Cable fault positioning method based on non-fault phase current cross correlation

Abstract

A cable fault positioning method based on non-fault phase current cross-correlation comprises the following steps: S1, indirect reconstruction of a fault traveling wave signal is carried out, and a reconstruction signal is obtained; s2, performing maximum overlapping discrete wavelet transform multi-scale decomposition on the reconstructed signal obtained in the step S1, and obtaining an optimal detail coefficient; s3, obtaining an autocorrelation function of the optimal detail coefficient obtained in the step S2, and obtaining a traveling wave propagation time difference; s4, determining the propagation velocity v of the traveling wave in the cable, and finally determining the distance L from the fault point to the measurement point; through the above steps, cable fault point positioning is realized.

Description

Technical Field

[0001] This invention relates to the field of power system and related equipment technology, and in particular to power cable fault detection and location technology, and especially to a method for accurate cable fault location based on indirect acquisition and cross-correlation analysis of parallel non-faulty phase current signals. Background Technology

[0002] With the increasing cable coverage of urban power grids, rapid and accurate location of cable faults is crucial for ensuring power supply reliability. Traditional cable fault location methods often rely on power outage detection or manual waveform analysis, which suffer from high safety risks, low efficiency, and significant susceptibility to human factors. While the traveling wave method boasts high theoretical accuracy, its traditional implementation requires direct coupling with high-voltage signals, and wavefront identification is significantly affected by noise, making it difficult to apply safely and reliably in live-line detection scenarios. Existing cable fault location technologies mainly suffer from the following problems: 1. High Safety Risks: Traditional traveling wave detection requires direct access to the high-voltage fault phase, posing electrical safety hazards and making it unsuitable for live-line detection of cables in urban areas. For example, the "Development and Application of Online Monitoring and Location System for Cable Faults in Distribution Networks" proposes a "four-in-one" online monitoring system (partial discharge + sheath circulation current + traveling wave line selection + ranging), but the traveling wave signal still needs to be collected from near the fault phase (high voltage is not completely isolated), and wavefront identification relies on "fixed threshold + manual assistance," resulting in a positioning error of approximately 80-120m under strong noise (20dB). The safety hazards of this technology stem from the physical method of signal acquisition. Its traveling wave ranging function relies on installing sensors on the fault phase cable or busbar to directly measure the transient traveling wave signal generated by the fault. Its principle can be simplified as follows: This type of direct contact measurement requires a direct electrical connection or tight electromagnetic coupling between the sensor, acquisition circuit and the high-voltage conductor, which exposes the equipment to high-voltage risks throughout the entire process of deployment, operation and maintenance. In essence, it involves taking on additional safety risks in order to obtain signals.

[0003] 2. Insufficient Positioning Accuracy: Due to noise interference and the subjectivity of manual wavefront identification, the accuracy of time difference extraction is limited, leading to significant positioning errors. Reliance on experience-based judgment and a lack of automated, intelligent signal processing and fault identification mechanisms are also significant issues. For example, in the field of cable fault location, as described in "Research on Fault Location of Underground Power Cables," positioning accuracy directly determines the efficiency and cost of fault diagnosis. Recent research (such as "Research on Fault Location of Underground Power Cables") aims to improve accuracy under strong noise conditions through complex signal processing combinations (such as SSA-VMD-NTEO), which itself demonstrates the severe challenges that traditional methods face in terms of accuracy.

[0004] Therefore, there is an urgent need for a safe, high-precision, and automated method for locating cable faults to meet the needs of intelligent operation and maintenance of urban cable networks. Summary of the Invention

[0005] The purpose of this invention is to address the technical shortcomings of existing live cable fault location technologies, such as high safety risks and insufficient location accuracy, which make it difficult to effectively identify fault types and locate fault locations. Therefore, this invention proposes a more advanced live cable fault location and detection technology to meet the needs of urban cable operation and maintenance.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A cable fault location method based on the mutual correlation of non-faulty phase currents includes the following steps: Step S1: Perform indirect reconstruction of the fault traveling wave signal and obtain the reconstructed signal. ; Step S2: Reconstruct the signal obtained in step S1 Perform maximally overlapping discrete wavelet transform multi-scale decomposition and obtain the optimal detail coefficients. ; Step S3: Obtain the optimal detail coefficients obtained in step S2. autocorrelation function And obtain the time difference of traveling wave propagation. ; Step S4: Determine the propagation speed v of the traveling wave in the cable, and finally determine the distance L from the fault point to the measurement point; The above steps are used to locate the cable fault point.

[0007] In step S1, the specific procedure is as follows: In a three-phase cable system, when a fault occurs in one phase, the transient current signals of the remaining non-faulty phases are collected. and The traveling wave current signal of the fault phase is reconstructed using Kirchhoff's current law. Specifically: (1); in, and This refers to the traveling wave component extracted from the non-faulty phase current.

[0008] In step S2, the optimal detail coefficients are selected based on the signal-to-noise ratio and energy concentration criteria. For use in subsequent analysis.

[0009] In step S3, the autocorrelation function is obtained. Use the following formula: (2); in, It refers to delay The optimal detail coefficients after that.

[0010] In step S3, by detection exist The first significant peak position within the interval Obtain the time difference of traveling wave propagation : (3).

[0011] In step S4, the distance L from the fault point to the measurement point is calculated based on the cable wave velocity v and the time difference Δt. (4).

[0012] The wave velocity v is determined by the inductance per unit length of the cable. With capacitor Decide: (5).

[0013] It also includes step 5: performing error analysis and confidence assessment, specifically by evaluating the total error range of the positioning results. Simultaneously, the signal-to-noise ratio (SNR) and peak value ratio (PKR) of the autocorrelation function are considered. Relative error of wave speed Considering multiple factors, a comprehensive confidence function C is constructed and calculated. The final output is the fault distance L and the possible error range ± And confidence levels (such as high, medium, and low) provide quantitative basis for operation and maintenance decisions.

[0014] In step 2, the number of decomposition layers J is adaptively determined based on the cable length and sampling rate, with the preferred frequency band being the high-frequency scale corresponding to the energy concentration of the traveling wave front. In step 4, if the cable type is known, the cable wave velocity v is directly obtained from the parameter library; otherwise, wave velocity calibration is performed using a cable segment of known length.

[0015] In step 3, a dynamic threshold is set to eliminate spurious peaks caused by noise. , where α∈(0,1), only when the peak value is greater than Only then is it identified as a valid peak value.

[0016] In step 5, a positioning error model is constructed: ; in, This is for time difference extraction error; And define a multi-factor confidence function: ; Where β1+β2+β3=1, and C∈[0,1], a higher value indicates a more reliable positioning result. Compared with the prior art, the present invention has the following technical effects: 1) This invention utilizes the parallel non-faulty phase current to indirectly reconstruct the fault signal, eliminating the need to install a dedicated sensor on the high-voltage side of the faulty phase. It is particularly suitable for live detection and online monitoring of urban cable lines, greatly improving operational safety. 2) This invention employs a dual strategy of MODWT optimal frequency band selection and cross-correlation peak detection, which effectively suppresses noise and achieves automatic and high-precision extraction of wavefront time difference, overcoming the random errors of manual identification; 3) The cross-correlation algorithm used in this invention has good noise suppression characteristics. Combined with the time-frequency focusing capability of wavelet transform, the system can still maintain stable performance in complex electromagnetic environments. 4) This invention requires no human intervention throughout the entire process from signal acquisition, processing, analysis to result evaluation, and outputs confidence indexes, providing a reliable technical means for fault self-healing and decision support of smart grids. 5) This invention constructs a complete mathematical model chain from physical principles (KCL), signal processing (MODWT, cross-correlation) to error assessment. The logic is clear and it is easy to implement in engineering and optimize performance. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a schematic diagram of indirect reconstruction of fault traveling wave signal based on parallel non-fault phase current in an embodiment of the present invention; Figure 3 This is a schematic diagram of MODWT multi-scale decomposition and optimal frequency band selection in an embodiment of the present invention; Figure 4 This is a cross-correlation analysis and extraction graph in an embodiment of the present invention; Figure 5 This is a simulation diagram of the output interface in an embodiment of the present invention; Figure 6 This is a comparison chart of the localization results of the traditional method and the method of the present invention under strong noise in the embodiments of the present invention. Detailed Implementation

[0018] like Figure 1 The cable fault location method shown includes the following steps: (1) Parallel non-faulty phase current detection and cross-correlation analysis. Step S1: Indirect acquisition and reconstruction of fault traveling wave signal: When a fault occurs in one phase (e.g., phase A) of a three-phase cable system, high-frequency current sensors (such as clamp-on current transformers with a bandwidth of not less than 1MHz) are used at measurement points such as the substation outgoing switch cabinet to simultaneously collect the transient current signals of two non-faulty phases (phase B and phase C). and High-frequency traveling wave components in the signal are extracted using a digital high-pass filter. and According to Kirchhoff's current law, the traveling wave signal of the faulty phase (phase A) is indirectly reconstructed as: ; This step electrically isolates the device from the high-voltage fault phase, ensuring the safety of the testing process; Step S2, Signal Preprocessing and Optimal Frequency Band Selection: For the reconstructed signal Perform maximum overlap discrete wavelet transform (MODWT) multi-scale decomposition. The number of decomposition layers (typically 4-7 layers) is adaptively determined based on cable parameters and sampling rate. The energy distribution of detail coefficients in each layer is analyzed, and the frequency bands with the highest travel wavefront energy and signal-to-noise ratio (e.g., detail coefficients corresponding to 12.5-100 kHz) are selected as optimal. ), for subsequent precise analysis; Step S3, Automatic Time Difference Extraction: Calculate the selected optimal frequency band signal. autocorrelation function Within the interval τ>0, detection The first significant local peak (corresponding to the correlation between the initial traveling wave and its reflected wave). By setting a dynamic threshold (e.g. (α is set to 0.5-0.7) to eliminate spurious peaks caused by noise, thereby automatically and robustly extracting the round-trip time difference of the traveling wave. .

[0019] Step S4, Fault Distance Calculation: Determine the propagation speed v of the traveling wave in the cable based on the cable type or pre-calibrated parameters (for commonly used XLPE cables, the typical value is approximately 1.52 × 10⁸ m / s). According to the formula: ; Calculate the distance L from the fault point to the measurement point; Step S5, Error Analysis and Confidence Assessment: Based on the error propagation theory, assess the total error range of the positioning results. Simultaneously, the signal-to-noise ratio (SNR) and peak value ratio (PSR) of the autocorrelation function are considered. Relative error of wave speed Considering multiple factors, a comprehensive confidence function C is constructed and calculated. The final output is the fault distance L and the possible error range ± And confidence levels (such as high, medium, and low) provide a quantitative basis for operation and maintenance decisions; Example 1: Locating a single-phase grounding fault in a 10kV XLPE cable; This embodiment is used to fully demonstrate the implementation process of the method of the present invention and to verify its positioning accuracy in typical scenarios.

[0020] Implementation Process: On a 10kV distribution network XLPE cable line in a certain city, phase A experienced a ground fault via a 100Ω transition resistor approximately 4.23 kilometers from the substation. At the substation's outgoing line cabinet, a clamp-on high-frequency current transformer with a bandwidth of 0-1 MHz was used to synchronously acquire the transient fault currents of phases B and C at a sampling rate of 10 MHz. Steps S1 to S5 were sequentially executed on the acquired signals. Specifically, in step S2, the 'db4' wavelet was used for MODWT decomposition, and the fourth-level detail coefficients (corresponding to a frequency band of approximately 31.25-62.5 kHz) were selected as the optimal analysis component.

[0021] Table 1: Parameters related to MODWT decomposition;

[0022] This embodiment clearly demonstrates the complete chain of the present invention from safety signal acquisition to intelligent analysis output. It achieves sub-one-hundredth level positioning accuracy without contacting the high-voltage fault phase, and provides quantitative support for the reliability of the results through confidence level assessment.

[0023] Example 2: Comparison of anti-interference performance under strong noise environment; This embodiment aims to highlight the robustness of the method of the present invention compared with traditional methods in strong electromagnetic interference environments through simulation comparison.

[0024] Implementation process: A fault model identical to that in Example 1 (4.23 km single-phase grounding) was constructed in the simulation platform. Based on this, 20 dB of Gaussian white noise was superimposed on the measurement signal to simulate the complex electromagnetic interference commonly found in urban cable corridors. The method of this invention (MODWT + cross-correlation) and the traditional fixed-threshold traveling wave detection method were used for processing, respectively.

[0025] Table 2: Comparison between the method of the present invention (MODWT + cross-correlation) and the traditional fixed threshold traveling wave detection method;

[0026] This embodiment demonstrates that the signal processing architecture employed in this invention possesses strong noise suppression capabilities. Even in a 20 dB noise environment where traditional methods have failed, this invention maintains high-precision positioning, showcasing excellent environmental adaptability.

[0027] In summary, this invention employs a radically different approach from existing technologies, proposing an indirect signal reconstruction method based on Kirchhoff's Current Law (KCL). When a fault occurs in a phase of the system (e.g., phase A), it is unnecessary to physically access the high-voltage faulty phase. Instead, transient current signals are collected from two electrically isolated and easily operable non-faulty phases (phases B and C), and the traveling wave information of the faulty phase is obtained through the following mathematical reconstruction: The physical meaning of this formula is very clear: in an ideal symmetrical three-phase system, the current traveling wave injected into the fault point at the instant of a fault must have its return current shared by the other two non-faulty phases. Therefore, by synchronously acquiring the current traveling wave components of the non-faulty phases with high precision... and The negative of the sum is the fault phase current. Therefore, this invention shifts the physical location of signal acquisition from the high-voltage fault phase to the safe, non-faulty phase. This shift is not simply an improvement in signal coupling methods, but rather a realization of "electrical isolation between the detection behavior and the high-voltage conductor" from a metrological perspective. This fundamentally eliminates the inherent risk of high-voltage electric shock associated with traditional methods, providing a truly inherently safe solution for uninterrupted power-off detection and online monitoring of urban cables.

[0028] In addition, to address the technical problem of insufficient positioning accuracy in existing technologies, the positioning method of this invention achieves a more fundamental and robust breakthrough in its core algorithm principle. Its superiority is mainly reflected in the following two aspects: (1) The core of existing high-precision methods lies in optimizing the detection of single-point energy surges after feature extraction. The process can be summarized as follows: first, the VMD parameters are optimized through intelligent algorithms to separate the relatively pure fault component IMF from the noise; then, the instantaneous energy of this component is calculated using the improved energy operator NTEO. By searching energy sequences The abrupt change point is used to calibrate the wavefront time m. Although NTEO introduces parameters... While it improves noise immunity, it is still essentially an energy calculation of a single sampling point and its neighborhood. Under extreme noise or waveform distortion, noise pulses may still produce pseudo-energy abrupt changes, leading to calibration errors.

[0029] In contrast, this invention employs waveform similarity analysis based on autocorrelation functions. Its core step is to analyze the optimal detail coefficients obtained through MODWT decomposition. Perform autocorrelation calculation: The physical meaning of this formula is: to calculate the signal. Compared to its own delayed τ-replica Similarity (inner product) over the entire time axis. The reflected wave from the fault point is essentially a delayed copy of the initial traveling wave. Therefore, the delay time τ is exactly equal to the round-trip time of the traveling wave to the fault point. hour, A peak will appear that indicates a high degree of similarity between the two.

[0030] The fundamental advantage lies in the fact that autocorrelation is a mathematical correlation operation applied to the entire waveform segment, rather than an energy judgment at a single point. Random noise is uncorrelated with the signal itself, and even less so with its delayed replicas; therefore, noise energy is extensively suppressed in autocorrelation operations. In contrast, genuine fault reflection waves are strongly correlated with the incident wave, and their peak values... This will become apparent. This shift from "point detection" to "surface analysis" gives the algorithm a more solid mathematical foundation for noise resistance.

[0031] (2) Existing advanced methods have to introduce complex parameter optimization steps to achieve high accuracy. For example, the SSA-VMD-NTEO method requires the use of the Slug Group Algorithm (SSA) with fuzzy entropy as the fitness function to search and optimize the two key parameters of VMD, namely the number of modes K and the penalty factor α, in order to overcome the randomness of VMD decomposition. After that, it is also necessary to select a suitable resolution parameter i for NTEO. This process is computationally intensive, and its effectiveness depends on the optimization algorithm itself and the selected fitness function.

[0032] The core algorithm chain of this invention possesses higher inherent stability and lower parameter tuning complexity. The MODWT transform itself has excellent mathematical properties, and the selection of its optimal detail coefficients is based on the signal-to-noise ratio and energy concentration criterion, which is a more direct and stable standard. More importantly, the subsequent autocorrelation analysis... It is a classic mathematical tool whose performance does not depend on any complex parameters that require repeated optimization through intelligent algorithms. Wavefront time difference The determination is made by searching Significant peaks were achieved, supplemented by dynamic thresholds. To eliminate spurious peaks, α here is a coefficient with a clear meaning and a stable value range (usually 0.5-0.7), which does not require complex optimization.

Claims

1. A cable fault location method based on the cross-correlation of non-faulty phase currents, characterized in that, Includes the following steps: Step S1: Perform indirect reconstruction of the fault traveling wave signal and obtain the reconstructed signal. ; Step S2: Reconstruct the signal obtained in step S1 Perform maximally overlapping discrete wavelet transform multi-scale decomposition and obtain the optimal detail coefficients. ; Step S3: Obtain the optimal detail coefficients obtained in step S2. autocorrelation function And obtain the time difference of traveling wave propagation. ; Step S4: Determine the propagation speed v of the traveling wave in the cable, and finally determine the distance L from the fault point to the measurement point; The above steps are used to locate the cable fault point.

2. The method according to claim 1, characterized in that, In step S1, the specific procedure is as follows: In a three-phase cable system, when a fault occurs in one phase, the transient current signals of the remaining non-faulty phases are collected. and The traveling wave current signal of the fault phase is reconstructed using Kirchhoff's current law. Specifically: （1）； in, and This refers to the traveling wave component extracted from the non-faulty phase current.

3. The method according to claim 1, characterized in that, In step S2, the optimal detail coefficients are selected based on the signal-to-noise ratio and energy concentration criteria. For use in subsequent analysis.

4. The method according to claim 3, characterized in that, In step S3, the autocorrelation function is obtained. Use the following formula: （2）； in, It refers to delay The optimal detail coefficients after that.

5. The method according to claim 4, characterized in that, In step S3, by detection exist The first significant peak position within the interval Obtain the time difference of traveling wave propagation : （3）。 6. The method according to any one of claims 1 to 5, characterized in that, In step S4, the distance L from the fault point to the measurement point is calculated based on the cable wave velocity v and the time difference Δt. （4）。 7. The method according to claim 6, characterized in that, The wave velocity v is determined by the inductance per unit length of the cable. With capacitor Decide: (5)。 8. The method according to claim 1, characterized in that, It also includes step 5: performing error analysis and confidence assessment, specifically by evaluating the total error range of the positioning results. Simultaneously, the signal-to-noise ratio (SNR) and peak value ratio (PKR) of the autocorrelation function are considered. Relative error of wave speed These multiple factors are used to construct and calculate the comprehensive confidence function C. The final output is the fault distance L and the possible error range ± And confidence level, providing quantitative basis for operation and maintenance decisions.

9. The method according to claim 1, characterized in that, In step 2, the number of decomposition layers J is adaptively determined based on the cable length and sampling rate, with the preferred frequency band being the high-frequency scale corresponding to the energy concentration of the traveling wave front. In step 4, if the cable type is known, the cable wave velocity v is directly obtained from the parameter library; otherwise, wave velocity calibration is performed using a cable segment of known length.

10. The method according to claim 1, characterized in that, In step 3, a dynamic threshold is set to eliminate spurious peaks caused by noise. , where α∈(0,1), only when the peak value is greater than Only then is it identified as a valid peak value.

Images