A High-Resistance Fault Location Method for Distribution Networks Based on Virtual Capacitor Polarity and FIR Filtering

Abstract

This invention discloses a high-resistivity fault location method for distribution networks based on virtual capacitor polarity and FIR filtering. First, the zero-sequence voltage of the bus and the zero-sequence current of the feeder are acquired. Then, a differential filter is used to process the zero-sequence voltage, removing the unbalanced component. Simultaneously, a sliding accumulation filter is used to process the zero-sequence current, removing the unbalanced component and suppressing measurement noise. The waveforms of the processed zero-sequence voltage and the processed zero-sequence current are multiplied to obtain a fault feature fusion waveform. A sliding accumulation filter is then used to process the fault feature fusion waveform, suppressing noise interference from the zero-sequence voltage and enhancing fault characteristics. A synchronous attenuation integral is used to generate a fault metric. Finally, the fault location is determined based on the negative polarity of the metric. This invention can adapt to high-interference environments under high-resistivity fault conditions, accurately extracting fault features for fault location, and has good applicability and engineering application value.

Description

Technical Field

[0001] This invention relates to a method for extracting high-resistivity fault features in distribution networks based on virtual capacitor polarity and FIR filtering, belonging to the field of power system relay protection technology. Background Technology

[0002] As the direct connection hub between the power system and the user side, the safe and stable operation of the distribution network is crucial to the reliability of power supply. Neutral point resonant grounding systems, with their excellent arc-extinguishing capability and low fault current characteristics, are widely used in my country's medium-voltage distribution networks. However, high-resistivity faults, due to their high transition resistance, weak fault current, and accompanying nonlinear arc characteristics, pose a challenge for fault detection and location in distribution networks. Statistics show that high-resistivity faults account for more than 10% of all grounding faults in distribution networks. They are often caused by tree contact, insulation aging, or surface media (such as sand and cement). If not located and isolated in time, they may cause fires or evolve into phase-to-phase short circuits, seriously threatening power grid safety. Especially with the high penetration of new energy sources, the complex three-phase current distribution caused by distributed power source integration further exacerbates the challenge of high-resistivity fault location.

[0003] Existing high-resistivity fault location methods are mainly based on three categories: steady-state components, transient components, and traveling wave signals. Methods based on steady-state components (such as harmonic methods and negative-sequence current methods) rely on the post-fault power frequency characteristics, but the compensation effect of the arc suppression coil significantly suppresses the zero-sequence current amplitude, causing a sharp drop in sensitivity when the transition resistance exceeds 1kΩ. For example, the 5th harmonic method is limited by its low harmonic content and is easily affected by background interference, while the negative-sequence current method is sensitive to the three-phase imbalance of the system and has a high false alarm rate. Methods based on transient components achieve location by capturing the transient resonance characteristics of the initial stage of the fault, such as the first half-wave method and the transient energy method. However, transient signals decay rapidly, have wide bandwidths, and are significantly affected by line distributed parameters and the initial phase angle of the fault. Existing methods are prone to failure under strong noise or unbalanced component interference. Summary of the Invention

[0004] To address the shortcomings of existing high-resistance fault feature extraction methods, such as poor resistance to zero-sequence unbalance components and measurement noise, limited applicability to various operating conditions, and difficulty in accurately extracting fault features, a new high-resistance fault feature extraction method for distribution networks based on virtual capacitor polarity and FIR filtering is proposed. This method is based on the principle of virtual capacitor polarity and utilizes a combination of FIR filters for joint filtering. It exhibits complete robustness to zero-sequence unbalance classification, strong noise suppression capability, simple calculation method, and preserves the phase and amplitude characteristics of the original waveform in the output results, thus facilitating fault location.

[0005] The present invention adopts the following technical solution.

[0006] Methods for extracting high-resistivity fault features in distribution networks based on virtual capacitor polarity and FIR filtering include:

[0007] Step 1: When the zero-sequence voltage value of the bus exceeds the limit, the fault recording device installed at each measuring point of the distribution network immediately saves the zero-sequence voltage and zero-sequence current data for the three power frequency cycles before the fault and the two power frequency cycles after the fault.

[0008] Step 2: Use a differential filter to process the zero-sequence voltage, remove the unbalanced components in the zero-sequence voltage, and obtain the processed waveform of the zero-sequence voltage.

[0009] Step 3: Simultaneously, a sliding accumulator filter is used to process the zero-sequence current to remove the unbalanced component in the zero-sequence current and suppress measurement noise, resulting in the processed waveform of the zero-sequence current.

[0010] Step 4: Multiply the waveform after zero-sequence voltage processing with the waveform after zero-sequence current processing to obtain the fault characteristic fusion waveform.

[0011] Step 5: Finally, a sliding accumulator filter is used to process the fault feature fusion waveform to suppress noise interference caused by zero-sequence voltage and enhance fault features, thus obtaining the final fault feature extraction waveform.

[0012] Preferably, in step 1, when the zero-sequence voltage value of the bus exceeds the limit, the fault recording device installed at each measuring point in the distribution network immediately saves the zero-sequence voltage and zero-sequence current data for the three power frequency cycles before the fault and the two power frequency cycles after the fault:

[0013] When the zero-sequence voltage value ΔU0(k) of the bus exceeds 0.15 times the rated voltage U of the bus... N That is, satisfying ΔU0(k)>0.15U N At that time, the fault recording device installed at each measuring point in the distribution network immediately saves the zero-sequence voltage and zero-sequence current data for the three power frequency cycles before the fault and the two power frequency cycles after the fault.

[0014] Preferably, step 2 uses a differential filter to process the zero-sequence voltage:

[0015] The zero-sequence voltage measurement value is:

[0016]

[0017] In the formula, u m This is the zero-sequence voltage measurement value. This represents the true value of the zero-sequence voltage fault component. This represents the true value of the zero-sequence voltage imbalance component. This is noise in zero-sequence voltage measurement.

[0018] For the zero-sequence voltage measurement value u m Use the actual window length as one power frequency cycle T. n The differential filter then has an output value u. out (i) is:

[0019]

[0020] In the formula, N is the number of sampling points within one power frequency cycle, i.e., the window length.

[0021] Assume the sampling frequency is f s The system's power frequency cycle corresponds to the frequency f. n =1 / T n The formula for calculating the window length is:

[0022] N = f s / f n

[0023] If the sampling point at the time of the fault is set to i start And ignoring the influence of measurement noise, the output value u of the differential filter is... out (i) is:

[0024]

[0025] Preferably, in step 3, a sliding accumulator filter is used to process the zero-sequence current:

[0026] The zero-sequence current measurement value is:

[0027]

[0028] In the formula, i m This is the zero-sequence current measurement value. This represents the true value of the zero-sequence current fault component. This represents the true value of the zero-sequence current imbalance component. This is noise in zero-sequence current measurement.

[0029] For the zero-sequence current measurement value i m Use the actual window length as one power frequency cycle T. n The differential filter, then its output value I sum (i) is:

[0030]

[0031] If the sampling point at the time of the fault is set to i start And ignoring the influence of measurement noise, the output value I of the sliding accumulator filter is... sum (i) is:

[0032]

[0033] Preferably, in step 4, the waveform after processing the zero-sequence voltage is multiplied by the waveform after processing the zero-sequence current:

[0034] uout (i) and I sum (i) The fault feature fusion waveform VC(i) obtained after multiplication is:

[0035]

[0036] Preferably, step 5 uses a sliding accumulator filter to process the fault feature fusion waveform:

[0037] The fault feature fusion waveform VC(i) is processed using a sliding accumulator filter to obtain the final fault feature extraction waveform VC. out (i):

[0038]

[0039] Preferably, step 6 uses synchronous decay integral to generate the fault measure FM:

[0040]

[0041] Preferably, step 7, finding paths with negative fault metrics to determine the fault location, includes:

[0042] Step 7.1: Based on the fault measurement of each measuring point, start from the beginning of the busbar to find the measuring point with a negative fault measurement. The path formed during the search is the fault path.

[0043] Step 7.2: If there is still a measuring point downstream of the measuring point at the end of the fault path, the fault location is between the measuring point at the end of the fault path and the adjacent measuring point downstream; if there is no measuring point downstream of the measuring point at the end of the fault path, the fault location is between the measuring point at the end of the fault path and the end of the line. Attached Figure Description

[0044] Figure 1 Flowchart of Fault Section Location Algorithm

[0045] Figure 210kV radial resonant grounding system schematic diagram

[0046] Figure 3 Influence of the balance component on this method

[0047] Figure 4 The effect of measurement noise on this method Detailed Implementation

[0048] Simulation Analysis

[0049] A simulation model of a 10.5kV resonant grounding system was built using the electromagnetic transient simulation software PSCAD / EMTDC, such as... Figure 2As shown. The cable line uses YJV22-3*400 model, and the overhead line uses JKLYJ-150 model. The arc suppression coil is overcompensated by 5%, and the inductance of the arc suppression coil is L = 0.58H. f1 to f3 are different single-phase ground fault points. Measurement points m1 and m4 are set at the beginning of feeders l1 and l2 respectively for fault selection; measurement points m2 and m3 are set at the beginning of the branch feeder of feeder l1 for fault location. The simulation sampling frequency is 10kHz, therefore the window length N = 200.

[0050] Case 1: Verification of the impact of unbalanced components

[0051] To investigate the impact of unbalanced components on the proposed method, an artificial unbalanced state was created by adjusting the three-phase parameters of feeder l3. Figure 2 At point f2, a single-phase ground fault occurred in phase A, with a transition resistance of 3500Ω and an initial phase angle of 288°. The fault occurred at 3.016 seconds. Figure 3 (a) and Figure 3 As shown in (b), before the fault occurs, both the zero-sequence voltage and the zero-sequence current of each feeder exhibit significant unbalanced components. At the moment of fault occurrence, due to the influence of the unbalanced components, the characteristics of the fault component are masked by the unbalanced components, making it difficult to identify the fault occurrence time based solely on the zero-sequence voltage; the amplitude of the unbalanced component of the zero-sequence current is also greater than that of the fault component, making it difficult to accurately identify the fault based solely on the zero-sequence current. The method proposed in this paper is based on the superposition theorem and uses a differential filter to filter out the unbalanced voltage component, thereby effectively preserving the zero-sequence voltage fault component within one power frequency cycle after the fault, such as... Figure 3 As shown in (a), a sliding accumulator filter is used to filter out the unbalanced current component, thereby effectively extracting the zero-sequence current fault characteristics within one power frequency cycle after the fault, such as... Figure 3 As shown in (c). For Figure 3 VC at each measuring point shown in (e) out Fault measurement calculations were performed, and the results are as follows: Figure 3 As shown in (f), the fault measure of the upstream measuring point m1 is negative, while the fault measures of the downstream measuring point and the intact feeder measuring point are positive. The fault location algorithm determines that the fault location is between measuring points m1 and m2, and the location is correct.

[0052] Experimental results demonstrate that the proposed method is fully robust to unbalanced components. Even in the presence of unbalanced components, the method can still accurately extract fault features and successfully locate the fault.

[0053] Case 2: Verification of the impact of measurement noise

[0054] To investigate the tolerance of the research method to measurement noise, the following experiment was conducted: Figure 2At point f1, a single-phase ground fault with a transition resistance of 3000Ω and an initial phase angle of 90° occurred in phase A, with a fault time of 0.205s. Since measurement noise has a relatively small impact on the zero-sequence voltage, white noise with a signal-to-noise ratio of 45dB within one power frequency cycle at the initial fault time was added to the zero-sequence voltage, such as... Figure 4 As shown in (a); since the measurement noise of zero-sequence current is determined by the magnitude of the three-phase load current and the measurement error of the transformer, and the amplitude of the zero-sequence current fault component is very small when a high-resistance fault occurs, it is easily masked by the measurement noise. Therefore, an average noise energy of 1.13A per unit sampling point is added to the zero-sequence current at each measuring point. 2 The white noise at each measuring point within one power frequency cycle at the initial moment of the fault is m1: 4.14dB, m2: -88.13dB, m3: -33.82dB, and m4: -19.33dB, respectively. Figure 4 As shown in (b), the fault components at each measuring point are completely masked by measurement noise. Figure 4 VC at each measuring point shown in (e) out Fault measurement calculations were performed, and the results are as follows: Figure 4 As shown in (f), the fault measure of the upstream measuring point m1 is negative, while the fault measures of the downstream measuring point and the intact feeder measuring point are positive. The fault location algorithm determines that the fault location is between measuring points m1 and m2, and the location is correct.

[0055] Table 1. Location results of different fault conditions under the influence of measurement noise.

[0056]

[0057] The average energy of noise per unit sampling point is 1.13A added to the zero-sequence current at each measuring point under the fault conditions shown in Table 1. 2 White noise was used, and the maximum signal-to-noise ratio of the zero-sequence current at each measuring point was taken. The fault measurement and location results at each measuring point under different operating conditions are shown in Table 1. As shown in Table 1, the fault location results under each noisy operating condition are correct. Accurate location can still be achieved under the fault condition with a transition resistance of 3500Ω and a maximum signal-to-noise ratio of 3.87dB. This fully verifies that the proposed method still has high fault location accuracy and robustness under the influence of measurement noise.

[0058] Working principle

[0059] Virtual capacitance dynamic polarity: at the instant HIF occurs (t=0) + The inductor current of the arc suppression coil cannot change abruptly. The zero-sequence voltage and zero-sequence current of each feeder are approximately capacitively constrained near the initial moment of the fault. However, after the electrical quantities enter steady state (τ→+∞), due to the overcompensation of the arc suppression coil, there is no significant difference in amplitude and phase between the zero-sequence current from the beginning of the faulty feeder to the upstream of the fault point and the zero-sequence current of the healthy feeder. Therefore, it cannot be used as a criterion for line selection and location.

[0060] It is worth noting that the virtual capacitor is essentially the coupling characteristic between the zero-sequence voltage and the zero-sequence current at the measurement point, and is not a capacitor in the physical sense.

[0061] Therefore, the fault path can be found by relying on the polarity difference of the coupling characteristics of transient zero-sequence voltage and transient zero-sequence current near the initial moment of the fault, thereby locating the fault section.

[0062] The beneficial effects of this invention are that, compared with existing technologies, it proposes a high-resistance fault location method for distribution networks based on virtual capacitor polarity and FIR filtering. This method utilizes the transient electrical quantity characteristics at the moment of fault occurrence, combined with joint filtering using an FIR filter, effectively overcoming the shortcomings of traditional methods in high-resistance fault location. By introducing the concept of virtual capacitor and combining it with the signal processing advantages of FIR filters, effective suppression of unbalanced components and measurement noise is achieved, improving the accuracy and robustness of fault location. This method is not only applicable to different types of distribution network fault conditions, but also exhibits good location performance even under conditions of large transition resistance and weak fault characteristics, providing a strong guarantee for the safe and reliable operation of distribution networks.

[0063] Furthermore, the method proposed in this invention is easy to operate and implement in terms of data acquisition, feature extraction, metric generation, and fault location, making it more suitable for localized fault identification equipment that requires real-time acquisition and real-time processing and analysis.

[0064] The applicant of this invention has provided a detailed explanation and description of embodiments of the invention based on the accompanying drawings. However, those skilled in the art should recognize that the above embodiments are merely preferred embodiments of the invention. The detailed explanation is intended to assist the reader in gaining a deeper understanding of the spirit of the invention, rather than to limit the scope of protection of the invention. On the contrary, any improvements or variations made based on the spirit of the invention should be included within the scope of protection of the invention.

Claims

1. A method for locating high-resistivity faults in distribution networks based on virtual capacitor polarity and FIR filtering, characterized in that, The method includes the following steps: Step 1: When the zero-sequence voltage value of the bus exceeds the limit, the fault recording device installed at each measuring point in the distribution network immediately saves the zero-sequence voltage and zero-sequence current data for the three power frequency cycles before the fault and the two power frequency cycles after the fault. Step 2: Use a differential filter to process the zero-sequence voltage, remove the unbalanced components in the zero-sequence voltage, and obtain the waveform of the zero-sequence voltage after processing; Step 3: Simultaneously use a sliding accumulator filter to process the zero-sequence current, remove the unbalanced component in the zero-sequence current and suppress measurement noise, and obtain the waveform of the zero-sequence current after processing; Step 4: Multiply the waveform after processing the zero-sequence voltage with the waveform after processing the zero-sequence current to obtain the fault characteristic fused waveform; Step 5: Use a sliding accumulator filter to process the fault feature fusion waveform, suppress noise interference caused by zero-sequence voltage and enhance fault features to obtain the final fault feature extraction waveform. Step 6: Generate a fault measure using the synchronous decay integral; Step 7: Locate the path with a negative fault measure to determine the fault location.

2. The method for locating high-resistance faults in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, The data storage conditions for the fault recording device in step 1 are as follows: When the bus zero-sequence voltage value Exceeding 0.15 times the rated bus voltage That is, satisfying At that time, the fault recording device installed at each measuring point in the distribution network immediately saves the zero-sequence voltage and zero-sequence current data for the three power frequency cycles before the fault and the two power frequency cycles after the fault.

3. The method for locating high-resistance faults in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 2 uses a differential filter to process the zero-sequence voltage: The zero-sequence voltage measurement value is: In the formula, This is the zero-sequence voltage measurement value. This represents the true value of the zero-sequence voltage fault component. This represents the true value of the zero-sequence voltage imbalance component. For zero-sequence voltage measurement noise; Zero-sequence voltage measurement value Use the actual window length as one power frequency cycle. The differential filter, then its output value for: In the formula, N is the number of sampling points within one power frequency cycle, i.e., the window length; Assuming the sampling frequency is The system's power frequency cycle corresponds to the frequency of The formula for calculating the window length is: If the sampling point at the time of the fault is set to And ignoring the influence of measurement noise, the output value of the differential filter is... for: 。 4. The high-resistivity fault feature localization method for distribution networks based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 3 uses a sliding accumulator filter to process the zero-sequence current: The zero-sequence current measurement value is: In the formula, This is the zero-sequence current measurement value. This represents the true value of the zero-sequence current fault component. This represents the true value of the zero-sequence current imbalance component. Noise for zero-sequence current measurement; Zero-sequence current measurement value Use the actual window length as one power frequency cycle. The differential filter, then its output value for: If the sampling point at the time of the fault is set to And ignoring the influence of measurement noise, the output value of the sliding accumulator filter is... for: 。 5. The method for locating high-resistance faults in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 4 involves multiplying the waveform after processing the zero-sequence voltage with the waveform after processing the zero-sequence current: and The resulting fault feature fusion waveform after multiplication for: 。 6. The method for extracting high-resistivity fault features in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 5 uses a sliding accumulator filter to process the fault feature fusion waveform: Fault feature fusion waveforms are obtained using a sliding accumulator filter. The process is performed to obtain the final fault feature extraction waveform. : 。 7. The method for extracting high-resistivity fault features in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 6 uses synchronous attenuation integral to generate a fault measure. : 。 8. The method for extracting high-resistivity fault features in a distribution network based on virtual capacitor polarity and FIR filtering according to claim 1, characterized in that, Step 7, finding paths with negative fault metrics to determine the fault location, includes: Step 7.1: Based on the fault measurement of each measuring point, start from the beginning of the busbar to find the measuring point with a negative fault measurement. The path formed during the search is the fault path. Step 7.2: If there is still a measuring point downstream of the measuring point at the end of the fault path, the fault location is between the measuring point at the end of the fault path and the adjacent measuring point downstream; if there is no measuring point downstream of the measuring point at the end of the fault path, the fault location is between the measuring point at the end of the fault path and the end of the line.

Images