A method and system for integrated intelligent detection of a full state of a surge arrester

Abstract

The application discloses a lightning arrester full-state integrated intelligent detection method and system, and belongs to the technical field of intelligent monitoring and diagnosis of power equipment. In view of the problems that the existing lightning arrester detection means is single, the state evaluation is one-sided, and the operating stress and performance degradation cannot be associated, the application proposes a detection scheme covering the full state of the electrical mechanical environment: deploying a multi-source sensor array to synchronously collect full leakage current, vibration and acoustic emission signals, surface and internal temperature fields, and environmental data; constructing a lightning arrester multi-physical field coupling digital twin, deeply fusing real-time sensing data and twin simulation, realizing high-precision extraction of resistive current, internal mechanical anomaly diagnosis and thermal stability evaluation; based on the multi-dimensional evaluation results, using an information fusion decision model, generating a panoramic state evaluation report covering insulation aging degree, mechanical integrity, thermal stability and comprehensive health index, and realizing fault early warning and residual life prediction.

Description

Technical Field

[0001] This invention relates to the field of surge arrester testing technology, specifically to an integrated intelligent testing method and system for surge arresters in all states. Background Technology

[0002] Metal oxide surge arresters (MOAs), with their excellent nonlinear characteristics and current-carrying capacity, have become key equipment for limiting overvoltage and protecting insulation coordination in modern power systems. The operational reliability of MOAs is directly related to the safety and stability of substations and even the entire regional power grid. However, in the process of long-term exposure to power frequency voltage, switching overvoltage, lightning strikes, and environmental stress (pollution, temperature and humidity changes), MOAs inevitably experience performance degradation. This mainly includes: electrical performance degradation due to aging and moisture absorption of the resistor elements, loosening of internal mechanical structures (such as compression springs and insulating rods) or cracking of porcelain bushings, and thermal stability problems such as localized overheating or poor heat dissipation.

[0003] Currently, state detection technologies for MOA can be mainly divided into the following categories, but each of them has significant technical shortcomings: Single electrical characteristic monitoring: The most common method is to assess the moisture and aging status of the MOA by measuring the total leakage current and extracting the resistive current component. However, this method is highly susceptible to interference from system background harmonics and capacitive coupling between adjacent phases, resulting in insufficient accuracy in resistive current extraction. More importantly, it cannot distinguish between overall aging and localized defects (such as deterioration of a single valve plate), nor can it identify purely mechanical faults.

[0004] Infrared thermal imaging temperature measurement: It can detect localized overheating on the surface of surge arresters in a non-contact manner. However, this technology can only provide temperature distribution information and cannot diagnose the specific cause of overheating (whether it is caused by internal valve plate deterioration, moisture, or external contamination). It is also not sensitive to early and minor heating defects and is often only detected when the fault has developed to a more serious stage.

[0005] Periodic power outage tests: such as DC reference voltage tests and leakage current tests. These offline testing methods not only require power outages, are inefficient, and have long cycles, but also cannot reflect the dynamic behavior of surge arresters under actual operating voltage and environmental stress, making them prone to missed or false alarms.

[0006] Discrete sensor acquisition: In existing technologies, vibration, acoustic emission, and temperature sensors are often deployed and analyzed independently, forming "information silos." The lack of a unified platform to deeply correlate and fuse multi-dimensional physical quantities such as electrical, mechanical, and thermal data makes it impossible to accurately diagnose complex faults caused by the combined effects of multiple stresses. For example, a loose mechanical spring may lead to poor valve contact, resulting in localized heating and increased resistive current; this causal chain is easily overlooked under single-parameter detection. Summary of the Invention

[0007] This invention provides an integrated intelligent detection method and system for the entire state of surge arresters, aiming to solve the problem that the one-sidedness and discreteness of existing detection methods lead to low confidence in the state assessment results, making it impossible to achieve comprehensive, accurate, and forward-looking diagnosis and early warning of the health status of MOA.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A fully integrated intelligent detection method for surge arresters, comprising: S1: When the surge arrester is in energized operation, multi-dimensional operating data is collected synchronously through an integrated sensor terminal. The data includes at least: full leakage current timing signal, broadband vibration and acoustic emission signal of the equipment body, temperature field data obtained by infrared thermal imager and distributed fiber optic temperature measurement, and ambient temperature, humidity and pollution data of the installation point. S2: Input the total leakage current signal and the system synchronization voltage signal into the electrical characteristic simulation module of the digital twin. Through adaptive harmonic separation and phase tracking algorithms, the simulation parameters are dynamically corrected and the fundamental wave and harmonic components of the resistive current under the operating voltage are extracted with high precision. S3: Input the vibration and acoustic signals into the mechanical state simulation module of the digital twin. Through time-frequency analysis and pattern recognition, the internal mechanical loosening, porcelain sleeve cracks or abnormal valve plate vibration are identified and located. S4: Input the temperature field data and environmental data into the thermal-fluid coupling simulation module of the digital twin to calculate the steady-state and transient temperature rise of the surge arrester, evaluate its thermal stability, and, in conjunction with the resistive current extracted in step S2, diagnose whether there is local overheating or abnormal heat dissipation. S5: Input the electrical characteristic parameters, mechanical diagnosis results, thermal assessment results, and environmental stress parameters obtained in steps S2-S4 into a multi-source information fusion decision model constructed based on DS evidence theory or Bayesian network. S6: The fusion decision model outputs an integrated comprehensive condition assessment result, which includes: insulation aging condition level, mechanical structure condition level, thermal stability condition level, comprehensive health index, potential failure modes and their confidence levels, and remaining service life prediction based on condition degradation trend.

[0009] Preferably, in step S1, the integrated sensor terminal adopts a composite sensor design, integrating a micro-current sensing coil, a vibration sensor, and an acoustic emission sensor into an insulating housing installed on the grounding lead of the surge arrester; the temperature field data is jointly provided by an infrared thermal imager installed on the surface of the surge arrester and a distributed fiber optic temperature sensor pre-embedded inside the equipment.

[0010] Preferably, in step S2, the electrical characteristic simulation module of the digital twin incorporates a nonlinear resistor VI characteristic model that considers the effects of temperature and aging; the adaptive harmonic separation employs an improved adaptive filtering algorithm based on real-time feedback of system voltage harmonic components to suppress interference from system background harmonics on resistive current extraction.

[0011] Preferably, in step S3, the vibration and acoustic emission signals are jointly analyzed, specifically including: analyzing the energy change of the vibration signal in a specific frequency band (such as 100Hz-1kHz), and spatiotemporally correlating it with the sudden events of the acoustic emission signal, so as to distinguish between external wind vibration and internal mechanical faults.

[0012] Preferably, the multi-source information fusion decision model in step S5 also includes a state degradation trend vector extracted from historical data as its input features; the model learns the contribution weights of features in each dimension under different fault modes through training, thereby achieving dynamic weighted fusion and conflict evidence resolution.

[0013] Furthermore, the present invention also provides an integrated intelligent detection system for all states of a surge arrester, comprising: The multi-dimensional sensing layer consists of the integrated sensor terminal, infrared thermal imager, distributed fiber optic temperature measurement system, and environmental monitoring unit as described in claim 2. Edge computing layer: Data acquisition and preprocessing unit deployed on site, with embedded lightweight digital twin core algorithm, used to perform localized calculations and preliminary diagnosis of steps S2-S4 in real time; The cloud-based intelligent layer includes a high-performance digital twin complete model, a multi-source information fusion decision model, and a big data storage and analysis platform. The edge computing layer uploads the pre-processed feature data and preliminary results to the cloud-based intelligent layer for in-depth analysis, model optimization, comprehensive evaluation, and prediction.

[0014] Preferably, the edge computing layer and the multi-dimensional perception layer communicate via a wireless or wired low-power network; the edge computing layer and the cloud intelligence layer synchronize data and interact with commands via a power wireless private network or a public mobile network; the system supports automatic alarms and maintenance work order pushes based on evaluation results.

[0015] Preferably, the adaptive harmonic separation algorithm in step S2 specifically adopts an adaptive notch filter based on recursive least squares (RLS), using the real-time harmonic detection result of the system voltage as a reference input to dynamically update the filter weights, thereby achieving adaptive tracking and suppression of capacitive fundamental and harmonic components in the leakage current.

[0016] Preferably, the multi-source information fusion decision model in step S5 further includes a state trend prediction module based on a long short-term memory network (LSTM). This module takes the fusion evaluation results of multiple consecutive monitoring cycles as the input sequence and outputs the changing trends of the comprehensive health index and various state levels over a future period of time to assist in the prediction of remaining lifespan.

[0017] Preferably, the cloud-based intelligent layer further includes a model self-evolution module. This module periodically utilizes historical sensing data accumulated in the big data storage and analysis platform and corresponding actual inspection and verification results to retrain offline and update online the multi-physics model parameters of the digital twin and the fusion weights of the multi-source information fusion decision model, so as to achieve continuous iterative optimization of the diagnostic knowledge base.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention is the first in the field of online surge arrester monitoring to achieve synchronous perception and deep fusion of "electrical, mechanical, thermal, and environmental" status information. It overcomes the limitations of single-parameter evaluation, clearly reveals the inherent causal relationships between different fault modes, and greatly improves the depth and accuracy of diagnosis. By constructing a high-fidelity multi-physics digital twin, it visualizes and extrapolates internal processes that are difficult to measure directly in the physical world (such as stress wave propagation and three-dimensional hotspot distribution) in virtual space, making the diagnostic results more interpretable and providing intuitive basis for maintenance personnel to formulate precise maintenance strategies. Utilizing the adaptive correction capability of the digital twin model, it effectively suppresses system background harmonics and temperature drift. The interference of environmental factors such as external vibration on the extraction of core feature parameters ensures the stability and reliability of the technology under complex field conditions. By analyzing the collaborative degradation trend of multi-dimensional state features and constructing a health index model that integrates historical data and current status, it can predict the failure risk and remaining service life of equipment in the next few months earlier and more accurately than traditional threshold alarm methods. The three-layer architecture of this solution takes into account the real-time performance of data processing and the intelligence of deep analysis. Edge computing ensures rapid response to early warnings, cloud computing provides unlimited model iteration and data analysis capabilities, and the maintenance work orders generated by the system are directly connected to the operation and maintenance process, realizing the full-link automation and intelligence of monitoring, diagnosis, early warning, and maintenance. Attached Figure Description

[0019] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0020] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0021] Reference Figure 1As shown, an integrated intelligent detection method for all states of a surge arrester includes: S1 (Full-Domain Synchronous Sensing Step): While the surge arrester is energized, a wideband, multi-physical quantity operational data reflecting its electrical, mechanical, and thermal states is synchronously collected via an integrated sensor terminal. The data includes at least: Electrical quantities: the full leakage current timing signal i(t) with a nanosecond sampling rate, and the synchronous system voltage signal u(t) obtained through an external voltage transformer.

[0022] Mechanical quantities: triaxial vibration acceleration signal a(t) and high-frequency acoustic emission signal ae(t) of the device body covering a wide frequency range (e.g., 0.5Hz to 20kHz).

[0023] Thermal parameters: Two-dimensional temperature field distribution TIR(x,y) on the outer surface of the surge arrester porcelain bushing acquired by an infrared thermal imager, and radial and axial multi-point temperature data T acquired by distributed fiber optic temperature sensors embedded inside the resistor column. DOFS (r,z).

[0024] Environmental quantity: Real-time ambient temperature (T) at the installation point amb Relative humidity (RH) and salt density / ash density data characterizing the degree of soiling.

[0025] S2, (Electrical Characteristic Digital Twin and High-Precision Resistive Current Extraction Steps): The total leakage current signal i(t) and the system synchronization voltage signal u(t) are input to the electrical characteristic simulation module of the cloud-edge collaborative digital twin. This module has a built-in dynamic VI characteristic model of the nonlinear resistive element that considers the effects of temperature and aging process. High-precision resistive current extraction is achieved through the following sub-steps: S2.1: An improved adaptive harmonic separation algorithm (such as an adaptive notch filter based on recursive least squares RLS) is adopted to use the real-time detected system voltage harmonic components as feedback to dynamically track and suppress capacitive current and nonlinear interference components in i(t).

[0026] S2.2: Using a phase tracking algorithm (such as zero-crossing detector combined with phase-locked loop PLL), the phase difference between the fundamental voltage wave and the fundamental resistive current wave is accurately obtained.

[0027] S2.3: Dynamically correct the model parameters of the digital twin (such as the nonlinear coefficient α of the resistor and the threshold voltage V). ref This allows the simulated leakage current output by the twin model to approximate the measured value, ultimately outputting a high-fidelity resistive current fundamental amplitude I. r1 and each harmonic component I r3 ,I r5 .

[0028] S3 (Mechanical State Digital Twin and Anomaly Diagnosis Step): The vibration signal a(t) and the high-frequency acoustic emission signal ae(t) are input into the mechanical state simulation module of the digital twin. This module constructs a mechanical structure dynamic model of the surge arrester porcelain bushing-internal valve plate column-compression spring based on the finite element method. Diagnosis is performed through the following sub-steps: S3.1: Perform joint time-frequency analysis (e.g., wavelet packet transform or short-time Fourier transform) on a(t) and ae(t) to extract characteristic parameters such as spectral energy distribution, energy proportion of specific frequency bands, and acoustic emission event count and energy.

[0029] S3.2: Perform pattern recognition and matching between the extracted features and the fault modes in the twin simulation model library (such as the characteristic frequency of spring loosening being 200-600Hz, and the characteristic acoustic emission waveform of porcelain sleeve crack being a burst-type high-frequency signal).

[0030] S3.3: Perform spatiotemporal correlation analysis on sudden energy changes of vibration signals and sudden events of acoustic emission signals to distinguish between external environmental interference such as wind vibration and electromagnetic noise and internal real mechanical fault sources, and perform spatial location inversion and localization of fault sources.

[0031] S4 (Digital Twin and Evaluation Step for Thermal Stability): The external surface temperature field data T... IR (x,y), internal multi-point temperature T DOFS (r,z) and a thermal-fluid coupling simulation module for the environmental data input digital twin. This module establishes a multiphysics coupling model considering Joule heating inside the resistor, solid-state heat conduction, and natural convection cooling between the outer surface and air. Thermal stability is evaluated through the following sub-steps: S4.1: The resistive current I extracted in step S2 r Using heat source excitation input, the steady-state temperature distribution T inside the surge arrester is calculated. sim (r,z) and transient temperature rise response.

[0032] S4.2: Compare the simulated temperature distribution with the measured T IR (x,y) and T DOFS By comparing each point (r,z), the thermal field residual distribution map is calculated to accurately locate the spatial position of temperature anomalies.

[0033] S4.3: Diagnose the cause of thermal anomalies by combining ambient temperature and historical resistive current trends. If the local hot spot highly coincides with the resistive current harmonic anomaly area, it is diagnosed as "local valve plate deterioration overheating"; if the hot spot is located at the flange connection and is accompanied by abnormal mechanical vibration, it may be diagnosed as "mechanical contact failure overheating".

[0034] S5 (Multi-source information fusion decision step): The electrical characteristic parameters (I) obtained in steps S2-S4 are... r1 ,I r3 The results of mechanical diagnosis (fault type, confidence level, location), thermal assessment (hot spot temperature, thermal field residual), and long-term recorded environmental stress parameters are all input into a multi-source information fusion decision model built on DS evidence theory or dynamic Bayesian network.

[0035] S6 (Comprehensive Assessment and Prediction Output Step): The fusion decision model synthesizes and reasones about evidence from different physical domains that may contain uncertainties and conflicts, ultimately outputting an integrated comprehensive state assessment result. The result includes at least: Insulation aging status rating (e.g., Excellent / Normal / Caution / Critical / Hazardous).

[0036] Mechanical structure condition level (e.g., good / slightly loose / severely loose / risk of cracks).

[0037] Thermal stability status level (e.g., normal / hot spot concern / overheating warning).

[0038] Overall Health Index (HI): A quantified, normalized (0-100%) comprehensive score that reflects the overall health of the equipment.

[0039] Potential failure modes and their confidence levels: For example, "internal pressure spring relaxation (confidence level 85%)", "local valve plate deterioration (confidence level 70%)".

[0040] Remaining useful life prediction: Based on the degradation trend (such as the rate of change of the HI index over time), combined with the Wiener process or grey prediction model trained with historical data of similar equipment, the remaining useful life probability range is given.

[0041] Specifically, in step S1, the integrated sensor terminal adopts a composite intelligent sensor design, integrating a high-sensitivity micro-current sensing coil, a low-noise triaxial MEMS vibration sensor, and a wideband acoustic emission sensor (AE) into an insulating housing with electromagnetic shielding. This housing is fixed to the grounding lead at the bottom of the surge arrester by a snap-fit ​​or clamp, achieving synchronous and co-position measurement and reducing signal transmission paths and interference. In step S2, the electrical characteristic model of the digital twin uses a simplified exponential model I=k⋅Vα for its resistive element VI characteristics, where the nonlinear coefficient αα... The parameters decrease as the degree of aging increases, and the model parameters are continuously optimized and corrected through real-time data. In step S3, the joint analysis of vibration and acoustic emission signals specifically includes feature layer fusion: calculating the cross-correlation function between the envelope spectrum of the vibration signal in the characteristic frequency band (100Hz-1kHz) and the energy release rate (ARR) of the acoustic emission signal to assess the intensity and periodicity of the internal mechanical dynamic "impact". The multi-source information fusion decision model described in step S5 also includes a state trend prediction module based on a long short-term memory network (LSTM) or gated recurrent unit (GRU). This module uses historical fusion evaluation results as input to predict the health status evolution trajectory over a period of time, providing a basis for remaining life prediction.

[0042] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A fully integrated intelligent detection method for surge arresters, characterized in that, Includes the following steps: S1: When the surge arrester is in energized operation, multi-dimensional operating data is collected synchronously through an integrated sensor terminal. The data includes at least: full leakage current timing signal, broadband vibration and acoustic emission signal of the equipment body, temperature field data obtained by infrared thermal imager and distributed fiber optic temperature measurement, and ambient temperature, humidity and pollution data of the installation point. S2: Input the full leakage current signal and the system synchronization voltage signal into the electrical characteristic simulation module of the digital twin. Through the adaptive harmonic separation and phase tracking algorithm, dynamically correct the simulation parameters and extract the fundamental wave and harmonic components of the resistive current under the operating voltage with high precision. S3: Input the vibration and acoustic signals into the mechanical state simulation module of the digital twin, and identify and locate internal mechanical loosening, porcelain sleeve cracks or abnormal valve plate vibration through time-frequency analysis and pattern recognition; S4: Input the temperature field data and environmental data into the thermal-fluid coupling simulation module of the digital twin to calculate the steady-state and transient temperature rise of the surge arrester, evaluate its thermal stability, and, in conjunction with the resistive current extracted in step S2, diagnose whether there is local overheating or abnormal heat dissipation. S5: Input the electrical characteristic parameters, mechanical diagnosis results, thermal assessment results, and environmental stress parameters obtained in steps S2-S4 into a multi-source information fusion decision model constructed based on DS evidence theory or Bayesian network. S6: The fusion decision model outputs an integrated comprehensive condition assessment result, which includes: insulation aging condition level, mechanical structure condition level, thermal stability condition level, comprehensive health index, potential failure modes and their confidence levels, and remaining service life prediction based on condition degradation trend.

2. The integrated intelligent detection method for all states of a surge arrester according to claim 1, characterized in that, In step S1, the integrated sensor terminal adopts a composite sensor design, integrating a micro-current sensing coil, a vibration sensor, and an acoustic emission sensor into an insulating housing installed on the grounding lead of the surge arrester; the temperature field data is jointly provided by an infrared thermal imager installed on the surface of the surge arrester and a distributed fiber optic temperature sensor embedded inside the equipment.

3. The integrated intelligent detection method for all states of a surge arrester according to claim 1, characterized in that, In step S2, the electrical characteristic simulation module of the digital twin incorporates a nonlinear resistor VI characteristic model that considers the effects of temperature and aging; the adaptive harmonic separation uses an improved adaptive filtering algorithm based on real-time feedback of system voltage harmonic components to suppress the interference of system background harmonics on resistive current extraction.

4. The integrated intelligent detection method for all states of a surge arrester according to claim 3, characterized in that, In step S3, the vibration and acoustic emission signals are jointly analyzed, specifically including: analyzing the energy change of the vibration signal in a specific frequency band (such as 100Hz-1kHz) and spatiotemporally correlating it with the sudden events of the acoustic emission signal in order to distinguish between external wind vibration and internal mechanical faults.

5. The integrated intelligent detection method for all states of a surge arrester according to claim 4, characterized in that, The multi-source information fusion decision model in step S5 also includes a state degradation trend vector extracted from historical data as its input features. The model learns the contribution weights of each dimension of features under different fault modes through training, thereby achieving dynamic weighted fusion and conflict evidence resolution.

6. A surge arrester all-state integrated intelligent detection system for implementing the method of any one of claims 1-5, characterized in that, include: The multi-dimensional sensing layer consists of the integrated sensor terminal, infrared thermal imager, distributed fiber optic temperature measurement system, and environmental monitoring unit as described in claim 2. Edge computing layer: Data acquisition and preprocessing unit deployed on site, with embedded lightweight digital twin core algorithm, used to perform localized calculations and preliminary diagnosis of steps S2-S4 in real time; The cloud-based intelligent layer includes a high-performance digital twin complete model, a multi-source information fusion decision model, and a big data storage and analysis platform. The edge computing layer uploads the pre-processed feature data and preliminary results to the cloud-based intelligent layer for in-depth analysis, model optimization, comprehensive evaluation, and prediction.

7. The integrated intelligent detection system for all states of a surge arrester according to claim 1, characterized in that, The edge computing layer and the multi-dimensional perception layer communicate via wireless or wired low-power networks; the edge computing layer and the cloud intelligence layer synchronize data and interact with commands via a private power wireless network or a public mobile network; the system supports automatic alarms and maintenance work order pushes based on evaluation results.

8. The integrated intelligent detection system for all states of a surge arrester according to claim 7, characterized in that, The adaptive harmonic separation algorithm in step S2 specifically employs an adaptive notch filter based on recursive least squares (RLS). It uses the real-time harmonic detection results of the system voltage as a reference input to dynamically update the filter weights, thereby achieving adaptive tracking and suppression of capacitive fundamental and harmonic components in the leakage current.

9. The integrated intelligent detection system for all states of a surge arrester according to claim 1, characterized in that, The multi-source information fusion decision model in step S5 also includes a state trend prediction module based on a long short-term memory network (LSTM). This module takes the fusion evaluation results of multiple consecutive monitoring cycles as the input sequence and outputs the changing trends of the comprehensive health index and various state levels over a future period of time to assist in the prediction of remaining lifespan.

10. The integrated intelligent detection system for all states of a surge arrester according to claim 9, characterized in that, The cloud-based intelligent layer also includes a model self-evolution module. This module periodically uses historical sensing data accumulated in the big data storage and analysis platform and corresponding actual inspection and verification results to retrain the multi-physics model parameters of the digital twin and the fusion weights of the multi-source information fusion decision model offline and update them online, so as to achieve continuous iterative optimization of the diagnostic knowledge base.

Images