Intelligent lightning arrester for line and state diagnosis method thereof

Abstract

The application provides a line intelligent lightning arrester, comprising: a lightning arrester body and an intelligent control box; the lightning arrester body is used for providing a low-impedance ground channel for overcurrent when overvoltage caused by lightning or switch operation occurs on a line, thereby limiting the overvoltage below a safe level, protecting subsequent electrical equipment from being damaged, and restoring to an insulation state after the overvoltage disappears, so as to ensure normal operation of a power system; the intelligent control box comprises an embedded intelligent sensing module and a diagnosis module; the embedded intelligent sensing module is used for collecting running state parameters of the lightning arrester body in real time; and the diagnosis module is used for evaluating an insulation state of the lightning arrester body by using a partial discharge mode recognition model and predicting an aging trend of a valve unit of the lightning arrester body by using a valve aging trend prediction model according to the running state parameters of the lightning arrester body; and the lightning arrester has the dual advantages of compact structure and high-performance protection.

Description

Technical Field

[0001] This invention relates to the field of surge arrester technology, specifically to an intelligent surge arrester for power lines and its condition diagnosis method. Background Technology

[0002] Metal oxide surge arresters are key devices for protecting power system equipment from damage caused by lightning overvoltage and switching overvoltage. With the development of ultra-high voltage and compact transmission lines, higher requirements are placed on the performance of surge arresters: on the one hand, they are required to have higher protection characteristics (i.e., higher nonlinear coefficient and gradient) to reduce size and save space in the line corridor; on the other hand, they are required to have condition self-sensing capabilities to realize the transformation from "periodic maintenance" to "condition maintenance".

[0003] At present, traditional 500kV and above line surge arresters have the following prominent problems: (1) The electric field gradient is generally low (about 250-280V / mm), resulting in the surge arrester height exceeding 4 meters, huge diameter, inconvenient installation and affecting the landscape; (2) There is a lack of effective built-in condition monitoring methods. Existing monitoring mostly uses external leakage current meters or periodic pre-tests, which cannot detect in real time the hidden dangers such as aging of internal valve plates, moisture and uneven current distribution of multiple columns in parallel; (3) The surge arrester body will generate extremely high ground potential and strong electromagnetic interference. The operating environment is harsh (large temperature difference, high electromagnetic intensity), and conventional electronic sensors are difficult to work stably in this environment for a long time.

[0004] Chinese patent CN116487137A discloses "an intelligent anti-aging device for surge arresters", which combines the relatively mature online detection technology for surge arrester aging and uses the automatic resistance method to reduce the charge rate of the protected surge arrester and slow down the aging process of the surge arrester that has already aged. However, it does not improve the structure and performance of the surge arrester, nor can it perceive the status of the surge arrester in real time. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing an intelligent surge arrester for power lines, comprising: Surge arrester body and intelligent control box; The surge arrester body is used to provide a low-impedance path to ground for overcurrent when overvoltage occurs on the line due to lightning or switching operation, thereby limiting the overvoltage to a safe level, protecting subsequent electrical equipment from damage, and restoring to the insulation state after the overvoltage disappears, ensuring the normal operation of the power system. The intelligent control box includes an embedded intelligent sensing module and a diagnostic module. The embedded intelligent sensing module is used to collect the operating status parameters of the surge arrester body in real time. The diagnostic module is used to evaluate the insulation status of the surge arrester body using a partial discharge mode recognition model and predict the aging trend of the valve plate unit of the surge arrester body using a valve plate aging trend prediction model based on the operating status parameters of the surge arrester body.

[0006] Furthermore, the surge arrester body includes a valve plate unit and an insulating shell surrounding the valve plate unit. The valve plate unit is composed of multiple zinc oxide resistors doped with Bi2O3 connected in series. The Bi2O3 doping concentration of each zinc oxide resistor gradually decreases from the edge to the axis. The insulating shell is made of high-temperature vulcanized silicone rubber material.

[0007] Furthermore, the embedded intelligent sensing module includes a temperature sensor, a leakage current sensor, and an ultra-high frequency partial discharge sensor. The leakage current sensor collects the leakage current signal of the surge arrester body mixed with electromagnetic interference in real time, the ultra-high frequency partial discharge sensor collects the ultra-high frequency partial discharge signal of the surge arrester body mixed with electromagnetic interference in real time, and the temperature sensor collects the ambient temperature in real time.

[0008] Furthermore, the diagnostic module incorporates a noise suppression algorithm, a partial discharge pattern recognition model, and a valve plate aging trend prediction model. The noise suppression algorithm eliminates electromagnetic interference in the leakage current signal and the ultra-high frequency partial discharge signal that are mixed with electromagnetic interference, resulting in a denoised leakage current signal and a denoised ultra-high frequency partial discharge signal. The denoised UHF partial discharge signal is input into the partial discharge pattern recognition model to obtain the insulation status recognition result of the surge arrester body. The insulation status recognition result is either normal insulation function or preset insulation defect type. The noise-reduced leakage current signal is input into the valve plate aging trend prediction model to obtain the current aging degree quantification value of the valve plate unit. Combined with the ambient temperature and humidity, the model outputs the change trend of the aging degree quantification value of the valve plate unit within a future preset time window.

[0009] Furthermore, the specific method for the noise suppression algorithm to eliminate electromagnetic interference in the leakage current signal and the ultra-high frequency partial discharge signal mixed with electromagnetic interference is as follows: The noise suppression algorithm includes a reference pre-set noise sample library and an adaptive filter; The adaptive filter is configured to perform adaptive noise suppression based on the least mean square algorithm. The adaptive filter minimizes the output mean square error by adjusting its weight coefficients in real time. Then, based on the real-time adjusted weight coefficients, it suppresses the corresponding signal noise components related to noise samples in the preset noise sample library from the leakage current signal and the UHF partial discharge signal mixed with electromagnetic interference. It extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the leakage current signal mixed with electromagnetic interference as the denoised leakage current signal, and extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the UHF partial discharge signal mixed with electromagnetic interference as the denoised UHF partial discharge signal.

[0010] Furthermore, the partial discharge pattern recognition model is obtained through the following method: The historical UHF partial discharge signals in the historical UHF partial discharge signal set are tagged with the insulation state identification results. The tagged historical UHF partial discharge signal set is divided into a training set and a validation set. The convolutional neural network is trained based on the training set and the validation set to obtain the partial discharge pattern recognition model.

[0011] Furthermore, the specific method for inputting the denoised UHF partial discharge signal into the partial discharge pattern recognition model to obtain the insulation state recognition result of the surge arrester body is as follows: The partial discharge pattern recognition model includes an input unit, a feature extraction unit, and a classification decision unit; The input unit receives a denoised UHF partial discharge signal. The feature extraction unit includes a convolutional layer and a pooling layer. The convolutional layer is used to extract high-frequency features and phase distribution features from the partial discharge signal using the ReLU activation function. The pooling layer is used to reduce the dimensionality of the high-frequency features and the phase distribution features using a max pooling strategy. The dimensionality-reduced high-frequency features and the phase distribution features are concatenated in the channel dimension to form partial discharge features. The classification decision unit includes a fully connected layer and a Softmax classifier. The fully connected layer linearly combines the partial discharge features using a weight matrix and a bias to generate a score vector. The Softmax classifier maps the score vector to the insulation state identification result.

[0012] Furthermore, the valve plate aging trend prediction model includes: Signal decomposition unit, resistive current calculation unit, health index calculation unit, trend prediction unit; The signal decomposition unit is used to preprocess the denoised leakage current signal using wavelet transform, and to decompose the preprocessed denoised leakage current signal into the fundamental component doped with third harmonic noise using an adaptive noise complete set empirical mode decomposition algorithm. The resistive current calculation unit is used to extract the resistive current signal from the fundamental component doped with third harmonic noise using a capacitor current compensation algorithm, and to perform temperature correction on the amplitude of the resistive current signal using the ambient temperature collected by the temperature sensor. The rate of change of the peak value of the resistive current signal over time is calculated to obtain the resistive current change rate. The ratio of the amplitude of the third harmonic noise in the resistive current signal to the amplitude of the fundamental wave is calculated to obtain the third harmonic content of the resistive current. The health index calculation unit is used to calculate the current health index of the valve plate unit based on the resistive current change rate, the resistive current third harmonic content and the partial discharge characteristic value, using a fuzzy comprehensive evaluation model. The health index represents the quantitative value of the aging degree of the valve plate unit. The trend prediction unit uses a long short-term memory neural network, taking a set of historical health indices as input, to output the trend of health index changes within a preset time window in the future.

[0013] A condition diagnosis method for an intelligent surge arrester for power lines includes: When an overvoltage occurs on the line due to lightning or switching operation, the surge arrester body provides a low-impedance path to ground for the overcurrent, thereby limiting the overvoltage to a safe level and protecting the subsequent electrical equipment from damage. After the overvoltage disappears, it returns to the insulation state to ensure the normal operation of the power system. The intelligent control box includes an embedded intelligent sensing module and a diagnostic module. The embedded intelligent sensing module collects the operating status parameters of the surge arrester body in real time. Based on the operating status parameters of the surge arrester body, the diagnostic module uses a partial discharge mode recognition model to evaluate the insulation status of the surge arrester body and a valve aging trend prediction model to predict the aging trend of the valve unit of the surge arrester body.

[0014] A computer program product includes a computer program / instructions that, when executed by a processor, implement the aforementioned method for diagnosing the condition of an intelligent surge arrester for lines.

[0015] The beneficial effects of this invention are as follows: 1. This invention optimizes the electric field distribution within the zinc oxide resistive element by employing a Bi₂O₃ doping concentration that decreases gradually from the edge to the axis, significantly improving the average electric field gradient (exceeding 330V / mm). This allows for a 20%-30% reduction in the radial dimension of the surge arrester body while maintaining the same voltage level and protection level, effectively saving material costs, reducing tower load, and facilitating installation in space-constrained lines. Simultaneously, the gradient doping structure suppresses local electric field distortion, enhances the nonlinear characteristics and energy absorption capacity of the varistor, and gives the surge arrester the dual advantages of a compact structure and high-performance protection.

[0016] 2. By integrating multi-dimensional sensors such as temperature, leakage current, and ultra-high frequency partial discharge sensors, the embedded intelligent sensing module can simultaneously collect key multi-physics parameters reflecting the operating status of the surge arrester. In particular, the ultra-high frequency partial discharge sensor can effectively capture weak discharge signals generated by early insulation defects, overcoming the shortcomings of traditional monitoring methods such as susceptibility to on-site electromagnetic interference and insufficient sensitivity, thus laying a data foundation for early fault diagnosis.

[0017] 3. The diagnostic module's built-in noise suppression algorithm innovatively combines a preset noise sample library with an adaptive filter based on the least mean square criterion. This algorithm can dynamically track and suppress complex electromagnetic interference aliased in leakage current and partial discharge signals, adaptively extracting clean fault feature signals with a satisfactory signal-to-noise ratio from a strong noise background, greatly improving the accuracy and reliability of subsequent diagnostic analysis.

[0018] 4. This invention utilizes a partial discharge pattern recognition model constructed using a convolutional neural network. This model can automatically learn and extract deep features from denoised UHF partial discharge signals, achieving end-to-end intelligent identification of insulation conditions. This method avoids the problems of reliance on expert experience and strong subjectivity in traditional diagnosis, and can quickly and accurately classify insulation conditions into normal or specific defect types, greatly improving the automation level and recognition accuracy of diagnosis.

[0019] 5. The valve plate aging trend prediction model constructed in this invention refines the signal through wavelet transform and adaptive noise complete ensemble empirical mode decomposition algorithm, and combines capacitor current compensation and temperature correction techniques to accurately calculate the resistive current change rate and third harmonic content, which characterize key aging features. Furthermore, a fuzzy comprehensive evaluation model is used to fuse multiple feature parameters to calculate an intuitive health index, achieving precise quantification of the aging degree. Finally, a long short-term memory neural network is used to learn the health index time series, enabling accurate prediction of the aging trend within a preset time window. This transforms the surge arrester maintenance strategy from periodic inspection or reactive maintenance to proactive predictive maintenance, effectively avoiding sudden failures, extending equipment life, and optimizing operation and maintenance resources.

[0020] 6. This invention deeply integrates advanced sensing technology, signal processing algorithms, and artificial intelligence models to construct an intelligent system that combines real-time monitoring, precise diagnosis, and trend prediction. It is not merely a lightning protection device, but also an intelligent line health management node, providing crucial technical support and data foundation for building smart grids and implementing condition-based maintenance. Attached Figure Description

[0021] Figure 1 This is a structural diagram of the intelligent surge arrester for power lines according to the present invention. Detailed Implementation

[0022] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0023] Glossary Leakage current sensor: a precision measuring device used for condition monitoring of high-voltage electrical equipment. Its core function is to acquire the full signal of the tiny leakage current flowing through the valve plate of a metal oxide surge arrester (MOA) in operation in real time and with high precision, and convert it into a standard electrical signal that can be processed by the subsequent intelligent analysis unit.

[0024] Ultra-high frequency partial discharge sensor: a special sensor used for online monitoring of the insulation status of high-voltage electrical equipment. Its core function is to detect and collect ultra-high frequency electromagnetic wave signals in the 300MHz to 3GHz frequency band excited by partial discharge activity inside the insulation.

[0025] High-temperature vulcanized rubber: is a material based on high molecular weight (usually 400,000 to 800,000) polyorganosiloxane, with added reinforcing fillers (such as fumed silica), vulcanizing agents (such as organic peroxides) and other additives, which is vulcanized at high temperature (100℃ to 300℃) to form an elastomer.

[0026] Convolutional Neural Networks (CNNs): A deep learning model specifically designed for processing grid-like topological data (such as images and time-series signal spectra). Its core idea is to efficiently extract features layer by layer from low to high levels through local connectivity, weight sharing, and spatial downsampling, ultimately completing classification or recognition tasks. It typically consists of a feature extractor (convolutional layers, pooling layers) and a classifier (fully connected layers, softmax) connected in series, and its workflow is an end-to-end feature learning and mapping process.

[0027] Adaptive noise suppression based on the Least Mean Square (LMS) algorithm: an advanced signal processing technique. Its core is to dynamically adjust filter parameters using the Least Mean Square (LMS) adaptive algorithm to estimate and cancel noise components in the signal in real time, thereby extracting a high signal-to-noise ratio (SNR) effective signal from a highly interfering environment. This algorithm system typically comprises three basic components: a reference input, derived from a pre-set database of power frequency interference and high-frequency noise samples. This database contains typical noise patterns extracted from historical data or prior knowledge, providing the adaptive filter with noise samples to be suppressed; a primary input, derived from the real, raw signal aliased with strong electromagnetic interference, collected by a leakage current sensor or an ultra-high frequency partial discharge sensor; and an adaptive filter, the core of which is a set of weight coefficient vectors, serving as the filter's "brain" and adjustable parameters. d(n) = s(n) + n0(n), where s(n) is the desired effective state signal (such as the actual leakage current or partial discharge pulse), n0(n) is the mixed strong noise, and d(n) is the primary input. The adaptive filter processes the reference input x(n) to generate a noise estimate y(n). The filter output y(n) is subtracted from the primary input d(n) to obtain the error signal e(n) = d(n) - y(n). This e(n) is the desired clean output signal. The LMS algorithm uses the error signal e(n) and the reference input x(n) to update the weight vector in real time according to the following rule: W(n+1) = W(n) + μ × e(n) × X(n), where μ is the step size factor, which controls the convergence speed and stability. The purpose of this formula is to minimize the mean square value of the error signal e(n). Through countless iterations, the weight vector W(n) will converge to an optimal value. At this point, the filter output y(n) can best approximate the real noise n0(n) in the primary input, so that the error signal e(n) is infinitely close to the pure signal s(n).

[0028] Capacitor current compensation algorithm: A algorithm specifically designed to precisely separate the resistive current characterizing the aging state of the valve plate from the total leakage current signal. Essentially, it uses mathematical models and electrical characteristic analysis to cancel the capacitive component in the leakage current, thereby extracting the pure resistive component and providing a reliable data foundation for condition diagnosis. In the equivalent circuit of a surge arrester, the valve plate can be considered as a parallel model of a capacitor and a nonlinear resistor under operating voltage. The total leakage current includes: capacitive current, generated by the geometric capacitance of the surge arrester, which leads the voltage by approximately 90° and is unrelated to aging; and resistive current, which is in phase with the voltage and directly reflects the degradation of the valve plate's nonlinear characteristics, serving as the core of the diagnosis. The algorithm flow is generally as follows: The phase of the bus voltage and the total current is acquired; a compensation vector with the same amplitude but opposite direction to the capacitive current is constructed; and the resistive current is obtained by subtracting the compensation vector from the total current.

[0029] Wavelet transform is a powerful time-frequency analysis tool. Its core idea is to use a short, scalable, and shiftable oscillating waveform—the wavelet—to scan a signal, thereby revealing local details of the signal's time and frequency characteristics. It is well-suited for analyzing non-stationary signals, i.e., signals whose statistical properties change over time, such as partial discharges, vibrations, speech, and biomedical signals. The ultra-high frequency signals generated by partial discharges are transient pulses, submerged in strong background electromagnetic noise. Wavelet transform can accurately pinpoint the timing of these discharge pulses and, by setting a threshold, filter out weaker noise components while retaining stronger discharge components, effectively improving the signal-to-noise ratio.

[0030] Adaptive Noise Complete Ensemble Empirical Mode Decomposition (CEEMDAN): An advanced, adaptive signal processing technique used to decompose complex non-stationary, nonlinear signals into a series of physically meaningful intrinsic mode functions (IMFs). It is an improved algorithm of Empirical Mode Decomposition (EMD), effectively solving the mode aliasing and reconstruction error problems of traditional EMD methods by introducing adaptive white noise and a complete reconstruction mechanism. In this invention, CEEMDAN works in conjunction with wavelet transform to form a signal preprocessing pipeline: The full leakage current signal or ultra-high frequency partial discharge signal is denoised; the signal is adaptively decomposed into a series of IMFs, including: the fundamental component (power frequency component, approximately 50Hz), the third harmonic component (150Hz component), other high-frequency harmonics, and residual noise; the decomposed pure fundamental and third harmonic components are fed to a resistive current calculation unit to extract the resistive current peak value, active power loss value, and third harmonic content.

[0031] Fuzzy comprehensive evaluation model: a decision analysis method based on fuzzy logic, specifically designed to handle complex system evaluation problems involving uncertainty, fuzziness, and multiple factors. Its core idea is to transform fuzzy, non-quantitative subjective evaluations into clear, quantifiable assessment results through mathematical methods.

[0032] Long Short-Term Memory (LSTM) neural network is a special type of recurrent neural network (RNN) specifically designed to process and predict time series data.

[0033] Example 1 In high-voltage transmission lines, surge arresters are crucial devices for protecting electrical equipment from damage caused by lightning overvoltages and switching overvoltages. However, traditional line surge arresters and their monitoring methods have many shortcomings, making it difficult to meet the requirements of smart grids for equipment miniaturization, intelligence, and high reliability.

[0034] Firstly, in terms of structural performance, traditional surge arresters are typically large in size and weight to achieve sufficient insulation strength and current carrying capacity. This not only increases the mechanical load on the tower and the difficulty of installation, but also occupies valuable line space. Their varistors often use uniformly doped zinc oxide resistance sheets, resulting in uneven electric field distribution, which limits the further improvement of its electric field gradient. Consequently, the pursuit of miniaturization often comes at the cost of sacrificing protective performance.

[0035] Secondly, in terms of condition monitoring and diagnosis, traditional methods mainly rely on periodic inspections and preventative tests, which cannot achieve real-time perception of operational status. Even if some surge arresters are equipped with simple leakage current monitoring functions, the lack of effective anti-interference measures results in a low signal-to-noise ratio of the acquired signals, making it difficult to extract characteristic quantities that truly reflect valve plate aging or insulation degradation from strong electromagnetic noise environments. In addition, existing diagnostic methods mostly rely on threshold alarms or manual experience judgment, lacking the ability to deeply identify potential faults and predict aging trends. This makes it impossible to achieve the transformation from periodic maintenance to predictive maintenance, resulting in a high risk of sudden equipment failure and huge operation and maintenance costs.

[0036] For the reasons mentioned above, this invention proposes an intelligent surge arrester for power lines, such as... Figure 1 As shown, it includes: The surge arrester body 1, the connector 2, and the intelligent control box 3 are electrically connected, and the connector 2 physically fixes the surge arrester body 1 and the intelligent control box 3. The surge arrester body 1 includes an insulating shell 5 and a valve plate unit 4 inside the insulating shell 5. The valve plate unit 4 is composed of multiple zinc oxide resistors doped with Bi2O3 connected in series. The Bi2O3 doping concentration of the zinc oxide resistors gradually decreases from the edge to the axis. The intelligent control box 3 includes an embedded intelligent sensing module 6 and a diagnostic module 7, which are electrically connected. The embedded intelligent sensing module 6 is used to collect the operating status parameters of the surge arrester body 1 in real time, and the diagnostic module 7 is used to evaluate the insulation status of the surge arrester body 1 and predict the aging trend of the valve plate unit 4 based on the operating status parameters of the surge arrester body 1.

[0037] In a preferred embodiment, the insulating housing 5 is made of high-temperature vulcanized silicone rubber material, and its radial dimension is reduced by 20%-30% compared to traditional surge arresters of the same voltage level. The high-temperature vulcanized silicone rubber material has high electrical strength, high temperature resistance, aging resistance and UV resistance, ensuring that the insulating housing can work stably for a long time in harsh environments (such as high and low temperatures, humidity).

[0038] In a preferred embodiment, the Bi₂O₃ doping concentration at the edge of the zinc oxide resistive element is 2.0-5.5 wt% higher than that at the center, ensuring an average electric field gradient >330 V / mm. By using a tilted equalizing ring design to optimize the axial electric field distribution, the insulation height is reduced to ≤2.8 m (compared to ≥4 m for traditional structures) for the same current carrying capacity. Combined with a compact silicone rubber composite insulator with optimized electric field design, the overall size is reduced by 20%-30% compared to traditional products.

[0039] By varying the doping concentration within a specific range (2.0-5.5 wt%), a smooth transition of the electric field gradient between the edge and center of the resistor element is ensured, avoiding electric field concentration and reducing the risk of partial discharge and breakdown. This design improves the nonlinear characteristics of the resistor element, enabling the surge arrester to respond more quickly and with lower residual voltage during overvoltage protection. The concentration gradient design reduces thermal stress on the resistor element, delays aging, and improves the long-term stability of the valve unit.

[0040] In a preferred embodiment, the embedded intelligent sensing module 6 includes a temperature sensor, a leakage current sensor, and an ultra-high frequency partial discharge sensor. These sensors are protected by a multi-physics shielded enclosure, ensuring long-term reliable operation in harsh environments with high voltage, strong electromagnetic interference, and extreme temperatures (-40°C to +85°C).

[0041] The invention will be further described in detail below with reference to an embodiment of a 500kV AC line high-gradient small-size intelligent surge arrester.

[0042] S1 High-gradient valve unit fabrication: Zinc oxide resistor sheets with a diameter of approximately 76-80 mm were selected. A special process was used to achieve a Bi₂O₃ doping concentration of 10.9 wt% in the edge region and 8.9 wt% in the central region. Testing showed that the gradient of a single valve unit could reach 380 V / mm. The number of valves connected in series was determined based on the target residual voltage level.

[0043] S2 Compact Insulating Shell Molding: A compact insulating shell is manufactured using high-temperature vulcanized silicone rubber material and precision injection molding. The shell undergoes electrical strength, weather resistance, and mechanical strength tests to ensure it meets requirements for high-potential, strong electromagnetic interference, and high / low temperature environments. Finite element analysis software (such as ANSYS) is used to simulate the overall electric field of the surge arrester, optimizing the grading ring size and skirt shape. The final designed surge arrester is approximately 2.82 meters high and 180 mm in diameter, representing a size reduction of approximately 28% compared to traditional products.

[0044] The S3 embedded intelligent sensing module integrates: a miniature fiber optic grating (FBG) temperature sensor; and a miniature broadband current sensor based on the Rogowski coil principle. Integration locations: the FBG sensor is attached to the side of the valve plate in a critical location; the current sensor is embedded near the grounding lead of the valve plate string; and the leakage current sensor is placed on the grounding side at the bottom of the surge arrester.

[0045] S4 Diagnostic Module Development: The hardware utilizes a low-power ARM Cortex-M7 core processor and is equipped with a high-precision ADC unit (analog-to-digital converter), supporting multi-channel synchronous data acquisition. The communication unit integrates both a LoRa wireless unit and a single-mode fiber optic interface. An LMS-based adaptive filtering algorithm is used to denoise the acquired leakage current signal, and this algorithm is integrated into the embedded system.

[0046] S5 Overall Assembly and Testing: The assembled 500kV intelligent surge arrester prototype must pass the following tests: (1) Electrical performance test: It passed the standard steep wave, lightning wave, switching wave impulse test and power frequency withstand test in the high voltage test chamber. The measured protection characteristics meet the design requirements, and the gradient is 352V / mm. (2) Sensor accuracy test: The current sensor was tested by injecting standard square wave current, and the accuracy was better than ±1%. (3) Communication and function test: Under simulated operation, the diagnostic module successfully received the sensor data, executed the diagnostic algorithm, and uploaded the "normal status" report to the background system through optical fiber. The LoRa communication as a backup link was also successfully tested.

[0047] By adopting gradient-doped valve plate groups and an inclined equalizing ring design, the electric field gradient of the surge arrester is significantly improved to >330V / mm, while the insulation height is reduced to ≤2.8m. The overall size is reduced by 20%-30% compared to traditional products, effectively saving corridor space and reducing installation difficulty. The embedded intelligent sensing module integrates miniature temperature, leakage current, and partial discharge sensors, combined with a multi-physics field protection package, enabling stable operation in harsh environments such as high voltage, strong electromagnetic interference, and high and low temperatures, solving the problem of traditional surge arresters lacking built-in condition monitoring methods. The intelligent diagnostic system, through intelligent algorithms, can process data locally and achieve condition assessment and fault early warning. Combined with LoRa and fiber optic dual-mode communication interfaces, it supports the IEC61850 standard protocol and uploads results to the monitoring center in real time, promoting the transformation from "periodic maintenance" to "condition-based maintenance" and greatly improving the reliability of power grid operation.

[0048] Example 2 A condition diagnosis method for an intelligent surge arrester for power lines includes: The leakage current sensor collects the leakage current signal of the surge arrester body 1 in real time, which is mixed with electromagnetic interference; the ultra-high frequency partial discharge sensor collects the ultra-high frequency partial discharge signal of the surge arrester body 1 in real time, which is mixed with electromagnetic interference; and the temperature sensor collects the ambient temperature in real time. The diagnostic module 7 has built-in noise suppression algorithm, partial discharge pattern recognition model and valve plate aging trend prediction model. The noise suppression algorithm performs electromagnetic interference elimination on the leakage current signal and the ultra-high frequency partial discharge signal mixed with electromagnetic interference to obtain the denoised leakage current signal and the denoised ultra-high frequency partial discharge signal. The denoised UHF partial discharge signal is input into the partial discharge mode recognition model to obtain the insulation status recognition result of the surge arrester body 1. The insulation status recognition result is either normal insulation function or preset insulation defect type (such as internal partial discharge, surface partial discharge, surface flashover). The noise-reduced leakage current signal is input into the valve plate aging trend prediction model to obtain the current aging degree quantification value of valve plate unit 4, and the change trend of the aging degree quantification value of valve plate unit 4 within a future preset time window is output in combination with the ambient temperature and ambient humidity.

[0049] As a preferred embodiment, the specific method for the noise suppression algorithm to eliminate electromagnetic interference in leakage current signals and ultra-high frequency partial discharge signals mixed with electromagnetic interference is as follows: The noise suppression algorithm includes a reference pre-set noise sample library (including power frequency interference and high frequency interference) and an adaptive filter; The adaptive filter is configured to perform adaptive noise suppression based on the least mean square algorithm. The adaptive filter minimizes the output mean square error by adjusting its weight coefficients in real time. Then, based on the real-time adjusted weight coefficients, it suppresses the corresponding signal noise components related to noise samples in the preset noise sample library from the leakage current signal and the UHF partial discharge signal mixed with electromagnetic interference. It extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the leakage current signal mixed with electromagnetic interference as the denoised leakage current signal, and extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the UHF partial discharge signal mixed with electromagnetic interference as the denoised UHF partial discharge signal.

[0050] The adaptive filter adjusts the weight coefficients in real time using the least mean square algorithm to suppress power frequency and high-frequency noise from aliased signals. This improves the signal-to-noise ratio (SNR) and reduces the impact of environmental electromagnetic interference. The algorithm extracts denoised signals with an SNR greater than a threshold from leakage current and UHF partial discharge signals, providing a reliable data foundation for subsequent diagnostics. A reference noise sample library allows the algorithm to adapt to different noise environments, improving robustness and ensuring effective operation even under strong interference conditions.

[0051] As a preferred embodiment, the partial discharge pattern recognition model is obtained through the following method: Historical UHF partial discharge signals from a set of historical UHF partial discharge signals are tagged with labels representing insulation state identification results. The tagged set of historical UHF partial discharge signals is then divided into a training set and a validation set. A convolutional neural network is trained using the training and validation sets to obtain a partial discharge pattern recognition model. This training process is a typical supervised training method and will not be elaborated further in this invention.

[0052] CNNs can automatically extract features from historical UHF partial discharge signals without manual feature engineering, improving the efficiency and accuracy of pattern recognition. Through supervised training, the model can accurately classify insulation states (such as normal insulation function or specific preset insulation defect types), reducing human intervention and lowering the false alarm rate.

[0053] As a preferred embodiment, the specific method for inputting the denoised UHF partial discharge signal into the partial discharge pattern recognition model to obtain the insulation state recognition result of the surge arrester body 1 is as follows: The partial discharge pattern recognition model includes an input unit, a feature extraction unit, and a classification decision unit; The input unit receives a denoised UHF partial discharge signal. The feature extraction unit includes a convolutional layer and a pooling layer. The convolutional layer is used to extract high-frequency features and phase distribution features from the partial discharge signal using the ReLU activation function. The pooling layer is used to reduce the dimensionality of the high-frequency features and the phase distribution features using a max pooling strategy. The dimensionality-reduced high-frequency features and the phase distribution features are concatenated in the channel dimension to form partial discharge features. The classification decision unit includes a fully connected layer and a Softmax classifier. The fully connected layer linearly combines the partial discharge features using a weight matrix and a bias to generate a score vector. The Softmax classifier maps the score vector to the insulation state identification result.

[0054] Convolutional layers extract high-frequency and phase distribution features of partial discharge signals using the ReLU activation function, while pooling layers reduce feature dimensionality while retaining key information. This improves feature representativeness and computational efficiency. Fully connected layers and a Softmax classifier map features to insulation status identification results, achieving end-to-end classification and outputting clear status labels (such as normal insulation function or specific preset insulation defect type), supporting rapid decision-making. The optimized model structure is suitable for embedded system deployment, enabling real-time monitoring and diagnosis, and meeting the needs of online applications.

[0055] As a preferred embodiment, the valve plate aging trend prediction model includes: Signal decomposition unit, resistive current calculation unit, health index calculation unit, trend prediction unit; The signal decomposition unit is used to preprocess the denoised leakage current signal using wavelet transform, and to decompose the preprocessed denoised leakage current signal into the fundamental component doped with third harmonic noise using an adaptive noise complete set empirical mode decomposition algorithm. The resistive current calculation unit is used to extract the resistive current signal from the fundamental component doped with third harmonic noise using a capacitor current compensation algorithm, and to perform temperature correction on the amplitude of the resistive current signal using the ambient temperature collected by the temperature sensor. The rate of change of the peak value of the resistive current signal over time is calculated to obtain the resistive current change rate. The ratio of the amplitude of the third harmonic noise in the resistive current signal to the amplitude of the fundamental wave is calculated to obtain the third harmonic content of the resistive current. The health index calculation unit is used to calculate the current health index of valve plate unit 4 based on the resistive current change rate, the resistive current third harmonic content, and the partial discharge characteristic value, using a fuzzy comprehensive evaluation model. The health index represents the quantified value of the aging degree of valve plate unit 4. In this invention, each factor (resistive current third harmonic content, resistive current change rate, and partial discharge characteristic) is evaluated individually to determine its degree of membership in each evaluation level, forming a membership vector. The membership vectors of all factors are combined to form a fuzzy relation matrix R. The weight vector A is synthesized with the fuzzy relation matrix R (e.g., using a weighted average operator) to obtain a comprehensive evaluation vector B. Assuming the calculation result is B = (0.1, 0.2, 0.6, 0.1), a final judgment is made based on the comprehensive evaluation vector B. Typically, the maximum membership principle is adopted, that is, the evaluation level with the largest value in the vector (i.e., the health index) is selected as the final result. In the comprehensive evaluation vector B=(0.1,0.2,0.6,0.1), the largest value is 0.6, which means the health index is 0.6 and the corresponding evaluation level is abnormal. Therefore, the health status of the surge arrester valve is determined to be abnormal.

[0056] The trend prediction unit employs a Long Short-Term Memory (LSTM) neural network, taking a set of historical health indices as input, and outputting the trend of health index changes within a preset future time window. After learning historical patterns from the set of historical health indices, the LSM neural network outputs a prediction of the health index trend within the preset future time window (e.g., the next 6 months). This prediction is a trend line that visually illustrates how the health index may change. When the prediction indicates that the health index will deteriorate rapidly and approach the warning threshold, the system will generate maintenance suggestions in advance.

[0057] Example 3 A computer program product includes a computer program / instructions that, when executed by a processor, implement the method of using the intelligent surge arrester for lines in Embodiment 2.

[0058] The contents not described in detail in this specification are prior art known to those skilled in the art. Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0060] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

Claims

1. An intelligent surge arrester for power lines, characterized in that, include: Surge arrester body (1) and intelligent control box (3); The surge arrester body (1) is used to provide a low-impedance path to ground for overcurrent when overvoltage occurs on the line due to lightning or switching operation, thereby limiting the overvoltage to a safe level, protecting the subsequent electrical equipment from damage, and restoring the insulation state after the overvoltage disappears to ensure the normal operation of the power system. The intelligent control box (3) includes an embedded intelligent sensing module (6) and a diagnostic module (7). The embedded intelligent sensing module (6) is used to collect the operating status parameters of the surge arrester body (1) in real time. The diagnostic module (7) is used to evaluate the insulation status of the surge arrester body (1) using a partial discharge mode recognition model and predict the aging trend of the valve plate unit (4) of the surge arrester body (1) using a valve plate aging trend prediction model based on the operating status parameters of the surge arrester body (1). The diagnostic module (7) has a built-in noise suppression algorithm, a partial discharge pattern recognition model and a valve plate aging trend prediction model. The noise suppression algorithm performs electromagnetic interference elimination on the leakage current signal and the ultra-high frequency partial discharge signal that are mixed with electromagnetic interference, and obtains the denoised leakage current signal and the denoised ultra-high frequency partial discharge signal. The partial discharge pattern recognition model includes an input unit, a feature extraction unit, and a classification decision unit. The input unit receives a denoised ultra-high frequency partial discharge signal. The feature extraction unit includes a convolutional layer and a pooling layer. The convolutional layer extracts high-frequency features and phase distribution features from the partial discharge signal using the ReLU activation function. The pooling layer reduces the dimensionality of the high-frequency features and phase distribution features using a max pooling strategy. The dimensionality-reduced high-frequency features and phase distribution features are concatenated along the channel dimension to form partial discharge features. The classification decision unit includes a fully connected layer and a Softmax classifier. The fully connected layer linearly combines the partial discharge features using a weight matrix and a bias to generate a score vector. The Softmax classifier maps the score vector to the insulation status recognition result, which is either normal insulation function or a preset insulation defect type. The valve plate aging trend prediction model includes a signal decomposition unit, a resistive current calculation unit, a health index calculation unit, and a trend prediction unit. The signal decomposition unit is used to preprocess the denoised leakage current signal using wavelet transform, and then uses an adaptive noise complete set empirical mode decomposition algorithm to decompose the preprocessed denoised leakage current signal into the fundamental component doped with third harmonic noise. The resistive current calculation unit is used to extract the resistive current signal from the fundamental component doped with third harmonic noise using a capacitor current compensation algorithm, and to perform temperature correction on the amplitude of the resistive current signal using the ambient temperature collected by the temperature sensor. The peak value of the resistive current signal after temperature correction is calculated as the rate of change of time to obtain the resistive current change rate. The ratio of the amplitude of the third harmonic noise in the resistive current signal to the amplitude of the fundamental wave is calculated to obtain the third harmonic content of the resistive current. The health index calculation unit is used to calculate the current health index of the valve plate unit (4) using a fuzzy comprehensive evaluation model based on the resistive current change rate, the third harmonic content of the resistive current and the partial discharge characteristic value. The health index represents the quantified value of the aging degree of the valve plate unit (4). The trend prediction unit is used to use a long short-term memory neural network, with the historical health index set as input, to output the trend of the health index change within a future preset time window.

2. The intelligent surge arrester for power lines according to claim 1, characterized in that: The surge arrester body (1) includes a valve plate unit (4) and an insulating shell (5) fitted outside the valve plate unit (4). The valve plate unit (4) is composed of multiple zinc oxide resistors doped with Bi2O3 connected in series. The Bi2O3 doping concentration of each zinc oxide resistor gradually decreases from the edge to the axis. The insulating shell (5) is made of high-temperature vulcanized silicone rubber material.

3. The intelligent surge arrester for power lines according to claim 1, characterized in that: The embedded intelligent sensing module (6) includes a temperature sensor, a leakage current sensor and an ultra-high frequency partial discharge sensor. The leakage current sensor collects the leakage current signal of the arrester body (1) mixed with electromagnetic interference in real time. The ultra-high frequency partial discharge sensor collects the ultra-high frequency partial discharge signal of the arrester body (1) mixed with electromagnetic interference in real time. The temperature sensor collects the ambient temperature in real time.

4. The intelligent surge arrester for power lines according to claim 1, characterized in that, The specific method for the noise suppression algorithm to eliminate electromagnetic interference in leakage current signals and ultra-high frequency partial discharge signals mixed with electromagnetic interference is as follows: The noise suppression algorithm includes a reference pre-set noise sample library and an adaptive filter; The adaptive filter is configured to perform adaptive noise suppression based on the least mean square algorithm. The adaptive filter minimizes the output mean square error by adjusting its weight coefficients in real time. Then, based on the real-time adjusted weight coefficients, it suppresses the corresponding signal noise components related to noise samples in the preset noise sample library from the leakage current signal and the UHF partial discharge signal mixed with electromagnetic interference. It extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the leakage current signal mixed with electromagnetic interference as the denoised leakage current signal, and extracts the signal with a signal-to-noise ratio greater than the signal-to-noise ratio threshold from the UHF partial discharge signal mixed with electromagnetic interference as the denoised UHF partial discharge signal.

5. The intelligent surge arrester for power lines according to claim 1, characterized in that, The partial discharge pattern recognition model is obtained through the following method: The historical UHF partial discharge signals in the historical UHF partial discharge signal set are tagged with the insulation state identification results. The tagged historical UHF partial discharge signal set is divided into a training set and a validation set. The convolutional neural network is trained based on the training set and the validation set to obtain the partial discharge pattern recognition model.

6. A condition diagnosis method for an intelligent surge arrester for power lines, characterized in that, include: When an overvoltage occurs on the line due to lightning or switching operation, the surge arrester body (1) provides a low-impedance path to ground for the overcurrent, thereby limiting the overvoltage to a safe level and protecting the subsequent electrical equipment from damage. After the overvoltage disappears, it returns to the insulation state to ensure the normal operation of the power system. The intelligent control box (3) includes an embedded intelligent sensing module (6) and a diagnostic module (7). The embedded intelligent sensing module (6) collects the operating status parameters of the surge arrester body (1) in real time. The diagnostic module (7) evaluates the insulation status of the surge arrester body (1) using a partial discharge mode recognition model based on the operating status parameters of the surge arrester body (1) and predicts the aging trend of the valve plate unit (4) of the surge arrester body (1) using a valve plate aging trend prediction model. The diagnostic module (7) has a built-in noise suppression algorithm, a partial discharge pattern recognition model and a valve plate aging trend prediction model. The noise suppression algorithm performs electromagnetic interference elimination on the leakage current signal and the ultra-high frequency partial discharge signal that are mixed with electromagnetic interference, and obtains the denoised leakage current signal and the denoised ultra-high frequency partial discharge signal. The partial discharge pattern recognition model includes an input unit, a feature extraction unit, and a classification decision unit. The input unit receives a denoised ultra-high frequency partial discharge signal. The feature extraction unit includes a convolutional layer and a pooling layer. The convolutional layer extracts high-frequency features and phase distribution features from the partial discharge signal using the ReLU activation function. The pooling layer reduces the dimensionality of the high-frequency features and phase distribution features using a max pooling strategy. The dimensionality-reduced high-frequency features and phase distribution features are concatenated along the channel dimension to form partial discharge features. The classification decision unit includes a fully connected layer and a Softmax classifier. The fully connected layer linearly combines the partial discharge features using a weight matrix and a bias to generate a score vector. The Softmax classifier maps the score vector to the insulation status recognition result, which is either normal insulation function or a preset insulation defect type. The valve plate aging trend prediction model includes a signal decomposition unit, a resistive current calculation unit, a health index calculation unit, and a trend prediction unit. The signal decomposition unit is used to preprocess the denoised leakage current signal using wavelet transform, and then uses an adaptive noise complete set empirical mode decomposition algorithm to decompose the preprocessed denoised leakage current signal into the fundamental component doped with third harmonic noise. The resistive current calculation unit is used to extract the resistive current signal from the fundamental component doped with third harmonic noise using a capacitor current compensation algorithm, and to perform temperature correction on the amplitude of the resistive current signal using the ambient temperature collected by the temperature sensor. The peak value of the resistive current signal after temperature correction is calculated as the rate of change of time to obtain the resistive current change rate. The ratio of the amplitude of the third harmonic noise in the resistive current signal to the amplitude of the fundamental wave is calculated to obtain the third harmonic content of the resistive current. The health index calculation unit is used to calculate the current health index of the valve plate unit (4) using a fuzzy comprehensive evaluation model based on the resistive current change rate, the third harmonic content of the resistive current and the partial discharge characteristic value. The health index represents the quantified value of the aging degree of the valve plate unit (4). The trend prediction unit is used to use a long short-term memory neural network, with the historical health index set as input, to output the trend of the health index change within a future preset time window.

7. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the condition diagnosis method for intelligent surge arresters for lines as described in claim 6.

Images