A GIS loop resistance detection method and system

Abstract

The application relates to the technical field of loop resistance detection, and discloses a GIS loop resistance detection method and system, which comprises the following steps: collecting total output current, shunt current and DC voltage drop signals in a GIS loop resistance detection process; performing hardware anti-interference pretreatment on the DC voltage drop signals; tracking power frequency deviation of the DC voltage drop signals, filtering 50Hz fundamental waves and odd harmonics, reserving DC voltage drop effective signals, decomposing the DC voltage drop effective signals into different sub-frequency bands, calculating oscillation intensity comprehensive scores and adaptive threshold values of the sub-frequency bands, using wavelet packet adaptive denoising to denoise the DC voltage drop effective signals, performing wavelet packet reconstruction on the denoising results, obtaining real DC voltage drops, combining the total output current and the shunt current, and completing GIS loop resistance detection. The application can adaptively set soft and hard threshold values, and improves the precision and reliability of loop resistance detection.

Description

Technical Field

[0001] This invention relates to the field of loop resistance detection technology, and specifically to a method and system for detecting the loop resistance of a GIS circuit. Background Technology

[0002] During long-term operation, GIS equipment is prone to problems such as contact surface oxidation and loose bolts, which can lead to poor contact and seriously threaten the safe and stable operation of the power grid. Therefore, conductive loop resistance testing is a core and mandatory inspection item in GIS equipment maintenance. Existing loop resistance testing methods are generally based on Ohm's law, calculating resistance by collecting the voltage across the equipment and the test current. Conventional testing methods require removing the grounding copper busbars on both sides of the equipment to avoid ground grid current shunting errors, which is cumbersome and inefficient. If the grounding wire is retained for testing, it is necessary to rely on multi-port resistance network theory and impedance matrix algorithms to analyze the contact resistance of each segment, which requires extremely high signal acquisition quality.

[0003] However, GIS equipment is deployed in the complex electromagnetic environment of substations, where weak microvolt-level voltage and milliampere-level current measurement signals are susceptible to multi-source interference. Nearby busbars and transformers can couple and generate 50Hz power frequency interference, while switch opening and closing operations can excite broadband high-frequency oscillation signals from 10kHz to 5MHz. This type of high-frequency interference can couple into the detection equipment through spatial radiation and line conduction, causing sampling data disorder, abnormal peaks in the measurement waveform, and even damage to the front-end input circuit. Currently, traditional wavelet denoising methods are commonly used to process interference signals, but conventional soft and hard thresholds are fixed and difficult to adapt to transient impact interference characteristics. When threshold parameters are not properly matched, excessive smoothing can lead to loss of effective signals, or insufficient denoising can result in residual interference, making it impossible to accurately separate weak DC signals mixed with noise, significantly reducing the accuracy and reliability of GIS loop resistance detection. Summary of the Invention

[0004] This invention provides a method and system for detecting the loop resistance of GIS, to solve the problem of insufficient accuracy and reliability of loop resistance detection caused by unreasonable soft and hard threshold settings when using wavelet denoising to denoise the sampled data for GIS loop resistance detection. The specific technical solution adopted is as follows: In a first aspect, one embodiment of the present invention provides a method for detecting the resistance of a GIS loop, the method comprising the following steps: Collect the total output current, shunt current, and DC voltage drop signals during the GIS loop resistance detection process; Combined with RC low-pass filter and power supply filter, it attenuates wideband high-frequency interference and completes hardware anti-interference preprocessing for DC voltage drop signals; The DC voltage drop signal is tracked by the power frequency deviation, and the 50Hz fundamental wave and odd harmonics are filtered out to retain the effective DC voltage drop signal. The effective DC voltage drop signal is then decomposed into different sub-frequency bands. Based on the wavelet packet coefficients and corresponding coefficient sequences of the sub-frequency bands, the comprehensive score of the oscillation intensity of the sub-frequency bands is calculated. Combined with the VisuShrink universal threshold under interference-free conditions, the adaptive threshold of the sub-frequency bands is calculated. Based on the adaptive threshold of the sub-frequency bands and the comprehensive score of oscillation intensity, wavelet packet adaptive denoising is used to denoise the effective DC voltage drop signal. The denoising result is then reconstructed using wavelet packets to obtain the true DC voltage drop. Combined with the total output current and the shunt current, the resistance detection of the GIS loop is completed.

[0005] Furthermore, odd harmonic suppression is achieved through a three-stage complex bandstop filter.

[0006] Furthermore, the specific calculation method for the comprehensive score of the oscillation intensity of the sub-frequency band is as follows: Calculate the peak factor of the sub-band based on the coefficient sequence corresponding to the sub-band, and assign values ​​to the indicator terms of the sub-band based on the peak factor of the sub-band. Hilbert transform is performed on the wavelet packet coefficients of the sub-band to obtain the analytic signal. The instantaneous envelope of the analytic signal is extracted, and exponential decay fitting is performed on the instantaneous envelope to obtain the decay time constant. Based on the decay time constant, the hyperbolic tangent term of the sub-band is calculated. The difference between the number 1 and the normalized value of the energy spectral entropy of the coefficient sequence corresponding to the sub-band is denoted as the correction term of the sub-band. The weighted sum of the indicator, hyperbolic tangent, and correction terms of the sub-band is denoted as the comprehensive score of the sub-band's oscillation intensity. The sum of the weight coefficients of the indicator, hyperbolic tangent, and correction terms of the sub-band is 1.

[0007] Furthermore, the peak factor of the sub-band is the ratio of the absolute value of the peak value of the sub-band to the root mean square value of the sub-band.

[0008] Furthermore, the specific method for assigning values ​​to the indicator items of the sub-frequency bands is as follows: When the peak factor of a sub-band is greater than the first determination threshold, the indicator of the sub-band is assigned a value of 1; otherwise, the indicator of the sub-band is assigned a value of 0.

[0009] Furthermore, the specific method for obtaining the hyperbolic tangent term of the sub-band is as follows: The ratio of the attenuation time constant of the sub-band to the preset reference attenuation time constant is used as the independent variable of the hyperbolic tangent function, and the calculated value of the hyperbolic tangent function is denoted as the hyperbolic tangent term of the sub-band.

[0010] Furthermore, the specific calculation method for the adaptive threshold of the sub-frequency band is as follows: The product of the logarithm of the number of coefficients in the coefficient sequence corresponding to the sub-band and the number 2 is denoted as the first product of the sub-band. The product of the arithmetic square root of the first product of the sub-band and the standard deviation of the noise in the sub-band is denoted as the basic threshold. The standard deviation of the noise in the sub-band is the ratio of the median of the absolute values ​​of the coefficients in the coefficient sequence corresponding to the sub-band to the constant 0.6745. Based on the basic threshold, the comprehensive score of the sub-band's oscillation intensity, the peak factor, and the coefficient sequence corresponding to the sub-band, the adaptive threshold of the sub-band is calculated.

[0011] Furthermore, when performing wavelet packet adaptive denoising, the specific classifications of not applying thresholds to sub-frequency bands, applying soft thresholding to sub-frequency bands, or applying hard thresholding are as follows: When the comprehensive score of the oscillation intensity of a sub-band is less than the first threshold of the comprehensive score, no threshold is applied to the sub-band, and all wavelet coefficients are retained. When the comprehensive score of the oscillation intensity of a sub-band is less than the second threshold of the comprehensive score but greater than or equal to the first threshold of the comprehensive score, soft thresholding is used. For any coefficient in the coefficient sequence corresponding to the sub-band, the difference between the absolute value of the coefficient and the adaptive threshold of the sub-band is recorded as the first difference of the coefficient. When the coefficient is greater than 0, the sign feature value of the coefficient is assigned to 1. When the coefficient is less than 0, the sign feature value of the coefficient is assigned to -1. When the coefficient is equal to 0, the sign feature value of the coefficient is assigned to 0. The product of the first difference of the coefficient and the maximum value of the number 0 and the sign feature value is recorded as the soft threshold of the coefficient. When the comprehensive score of the oscillation intensity of a sub-band is greater than or equal to the second threshold of the comprehensive score, hard thresholding is used to retain the coefficients in the coefficient sequence corresponding to the sub-band whose absolute value is greater than the adaptive threshold of the sub-band.

[0012] Furthermore, the specific method for combining the total output current and the shunt current to complete the GIS loop resistance detection includes: The sum of the shunt currents on both sides is subtracted from the total output current and recorded as the actual measured current; the actual DC voltage drop is divided by the actual measured current and recorded as the GIS loop resistance.

[0013] Secondly, embodiments of the present invention also provide a GIS loop resistance detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0014] The beneficial effects of this invention are: This application employs a combined "hardware pre-suppression + digital dynamic tracking" defense strategy. At the physical level, it eliminates common-mode interference, high-frequency conducted noise, and ultra-high-frequency radio frequency interference, preventing problems such as ADC saturation and instrument overload caused by interference signals. At the digital level, it tracks 50Hz power frequency drift in real time and accurately isolates it. This method thoroughly addresses various interferences in the DC voltage drop signal, resolving the technical pain points of useful signals being submerged and waveform distortion occurring in the strong electromagnetic environment of substations. It obtains effective DC voltage drop signals, providing a clean signal foundation for accurate calculation of contact resistance. Then, considering that the arc reignition at the moment of circuit breaker contact separation will produce a nanosecond-level voltage drop, the method evaluates the deviation of the impact intensity from the overall average signal energy of the frequency band by assessing the instantaneous peak value in the sub-band, the significance of the "long-tailed oscillation" characteristic of the signal in the sub-band, and the energy distribution. It calculates the comprehensive score of the oscillation intensity of the sub-band. The higher the comprehensive score of the oscillation intensity of the sub-band, the more severe the opening and closing interference of the sub-band, and the greater the difference between the signal corresponding to the sub-band and the real signal. Furthermore, by combining the VisuShrink universal threshold under interference-free conditions, the adaptive threshold of the sub-band is calculated. Based on the adaptive threshold of the sub-band and the comprehensive score of oscillation intensity, wavelet packet adaptive denoising is used to denoise the effective signal of DC voltage drop. Wavelet packet reconstruction is performed on the denoising result to obtain the true DC voltage drop. Combined with the total output current and the shunt current, the GIS loop resistance detection is completed. This solves the problem that when using wavelet denoising to denoise the interfered data sampled for GIS loop resistance detection, the unreasonable setting of soft and hard thresholds leads to insufficient accuracy and reliability of loop resistance detection. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic flowchart of a GIS loop resistance detection method provided in one embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1The diagram illustrates a flowchart of a GIS loop resistance detection method according to an embodiment of the present invention, which includes the following steps: Step S001: Collect the total output current, shunt current and DC voltage drop signals during the GIS loop resistance detection process.

[0019] The GIS loop resistance detection device consists of three parts: a main unit, a wireless probe, and supporting accessories.

[0020] The core of the main unit is a high-current source structure based on a phase-shifted full-bridge circuit. The main power supply supports parallel operation of multiple units, employs a constant current and voltage-limiting working mode, and its output current can be continuously and arbitrarily adjusted. The maximum output current of the main power supply can reach 500A, ensuring that the current flowing through the tested equipment after three-phase shunt and ground grid shunt meets the requirements of on-site testing standards. The GIS loop resistance detection device integrates integrated voltage and current probes, effectively simplifying on-site wiring layout. The probes adopt a wireless data transmission architecture, further improving the safety and testing efficiency of substation on-site operations.

[0021] The device injects a 300A-500A DC current into the grounding busbars at both ends of the GIS equipment through its built-in phase-shifted full-bridge high-current source. The device relies on a built-in high-precision shunt or sensor to collect the total output current of the system in real time. Simultaneously, a wireless current clamp meter based on the closed-loop Hall effect principle is used to collect the shunt current on both sides. Specifically, the wireless current clamp meter is connected to the grounding copper busbar that needs to maintain conductivity, i.e., each shunt branch, to achieve accurate acquisition of the shunt signal.

[0022] For the weak DC voltage drop signal in the main circuit, a high common-mode rejection ratio differential probe is used for acquisition, as this weak DC voltage drop signal is usually submerged in a complex electromagnetic interference background. Voltage test leads are clamped to both ends of the circuit breaker or disconnector under test, or to the equipment grounding busbar; the specific wiring method can be flexibly switched according to the on-site operating mode. After completing all wiring and equipment debugging, synchronous acquisition of the total output current, shunt current, and DC voltage drop signals can be carried out.

[0023] It should be noted that in actual GIS field testing, due to differences in the retention status of grounding bars and the number of shunt branches, this method supports three modes: single-sided shunt, double-sided shunt, and multi-sided shunt. In the single-sided shunt mode, only one side of the grounding bar has significant shunt, in which case a single wireless current clamp is used to collect the shunt current on that side. circuit resistance The calculation formula is In the bilateral shunt mode, if both grounding bars are not removed and shunt current exists on both sides, then two wireless current clamps are used to collect the shunt current separately. and circuit resistance The formula is For three-phase common GIS or complex grounding networks, there may be a multilateral current sharing pattern with three or more branch lines. In this case, use Wireless current clamps are used to collect the shunt current of each branch. , This indicates the number of shunt branches, i.e., the number of wireless current clamps, and the loop resistance. The calculation formula is Of the above formulas, This represents the actual DC voltage drop obtained after subsequent processing steps. This represents the total output current of the host. Various shunt modes do not change the core signal acquisition and processing flow of this method; they only dynamically select the appropriate shunt current superposition term based on the topology of the on-site grounding network, thereby achieving accurate loop resistance detection without disassembling the grounding copper busbar.

[0024] At this point, the total output current, shunt current, and DC voltage drop signals are obtained during the GIS loop resistance detection process.

[0025] Step S002, in conjunction with RC low-pass filter and power supply filter, attenuates wideband high-frequency interference, completing the hardware anti-interference preprocessing of DC voltage drop signal.

[0026] The DC voltage drop signal is affected by the complex electromagnetic environment at the site, directly reducing the accuracy of GIS circuit resistance detection. Interference at the substation site is mainly generated by various sources such as operating equipment and switch opening and closing operations. It can intrude into the measurement circuit through multiple paths, including electromagnetic coupling and conductive coupling, and strong interference signals can easily overwhelm the signal. The weak DC voltage drop signal is a significant contributor to the overall signal quality, ultimately leading to data distortion and detection failure. Therefore, it is necessary to accurately separate and extract the weak DC voltage drop signal and the shunt current signal from the noisy mixed signal under a strong electromagnetic field environment. This strong electromagnetic field environment consists of 50Hz power frequency induced current and high-frequency transient electromagnetic interference.

[0027] The main sources of interference in the field can be divided into two categories: steady-state power frequency interference and transient broadband radiation interference.

[0028] The steady-state power frequency interference primarily originates from nearby operating buses and power transformers. The signal components are a 50Hz fundamental frequency and its odd harmonics, with interference amplitudes reaching up to 10V, exhibiting a standard sine wave. The coupling paths for this type of interference include: distributed capacitance between the test cable and the ground; significant capacitive coupling in long-distance cabling scenarios, causing 50Hz power frequency interference to enter the measurement circuit; stray interference currents in the substation grounding grid; inconsistent grounding methods between the test instrument and the device under test, generating ground potential differences and inducing ground loop interference; and floating operation of the measurement system, with potential deviations between the oscilloscope probe grounding terminal and the ground of the system under test, introducing widespread common-mode interference. These issues cause multiple adverse effects: the power frequency interference amplitude is far higher than the effective signal of a μV-level DC voltage drop, completely masking the target measurement signal; strong interference voltages exceed the input range of the analog-to-digital converter, causing converter saturation distortion and abnormal signal acquisition; and under extreme conditions, strong interference can trigger instrument overload protection, directly interrupting the test.

[0029] The transient broadband radiated interference is mainly generated by the operation of energized bay switches near the substation. The signal frequency covers 10kHz-5MHz, with energy concentrated in the 100kHz-500kHz frequency band. The waveform is a pulse signal with a steep rising edge, short duration, and instantaneous interference voltage reaching hundreds of volts. The coupling paths of this type of interference include: electromagnetic pulses generated by switching operations are coupled to the probe front end in the form of spatial radiation, forming an antenna coupling effect; the interference signal is conducted through the grounding wire, intruding into the instrument's power ground and signal ground, inducing common-mode interference; and the close proximity of switching equipment and test cables forms strong electromagnetic coupling interference. The adverse effects of this type of interference include: high-frequency spike pulses superimposed on the sampling signal, causing complete failure of single-point and single-cycle sampling data; high-frequency interference generates ringing effects inside the measurement system, masking the true resistance voltage drop characteristics; and high-frequency noise superimposed on the effective signal causes distortion, making it impossible to accurately extract the DC component.

[0030] To address the complex electromagnetic interference issues faced by DC voltage drop signals, this application constructs a joint anti-interference architecture combining hardware pre-suppression and digital dynamic tracking to ensure 0.1 interference resistance under complex substation operating conditions. High-precision contact resistance calculation capability.

[0031] Before the analog DC voltage drop signal is input to the analog-to-digital converter, a high common-mode rejection ratio differential probe is used in conjunction with a high-frequency EMI filter network to suppress common-mode interference and high-frequency conducted noise at the physical level. This avoids the equivalent antenna effect of the measurement link from the source, prevents the distorted DC voltage drop signal from entering the subsequent circuit, and avoids the limitations of digital filtering in correcting distorted DC voltage drop signals.

[0032] Among them, the high common-mode rejection ratio differential probe is realized by relying on the composite architecture of ACPL-C790 optocoupler isolation and closed-loop Hall magnetic balance acquisition. The implementation process is a well-known technology and will not be described in detail. The high-frequency EMI filter network is composed of DC voltage drop signal conditioning circuit RC low-pass filter and power supply end large-capacity capacitor energy storage filter.

[0033] At the ACPL-C790 optocoupler isolation front end, weak voltage and current signals are first attenuated and current-limited by a precision resistor network. In the isolation transmission layer, the DC voltage drop signal is transmitted in isolated signal form through the ACPL-C790 chip, achieving complete electrical isolation between the high-voltage side of the GIS equipment under test and the low-voltage side of the DSP digital processing system. Simultaneously, the optocoupler isolation chip possesses ultra-high common-mode rejection capability, ensuring that even with several kilovolts of induced common-mode voltage at the front end, a clean differential-mode DC voltage drop measurement signal can still be transmitted to the subsequent stages.

[0034] All DC dropout signal channels connected to the DSP are equipped with one or more stages of RC filter circuits at the front end, forming a hardware low-pass filter unit. This effectively attenuates high-frequency interference components higher than DC and ultra-low frequency target DC dropout signals, significantly suppressing 10kHz-5MHz wideband oscillation interference generated by isolating switch operation. By properly configuring the circuit cutoff frequency, megahertz-level ultra-high frequency RF interference can be accurately filtered out, eliminating problems such as high-frequency DC dropout signal aliasing and ADC sampling saturation. For example, a 10MHz cutoff frequency can be set for this purpose.

[0035] By combining a high common-mode rejection ratio differential probe with a high-frequency EMI filter network, high-voltage induced interference and high-frequency radiated noise can be effectively filtered out, thus completing the purification preprocessing of the front-end analog DC voltage drop signal.

[0036] This completes the hardware anti-interference preprocessing of the DC voltage drop signal.

[0037] Step S003: Track the power frequency deviation of the DC voltage drop signal, filter out the 50Hz fundamental wave and odd harmonics, retain the effective DC voltage drop signal, obtain the effective DC voltage drop signal, decompose the effective DC voltage drop signal into different sub-frequency bands, calculate the comprehensive score of the oscillation intensity of the sub-frequency band based on the wavelet packet coefficients and the corresponding coefficient sequences of the sub-frequency bands, calculate the adaptive threshold of the sub-frequency bands by combining the VisuShrink universal threshold when there is no interference, use wavelet packet adaptive denoising to denoise the effective DC voltage drop signal based on the adaptive threshold of the sub-frequency bands and the comprehensive score of oscillation intensity, reconstruct the denoising result using wavelet packets to obtain the true DC voltage drop, and complete the GIS loop resistance detection by combining the total output current and the shunt current.

[0038] By employing an improved second-order generalized integral phase-locked loop (SOGI-PLL) and a complex band-stop filter (ComplexBR) in synergy, the system dynamically tracks the 50Hz power frequency drift and various harmonics, achieving zero-phase notch suppression and fully preserving the effective DC component.

[0039] Specifically, the actual power frequency of the power grid exhibits a dynamic frequency offset of 49.5Hz-50.5Hz, and fixed-window moving average filtering is prone to spectral leakage due to asynchronous sampling. Therefore, SOGI-PLL is used to track the slight frequency offset of the power frequency in real time, ensuring that the filter's center frequency accurately matches the real-time interference frequency.

[0040] Using a DC voltage drop signal acquired by a high common-mode rejection ratio differential probe as input, the system outputs a precise 50Hz fundamental phase and real-time instantaneous frequency. Orthogonal components are generated using SOGI, and the real-time frequency and phase information are obtained through Park transform calculation.

[0041] Based on the phase-locked loop (PLL) results, a dual quadrature notch filter is dynamically configured. The time-domain DC voltage drop signal is converted into a complex analytic signal, and the two quadrature components output by SOGI are used as the real and imaginary parts of the complex signal, respectively. This removes the frequency domain symmetry limitation and enables independent processing of positive and negative frequency components.

[0042] Within the unit circle of the Z-domain, the filter transfer function parameters are dynamically updated, with the real-time instantaneous frequency obtained by phase-locked loop as the center frequency. and The conjugate frequency point is set as a null to cancel power frequency interference; in and A pole is positioned at a certain point to achieve zero-phase filtering of the 50Hz fundamental frequency interference. The pole mode length is... The value is 0.98, and the system sampling rate is 10 megabits per second. Indicates the center frequency. Represents the natural constant. Represents the imaginary unit. and They represent the center frequencies respectively. The positive and negative frequency components.

[0043] Simultaneously activate three sub-stage complex bandstop filters, with their center frequencies locked at 3... 5 and 7 The pole coefficient is uniformly set to 0.98. By reducing the quality factor and widening the suppression bandwidth, and adapting to the harmonic frequency shift, full-domain suppression of odd harmonics at the power frequency is achieved, thereby preserving the effective DC voltage drop signal. The effective DC voltage drop signal is recorded as the effective DC voltage drop signal. Indicates time The real-time instantaneous frequency; the improved second-order generalized integral phase-locked loop SOGI-PLL and complex band-stop filter ComplexBR are well-known technologies and will not be described in detail here.

[0044] Furthermore, the effective signal of the DC voltage drop is further processed.

[0045] To accurately separate the high-frequency oscillations of circuit breaker opening and closing from the time-frequency characteristics of the actual resistance voltage drop in the effective DC voltage drop signal, wavelet packet adaptive denoising is required. When performing wavelet packet adaptive denoising, it is necessary to consider that the oscillations generated by circuit breaker opening and closing operations are not only high-frequency signals but also oscillate over a wide frequency range. However, existing wavelet packet adaptive denoising algorithms only set thresholds based on the local standard deviation of sub-bands, resulting in poor filtering performance. Therefore, it is necessary to evaluate the high-frequency and wide-frequency oscillation levels of each sub-band decomposed from the effective DC voltage drop signal separately to achieve adaptive threshold configuration.

[0046] The high-frequency oscillations generated by the opening and closing of circuit breakers are not stationary white noise, but rather impact interference with strong transient characteristics, wide bandwidth, non-Gaussian nature, and time-varying energy accumulation. On the effective DC voltage drop signal, this manifests as a mixture of multi-scale spikes and damped ringing. Traditional wavelet thresholding methods based on local standard deviations, assuming a Gaussian distribution and uniform energy distribution, cannot distinguish between the high-frequency components of the "true resistance step response" and the "false resonance caused by switching oscillations," easily leading to over-smoothing or insufficient noise reduction. Specifically, the frequency range of the high-frequency oscillations is 10kHz–5MHz; over-smoothing damages the rising edge of the resistance, while insufficient noise reduction leaves residual spikes contaminating the DC voltage drop.

[0047] First, the effective signal of DC voltage drop is decomposed. Specifically, the Symlet-8 wavelet basis is selected, and the empirical decomposition level is set to 4 levels, resulting in 16 sub-bands, each of which corresponds to a set of coefficient sequences.

[0048] The process of decomposing the effective DC voltage drop signal into sub-bands is a well-known technique and will not be elaborated further.

[0049] Because the arc reignites the instantaneously after the circuit breaker contacts separate, a voltage drop on the order of nanoseconds will occur. Therefore, the impulse intensity can be assessed by the instantaneous peak value in the sub-band, and the peak factor of the sub-band can be calculated.

[0050] Preferably, as an embodiment of this application, the calculation result of dividing the absolute value of the peak value of the sub-band by the root mean square value of the sub-band is recorded as the peak factor of the sub-band.

[0051] The root mean square (RMS) value is the arithmetic square root of the average of the squares of all values ​​within a sub-band. The RMS value of a sub-band can reflect the average energy level of the signal. The larger the peak factor of a sub-band, the greater the deviation of the instantaneous amplitude of transient interference spikes from the overall average signal energy of that frequency band within the same sub-band.

[0052] On the other hand, multiple reflections of electromagnetic waves within the GIS cavity will form a decaying oscillating envelope. By performing a Hilbert transform on the wavelet packet coefficients of the sub-frequency band, an analytic signal is obtained. The instantaneous envelope of the analytic signal is extracted, and an exponential decay fitting is performed on the instantaneous envelope to obtain the decay time constant.

[0053] The instantaneous envelope is the outer contour line of the waveform oscillation; Hilbert transform, extraction of instantaneous envelope, and exponential decay fitting are all well-known techniques and will not be described in detail here.

[0054] The formula for the exponential decay fit is: in, Indicates the first The attenuation time constant of each sub-band; Indicates the first The Hilbert transform of each sub-band is used to obtain the maximum envelope value at the first instant in the instantaneous envelope curve; Indicates time The instantaneous envelope value; Represents the natural constant.

[0055] The larger the decay time constant, the longer the oscillation lasts, the more significant the "long tail oscillation" characteristic of the signal in the sub-band, the more the sub-band conforms to the characteristics of electromagnetic interference, and the more it should be processed with a soft threshold.

[0056] Secondly, the interference energy is highly concentrated in a few sub-bands, while the actual resistance response is flatter in the low-frequency range, indicating that the band energy exhibits clustering. Therefore, based on the Shannon entropy calculation formula, the normalized value of the energy spectral entropy of the coefficient sequence corresponding to each sub-band is calculated to assess the degree of energy clustering in the sub-bands.

[0057] The calculation of Shannon entropy is a well-known technique and will not be elaborated further; the normalized value of energy spectrum entropy is the sum of energy spectrum entropy and... The ratio, This indicates the number of sub-bands from which the effective signal of the DC voltage drop is decomposed.

[0058] The closer the normalized value of the energy spectrum entropy is to 0, the more prominently the energy is concentrated at a few frequency points. When calculating the Shannon entropy, if the value of the coefficient sequence corresponding to the frequency band is 0, the value of the corresponding probability distribution is recorded as 0, and the product of the probability distribution and the logarithm is assigned to 0.

[0059] It should be noted that assessing oscillation solely based on the impact intensity will lead to misjudgment at the rising edge of the resistor; assessing only the energy accumulation will ignore the sustained characteristics of ringing. Therefore, a comprehensive score of the oscillation intensity of the sub-band is calculated based on the peak factor, decay time constant, and normalized value of the energy spectral entropy of the sub-band.

[0060] Specifically, when the peak factor of the sub-band is greater than the first judgment threshold, the indicator of the sub-band is assigned a value of 1; when the peak factor of the sub-band is less than or equal to the first judgment threshold, the indicator of the sub-band is assigned a value of 0. The ratio of the attenuation time constant of the sub-band to the preset reference attenuation time constant is used as the independent variable of the hyperbolic tangent function, and the calculated value of the hyperbolic tangent function is recorded as the hyperbolic tangent term of the sub-band. The difference between the number 1 and the normalized value of the energy spectrum entropy of the sub-band is recorded as the correction term of the sub-band. The weighted sum of the indicator term, hyperbolic tangent term, and correction term of the sub-band is recorded as the comprehensive score of the oscillation intensity of the sub-band, and the sum of the weight coefficients of the indicator term, hyperbolic tangent term, and correction term of the sub-band is 1.

[0061] In this embodiment, Shapley value analysis is used to determine the weighting coefficients of the indicator, hyperbolic tangent, and correction terms of the sub-band. Specifically, the weighting coefficients of the indicator, hyperbolic tangent, and correction terms of the sub-band are 0.40, 0.35, and 0.25, respectively. Shapley value analysis is a well-known technique and will not be described in detail here.

[0062] In this embodiment, the first judgment threshold is set to 4.5, and the reference decay time constant is set to 150. .

[0063] The higher the overall score of the sub-band oscillation intensity, the more severe the switching interference experienced by the sub-band, and the greater the difference between the signal corresponding to the sub-band and the real signal.

[0064] Using the VisuShrink universal threshold under interference-free conditions as the base threshold, and combining the comprehensive score of sub-band oscillation intensity, peak factor, and the coefficient sequence corresponding to the sub-band, the adaptive threshold of the sub-band is calculated. The formula for calculating the adaptive threshold is: in, Indicates the first Adaptive threshold for each sub-band; Indicates the first The basic threshold for each sub-band; This indicates the upper limit of the preset interference enhancement factor. In this embodiment, the upper limit of the interference enhancement factor is set to 2.5. Represents the hyperbolic tangent function; This indicates the preset steepness of the enhancement curve; in this embodiment, the value of the steepness of the enhancement curve is 8.0. Indicates the first Comprehensive score of oscillation intensity of each sub-band; This represents the preset interference activation threshold. In this embodiment, the interference activation threshold is set to 0.55. Represents the logarithmic function with the natural constant as the base; This represents the preset impact attenuation coefficient. In this embodiment, the impact attenuation coefficient is set to 0.3. The impact attenuation coefficient can prevent the peak factor of the sub-band from being too large and thus missing the real edge. Indicates the first Peak factor of each sub-band.

[0065] The specific method for determining the basic threshold is as follows: the product of the logarithm of the number of coefficients in the coefficient sequence corresponding to the sub-band and the number 2 is recorded as the first product of the sub-band; the product of the arithmetic square root of the first product of the sub-band and the standard deviation of the noise within the sub-band is recorded as the basic threshold. Wherein, when calculating the logarithm of the number of coefficients, the base of the logarithm is the natural constant; the standard deviation of the noise within the sub-band is the ratio of the median of the absolute values ​​of the coefficients in the coefficient sequence corresponding to the sub-band to the constant 0.6745.

[0066] Based on the adaptive threshold and oscillation intensity comprehensive score of the sub-frequency band, wavelet packet adaptive denoising is used to denoise the effective DC voltage drop signal, and the denoised coefficient sequence is obtained. Wavelet packet reconstruction is performed on all the denoised coefficient sequences corresponding to the effective DC voltage drop signal to obtain the denoised effective DC voltage drop signal, i.e., the real DC voltage drop.

[0067] In the process of wavelet packet adaptive denoising, the specific classifications of not applying thresholds to sub-frequency bands, applying soft thresholds to sub-frequency bands, or applying hard thresholds are as follows: When the comprehensive score of the sub-band's oscillation intensity is less than the first threshold, no threshold is applied to the sub-band, and the original wavelet coefficients are fully preserved, retaining only the useful features of low-frequency resistance and DC. When the comprehensive score of the sub-band's oscillation intensity is less than the second threshold but greater than or equal to the first threshold, soft thresholding is applied. For any coefficient in the coefficient sequence corresponding to the sub-band, the difference between the absolute value of the coefficient and the adaptive threshold of the sub-band is recorded as the first difference of the coefficient. When the coefficient is greater than 0, the sign eigenvalue of the coefficient is assigned a value of 1; when the coefficient is less than 0, the sign eigenvalue of the coefficient is assigned a value of -1; when the coefficient is equal to 0, the sign eigenvalue of the coefficient is assigned a value of 0. The product of the first difference of the coefficient and the maximum value of the digit 0 and the sign eigenvalue is recorded as the soft threshold of the coefficient. When the comprehensive score of the sub-band's oscillation intensity is greater than or equal to the second threshold, hard thresholding is applied. For all coefficients in the coefficient sequence corresponding to the sub-band, only coefficients with an absolute value greater than the adaptive threshold of the sub-band are retained, completely eliminating spikes.

[0068] Among them, the soft threshold can preserve the continuity of weak signals and is suitable for ringing envelopes; the hard threshold can accurately remove isolated spikes and is suitable for nanosecond-level impacts; not processing low interference bands can prevent "overcorrection"; in this embodiment, the values ​​of the first threshold and the second threshold of the comprehensive score are 0.4 and 0.75, respectively.

[0069] Finally, based on the total output current, shunt current, and the effective signal of the denoised DC voltage drop, the GIS loop resistance is calculated, and the GIS loop resistance detection in the bilateral shunt mode is completed. Specifically, the total output current is subtracted from the sum of the shunt currents on both sides, and the result is recorded as the actual measured current; the actual DC voltage drop is divided by the actual measured current, and the result is recorded as the GIS loop resistance.

[0070] The bilateral diversion mode is used when the grounding bars on both sides of the GIS equipment have not been removed and diversion exists on both sides.

[0071] This completes the GIS loop resistance test.

[0072] Based on the same inventive concept as the above method, this embodiment of the invention also provides a GIS loop resistance detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described GIS loop resistance detection methods.

[0073] 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 principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the resistance of a GIS loop, characterized in that, The method includes the following steps: Collect the total output current, shunt current, and DC voltage drop signals during the GIS loop resistance detection process; Combined with RC low-pass filter and power supply filter, it attenuates wideband high-frequency interference and completes hardware anti-interference preprocessing for DC voltage drop signals; The DC voltage drop signal is tracked by the power frequency deviation, and the 50Hz fundamental wave and odd harmonics are filtered out to retain the effective DC voltage drop signal. The effective DC voltage drop signal is then decomposed into different sub-frequency bands. Based on the wavelet packet coefficients and corresponding coefficient sequences of the sub-frequency bands, the comprehensive score of the oscillation intensity of the sub-frequency bands is calculated. Combined with the VisuShrink universal threshold under interference-free conditions, the adaptive threshold of the sub-frequency bands is calculated. Based on the adaptive threshold of the sub-frequency bands and the comprehensive score of oscillation intensity, wavelet packet adaptive denoising is used to denoise the effective DC voltage drop signal. The denoising result is then reconstructed using wavelet packets to obtain the true DC voltage drop. Combined with the total output current and the shunt current, the resistance detection of the GIS loop is completed.

2. The method for detecting the resistance of a GIS loop according to claim 1, characterized in that, Odd harmonic suppression is achieved through a three-stage complex bandstop filter.

3. The method for detecting the resistance of a GIS loop according to claim 1, characterized in that, The specific calculation method for the comprehensive score of the oscillation intensity of the sub-frequency band is as follows: Calculate the peak factor of the sub-band based on the coefficient sequence corresponding to the sub-band, and assign values ​​to the indicator terms of the sub-band based on the peak factor of the sub-band. Hilbert transform is performed on the wavelet packet coefficients of the sub-band to obtain the analytic signal. The instantaneous envelope of the analytic signal is extracted, and exponential decay fitting is performed on the instantaneous envelope to obtain the decay time constant. Calculate the hyperbolic tangent term of the sub-band based on the decay time constant; The difference between the number 1 and the normalized value of the energy spectral entropy of the coefficient sequence corresponding to the sub-band is denoted as the correction term of the sub-band. The weighted sum of the indicator, hyperbolic tangent, and correction terms of the sub-band is denoted as the comprehensive score of the sub-band's oscillation intensity. The sum of the weight coefficients of the indicator, hyperbolic tangent, and correction terms of the sub-band is 1.

4. The method for detecting the resistance of a GIS loop according to claim 3, characterized in that, The peak factor of the sub-band is the ratio of the absolute value of the peak value of the sub-band to the root mean square value of the sub-band.

5. The method for detecting the resistance of a GIS loop according to claim 3, characterized in that, The specific method for assigning values ​​to the indicator items of the sub-frequency bands is as follows: When the peak factor of a sub-band is greater than the first determination threshold, the indicator of the sub-band is assigned a value of 1; otherwise, the indicator of the sub-band is assigned a value of 0.

6. The method for detecting the resistance of a GIS loop according to claim 3, characterized in that, The specific method for obtaining the hyperbolic tangent term of the sub-frequency band is as follows: The ratio of the attenuation time constant of the sub-band to the preset reference attenuation time constant is used as the independent variable of the hyperbolic tangent function, and the calculated value of the hyperbolic tangent function is denoted as the hyperbolic tangent term of the sub-band.

7. The method for detecting the resistance of a GIS loop according to claim 1, characterized in that, The specific calculation method for the adaptive threshold of the sub-frequency band is as follows: The product of the logarithm of the number of coefficients in the coefficient sequence corresponding to the sub-band and the number 2 is denoted as the first product of the sub-band. The product of the arithmetic square root of the first product of the sub-band and the standard deviation of the noise in the sub-band is denoted as the basic threshold. The standard deviation of the noise in the sub-band is the ratio of the median of the absolute values ​​of the coefficients in the coefficient sequence corresponding to the sub-band to the constant 0.6745. Based on the basic threshold, the comprehensive score of the sub-band's oscillation intensity, the peak factor, and the coefficient sequence corresponding to the sub-band, the adaptive threshold of the sub-band is calculated.

8. The method for detecting the resistance of a GIS loop according to claim 1, characterized in that, When performing wavelet packet adaptive denoising, the specific classifications of not applying thresholds to sub-frequency bands, applying soft thresholds to sub-frequency bands, or applying hard thresholds are as follows: When the comprehensive score of the oscillation intensity of a sub-band is less than the first threshold of the comprehensive score, no threshold is applied to the sub-band, and all wavelet coefficients are retained. When the comprehensive score of the oscillation intensity of the sub-band is less than the second threshold of the comprehensive score but greater than or equal to the first threshold of the comprehensive score, soft thresholding is used. For any coefficient in the coefficient sequence corresponding to the sub-band, the difference between the absolute value of the coefficient and the adaptive threshold of the sub-band is recorded as the first difference of the coefficient. When the coefficient is greater than 0, the sign eigenvalue of the coefficient is assigned to 1; when the coefficient is less than 0, the sign eigenvalue of the coefficient is assigned to -1; when the coefficient is equal to 0, the sign eigenvalue of the coefficient is assigned to 0; the product of the first difference of the coefficient and the maximum value of the number 0 and the sign eigenvalue is recorded as the soft threshold of the coefficient. When the comprehensive score of the oscillation intensity of a sub-band is greater than or equal to the second threshold of the comprehensive score, hard thresholding is used to retain the coefficients in the coefficient sequence corresponding to the sub-band whose absolute value is greater than the adaptive threshold of the sub-band.

9. The method for detecting the resistance of a GIS loop according to claim 1, characterized in that, The specific method for combining the total output current and the shunt current to complete the GIS loop resistance detection includes: The sum of the shunt currents on both sides is subtracted from the total output current and recorded as the actual measured current; the actual DC voltage drop is divided by the actual measured current and recorded as the GIS loop resistance.

10. A GIS loop resistance detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-9.

Images