GIS equipment partial discharge on-line monitoring method and device

Abstract

The invention provides an online monitoring method and device for partial discharge of GIS equipment, and relates to the field of online monitoring of high-voltage electrical equipment, and the method comprises the steps: collecting a partial discharge signal in the GIS equipment, and synchronously obtaining the data of an environment noise sensor; processing the acquired ultrahigh frequency signal based on the data of the environmental noise sensor to obtain a preprocessed analog signal; converting the preprocessed analog signal into a digital signal through a high-precision analog-to-digital converter; key characteristic parameters of the partial discharge signals are extracted from the digital signals; dynamically adjusting a discharge detection threshold according to the extracted key characteristic parameters by applying a hybrid optimization algorithm; and based on the adjusted discharge detection threshold, setting an event trigger condition, and when the detected partial discharge signal amplitude and the trigger frequency reach the event trigger condition, sending out an alarm signal. According to the invention, the partial discharge phenomenon can be detected and alarmed in time, so that the operation reliability and safety of the GIS equipment are improved.

Description

Technical Field

[0001] The present invention relates to the field of online monitoring of high-voltage electrical equipment, and in particular to an online monitoring method and device for partial discharge of GIS equipment. Background Art

[0002] Gas Insulated Switchgear (GIS) is a key equipment widely used in high-voltage power transmission and distribution systems. Due to its compact structure, high reliability and excellent insulation performance, GIS has been widely used in urban power grids, industrial parks and important infrastructure. GIS equipment mainly consists of circuit breakers, disconnectors, capacitors and reactors in metal-enclosed cabinets. It uses sulfur hexafluoride (SF6) gas as an insulating and arc-extinguishing medium, which effectively reduces the size and floor space of the equipment, while improving the operating stability and safety of the system.

[0003] During the operation of GIS equipment, partial discharge (PD) is one of the key factors affecting the equipment life and operational reliability. The online monitoring technology of partial discharge aims to evaluate the insulation status of the equipment and prevent potential failures and accidents by detecting and analyzing partial discharge signals in real time. At present, commonly used monitoring methods include acoustic emission, ultrasonic and electromagnetic methods, which rely on sensors to collect and process various signals generated by discharge. However, the existing technology has certain shortcomings in signal noise suppression, feature extraction accuracy and real-time response capabilities, resulting in limited accuracy and reliability of monitoring results. In addition, complex environmental conditions and the variability of equipment operating conditions also increase the difficulty of designing and implementing online monitoring systems. More efficient and accurate monitoring technologies are urgently needed to improve the maintenance and management level of GIS equipment. Summary of the invention

[0004] In view of this, the present invention proposes an online monitoring method and device for partial discharge of GIS equipment, which realizes timely detection and alarm of partial discharge phenomena by optimizing signal acquisition, noise reduction processing and extraction of key characteristic parameters, thereby improving the operational reliability and safety of GIS equipment. This technology aims to overcome the shortcomings of existing monitoring methods in terms of noise suppression, feature extraction accuracy and real-time response capability, ensure the high accuracy of monitoring data and the efficient operation of the system, and thus optimize the maintenance and management level of GIS equipment.

[0005] The technical solution of the present invention is achieved in this way:

[0006] In one aspect, the present invention provides a method for online monitoring of partial discharge of GIS equipment, comprising:

[0007] S1. Use UHF sensors to collect local discharge signals inside GIS equipment in real time and simultaneously obtain data from environmental noise sensors;

[0008] S2, performing noise reduction processing on the collected UHF signal based on the data of the environmental noise sensor, and performing time domain alignment on the signal according to a synchronization method to obtain a preprocessed analog signal;

[0009] S3, converting the preprocessed analog signal into a digital signal through a high-precision analog-to-digital converter;

[0010] S4, extracting key characteristic parameters of partial discharge signal from digital signal;

[0011] S5, applying a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key feature parameters;

[0012] S6. Based on the adjusted discharge detection threshold, an event trigger condition is set, and when the detected partial discharge signal amplitude and trigger times reach the event trigger condition, an alarm signal is issued.

[0013] Based on the above technical solution, preferably, step S2 includes:

[0014] S21, analyzing data collected by the environmental noise sensor to determine the time-frequency characteristics of the environmental noise;

[0015] S22, using a multi-level adaptive wavelet threshold noise reduction algorithm to perform noise reduction processing on the collected UHF signal;

[0016] S23, performing time domain alignment on the noise-reduced signal according to a predetermined synchronization method, including: determining a reference signal or a reference time point; applying an interpolation algorithm to adjust the sampling rate and time axis of the signal to synchronize it with the reference signal; using a cross-correlation method to evaluate and adjust the time offset between each signal; generating an alignment signal under a unified time reference;

[0017] S24, generating a pre-processed analog signal after noise reduction and time domain alignment processing.

[0018] Based on the above technical solution, preferably, step S22 includes:

[0019] S221, assuming that the input signal is y(t), wherein y(t)=x(t)+n(t), x(t) is the original partial discharge signal, and n(t) is the environmental noise;

[0020] S222, perform multi-level wavelet decomposition on the input signal y(t) to obtain the wavelet coefficients W of each layer j,k , where j represents the number of decomposition levels and k represents the coefficient position;

[0021] S223, calculate the adaptive threshold T according to the noise time-frequency characteristics of each layer j :

[0022]

[0023] Among them, σ j is the estimated noise standard deviation of the j-th layer wavelet coefficients, and N is the total length of the signal;

[0024] S224, introduce nonlinear scaling function φ(W j,k )Threshold processing is performed on the wavelet coefficients:

[0025]

[0026] Wherein, tanh represents the hyperbolic tangent function; a is the adjustment factor;

[0027] S225, the processed wavelet coefficients Perform wavelet reconstruction to obtain the denoised signal

[0028] Based on the above technical solution, preferably, the key characteristic parameters of the partial discharge signal include:

[0029] Maximum discharge, which indicates the maximum energy value in the partial discharge signal;

[0030] Average discharge, which indicates the average energy value of the partial discharge signal;

[0031] Discharge phase, which indicates the electrical phase position where partial discharge occurs;

[0032] Discharge frequency refers to the number of discharges that occur per unit time;

[0033] Energy distribution, which indicates the distribution of partial discharge signals at different energy levels;

[0034] Discharge duration, which indicates the duration of a single discharge event;

[0035] Signal rise time, which indicates the time from the start of the partial discharge signal to the peak value;

[0036] Signal decay rate refers to the rate at which the partial discharge signal recovers from the peak value to the baseline.

[0037] Based on the above technical solution, preferably, step S5 includes:

[0038] S51, initializing the discharge detection threshold, and setting the maximum number of main iterations and the optimization target;

[0039] S52, in each main iteration, based on the discharge detection threshold and key characteristic parameters of the current main iteration, calculating the detection sensitivity and accuracy index;

[0040] S53, applying the first optimization algorithm to globally adjust the current discharge detection threshold to explore potential high-quality threshold areas;

[0041] S54: In the high-quality threshold region determined by the first optimization algorithm, the second optimization algorithm is applied to perform local refinement to obtain an adjusted discharge detection threshold:

[0042] S541, initializing the local threshold of the second optimization algorithm, setting the maximum number of local iterations and the local optimization objective function;

[0043] S542, in each local iteration, calculating the local optimization target based on the current local threshold and key feature parameters;

[0044] S543, adjusting the formula according to the second optimization algorithm and updating the local threshold;

[0045] S544, evaluating whether the local threshold meets the local optimization target, if not, proceeding to step S542; if satisfied, outputting the adjusted discharge detection threshold;

[0046] S55, combining the adjustment results of the first optimization algorithm and the second optimization algorithm to determine a final updated discharge detection threshold;

[0047] S56, evaluating whether the updated discharge detection threshold meets the predetermined optimization target. If not, returning to step S52 to continue iteration; if satisfied, terminating the optimization process.

[0048] On the basis of the above technical solution, preferably, in each main iteration, the high-quality threshold region update formula of the first optimization algorithm is:

[0049]

[0050] In the formula, Θ t is the high-quality threshold region of the tth main iteration; λ is the attenuation factor; U(Θ t ) is the optimization objective function under the current high-quality threshold area, and its value is obtained by the weighted sum of detection accuracy and detection sensitivity under the current threshold; μ is the amplitude factor, which adjusts the amplitude of the sine function; κ is the frequency factor, which controls the oscillation frequency of the sine function; φ(P) is a function of the key characteristic parameter P, which is used to adjust the oscillation angle.

[0051] On the basis of the above technical solution, preferably, in each local iteration, the adjustment formula of the second optimization algorithm is:

[0052]

[0053] In the formula, θ′ t is the discharge detection threshold in the kth local iteration; γ is the learning rate, which controls the adjustment step size; δ is the adjustment factor, which balances the gradient influence; U local It is the local optimization objective function; The local optimization objective function U local About the discharge detection threshold θ′ k The numerical approximation of the second-order derivative of the key characteristic parameter P represents the curvature information; β is the scaling factor used to control the nonlinear response of the denominator; f(D) is the function of the discharge duration D.

[0054] On the basis of the above technical solution, preferably, in step S55, the comprehensive adjustment formula is:

[0055]

[0056] In the formula, θ t+1 is the final updated discharge detection threshold; α1, α2, α3 are weight factors, which respectively control the influence of each adjustment; η is the gradient amplification factor; To optimize the objective function U with respect to the discharge detection threshold θ′ k and the gradient of the key feature parameter P, indicating the optimization direction; tanh represents the hyperbolic tangent function; log is the natural logarithm function; g(R) is the function of the signal attenuation rate R.

[0057] On the basis of the above technical solutions, preferably, the UHF sensor is a UHF sensor based on a microstrip antenna structure, with an operating frequency band covering 300MHz to 3GHz, a multi-layer microstrip antenna design, an integrated low-noise amplifier, a broadband response characteristic, a compact packaging structure, and a temperature compensation function.

[0058] On the other hand, the present invention also provides an online monitoring device for partial discharge of GIS equipment, the device is used to execute any of the above methods, the device comprising:

[0059] UHF acquisition module, used to collect partial discharge signals inside GIS equipment in real time and simultaneously obtain data from environmental noise sensors;

[0060] A noise reduction processing module is used to perform noise reduction processing on the collected UHF signal based on the data of the environmental noise sensor, and to perform time domain alignment on the signal according to a predetermined synchronization method to generate a preprocessed analog signal;

[0061] An analog-to-digital conversion module, including a high-precision analog-to-digital converter, for converting the preprocessed analog signal into a digital signal;

[0062] Feature extraction module, used to extract key characteristic parameters of partial discharge signals from digital signals, including maximum discharge amount, average discharge amount, discharge phase, discharge frequency, energy distribution, discharge duration, signal rise time and signal attenuation rate;

[0063] The threshold dynamic adjustment module is configured with a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key characteristic parameters to achieve a balance between detection sensitivity and accuracy;

[0064] The event trigger and alarm module sets the event trigger conditions based on the adjusted discharge detection threshold, and issues an alarm signal when the amplitude and trigger times of the detected partial discharge signal reach the predetermined conditions;

[0065] Control and management unit, used to control and coordinate the operation of each module;

[0066] Data storage module, used to save historical monitoring data and event records;

[0067] Communication interface module, including wired or wireless communication interface, used to exchange data with external monitoring system to ensure real-time transmission and remote access of monitoring data;

[0068] User interface module, including display screen and control buttons, to facilitate users to configure equipment, view monitoring results and maintain the system;

[0069] The software control system, embedded in the control and management unit, is responsible for the control logic of the overall system, the implementation of data processing algorithms, and the response to user instructions.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] (1) The present invention collects partial discharge signals in real time through ultra-high frequency sensors and performs adaptive noise reduction processing in combination with environmental noise data, and adopts a hybrid optimization algorithm to dynamically adjust the discharge detection threshold, thereby achieving high-precision real-time monitoring of partial discharge of GIS equipment. This solution not only improves the accuracy and reliability of partial discharge detection, but also can adaptively adjust monitoring parameters according to changes in environmental noise, effectively reducing the false alarm rate and improving the practicality of the system;

[0072] (2) The present invention adopts a multi-level adaptive wavelet threshold denoising algorithm, combines a nonlinear scaling function to process the wavelet coefficients, and introduces an adjustment factor to achieve differentiated suppression of noise in different frequency bands. This denoising method can effectively suppress the interference of environmental noise while retaining the characteristics of the partial discharge signal, thereby improving the signal-to-noise ratio of the signal;

[0073] (3) The hybrid optimization algorithm proposed in the present invention adopts a two-layer optimization structure. The first optimization algorithm explores the high-quality threshold area globally, and then uses the second optimization algorithm to perform local refinement, thereby achieving precise adjustment of the discharge detection threshold. The algorithm effectively balances detection sensitivity and accuracy, and improves the system's ability to identify different types of local discharges;

[0074] (4) The UHF sensor designed by the present invention adopts a multi-layer microstrip antenna structure, integrates a low-noise amplifier and has a temperature compensation function, achieving a broadband response characteristic of 300MHz to 3GHz. The sensor structure improves the sensitivity and stability of signal acquisition and ensures the reliability of monitoring data;

[0075] (5) The monitoring device of the present invention adopts a modular design, integrating functional modules such as signal acquisition, noise reduction processing, feature extraction, and threshold adjustment, and is equipped with a complete data storage and communication interface. This design improves the maintainability and scalability of the system, and facilitates implementation and subsequent maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0077] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0078] Figure 2 It is the optimization flow chart of the present invention;

[0079] Figure 3 It is a device framework diagram of the present invention. DETAILED DESCRIPTION

[0080] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0081] like Figure 1 As shown, first, the present invention provides a method for online monitoring of partial discharge of GIS equipment, comprising:

[0082] S1. Use UHF sensors to collect local discharge signals inside GIS equipment in real time and simultaneously obtain data from environmental noise sensors;

[0083] S2, performing noise reduction processing on the collected UHF signal based on the data of the environmental noise sensor, and performing time domain alignment on the signal according to a synchronization method to obtain a preprocessed analog signal;

[0084] S3, converting the preprocessed analog signal into a digital signal through a high-precision analog-to-digital converter;

[0085] S4, extracting key characteristic parameters of partial discharge signal from digital signal;

[0086] S5, applying a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key feature parameters;

[0087] S6. Based on the adjusted discharge detection threshold, an event trigger condition is set, and when the detected partial discharge signal amplitude and trigger times reach the event trigger condition, an alarm signal is issued.

[0088] Specifically, in one embodiment of the present invention, step S1 includes:

[0089] The UHF sensor is designed based on a multi-layer microstrip antenna structure, with an operating frequency range of 300MHz to 3GHz, and can effectively capture high-frequency electromagnetic wave signals generated by partial discharge. The sensor integrates a low-noise amplifier and has a temperature compensation function to ensure high sensitivity and stability of signal acquisition under different ambient temperatures. The UHF sensor is installed in key parts of GIS equipment, such as inside the switch valve body or near the connection point, to maximize the reception of high-frequency signals emitted by partial discharge.

[0090] At the same time, the system is equipped with an environmental noise sensor to synchronously collect noise data in the environment around the equipment. The environmental noise sensor is usually installed at a certain distance from the UHF sensor to effectively distinguish the partial discharge signal from the background noise. The data acquisition uses a high-speed data acquisition module to ensure that the data of the UHF sensor and the environmental noise sensor can be synchronously acquired under the same time reference. In order to achieve accurate synchronization, the system uses hardware triggering or software synchronization mechanism to ensure the consistency of the data of each sensor in time.

[0091] The collected partial discharge signals and environmental noise data are transmitted to the noise reduction processing module through a wired or wireless communication interface for subsequent signal processing and analysis. The entire data collection process is carried out in real time, ensuring that the monitoring system can respond to partial discharge events inside the GIS equipment in a timely manner, improving the real-time and accuracy of monitoring.

[0092] Specifically, in one embodiment of the present invention, step S2 includes:

[0093] S21. Analyze the data collected by the environmental noise sensor to determine the time-frequency characteristics of the environmental noise.

[0094] First, the environmental noise sensor collects the noise data in the environment around the GIS equipment in real time. These data are usually obtained in the form of time domain signals. The collected original noise signal is transmitted to the signal analysis unit through a high-precision data acquisition module. In the signal analysis unit, the fast Fourier transform (FFT) is used to convert the time domain signal into the frequency domain to obtain the spectrum information of the noise signal.

[0095] Next, the time-frequency characteristics of the noise signal are analyzed in detail using time-frequency analysis methods such as short-time Fourier transform (STFT) or wavelet transform. These methods can provide the frequency distribution of the noise signal in different time windows, thereby identifying the main frequency bands of the noise and their changing trends. Through time-frequency analysis, the main frequency components of the environmental noise, the frequency change rate, and the time fluctuation characteristics of the noise intensity can be determined.

[0096] Finally, the environmental noise data after time-frequency characteristic analysis will be used as benchmark information for subsequent noise reduction processing modules.

[0097] S22, using a multi-level adaptive wavelet threshold noise reduction algorithm to perform noise reduction on the collected UHF signal.

[0098] Step S22 includes:

[0099] S221, assuming that the input signal is y(t), wherein y(t)=x(t)+n(t), x(t) is the original partial discharge signal, and n(t) is the environmental noise;

[0100] S222, perform multi-level wavelet decomposition on the input signal y(t) to obtain the wavelet coefficients W of each layer j,k , where j represents the number of decomposition levels and k represents the coefficient position.

[0101] S223, calculate the adaptive threshold T according to the noise time-frequency characteristics of each layer j :

[0102]

[0103] Among them, σ j is the estimated noise standard deviation of the j-th layer wavelet coefficients, and N is the total length of the signal; the threshold calculation method can effectively distinguish between signals and noise and adapt to the noise characteristics of different levels.

[0104] S224, introduce nonlinear scaling function φ(W j,k)Threshold processing is performed on the wavelet coefficients:

[0105]

[0106] Among them, tanh represents the hyperbolic tangent function; a is the adjustment factor; this nonlinear scaling function further suppresses noise and improves the noise reduction effect while maintaining the main characteristics of the signal.

[0107] S225, the processed wavelet coefficients Perform wavelet reconstruction to obtain the denoised signal Through multi-level wavelet reconstruction, the denoised partial discharge signal is restored and the main characteristic information of the signal is retained.

[0108] The present invention uses wavelet transform to perform multi-level decomposition on the original signal, decomposes the signal into different frequency levels, and effectively separates the noise component and useful partial discharge information in the signal. Multi-level decomposition can capture the characteristics of the signal at different scales and improve the precision of noise reduction. According to the statistical characteristics of the wavelet coefficients of each layer, the adaptive threshold T is calculated. j , ensuring that the threshold can dynamically adapt to the noise intensity at different levels. The introduction of adaptive threshold improves the noise suppression ability of the noise reduction algorithm and reduces signal distortion. By introducing the nonlinear scaling function φ(W j,k ) to finely process the wavelet coefficients. This function effectively weakens the noise component and enhances the signal-to-noise ratio while retaining the main signal characteristics. Compared with the traditional soft threshold and hard threshold methods, the nonlinear processing method can better balance the noise reduction effect and signal fidelity. The processed wavelet coefficients are reconstructed to restore the denoised partial discharge signal. Multi-level wavelet reconstruction can integrate the noise reduction effects of each level and generate high-quality noise reduction signals.

[0109] S23. Performing time domain alignment on the denoised signal according to a predetermined synchronization method, including: determining a reference signal or a reference time point; applying an interpolation algorithm to adjust the sampling rate and time axis of the signal to synchronize it with the reference signal; using a cross-correlation method to evaluate and adjust the time offset between each signal; and generating an alignment signal under a unified time reference.

[0110] First, select a reference signal or set a reference time point as the standard for aligning all acquired signals. The reference signal is usually selected from a specific frequency band or a specific time in the signal after noise reduction processing to ensure the accuracy of alignment.

[0111] For signals collected by different sensors, since the sampling rates may be slightly different, the system uses interpolation algorithms (such as linear interpolation, spline interpolation, etc.) to adjust the sampling rate of each signal to keep it consistent with the sampling rate of the reference signal. At the same time, the time axis of the signal is adjusted to align the starting time of all signals to ensure subsequent processing under a unified time base.

[0112] To eliminate the possible slight time offsets between the signals, the system applies cross-correlation analysis to each pair of signals. By calculating the correlation coefficient between the signals, the optimal time delay is determined, and the time axis of each signal is adjusted accordingly to maximize the similarity between the signals. This process ensures that all signals are highly consistent in time.

[0113] After completing the time axis adjustment, the system unifies all signals to the set time base and generates signal data that has been aligned in the time domain.

[0114] S24, generating a pre-processed analog signal after noise reduction and time domain alignment processing.

[0115] The UHF signal after noise reduction in step S22 is integrated with the signal after time domain alignment in step S23 to ensure that all signal data are at the same time reference and noise suppression level. The final pre-processed analog signal is generated by comprehensive processing of the integrated signal. This signal not only has a high signal-to-noise ratio, but also maintains the key characteristics of the partial discharge signal, and is suitable for subsequent analog-to-digital conversion and feature extraction. After the pre-processed signal is generated, data verification is performed to check the integrity and consistency of the signal. By comparing the signal waveform and statistical characteristics before and after noise reduction, it is ensured that the pre-processed signal meets the requirements. The pre-processed analog signal is stored in the cache or storage module of the system and transmitted to the subsequent processing module, such as the analog-to-digital conversion module and the feature extraction module, through a wired or wireless communication interface.

[0116] Specifically, in one embodiment of the present invention, step S3 includes:

[0117] Use an analog-to-digital converter with high resolution, low noise and high linearity to meet the weak characteristics and high dynamic range requirements of partial discharge signals. An ADC with a resolution of at least 24 bits is often used, and an appropriate ADC architecture (such as Sigma-Delta or successive approximation register) is selected to optimize conversion accuracy and sampling efficiency.

[0118] Design high-quality signal interface parts, including matching circuits, filters, and shielded transmission lines, to ensure that the analog signals input to the ADC are pure and free of interference. The pre-processed analog signals are transmitted to the ADC module through high-quality shielded cables or differential signal transmission lines to prevent external electromagnetic interference from affecting the signals. Through front-end amplification and filtering, the signal-to-noise ratio of the signal is improved, the error in the analog-to-digital conversion process is reduced, and the signal input to the ADC is pure.

[0119] Equipped with high-precision clock sources, such as temperature-controlled crystal oscillators (TCXOs) or phase-locked loop (PLL) clocks, to provide stable sampling clocks and ensure synchronous sampling of multi-channel signals. Use high-stability clock components such as phase-locked loops (PLLs) or temperature-controlled crystal oscillators (TCXOs) to ensure the accuracy and stability of the sampling clock and avoid the influence of clock deviation on signal conversion.

[0120] According to the spectrum characteristics of the partial discharge signal, the sampling rate of the ADC is reasonably set to ensure the high time resolution and frequency response of the signal. Generally, the sampling rate must meet the Nyquist sampling theorem and be at least twice the highest frequency of the signal. In order to capture the details of the ultra-high frequency signal, the sampling rate of the ADC in the system is set between 10kS / s and 100kS / s. By dynamically adjusting the sampling rate, it can flexibly respond to the signal change requirements in different monitoring scenarios and ensure the high time resolution and frequency response of the signal.

[0121] After the pre-processed analog signal is input to the ADC, it is converted by a successive approximation register (SAR), Sigma-Delta or other high-precision conversion architecture ADC to generate a corresponding digital signal. The quantization error in the conversion process is minimized through noise shaping and oversampling technology to ensure the accuracy of the digital signal. By selecting a high-resolution, low-noise ADC, high-precision sampling of weak partial discharge signals is achieved to ensure accurate capture of signal characteristics.

[0122] The converted digital signal is transmitted through a high-speed digital interface (such as SPI, I 2 C or LVDS) to the subsequent digital signal processing module. Differential signal transmission and error detection mechanism are used to ensure the reliability and integrity of data transmission and prevent data loss or errors during transmission.

[0123] Data cache and buffer are configured between the ADC module and the subsequent processing unit to cope with instantaneous data flow fluctuations and avoid data overflow or loss. The cache adopts double buffer or ring buffer design to achieve continuous storage and efficient reading of data, ensuring stable storage and efficient reading of continuous data streams.

[0124] Specifically, in one embodiment of the present invention, step S4 includes:

[0125] The digital signal is detrended to eliminate the offset and drift in the signal. Then, the signal is normalized to standardize the signal amplitude to a uniform range.

[0126] The continuous digital signal is divided into predetermined time windows to ensure that each window contains a complete partial discharge event.

[0127] Time domain feature extraction:

[0128] Maximum discharge: In each time window, the peak energy of the partial discharge signal is identified and the maximum energy value is recorded.

[0129] Average discharge amount: Calculate the average energy value of the local discharge signal in each time window to reflect the overall energy level.

[0130] Discharge phase: By analyzing the electrical phase position of the signal, the specific electrical phase where the partial discharge occurs is determined to identify the location of the discharge source.

[0131] Discharge frequency: Count the number of local discharges that occur per unit time, reflecting the frequency of discharge activities.

[0132] Discharge duration: Measures the length of time from the start to the end of a single discharge event to evaluate the continuity of the discharge process.

[0133] Signal rise time: Calculates the time from the start of a partial discharge signal to when it reaches its peak value, used to evaluate the response speed of the discharge signal.

[0134] Signal decay rate: measures the rate at which a signal falls from its peak value to the baseline, reflecting the decay characteristics of the signal.

[0135] Frequency domain feature extraction:

[0136] Energy distribution: The time domain signal is converted to the frequency domain through fast Fourier transform (FFT) or wavelet transform to analyze the energy distribution of the partial discharge signal at different frequencies.

[0137] Spectral density: Calculate the power spectral density of the signal in each frequency band, evaluate the spectral characteristics of the signal, and assist in identifying different types of partial discharge.

[0138] Advanced feature extraction and combination:

[0139] Feature fusion: The features extracted from the time domain and frequency domain are fused to form a comprehensive feature vector to improve the descriptive ability and discrimination of the features.

[0140] Feature selection and dimensionality reduction: Apply principal component analysis (PCA) or correlation analysis methods to screen out the most representative and discriminative features, reduce redundant information, and optimize subsequent processing efficiency.

[0141] Calculate key characteristic parameters: Based on the extracted features, calculate and record key parameters such as maximum discharge amount, average discharge amount, discharge phase, discharge frequency, energy distribution, discharge duration, signal rise time and signal attenuation rate.

[0142] Data storage and management: The calculated characteristic parameters are stored in the system database to facilitate subsequent dynamic adjustment of thresholds, fault diagnosis and historical data analysis.

[0143] Specifically, Figure 2 As shown, in one embodiment of the present invention, step S5 includes:

[0144] S51, initializing the discharge detection threshold, and setting the maximum number of main iterations and the optimization target.

[0145] S52. In each main iteration, based on the discharge detection threshold and key characteristic parameters of the current main iteration, the detection sensitivity and accuracy index are calculated, that is, the optimization objective function U is optimized; the optimization objective function U is as follows:

[0146] U(θ)=w1·Sen(θ)+w2·Acc(θ)

[0147] Among them, w1 and w2 are weight coefficients, and Sen(θ) is the detection sensitivity:

[0148]

[0149] Acc(θ) is the detection accuracy:

[0150]

[0151] Specifically, when calculating the optimization objective function, the signal is detected according to the current output discharge detection threshold, the events exceeding the threshold are counted, and the real discharge events are identified using key characteristic parameters.

[0152] S53. Apply the first optimization algorithm to globally adjust the current discharge detection threshold and explore potential high-quality threshold areas. Specifically, adjust the discharge detection threshold according to the value of the optimization objective function, and record all threshold points whose corresponding objective function values ​​are greater than the preset target value according to the searched discharge detection threshold. Cluster these threshold points to obtain k cluster centers, and determine k sub-intervals with a preset radius r with the cluster center as the center. The union of these k sub-intervals is the high-quality threshold area.

[0153] S54: In the high-quality threshold region determined by the first optimization algorithm, the second optimization algorithm is applied to perform local refinement to obtain an adjusted discharge detection threshold:

[0154] S541, initialize the local threshold of the second optimization algorithm, set the maximum number of local iterations and the local optimization objective function; specifically, the local optimization objective function calculation formula is as follows:

[0155] U local (θ′ k )=β1·Sen(θ′ k )+β2·Acc(θ′ k )+β3·Stab(θ′ k )

[0156] Among them, Sen(θ′ k ),Acc(θ′ k ) is calculated in the same way as the overall optimization objective function; Stab(θ′ k ) is the threshold stability index: β1, β2, β3 are weight coefficients; γ is the stability adjustment parameter.

[0157] S542, in each local iteration, calculating the local optimization target based on the current local threshold and key feature parameters;

[0158] S543, adjusting the formula according to the second optimization algorithm and updating the local threshold;

[0159] S544, evaluate whether the local threshold satisfies the local optimization goal. If not, proceed to step S542; if satisfied, output the adjusted discharge detection threshold; the local optimization goal is to reach the maximum number of local iterations, or the local optimization objective function reaches the maximum.

[0160] S55, combining the adjustment results of the first optimization algorithm and the second optimization algorithm to determine a final updated discharge detection threshold;

[0161] S56, evaluate whether the updated discharge detection threshold meets the predetermined optimization goal. If not, return to step S52 to continue iteration; if satisfied, terminate the optimization process. The predetermined optimization goal includes reaching the maximum number of main iterations, or: taking the updated discharge detection threshold as the threshold, calculating the optimization objective function U, and determining whether the difference between the U values ​​of two consecutive main iterations is less than the preset difference. If so, it means that the predetermined optimization goal is met.

[0162] Specifically, in this embodiment, in each main iteration, the high-quality threshold region update formula of the first optimization algorithm is:

[0163]

[0164] In the formula, Θ t is the high-quality threshold region of the tth main iteration; λ is the attenuation factor; U(Θ t) is the optimization objective function under the current high-quality threshold region, and its value is obtained by weighted summation of detection accuracy and detection sensitivity under the current threshold; μ is the amplitude factor, which adjusts the amplitude of the sine function; κ is the frequency factor, which controls the oscillation frequency of the sine function; φ(P) is a function of the key characteristic parameter P, which is used to adjust the oscillation angle. Among them, It is the derivative of the optimization objective function. If this derivative is greater than 0, it means that increasing the threshold will improve performance, and the threshold should be increased in the next step. If this derivative is less than 0, it means that reducing the threshold will improve performance, and the threshold should be reduced in the next step. That is, the size of this derivative value determines the adjustment step size and adjustment direction.

[0165] In each local iteration, the adjustment formula of the second optimization algorithm is:

[0166]

[0167] In the formula, θ′ t is the discharge detection threshold in the kth local iteration; γ is the learning rate, which controls the adjustment step size; δ is the adjustment factor, which balances the gradient influence; U local It is the local optimization objective function; The local optimization objective function U local About the discharge detection threshold θ′ k The numerical approximation of the second-order derivative of the key characteristic parameter P represents the curvature information; β is the scaling factor used to control the nonlinear response of the denominator; f(D) is the function of the discharge duration D.

[0168] In step S55, the comprehensive adjustment formula is:

[0169]

[0170] In the formula, θ t+1 is the final updated discharge detection threshold; α1, α2, α3 are weight factors, which respectively control the influence of each adjustment; η is the gradient amplification factor; To optimize the objective function U with respect to the discharge detection threshold θ′ k and the gradient of the key feature parameter P, indicating the optimization direction; tanh represents the hyperbolic tangent function; log is the natural logarithm function; g(R) is the function of the signal attenuation rate R. Among them, U(θ′ k ,P) is the sum of θ′ k As the detection threshold, the signal is detected, the events exceeding the threshold are counted, and the key characteristic parameters are used to identify the real discharge events, so as to calculate the detection sensitivity and detection accuracy, and then calculate the value of the optimization objective function U(θ′ k ,P).

[0171] The hybrid optimization algorithm of the present invention is described with a specific example:

[0172] The hybrid optimization algorithm of this embodiment integrates the ant colony optimization algorithm ACO and the bat optimization algorithm BA to design an efficient hybrid optimization algorithm with global search capability to achieve optimal adjustment of the detection threshold.

[0173] 1. Overall process of hybrid optimization algorithm (H-ACO-BA)

[0174] The hybrid optimization algorithm combines the path building capability of ACO and the local search capability of BA, aiming to fully utilize the advantages of the two algorithms and improve optimization efficiency and accuracy. The specific process is as follows:

[0175] Initialization phase:

[0176] Ant colony initialization: set the number of ant colonies, initial pheromone value, heuristic function and other parameters. Each ant randomly selects its initial position at the beginning.

[0177] Bat initialization: Set the initial parameters of the bat population, frequency range, speed, position, etc. Each bat is randomly distributed in the search space.

[0178] Iterative search phase:

[0179] Ant colony part:

[0180] Path construction: Each ant selects a path based on pheromone concentration and heuristic information to construct candidate solutions.

[0181] Pheromone update: According to the high-quality paths found by ants, the pheromone concentration is updated to enhance the attractiveness of the high-quality paths.

[0182] Bat part:

[0183] Frequency Update: Adjust the bat's speed and position based on its frequency.

[0184] Position update: Use the improved position update formula to perform local search around the current optimal position to improve the accuracy of the solution.

[0185] Sound wave regulation: By adjusting the frequency and loudness of bats’ sound waves, we can balance exploration and exploitation and avoid falling into local optimality.

[0186] Hybrid optimization steps:

[0187] The search results of ACO and BA are integrated to achieve collaborative optimization of global and local search through pheromone interaction and bat position update.

[0188] According to the optimization objective function (such as detection sensitivity and accuracy indicators), the optimization effect of the current threshold is evaluated to decide whether to continue iterating.

[0189] Termination conditions:

[0190] When the maximum number of main iterations is reached or the optimization objective function meets a predetermined condition, the optimization process is terminated and the final discharge detection threshold is output.

[0191] 2. Specific implementation steps

[0192] 2.1 Initialization phase

[0193] Initialize the discharge detection threshold, and set the maximum number of main iterations and optimization target.

[0194] Set the initial discharge detection threshold θ0. Determine the maximum number of main iterations T max . Define the optimization objective function U to evaluate the detection sensitivity and accuracy of the current threshold.

[0195] Initialize hybrid optimization algorithm parameters

[0196] Ant Colony Optimization (ACO) parameter settings:

[0197] Number of ants Ants . Initial value of pheromone τ0. Heuristic function η z =1 / (1+|P z |), where P z is the zth key characteristic parameter. Pheromone volatility ρ. Pheromone importance factor α and heuristic function importance factor β.

[0198] Bat Optimization (BA) parameter settings:

[0199] Bat population number Bats . Frequency range f min and f max . Learning rate γ. Adjustment factor δ. Scaling factor β. Number of local iterations T. Local optimization objective function U local .

[0200] 2.2 Global Search Phase (ACO Part)

[0201] Ant colony optimization algorithm is applied to adjust the threshold globally.

[0202] Each ant has a pheromone concentration τ i,j and the heuristic function η i,j , select the path according to the probability formula and construct the candidate threshold region Θ:

[0203]

[0204] Among them, α and β are the weight coefficients of the importance of pheromone and heuristic function respectively.

[0205] According to the pheromone update formula, the pheromone concentration is updated according to the optimal path found by the ants:

[0206] τ i,j (t+1)=(1-ρ)·τ i,j (t)+Δτ i,j (t)

[0207] Among them, Δτ i,j (t) is the pheromone increment contributed by the optimal ant path.

[0208] According to the ant colony search results, determine the potential high-quality threshold area Θ t .

[0209] 2.3 Local Search Phase (BA Part)

[0210] The bat optimization algorithm is applied for local refinement in the high-quality threshold area.

[0211] Use the initial path generated by ACO as the initial position of BA to increase the diversity of the search starting point:

[0212]

[0213] in, is the kth initial path generated by ACO.

[0214] Dynamically adjust the frequency range of the bat, according to the number of iterations t and the total number of iterations T:

[0215]

[0216] Speed ​​update formula:

[0217]

[0218] Among them, x * is the current global optimal position, The frequency of bats.

[0219] Position update formula:

[0220]

[0221] in, is the phase angle of the bat, τ i,j (t) is the pheromone concentration of path i to j in ACO, which is used to adjust the search step size of BA.

[0222] Adjust the frequency and loudness of bat sounds to balance exploration and exploitation and avoid falling into local optimality.

[0223] 2.4 Integration and Optimization Phase

[0224] The adjustment results of ACO and BA are combined to determine the final updated discharge detection threshold.

[0225] Feedback the local search results of BA to the pheromone update of ACO to enhance the dynamic adjustment ability of pheromone concentration.

[0226] According to the comprehensive adjustment formula, the threshold after BA optimization is further adjusted to obtain the final updated discharge detection threshold θ t+1 .

[0227] 2.5 Termination Condition Judgment

[0228] Evaluate whether the updated discharge detection threshold satisfies the predetermined optimization target. If not, return to step S52, set t=t+1, and continue the next main iteration. If it is satisfied, terminate the optimization process and output the final discharge detection threshold.

[0229] The present invention integrates the global path construction capability of ACO and the local search capability of BA, effectively avoiding the defect that a single algorithm is prone to fall into local optimality, and improving the globality and accuracy of the optimization results. Through the adaptive adjustment of ACO and BA parameters, the hybrid optimization algorithm can dynamically respond to changes in signal characteristics under different monitoring environments, maintaining efficient search capabilities and optimization effects. In the process of dynamically adjusting the discharge detection threshold, the hybrid optimization algorithm can more accurately balance detection sensitivity and accuracy, reduce false alarms and missed alarms, and improve the reliability of the overall monitoring system. Through the fusion and improvement of the algorithm, a higher optimization efficiency is achieved, the iteration time of the threshold adjustment is shortened, and the strict real-time requirements of the online monitoring system are met.

[0230] Specifically, in one embodiment of the present invention, step S6 includes:

[0231] S61. Define the components of event triggering conditions.

[0232] The event triggering conditions are mainly composed of the following two elements:

[0233] Partial discharge signal amplitude threshold A th : Set the threshold that the signal amplitude must exceed based on the adjusted discharge detection threshold.

[0234] Trigger count threshold N th : Within a certain time window, the number of times the partial discharge signal exceeds the amplitude threshold must reach or exceed a predetermined value.

[0235] S62: Determine the time window and the number of triggers.

[0236] Time window T w : Set a fixed time period (such as 1 minute, 5 minutes, etc.) as the evaluation window for the event trigger condition.

[0237] Trigger times N th :Based on the operating environment and historical data of the GIS equipment, determine the number of partial discharge signal triggering times that needs to be achieved within the time window.

[0238] S63. Dynamically adjust the triggering conditions.

[0239] Dynamically adjust A according to system operation status and environmental changes th and N th To adapt to different monitoring needs and ensure the accuracy and timeliness of alarms.

[0240] S64, retrieve the digital signal, monitor the amplitude of the partial discharge signal in real time, and obtain key characteristic parameters of the partial discharge signal.

[0241] S65: Signal amplitude detection.

[0242] Amplitude comparison: compare the amplitude of the partial discharge signal A(t) monitored in real time with the set amplitude threshold A th For comparison:

[0243] If A(t)≥A th , a trigger event is recorded.

[0244] Trigger count statistics: within the set time window T w Within, the number of partial discharge signals exceeding the amplitude threshold value N(t) is counted.

[0245] S66: Evaluate trigger conditions.

[0246] Conditional judgment: At the end of each time window, determine whether the statistical trigger count N(t) reaches or exceeds the trigger count threshold N th :

[0247] If N(t)≥Nth, the trigger condition is met.

[0248] Multi-parameter judgment: In addition to amplitude and number, other key characteristic parameters (such as discharge frequency, energy distribution, etc.) can also be combined for comprehensive evaluation to improve the accuracy of judgment.

[0249] S67. Generation and sending of alarm signals.

[0250] Alarm signal generation: When the trigger conditions are met, the system generates an alarm signal, including the following forms:

[0251] Visual alarm: Display alarm information on the display screen of the user interface module, highlighting the alarm status. Auditory alarm: Sound prompts are issued through devices such as buzzers to alert on-site personnel. Remote alarm: Alarm information is sent to the remote monitoring system or mobile device through the communication interface module to achieve immediate response.

[0252] Alarm information content: The alarm information should contain the following content to facilitate quick positioning and processing:

[0253] Alarm type: partial discharge signal abnormality alarm. Alarm time: specific alarm trigger timestamp. Alarm parameters: key characteristic parameter values ​​exceeding the threshold, such as signal amplitude, trigger times, etc. Equipment information: GIS equipment number or location identifier where the alarm occurred.

[0254] In addition, if Figure 3 As shown, the present invention also provides an online monitoring device for partial discharge of GIS equipment, the device is used to execute any of the above methods, and the device includes:

[0255] The UHF acquisition module is used to collect the partial discharge signals inside the GIS equipment in real time and synchronously obtain the data of the environmental noise sensor.

[0256] It is equipped with a UHF sensor based on a microstrip antenna structure, with an operating frequency range of 300MHz to 3GHz, a multi-layer microstrip antenna design, an integrated low-noise amplifier, a broadband response characteristic, a compact packaging structure, and a temperature compensation function. The UHF signal generated by the partial discharge inside the GIS equipment is captured through the microstrip antenna. The environmental noise sensor is used to collect noise data from the surrounding environment and provide noise background information to assist in signal noise reduction processing.

[0257] The noise reduction processing module is used to perform noise reduction processing on the collected ultra-high frequency signal based on the data of the environmental noise sensor, and to align the signal in the time domain according to a predetermined synchronization method to generate a pre-processed analog signal.

[0258] It has a multi-level adaptive wavelet threshold denoising algorithm, which performs multi-level wavelet decomposition on the input signal y(t) = x(t) + n(t) and calculates the adaptive threshold T j , introduce a nonlinear scaling function φ(W j,k )Threshold processing is performed on the wavelet coefficients, and finally the noise-reduced signal is obtained through wavelet reconstruction.

[0259] After noise reduction, select a reference signal or a reference time point as the basis for alignment. Adjust the sampling rate and time axis of the signal to synchronize it with the reference signal. Evaluate and adjust the time offset between the signals to ensure that all signals are aligned under the same time base.

[0260] The analog-to-digital conversion module includes a high-precision analog-to-digital converter, which is used to convert the pre-processed analog signal into a digital signal.

[0261] It is equipped with a high-precision analog-to-digital converter (ADC) with high sampling rate and high resolution to ensure the capture of subtle partial discharge signal changes. It supports multi-channel input to meet the needs of collecting multiple sensor signals at the same time.

[0262] The feature extraction module is used to extract the key characteristic parameters of the partial discharge signal from the digital signal, including the maximum discharge amount, average discharge amount, discharge phase, discharge frequency, energy distribution, discharge duration, signal rise time and signal attenuation rate.

[0263] Specifically, signal processing and feature analysis algorithms are used to calculate the above key feature parameters. Technical means such as filtering, window analysis, and frequency domain conversion are used to ensure the accuracy and real-time performance of feature extraction.

[0264] The threshold dynamic adjustment module is configured with a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key feature parameters to achieve a balance between detection sensitivity and accuracy.

[0265] Its built-in hybrid optimization algorithm combines the ant colony optimization algorithm and the bat optimization algorithm. ACO is used for global search to explore potential high-quality threshold areas. BA performs local refinement within the high-quality threshold area determined by ACO to improve optimization accuracy.

[0266] The event trigger and alarm module sets the event trigger condition based on the adjusted discharge detection threshold, and sends out an alarm signal when the amplitude and trigger times of the detected partial discharge signal reach the predetermined conditions.

[0267] The event trigger conditions include the amplitude threshold and the trigger number threshold, and the real-time amplitude A(t) of the partial discharge signal is monitored and the set threshold A th Compare and count in time window T w The number of times N(t) exceeds the threshold.

[0268] Judgment condition: If N(t)≥N th , then trigger the alarm.

[0269] Alarm signal generation and transmission: Visual alarm: Display alarm information on the display screen of the user interface module, highlighting the alarm status. Auditory alarm: Sound prompts are issued through devices such as buzzers. Remote alarm: Send alarm information to the remote monitoring system or mobile device through the communication interface module to achieve immediate response.

[0270] Alarm information content: alarm type, alarm time, key characteristic parameter value exceeding the threshold, equipment number or location identification, etc.

[0271] Alarm records and log management: Record detailed information of each alarm event and generate alarm logs to facilitate subsequent analysis and fault diagnosis.

[0272] The control and management unit is used to control and coordinate the operation of each module.

[0273] It uses a microprocessor or embedded controller to execute control logic and management tasks. This unit is used to control the distribution and scheduling of signals to ensure that each module works in coordination according to the predetermined workflow. It monitors the operating status of each module in real time, detects abnormal conditions and takes appropriate measures. It manages system resources, such as processor time, storage space, power distribution, etc., to optimize system performance.

[0274] The data storage module is used to save historical monitoring data and event records.

[0275] The storage medium may adopt non-volatile memory, such as solid state drive (SSD), built-in flash memory, etc., to ensure the integrity of data in the event of power failure.

[0276] Save the collected original signal, pre-processed signal and extracted characteristic parameters. Save detailed information of alarm events, including time, parameters, processing status, etc. Generate system operation logs, record operation history and abnormal events, and support system maintenance and fault analysis.

[0277] The communication interface module includes a wired or wireless communication interface for exchanging data with an external monitoring system to ensure real-time transmission and remote access of monitoring data.

[0278] Interface types are divided into: Wired communication interface: such as Ethernet interface (Ethernet), serial port (RS-232 / RS-485), etc. Wireless communication interface: such as Wi-Fi, Bluetooth, 4G / 5G, etc., select the appropriate wireless communication technology according to the application scenario. Communication protocols can use TCP / IP, HTTP / HTTPS, MQTT, etc.

[0279] The module transmits the collected monitoring data to the remote monitoring center or cloud platform in real time, and receives remote instructions to implement functions such as equipment configuration and parameter adjustment.

[0280] The user interface module includes a display screen and control buttons, which facilitate users to configure the equipment, view monitoring results and maintain the system.

[0281] The display screen uses a liquid crystal display (LCD) or an organic light-emitting diode display (OLED): used to display real-time monitoring data, alarm information, system status, etc. The control button is a physical button or a touch button: used for manual operation of the equipment, such as start, stop, reset, menu navigation, etc. Interface design: user-friendly graphical interface (GUI): provides intuitive monitoring data display, alarm history, system settings and other functions. Visual charts: display the real-time change trend of key characteristic parameters to help users quickly understand the system status.

[0282] The software control system, embedded in the control and management unit, is responsible for the control logic of the overall system, the implementation of data processing algorithms, and the response to user instructions.

[0283] The control system is equipped with modular programming, real-time operating system, signal processing algorithm, optimization algorithm, alarm algorithm, command interpretation and execution, and feedback mechanism.

[0284] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The on-line monitoring method for partial discharge of GIS equipment is characterized by: include: S1. Use UHF sensors to collect local discharge signals inside GIS equipment in real time and simultaneously obtain data from environmental noise sensors; S2, performing noise reduction processing on the collected UHF signal based on the data of the environmental noise sensor, and performing time domain alignment on the signal according to a synchronization method to obtain a preprocessed analog signal; S3, converting the preprocessed analog signal into a digital signal through a high-precision analog-to-digital converter; S4, extracting key characteristic parameters of partial discharge signal from digital signal; S5, applying a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key feature parameters; S6. Based on the adjusted discharge detection threshold, an event trigger condition is set, and when the detected partial discharge signal amplitude and trigger times reach the event trigger condition, an alarm signal is issued.

2. The method for online monitoring of partial discharge of GIS equipment according to claim 1, characterized in that: Step S2 includes: S21, analyzing data collected by the environmental noise sensor to determine the time-frequency characteristics of the environmental noise; S22, using a multi-level adaptive wavelet threshold noise reduction algorithm to perform noise reduction processing on the collected UHF signal; S23, performing time domain alignment on the noise-reduced signal according to a predetermined synchronization method, including: determining a reference signal or a reference time point; applying an interpolation algorithm to adjust the sampling rate and time axis of the signal to synchronize it with the reference signal; using a cross-correlation method to evaluate and adjust the time offset between each signal; generating an alignment signal under a unified time reference; S24, generating a pre-processed analog signal after noise reduction and time domain alignment processing.

3. The method for online monitoring of partial discharge of GIS equipment according to claim 2, characterized in that: Step S22 includes: S221, assuming that the input signal is y(t), wherein y(t)=x(t)+n(t), x(t) is the original partial discharge signal, and n(t) is the environmental noise; S222, perform multi-level wavelet decomposition on the input signal y(t) to obtain the wavelet coefficients W of each layer j,k , where j represents the number of decomposition levels and k represents the coefficient position; S223, calculate the adaptive threshold T according to the noise time-frequency characteristics of each layer j : Among them, σ j is the estimated noise standard deviation of the j-th layer wavelet coefficients, and N is the total length of the signal; S224, introduce nonlinear scaling function φ(W j,k )Threshold processing is performed on the wavelet coefficients: Wherein, tanh represents the hyperbolic tangent function; a is the adjustment factor; S225, the processed wavelet coefficients Perform wavelet reconstruction to obtain the denoised signal 4. The method for online monitoring of partial discharge of GIS equipment according to claim 1, characterized in that: The key characteristic parameters of partial discharge signals include: Maximum discharge, which indicates the maximum energy value in the partial discharge signal; Average discharge, which indicates the average energy value of the partial discharge signal; Discharge phase, which indicates the electrical phase position where partial discharge occurs; Discharge frequency refers to the number of discharges that occur per unit time; Energy distribution, which indicates the distribution of partial discharge signals at different energy levels; Discharge duration, which indicates the duration of a single discharge event; Signal rise time, which indicates the time from the start of the partial discharge signal to the peak value; Signal decay rate refers to the rate at which the partial discharge signal recovers from the peak value to the baseline.

5. The method for online monitoring of partial discharge of GIS equipment according to claim 4, characterized in that: Step S5 includes: S51, initializing the discharge detection threshold, and setting the maximum number of main iterations and the optimization target; S52, in each main iteration, based on the discharge detection threshold and key characteristic parameters of the current main iteration, calculating the detection sensitivity and accuracy index; S53, applying the first optimization algorithm to globally adjust the current discharge detection threshold to explore potential high-quality threshold areas; S54: In the high-quality threshold region determined by the first optimization algorithm, the second optimization algorithm is applied to perform local refinement to obtain an adjusted discharge detection threshold: S541, initializing the local threshold of the second optimization algorithm, setting the maximum number of local iterations and the local optimization objective function; S542, in each local iteration, calculating the local optimization target based on the current local threshold and key feature parameters; S543, adjusting the formula according to the second optimization algorithm and updating the local threshold; S544, evaluating whether the local threshold meets the local optimization target, if not, proceeding to step S542; if satisfied, outputting the adjusted discharge detection threshold; S55, combining the adjustment results of the first optimization algorithm and the second optimization algorithm to determine a final updated discharge detection threshold; S56, evaluating whether the updated discharge detection threshold meets the predetermined optimization target. If not, returning to step S52 to continue iteration; if satisfied, terminating the optimization process.

6. The method for online monitoring of partial discharge of GIS equipment according to claim 5, characterized in that: In each main iteration, the quality threshold region update formula of the first optimization algorithm is: In the formula, Θ t is the high-quality threshold region of the tth main iteration; λ is the attenuation factor; U(Θ t ) is the optimization objective function under the current high-quality threshold area, and its value is obtained by weighted summation of detection accuracy and detection sensitivity under the current threshold; μ is the amplitude factor, which adjusts the amplitude of the sine function; κ is the frequency factor, which controls the oscillation frequency of the sine function; φ(P) is a function of the key characteristic parameter P and is used to adjust the oscillation angle.

7. The method for online monitoring of partial discharge of GIS equipment according to claim 6, characterized in that: In each local iteration, the adjustment formula of the second optimization algorithm is: In the formula, θ′ t is the discharge detection threshold in the kth local iteration; γ is the learning rate, which controls the adjustment step size; δ is the adjustment factor, which balances the gradient influence; U local It is the local optimization objective function; The local optimization objective function U local About the discharge detection threshold θ′ k The numerical approximation of the second-order derivative of the key characteristic parameter P represents the curvature information; β is the scaling factor used to control the nonlinear response of the denominator; f(D) is the function of the discharge duration D.

8. The method for online monitoring of partial discharge of GIS equipment according to claim 7, characterized in that: In step S55, the comprehensive adjustment formula is: In the formula, θ t+1 is the final updated discharge detection threshold; α1, α2, and α3 are weight factors, which respectively control the influence of each adjustment; η is the gradient amplification factor; To optimize the objective function U with respect to the discharge detection threshold θ′ k and the gradient of the key feature parameter P, indicating the optimization direction; tanh represents the hyperbolic tangent function; log is the natural logarithm function; g(R) is the function of the signal attenuation rate R.

9. The method for online monitoring of partial discharge of GIS equipment according to claim 1, characterized in that: The UHF sensor is a UHF sensor based on a microstrip antenna structure. The operating frequency band covers 300MHz to 3GHz. It adopts a multi-layer microstrip antenna design, integrates a low-noise amplifier, has broadband response characteristics, adopts a compact packaging structure, and has temperature compensation function.

10. GIS equipment partial discharge online monitoring device, characterized in that: The device is used to perform the method according to any one of claims 1 to 9, and the device comprises: UHF acquisition module, used to collect partial discharge signals inside GIS equipment in real time and simultaneously obtain data from environmental noise sensors; A noise reduction processing module is used to perform noise reduction processing on the collected UHF signal based on the data of the environmental noise sensor, and to perform time domain alignment on the signal according to a predetermined synchronization method to generate a preprocessed analog signal; An analog-to-digital conversion module, including a high-precision analog-to-digital converter, for converting the preprocessed analog signal into a digital signal; Feature extraction module, used to extract key characteristic parameters of partial discharge signals from digital signals, including maximum discharge amount, average discharge amount, discharge phase, discharge frequency, energy distribution, discharge duration, signal rise time and signal attenuation rate; The threshold dynamic adjustment module is configured with a hybrid optimization algorithm to dynamically adjust the discharge detection threshold according to the extracted key characteristic parameters to achieve a balance between detection sensitivity and accuracy; The event trigger and alarm module sets the event trigger conditions based on the adjusted discharge detection threshold, and issues an alarm signal when the amplitude and trigger times of the detected partial discharge signal reach the predetermined conditions; Control and management unit, used to control and coordinate the operation of each module; Data storage module, used to save historical monitoring data and event records; Communication interface module, including wired or wireless communication interface, used to exchange data with external monitoring system to ensure real-time transmission and remote access of monitoring data; User interface module, including display screen and control buttons, to facilitate users to configure equipment, view monitoring results and maintain the system; The software control system, embedded in the control and management unit, is responsible for the control logic of the overall system, the implementation of data processing algorithms, and the response to user instructions.

Images