A method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms

Abstract

This invention relates to the field of circuit breaker fault detection, specifically to a method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms. The method employs a heterogeneous architecture of FPGA and DSP to acquire high-frequency zero-sequence transient signals. By analyzing the extreme value distribution characteristics, the time-frequency local volatility of the waveform is calculated, and this volatility is converted into local inertial weights using a nonlinear function. In the signal decomposition iteration, the system dynamically adjusts the weighted fusion ratio of the standard screening path and the inertial holding path using the inertial weights, thereby constructing an adaptive screening model to extract pure intrinsic mode components. Finally, a Hilbert transform is performed on the extracted components, and the fault direction is determined based on the instantaneous phase difference between voltage and current. This technical solution effectively solves the mode aliasing problem in traditional empirical mode decomposition, significantly improving the system's feature extraction capability and identification accuracy for single-phase grounding, open-circuit, and high-resistance faults in high-noise environments.

Description

Technical Field

[0001] This invention relates to the field of circuit breaker fault detection, and more specifically to a method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms. Background Technology

[0002] Feeder terminals (FTUs) and pole-mounted circuit breakers (PTUs) are key equipment in smart distribution networks and are widely used for monitoring and controlling distribution lines. When a single-phase ground fault occurs in a distribution network, transient signals of zero-sequence voltage and zero-sequence current are generated in the line. These signals typically exhibit nonlinear and non-stationary characteristics and are often accompanied by high-frequency oscillations or arc noise. To achieve accurate fault detection and location, advanced signal processing techniques are often required to perform time-frequency analysis on these complex transient signals to extract effective fault characteristic components.

[0003] Currently, Empirical Mode Decomposition (EMD) has become the mainstream method for processing non-stationary transient signals due to its ability to adaptively decompose complex signals into several intrinsic mode functions (IMFs) without requiring preset basis functions. The core of this algorithm lies in performing layer-by-layer screening and decomposition by fitting upper and lower envelopes through extreme points.

[0004] However, in actual power distribution network fault detection scenarios, the distribution of extreme points in transient signals is often extremely uneven due to the intermittent nature of fault arcs or strong noise interference. When using traditional EMD algorithms to construct envelopes using cubic spline interpolation, this uneven distribution of extreme points easily leads to overshoot or fitting distortion in the envelope. This directly causes severe mode aliasing, where fluctuation components of different scales are mixed in the same mode, or components of the same scale are dispersed into different modes. Mode aliasing severely damages the physical authenticity of the decomposition results, leading to inaccurate extracted fault features, which may in turn cause circuit breakers to misjudge the fault direction or fail to operate. Summary of the Invention

[0005] To address the problem of potential misjudgment of fault direction by circuit breakers in traditional EMD algorithms during mode aliasing, this invention proposes a ground fault detection method for pole-mounted circuit breakers based on transient waveforms. The method includes: acquiring transient zero-sequence voltage and zero-sequence current signals from a distribution network line to obtain a transient signal sequence; performing adaptive signal decomposition on the transient signal sequence to extract the first-order intrinsic mode component; performing a Hilbert transform on the first-order intrinsic mode component; and determining the fault direction based on the instantaneous phase difference after the transform. The adaptive signal decomposition includes iterative screening of the current sequence to be processed, specifically: constructing the upper and lower envelopes of the sequence to be processed; obtaining the mean lines of the upper and lower envelopes; subtracting the mean lines from the sequence to be processed to obtain a standard residual sequence; obtaining multiple extreme points of the sequence to be processed; obtaining inertial weights characterizing local waveform fluctuations based on the distribution characteristics of the multiple extreme points; and weighted fusing the standard residual sequence with the sequence to be processed to obtain the sequence to be processed for the next iteration, wherein the inertial weights are the fusion weights of the standard residual sequence during the fusion process.

[0006] This invention improves the traditional adaptive signal decomposition screening process by introducing inertial weights based on the distribution characteristics of extreme points. Compared with the existing technology that directly uses standard residuals for simple iteration, this invention can adaptively adjust the sequence retention ratio according to the intensity of local waveform fluctuations during the screening process. This mechanism effectively suppresses mode mixing and endpoint effects, improves the accuracy and purity of the first-order intrinsic mode component extraction, thereby ensuring that the instantaneous phase difference obtained by the subsequent Hilbert transform is more accurate, and significantly improves the reliability of ground fault direction determination for pole-mounted circuit breakers in distribution networks.

[0007] Furthermore, the method for calculating the inertia weight is as follows: ; in This represents the inertia weight; This is a preset critical threshold. This is a preset sensitivity factor used to control the steepness of the weight changes; The extreme value interval volatility is used to characterize the distribution characteristics of extreme points.

[0008] This invention employs a specific function model to construct a method for calculating inertia weights. Compared to existing technologies that typically use linear or step-like weight adjustment strategies, this calculation formula can dynamically and smoothly adjust the weights nonlinearly based on the volatility of extreme value intervals. Combined with a sensitivity factor to control the steepness of weight changes, the algorithm can more precisely balance convergence speed and decomposition accuracy when dealing with transient waveforms of varying complexity, enhancing its adaptability to non-stationary signals.

[0009] Furthermore, the method for calculating the extreme value interval volatility is specifically as follows: ; in This represents the volatility of the extreme value interval; Indicates the index of the extreme point; Indicates the first The extreme point and the first The time difference between extreme points; Indicates the first The extreme point and the first The time difference between extreme points; This indicates the magnitude of the current extreme point; This represents the global maximum amplitude within the entire sampling window; This represents the smallest positive number that is specified.

[0010] This invention proposes a volatility calculation index that comprehensively considers both the density of the time distribution of extreme points and the relative strength of their amplitudes. Compared with existing technologies that rely on only a single dimension of features, this index can more comprehensively and three-dimensionally characterize the local fluctuation characteristics of transient signals. By combining the time difference ratio of adjacent extreme points with the current amplitude ratio, it can keenly capture the waveform distortion details at the moment of fault occurrence, providing a high signal-to-noise ratio feature input for the calculation of inertia weights.

[0011] Furthermore, if the calculated inertia weight value is greater than a preset confidence threshold, the inertia weight is forcibly set to 1.

[0012] Furthermore, the calculation method for the weighted fusion is as follows: ; in This represents the sequence to be processed in the next iteration; This represents the inertia weight; This represents the standard residual sequence; This represents the sequence to be processed in the current round.

[0013] This invention utilizes inertial weights to establish a dynamic balance between the standard residual sequence and the current sequence to be processed. Unlike the traditional EMD algorithm, which completely replaces the previous sequence, this fusion mechanism is equivalent to introducing a correction function during the sieving process. This smooths out abrupt changes during sieving, effectively prevents signal distortion caused by over-sieving, preserves more of the true physical characteristics of fault transients, and makes the extracted modal components more physically meaningful.

[0014] Furthermore, constructing the upper and lower envelopes of the sequence to be processed also includes: obtaining the upper envelope by connecting all local maxima points of the sequence to be processed based on cubic spline interpolation; and obtaining the lower envelope by connecting all local minima points of the sequence to be processed based on cubic spline interpolation.

[0015] Furthermore, it also includes using the average of the upper and lower envelopes as the mean line.

[0016] Furthermore, determining the fault direction based on the transformed instantaneous phase difference also includes: if the instantaneous phase difference between the first-order inherent mode components of the transient zero-sequence voltage and zero-sequence current is within a preset tolerance range of 180°, then the fault is determined to be located on the load side of the circuit breaker; if the instantaneous phase difference is within a preset tolerance range of 0°, then the fault is determined to be located on the power supply side of the circuit breaker.

[0017] Furthermore, the preset tolerance range is 30° to 60°.

[0018] Furthermore, the detection method is executed based on an intelligent controller containing a dual-core processing unit of FPGA and DSP; the FPGA is responsible for acquiring and buffering the transient signal sequence; and the DSP is responsible for performing iterative sieving, Hilbert transform, and fault determination.

[0019] This invention employs a dual-core hardware architecture that combines FPGA and DSP, which, compared to single-core processor solutions, fully leverages the parallel high-speed acquisition and large-capacity cache capabilities of the FPGA, as well as the powerful complex floating-point arithmetic capabilities of the DSP. This specific division of labor and cooperation mechanism ensures the lossless acquisition of massive amounts of transient data and enables the rapid completion of complex adaptive decomposition and Hilbert transform algorithms, achieving high real-time performance, high accuracy, and high throughput in pole-mounted circuit breaker fault detection.

[0020] The technical effects of this invention are as follows: This invention solves the mode mixing problem of traditional EMD by constructing extreme value interval volatility and dynamically calculating inertia weights, and adaptively weighting and fusing the standard residual and the sequence to be processed in the iterative screening process. This method can more accurately extract the first-order intrinsic mode components of transient zero-sequence signals, and combined with the instantaneous phase difference obtained by Hilbert transform, significantly improves the accuracy and anti-interference capability of grounding fault direction determination for pole-mounted circuit breakers in distribution networks. Attached Figure Description

[0021] Figure 1 This is a schematic flowchart illustrating a method for detecting grounding faults in a pole-mounted circuit breaker based on transient waveforms, according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the distribution of the zero-sequence voltage signal and its local maxima and minima collected in an embodiment of the present invention. Figure 3 This is a schematic illustration of the time series diagram of the inertial weight distribution calculated based on volatility in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the upper and lower envelopes and standard mean line of the zero-sequence voltage signal in an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the standard EMD decomposition results of an embodiment of the present invention where modal aliasing exists; Figure 6 This is a schematic diagram illustrating the improved EMD decomposition results after suppressing modal aliasing in an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the phase difference distribution comparison of the extracted components of standard EMD and improved EMD within the fault window in an embodiment of the present invention. Detailed Implementation

[0022] 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, not all, of the embodiments of the present invention. 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.

[0023] The embodiments of the present invention mainly operate in the intelligent controller of the distribution network feeder terminal unit (FTU) or pole-mounted circuit breaker, and adopt a heterogeneous computing architecture to meet the real-time requirements of transient high-frequency signal processing.

[0024] Specifically, the core processing unit of the intelligent controller adopts a dual-core architecture of Field Programmable Gate Array (FPGA) + Digital Signal Processor (DSP). The FPGA, as the core of the data perception layer, is responsible for high-speed parallel data throughput and caching; the DSP, as the core of the computational decision layer, is responsible for executing complex adaptive sieving algorithms and logical judgments.

[0025] In terms of physical connection, the system includes at least one set of high-precision electromagnetic or electronic instrument transformers for real-time acquisition of zero-sequence voltage and zero-sequence current signals of the distribution network lines. The output of the instrument transformers is connected to a high-sampling-rate analog-to-digital converter (ADC) via a pre-amplifier anti-aliasing filter. In this embodiment, the sampling rate of the ADC is preferably set to 10kHz to 50kHz to ensure that the high-frequency transient components at the moment of fault occurrence can be captured. The FPGA is internally configured with a first-in-first-out (FIFO) buffer queue to construct a time-domain discrete sequence containing several cycles (e.g., 20ms to 40ms before and after) before and after the fault, which is then read by the DSP via a high-speed parallel bus.

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] An embodiment of a ground fault detection method for pole-mounted circuit breakers based on transient waveforms: like Figure 1 As shown, the present invention provides a method for detecting ground faults in pole-mounted circuit breakers based on transient waveforms, comprising: S1. Collect zero-sequence transient signals of distribution network lines and calculate extreme value interval volatility based on extreme value distribution characteristics.

[0028] First, the DSP processor of the intelligent controller reads the preprocessed discrete sequence of transient zero-sequence current from the FPGA via the bus. The processor executes the extreme point search logic, traverses the discrete sequence, identifies all local maxima and local minima, and records their corresponding time coordinates and amplitudes.

[0029] like Figure 2 As shown, the system performs extremum search on the acquired zero-sequence voltage transient signal. The continuously changing curve in the figure represents the original zero-sequence voltage signal; the triangular markers above the waveform represent identified local maxima, and the inverted triangular markers below the waveform represent identified local minima. It can be seen that during the transient process after the fault occurs, the distribution of extremum points exhibits significant non-uniformity.

[0030] To address the mode aliasing problem caused by uneven distribution of extreme points in traditional Empirical Mode Decomposition (EMD), one embodiment avoids direct decomposition. Instead, the system calculates the extreme point interval volatility, reflecting waveform disorder, based on the rate of change of the time difference between adjacent extreme points and combined with relative amplitude weights. Let... The index of the extreme point. Indicates the first The extreme point and the first The time difference between extreme points Indicates the first The extreme point and the first The time difference between extreme points This represents the amplitude at the current extreme point. This represents the global maximum amplitude within the entire sampling window. The extreme value interval volatility... The calculation method is as follows: ; First half of the formula This reflects the density variation over time. When adjacent extreme values ​​are spaced... When a drastic change occurs, such as a sudden change from sparse to dense, this term increases significantly; the latter half of the formula... This reflects the energy proportion in the amplitude dimension; to prevent division by zero during calculation, a very small positive number is added to the denominator. ,For example .

[0031] As can be seen from the above formula, when the transient signal is in the high-frequency oscillation region, the time interval between adjacent extreme points changes rapidly, and if the amplitude is large at this time, the calculated... The value will tend towards the maximum value, indicating that the area is a high-risk zone for aliasing and requires subsequent intervention. Conversely, if the waveform is stable and the time interval changes little, then... Approaching zero.

[0032] In another embodiment, in scenarios where computing power allows or the noise environment is extremely harsh, The variance of the extreme value interval within a sliding window can be used for calculation. Specifically, a coverage is defined. A sliding window containing extrema is used to calculate the standard deviation of the time intervals between all adjacent extrema within the window. .like Above the historical average times, for example If the volatility of the region is high, then it is determined that the region has high volatility. Compared to the first embodiment, it is not sensitive to individual outliers and has stronger noise resistance.

[0033] S2. Based on the nonlinear mapping function, the extreme value interval volatility is transformed into local inertial weights to establish a signal processing mechanism.

[0034] After obtaining the extreme value interval volatility, the processor maps it to normalized inertial weights. This weight is used to establish the following mechanism: the more volatile the fluctuation, the higher the weight. The larger the value, the lower the system's confidence in the envelope generated by the current spline interpolation, thus giving the previous signal a stronger inertia.

[0035] In this embodiment, a variant of the Sigmoid function is used for inverse mapping. Inertia weights. The calculation formula is as follows: ; in This is the critical threshold, used to define the boundary of volatility; This is a sensitivity factor used to control the steepness of the weight change. In this embodiment, Preferred setting is to This value is negatively correlated with the system sampling rate; that is, the higher the sampling rate, the smaller this value should be. Preferred setting is to between.

[0036] As can be seen from the formula, when the calculated volatility... Much smaller than the threshold At this point, the exponent term approaches 0, the denominator approaches 1, and finally... When the value approaches 1, the system fully trusts the current envelope; when Significantly exceeding the threshold At that time, the exponent term increases rapidly, and the denominator tends to infinity, leading to When the value approaches 0, the system no longer trusts the current envelope and enters a hold state.

[0037] Furthermore, to optimize computational efficiency, the system incorporates the following control logic: if the calculated inertia weight is greater than a confidence threshold, i.e. The processor will then force it to be set to This reduces unnecessary floating-point operations.

[0038] like Figure 3 As shown, the processor maps the calculated volatility to normalized inertial weights. The solid curve in the figure represents the change of inertial weights over time, with values ​​ranging from 0 to 1; the horizontal dashed line represents the preset confidence threshold. When the signal is in a stable phase, the weight curve is above the threshold and approaches 1, indicating a high-confidence envelope. However, in the high-frequency oscillation region after a fault occurs, the weight curve drops sharply below the threshold and approaches 0, indicating that the system reduces its confidence in the envelope and instead enhances the inertial retention of historical signals, thus accurately identifying high-risk aliasing areas.

[0039] S3. Use inertial weights to weight and fuse the upper and lower envelopes with historical signals to construct a corrected mean line and extract modal components.

[0040] In the adaptive signal decomposition sieving iteration process, the system no longer uses a simple sieving method by subtracting the mean line, but instead constructs an adaptive sieving model based on inertial weights. This model dynamically determines the next round of sieving sequence based on the local fluctuation characteristics of the signal. The composition of.

[0041] First, the processor obtains the upper envelope based on cubic spline interpolation. and lower envelope And calculate the standard mean. .like Figure 4As shown, the envelope of the signal is constructed using cubic spline interpolation. The two outermost solid curves in the figure represent the upper and lower envelopes, respectively, which tightly enclose the original oscillating signal; the dashed curve in the middle is the calculated standard mean. This standard mean reflects the low-frequency trend of the signal and will serve as the basis for calculating the standard screening path in subsequent steps.

[0042] Subsequently, two candidate screening paths are defined: Standard screening path: Calculate the standard residual signal after removing the mean. ; Inertial preservation path: directly retain the original signal shape from the previous round. .

[0043] The inertia weight calculated based on the aforementioned steps The system directly calculates the signal sequence for the next round using the following weighted fusion method. : ; in This is the input signal for the previous round of screening iteration.

[0044] When the signal is in the low-frequency stable region When the latter part of the formula approaches zero, at this point... The algorithm adaptively degenerates into standard EMD screening, normally filtering out low-frequency trends; When the signal is in the high-frequency arc oscillation region The first part of the formula is suppressed, at this time The algorithm automatically skips the screening operation in this area, forcibly preserving the signal pattern from the previous round.

[0045] This mechanism requires no additional compensation calculations and directly preserves the high-frequency, intermittent arc noise completely in the current first-order intrinsic mode component. This fundamentally blocks modal aliasing.

[0046] The processor uses the calculated The above process is repeated for the new sequence to be processed. When the convergence condition of standard deviation being less than a preset value, such as 0.2 to 0.3, is met, the screening is stopped and the first-order intrinsic mode components are extracted. .

[0047] To verify the effectiveness of the present invention in suppressing modal aliasing. Figure 5 and Figure 6 The decomposition results of the two methods are compared. For example... Figure 5As shown, when using traditional standard EMD decomposition, within the time period indicated by the shaded rectangle in the figure, high-frequency oscillation components are incorrectly leaked into low-frequency components, resulting in severe distortion of the low-frequency waveform, i.e., mode aliasing occurs. And as... Figure 6 As shown, after applying the improved EMD method based on inertia weights according to this invention, within the same shaded rectangular region, the high-frequency arc characteristics are completely preserved in the first-order modal component, while the low-frequency component maintains a smooth sinusoidal shape. This indicates that the present invention effectively achieves correct decoupling of high and low frequency components, ensuring the physical authenticity of the extracted signal.

[0048] S4. Perform Hilbert transform on the extracted transient components and determine the fault direction based on the phase difference statistics.

[0049] Obtain the transient zero-sequence voltage after inertial screening correction. and transient zero-sequence current Then, the system executes the final decision on the direction of the fault.

[0050] The processor performs Hilbert transforms on the two components respectively, extracts their instantaneous phase sequences, and calculates the phase difference between them within the effective fault time window. That is: ,in and These are the instantaneous phases of the transient zero-sequence voltage and current, respectively.

[0051] System preset tolerance angle Preferred range to And make a judgment based on the following logic: if the phase difference Stable at If the phase difference is within the specified interval, i.e., the reverse phase interval, then the fault point is determined to be located on the load side of the circuit breaker, and the system immediately drives the circuit breaker operating mechanism to perform a trip operation; if the phase difference is... Stable at or If the fault is within the same phase interval, the fault point is determined to be on the power supply side, and the circuit breaker remains closed or only sends an alarm signal.

[0052] like Figure 7 As shown, a statistical analysis of the phase difference within the fault window is performed. The figure displays the histograms of the phase difference probability density distribution calculated by the standard EMD method and the improved EMD method, where the vertical dashed line represents the theoretical value at 180°. It can be seen that, compared to the standard EMD method, the phase difference distribution obtained by the improved EMD method of this invention is more concentrated around the theoretical value at 180°, its average value is closer to the actual fault characteristics, and its dispersion is smaller. This demonstrates that the signal processed by the method of this invention can provide a more accurate fault direction criterion, significantly improving the reliability of circuit breaker operation.

[0053] It should be noted that although the above embodiments are mainly described using single-phase grounding faults as an example, the signal processing method of the present invention is also applicable to the monitoring of open circuit faults or the identification of high-resistance grounding faults in distribution networks. In these scenarios, the signals also exhibit non-stationary and nonlinear transient characteristics. By utilizing the inertial weight correction mechanism of the present invention, weak fault feature components can also be effectively extracted.

[0054] Furthermore, the adaptive signal decomposition of the present invention is not limited to empirical mode decomposition (EMD). In other embodiments, it can also be based on EMD-like algorithms such as variational mode decomposition (VMD) or local mean decomposition (LMD), and by introducing the inertial weight and process intervention mechanism proposed in this invention, the same technical effect can be achieved.

Claims

1. A method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms, characterized in that, The method includes: acquiring transient zero-sequence voltage and zero-sequence current signals of distribution network lines to obtain a transient signal sequence; Adaptive signal decomposition is performed on the transient signal sequence to extract the first-order intrinsic mode component, Hilbert transform is performed on the first-order intrinsic mode component, and the fault direction is determined based on the instantaneous phase difference after the transformation. The adaptive signal decomposition includes iterative sieving of the current sequence to be processed, specifically: Construct the upper and lower envelopes of the sequence to be processed; obtain the mean lines of the upper and lower envelopes; subtract the mean lines from the sequence to be processed to obtain the standard residual sequence; Multiple extreme points of the sequence to be processed are obtained, and inertial weights characterizing local waveform fluctuations are obtained based on the distribution characteristics of the multiple extreme points. The standard residual sequence and the sequence to be processed are weighted and fused to obtain the sequence to be processed in the next iteration, wherein the inertial weights are the fusion weights of the standard residual sequence in the fusion process. The calculation method for the inertia weight is as follows: ； in This represents the inertia weight; This is a preset critical threshold. This is a preset sensitivity factor used to control the steepness of the weight changes; The extreme value interval volatility is used to characterize the distribution characteristics of extreme points.

2. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, The specific method for calculating the extreme value interval volatility is as follows: ； in This represents the volatility of the extreme value interval; Indicates the index of the extreme point; Indicates the first The extreme point and the first The time difference between extreme points; Indicates the first The extreme point and the first The time difference between extreme points; This indicates the magnitude of the current extreme point; This represents the global maximum amplitude within the entire sampling window; This represents the smallest positive number that is specified.

3. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, If the calculated inertia weight value is greater than the preset confidence threshold, then the inertia weight is forcibly set to 1.

4. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, The calculation method for the weighted fusion is as follows: ； in This represents the sequence to be processed in the next iteration; This represents the inertia weight; This represents the standard residual sequence; This represents the sequence to be processed in the current round.

5. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, Constructing the upper and lower envelopes of the sequence to be processed includes: The upper envelope is obtained by connecting all local maxima points of the sequence to be processed using cubic spline interpolation. The lower envelope is obtained by connecting all local minima of the sequence to be processed using cubic spline interpolation.

6. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 5, characterized in that, It also includes using the mean of the upper and lower envelopes as the mean line.

7. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, The fault direction is determined based on the transformed instantaneous phase difference, including: If the instantaneous phase difference between the first-order intrinsic mode components of the transient zero-sequence voltage and zero-sequence current is within a preset tolerance range of 180°, then the fault is determined to be located on the load side of the circuit breaker. If the instantaneous phase difference is within the preset tolerance range of 0°, the fault is determined to be located on the power supply side of the circuit breaker.

8. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 7, characterized in that, The preset tolerance range is 30° to 60°.

9. The method for detecting grounding faults in pole-mounted circuit breakers based on transient waveforms according to claim 1, characterized in that, The detection method is executed based on an intelligent controller containing a dual-core processing unit of FPGA and DSP; The FPGA is responsible for acquiring and buffering the transient signal sequence; The DSP is responsible for performing iterative sieving, Hilbert transform, and fault determination.

Images