arrester:

[0100] Φ0.2mm high-temperature-resistant OFDR optical fiber (spatial resolution 1cm) is adopted, which is spirally wound along the axial direction, covering the entire valve plate contact surface; FBG grating sensors (wavelength accuracy ±5pm) are set every 10cm to form a temperature-strain dual-parameter sensing chain; A miniaturized edge computing node (computing power ≥4TOPS) is integrated at the top fitting, with an optical fiber demodulation module (sampling rate 2kHz) built-in;

[0101] Construct a temperature-strain joint decoupling model: , where is the temperature sensitivity coefficient, the initial value is 10.8pm / ℃, is the strain-temperature cross-influence factor, the initial value is , is the strain sensitivity coefficient, the initial value is ; is the current wavelength, which represents the wavelength of the fiber grating sensor under the current environment, is the reference wavelength, which represents the wavelength of the fiber grating sensor under standard conditions; is the temperature change, is the wavelength change, L is the length of the optical fiber; is the temperature coefficient, which represents the thermal expansion coefficient of the material;

[0102] Diagnosis is performed using a lightweight defect recognition model. Specifically, an improved one-dimensional deep separable convolution (convolution kernel width = 5, channel number = 64) is used to extract axial abnormal signals. A Squeeze-and-Excitation module is introduced in the spatial dimension to weight and focus on defect-sensitive areas. A random forest algorithm (tree depth = 8, feature subset ratio = √d) is used to realize defect classification.

[0103] Dynamic task allocation strategy: edge nodes handle more than 80% of regular data (time delay requirement < 50 ms), and cloud platforms analyze complex working condition data (such as damage assessment after lightning impact); edge node model parameters are updated weekly (incremental learning data volume ≥ 10,000 groups), and when a new defect mode is detected, a federal learning process is automatically triggered (participating nodes ≥ 50);

[0104] Construction of a lightning arrester digital twin: mapping fiber monitoring data to a three-dimensional model (grid accuracy ≤ 1 mm³) and using HSV color space to encode temperature-strain composite indicators:

[0105] Red channel: temperature abnormality (0-120℃ mapped to 0-255); Blue channel: strain over-limit rate (0-2000με mapped to 0-255); Defect growth simulation: crack propagation path prediction based on phase field method (time step = 1h), when local strain exceeds 80% of material yield strength, automatically generate repair coordinates (latitude and longitude error < 5cm).

[0106] A lightning arrester operation state online monitoring and evaluation system, including a multi-source data acquisition module, an environment compensation and filtering module, a dynamic feature evaluation module, a collaborative diagnosis module, and a digital twin decision module:

[0107] The multi-source data acquisition module integrates wideband current sensors, temperature gradient monitoring chains, three-axis acceleration sensors, and micro weather stations to collect real-time electrical, mechanical, thermodynamic, and environmental multi-dimensional data of lightning arresters, providing original signals for subsequent analysis;

[0108] The environment compensation and filtering module uses a leakage current correction model (temperature and humidity / pressure compensation) and a pollution compensation algorithm (hyperbolic tangent nonlinear correction) to dynamically suppress interference (such as 20-50kHz frequency band noise in rainy and foggy weather) using an adaptive filter, improving data reliability;

[0109] The dynamic feature evaluation module extracts five types of features such as resistive current harmonic distortion rate and vibration energy proportion. An improved entropy weight method is used to calculate dynamic weights (such as temperature abnormality weight increase of 35%-60%), and covariance correction and parameter adaptive adjustment are used to generate a comprehensive health score;

[0110] The cooperative diagnosis module edge node processes regular data (time delay < 50 ms), and the cloud platform analyzes complex working conditions (such as lightning damage); micro defects are captured by using a distributed optical fiber array (OFDR+FBG), and high-precision defect classification and positioning are realized by combining a lightweight deep learning model (such as a deep separable convolution) and federated learning;

[0111] The digital twin decision module constructs a three-dimensional digital twin of the lightning arrester, encodes temperature-strain data in an HSV color space, predicts a crack propagation path based on a phase field method, automatically generates maintenance coordinates (error < 5 cm), and triggers a multi-level early warning strategy (emergency isolation, manual review, and regular monitoring).

[0112] The preferred embodiments disclosed above are only used to help explain the present application. The preferred embodiments do not describe all the details and limit the present application to the specific embodiments. Obviously, many modifications and variations can be made according to the content of the present application. The present application is selected and described in detail to better explain the principles and practical applications of the present application, so that those skilled in the art can well understand and use the present application. The present application is limited by the claims and their full scope and equivalents.

Claims

1. A method for online monitoring and evaluation of the operating status of surge arresters, characterized in that, Includes the following steps: S1. Collaborative calibration: Data is collected through a multi-sensor network to establish an environmental compensation model, optimize the accuracy of leakage current measurement, and dynamically adjust the filter to suppress interference signals. S2. Health assessment integrates multiple dimensions of electrical, mechanical, thermodynamic, and environmental characteristics. It uses dynamic entropy weighting and adaptive membership functions to achieve accurate state classification and improves parameter coupling analysis capabilities through covariance correction. S3. Local defect identification: Utilizing distributed fiber optic sensor networks and digital twin technology, real-time location and growth prediction of micro-defects are achieved, and edge-cloud collaborative computing is combined to optimize defect classification efficiency. The specific process of S1 is as follows: S101, Distributed sensing network construction: Four sets of broadband current sensors are arranged in a ring within 0.5m of the top of the surge arrester body to form an anti-interference differential measurement array; Five sets of thin-film temperature sensors are equidistantly embedded along the axial direction of the surge arrester skirt to form an axial temperature gradient monitoring chain. A triaxial accelerometer is mounted on the base to collect mechanical vibration spectra in real time from 0.1 to 2000 Hz; The top integrates a mini weather station to simultaneously monitor ambient temperature, humidity, wind speed, and air pressure; S102. Establish a leakage current correction model: ,in The corrected leakage current. The original leakage current, This is the temperature compensation coefficient. Here is the humidity correction factor, and T is the real-time temperature. The rated operating temperature of the equipment. These are real-time air pressure and standard air pressure, respectively. Here, RH is the pressure normalization factor, and RH is the real-time relative humidity. S103. Based on the leakage current correction model, a surface contamination compensation factor is added, and non-sensory contamination degree estimation is achieved through multi-sensor data fusion. S104, Intelligent Filtering: Employs a parameter-adaptive finite impulse response filter bank, with the cutoff frequency dynamically adjusted based on the real-time ambient temperature. Specifically: For every 1°C increase in temperature, the cutoff frequency shifts upward by 0.5 Hz; For every 10% increase in RH humidity, the stopband attenuation increases by 3dB. Used to suppress interference signals in the 20-50kHz frequency band caused by rain and fog.

2. The method for online monitoring and evaluation of the operating status of a surge arrester according to claim 1, characterized in that, The specific operation steps of S103 are as follows: Establish a pollution rate prediction equation based on environmental parameters: ,in For surface contamination density, The temperature decay factor, Humidity influence coefficient This is the wind speed suppression coefficient, with an initial value of 0.

08. The relative humidity over time τ Let τ be the wind speed. Let be the surface temperature over time τ, and e be the base of the natural logarithm. Let be the integral variable, representing each moment from the start time to the current time t. Represents the time variable Integrate the pollution rate from time 0 to time t; Introduce a pollution compensation term into the leakage current correction model: ,in The corrected leakage current. The pollution sensitivity coefficient, Given the current surface contamination density, Critical pollution density, It is the hyperbolic tangent function; A sliding window optimizer is deployed on the edge computing node to calculate the pollution compensation effect every 30 minutes. The calculation logic is as follows: ;in To represent the pollution compensation effect, N is the number of data points within the sliding window. Here, is the partial derivative, representing the corrected leakage current. The rate of change of pollution density p Let i be the time interval for the i-th data point; when Time-triggered parameter self-tuning: ;in A new pollution sensitivity coefficient; This is the original pollution sensitivity coefficient; A symbolic function used to determine... The positive and negative, if If it is positive, then the value is 1; if If it is 0, then the value is 0. If the number is negative, its value is -1; Data from weather stations and temperature sensors is acquired; the pollution model output is used as the feature input for health assessment; and the compensation result is fed back to the adaptive filter. Automatically enhances filtering in the 50-100Hz frequency band.

3. The method for online monitoring and evaluation of the operating status of a surge arrester according to claim 1, characterized in that, The specific operation steps of S2 are as follows: S201. Multi-dimensional feature engineering construction, extraction of state indicators, including: Electrical characteristics: resistive current third harmonic distortion rate, discharge pulse repetition rate; Mechanical characteristics: the proportion of energy in the 100-400Hz frequency band of the base vibration signal; Thermodynamic characteristics: The product of the maximum axial temperature gradient and its standard deviation; Environmental characteristics: a weighted composite index of temperature and humidity change rate and wind speed; S202, Dynamic weight allocation mechanism: This mechanism calculates the index weights in real time using the dynamic entropy weight method, and also calculates the dispersion of each feature parameter. ; Generate dynamic weights: ;in Let be the dispersion of the j-th feature parameter. The dynamic weight of the j-th feature; Let be the standardized value of the j-th feature of the i-th sample, and n and m be the total number of samples and the total number of features, respectively; so that the weight allocation is automatically adjusted as the equipment ages. S203, Fuzzy decision-making mechanism, constructing an S-shaped membership function: ,in Let j be the membership degree of the j-th feature. This is the sensitivity adjustment factor, initially set to 2.

5. The historical baseline values ​​for each feature quantity are used; and a dynamic membership function parameter adaptive adjustment mechanism is introduced. Calculate overall health score Establish a multi-level early warning system, where S represents the overall health level. The weight of the j-th feature is... Let be the membership degree of the j-th feature; when When this triggers emergency isolation of the device, When this happens, notify a human to conduct a manual review and decision-making process. At the same time, maintain the regular monitoring mode.

4. The method for online monitoring and evaluation of the operating status of a surge arrester according to claim 3, characterized in that, The specific steps for introducing the adaptive adjustment mechanism of dynamic membership function parameters in S203 are as follows: Deploy a sliding window statistics module on the edge computing node to perform statistics on each feature parameter. Calculate the data distribution characteristics within the calculation window: , Dynamically update the membership function threshold: ;in Features The mean and standard deviation of the sliding window; N is the number of data points within the sliding window. Let x be the value of the feature x of the k-th data point. The updated membership function threshold. The threshold value is the original membership function threshold. Dual-modal regulation strategy: Steady-state operating condition, i.e., health status change rate < 0.01 / h: ;in These are the new feature weight coefficients, used to adjust the feature weights under steady-state conditions. These are the original feature weight coefficients. A symbolic function used to determine... Positive and negative, For health score S, the characteristics The partial derivatives; Transient operating conditions, i.e., health status change rate ≥ 0.01 / h: , The parameter is the increment over the most recent hour; A covariance matrix correction mechanism is introduced, and the covariance matrix of the characteristic parameters is calculated every 6 hours. : Where N is the sample size, Let i be the i-th feature value of the κ-th sample. Let be the mean of the i-th feature; Modify the membership function: ;in Let be the membership degree of the i-th feature after correction. As a sensitivity adjustment factor, Let i be the covariance between feature i and feature j. For all The features are multiplied together. are the standard deviations of the i-th and j-th features, respectively; Obtain the entropy weight calculated by the dynamic weight allocation module, when a certain parameter weight... When, its corresponding The adjustment step size was increased to 0.15; An automatic parameter freeze mechanism is triggered under abnormal operating conditions: when a parameter changes abruptly by more than 3σ, its membership parameter is locked to the value of the previous time period. The optimization results are fed back to manual alerts when When the cumulative adjustment exceeds 15%, a parameter calibration work order is generated.

5. The method for online monitoring and evaluation of the operating status of a surge arrester according to claim 1, characterized in that, The specific operation steps of S3 are as follows: S301, Fiber optic sensor network deployment: A distributed fiber optic array is embedded between the valve plate layers inside the surge arrester. Φ0.2mm OFDR optical fiber is used, spirally wound along the axis to cover the entire valve plate contact surface; FBG grating sensors are set at 10cm intervals to form a temperature-strain dual-parameter sensing chain; A miniaturized edge computing node is integrated at the top fitting, with a built-in fiber demodulation module and a sampling rate of 2kHz. S302. Analysis of multi-physics coupled signals, construction of a temperature-strain joint decoupling model: ,in This is the temperature sensitivity coefficient. The strain-temperature cross-influence factor, The strain sensitivity coefficient, For the current wavelength, Reference wavelength; The change in temperature The wavelength change is L, and the fiber length is L. Temperature coefficient; S303, Real-time edge-side diagnosis, uses a lightweight defect recognition model for diagnosis. Specifically, it uses an improved one-dimensional depthwise separable convolution with a kernel width of 5 and 64 channels to extract axial abnormal signals. A Squeeze-and-Excitation module is introduced in the spatial dimension to focus on defect-sensitive areas with weights, and a random forest algorithm is used to classify defects. S304, Collaborative optimization mechanism, dynamic task allocation strategy: edge nodes process more than 80% of routine data with a latency requirement of <50ms, while the cloud platform centrally analyzes complex working condition data, including damage assessment after lightning strikes. The edge node model parameters are updated weekly, with an incremental learning data volume of ≥10,000 sets. When a new defect pattern is detected, the federated learning process is automatically triggered, involving ≥50 nodes. S305, 3D defect visualization, constructing a digital twin of the surge arrester: mapping fiber optic monitoring data to a 3D model, and using the HSV color space to encode temperature-strain composite indicators: Red channel: Temperature anomaly, 0-120℃ is mapped to 0-255; Blue channel: Strain exceedance rate, 0-2000με mapped to 0-255; Defect growth simulation: Based on the phase field method, crack propagation path is predicted, and repair coordinates are automatically generated when the strain exceeds 80% of the material's yield strength.

6. A system based on the online monitoring and evaluation method for the operating status of a surge arrester according to any one of claims 1-5, characterized in that, include: The multi-source data acquisition module integrates a wideband current sensor, a temperature gradient monitoring chain, a triaxial accelerometer, and a miniature weather station to collect real-time electrical, mechanical, thermodynamic, and environmental data of the surge arrester. The environmental compensation and filtering module is used to dynamically suppress interference and improve data reliability by combining a leakage current correction model and a pollution compensation algorithm with an adaptive filter. The dynamic feature evaluation module is used to extract multi-dimensional features of electrical, mechanical, thermodynamic and environmental aspects. It uses the dynamic entropy weight method to calculate dynamic weights and generates a comprehensive health score through covariance correction and parameter adaptive adjustment. The collaborative diagnostic module processes routine data through edge nodes and analyzes complex operating conditions through the cloud platform; it uses a distributed fiber optic array to capture microscopic defects and combines a lightweight deep learning model and federated learning to achieve high-precision defect classification and localization. The digital twin decision-making module constructs a three-dimensional digital twin of the surge arrester, encodes temperature-strain data using the HSV color space, predicts crack propagation paths based on the phase field method, automatically generates maintenance coordinates, and triggers multi-level early warning strategies.

Images