A method and system for cable fault detection and location

Abstract

This invention belongs to the field of power fault detection technology, specifically a cable fault detection and location method and system. It aims to solve the core problem that transient traveling waves of 0.4kV cable faults are easily submerged by noise and the wavefronts cannot be identified under strong electromagnetic interference environments. After fault isolation, a mobile inspection is performed along the cable path, simultaneously collecting mixed signals and geographical location information. The three-phase sampling sequence is decoupled using Karrenbauer phase-mode transformation to eliminate inter-phase electromagnetic coupling interference, and fault characteristic components are selected. The characteristic components are then subjected to Db45 wavelet multi-scale decomposition to extract high-frequency detail coefficients. Effective fault wavefronts are screened from strong noise using energy, amplitude, and steepness criteria, and the time difference between the incident and reflected waves is extracted. The fault distance is calculated by combining corrected wave velocity and system delay compensation, and real-world geographic coordinates are generated by fusing RTK and IMU positioning information. This achieves highly reliable identification and accurate location of fault traveling wavefronts under strong interference environments.

Description

Technical Field

[0001] This invention belongs to the field of power fault detection technology, specifically a cable fault detection and location method and system. Background Technology

[0002] The statements in this section only refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] The 0.4kV outgoing cable on the low-voltage side of a 10kV distribution transformer is a critical component of the distribution network. Its complex branching and the common practice of multiple circuit cables being laid in parallel within the same trench are significant challenges. Due to the dense connection of frequency converters and high-power equipment, the operating environment experiences severe interference from power frequency harmonics and spatial electromagnetic clutter. When a phase-to-phase short circuit or single-phase ground fault occurs in the cable, the breakdown arc at the fault point spontaneously generates a transient traveling wave signal, which is an ideal information source for fault location.

[0004] However, in practical engineering applications, the amplitude of fault transient traveling waves is usually only in the millivolt range, making them easily drowned out by strong background electromagnetic noise. On the one hand, existing cable fault location equipment generates a large number of false pulses when processing 0.4kV low-voltage cable signals due to electromagnetic coupling between the three phases, which mix with the real fault wavefronts. On the other hand, conventional equipment lacks the ability to decompose signals at multiple scales and to filter them using refined criteria. In the face of strong interference environments, it is difficult to identify the real traveling wavefronts, resulting in the fault ranging program failing to start or producing significant location errors. Summary of the Invention

[0005] This invention provides a cable fault detection and location method and system, aiming to solve the problems in the prior art where the transient traveling wave of a 0.4kV cable fault is easily submerged by noise due to strong electromagnetic interference such as multiple cables in the same trench and frequency converter harmonics, the real traveling wave head cannot be effectively extracted and identified, and ranging and location are difficult to perform. It achieves high-reliability fault wave head capture and accurate location under strong interference environment.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The first aspect of this invention discloses a cable fault detection and location method, comprising the following steps:

[0008] After a fault trip occurs in the low-voltage side outgoing cable of the transformer, disconnect the circuit breaker of the faulty branch and keep the transformer body and other non-faulty outgoing circuits energized and operating normally.

[0009] The system moves along the cable laying path for inspection, collecting mixed time-domain signals from the cable and geographical location information of the current location; the mixed time-domain signals carry transient traveling wave signals generated by fault point puncture arc;

[0010] The collected three-phase time-domain sampling sequence was subjected to Karrenbauer phase-mode transformation, and zero-mode and line-mode components were obtained after decoupling. A fault characteristic component was selected according to the fault type to eliminate inter-phase electromagnetic coupling interference between the three-phase cables.

[0011] The fault feature components are decomposed by Db45 wavelet multi-scale decomposition to extract high-frequency detail coefficients. The high-frequency detail coefficients are then screened by a triple quantitative criterion of energy threshold, amplitude threshold and steepness threshold. Effective fault traveling wave fronts are identified from the background of strong electromagnetic interference, and the time difference between incident wave and reflected wave is extracted.

[0012] Based on the time difference, the traveling wave velocity after multi-dimensional correction, and the inherent hardware delay of the system, the distance from the fault point to the transformer output terminal is calculated. By performing multi-source fusion with geographic location information, the actual geographic coordinates of the fault point are generated.

[0013] Furthermore, during the acquisition of signals from the cable, a fixed-frequency reference synchronization signal injected into the cable by the transmitting subsystem is also acquired. The reference synchronization signal is used to provide a unified clock reference for both the transmitting and receiving ends and to suppress power frequency harmonic interference in the frequency domain.

[0014] Furthermore, the fault type selection rules are as follows: During the phase-mode transformation decoupling, the three-phase sampling sequence is transformed by Karrenbauer to obtain zero-mode components and line-mode components; among them, the global spatial induced common-mode interference is mapped to the zero-mode component, and the fault arc differential-mode pulse is mapped to the line-mode component; the zero-mode component containing only spatial interference is discarded, and the line-mode component is selected as the fault characteristic component; the corresponding single-phase line-mode component is extracted for single-phase grounding faults, and the two-phase coupled line-mode component is extracted for phase-to-phase short-circuit faults.

[0015] Furthermore, the triple quantitative criteria include:

[0016] The energy criterion is as follows: the total energy of the high-frequency detail coefficients exceeds a preset energy threshold.

[0017] The amplitude criterion is as follows: the wavefront amplitude is not less than the set value;

[0018] The steepness criterion is as follows: the duration of the rising edge of the wave head does not exceed a set value.

[0019] When all three criteria are met simultaneously, the wave is determined to be a valid faulty traveling wave head.

[0020] Furthermore, the traveling wave velocity is obtained through a four-level progressive correction model:

[0021] The first level involves selecting the theoretical reference wave velocity based on the type of cable insulation material.

[0022] The second stage involves conducting an open-circuit reflection test based on the actual total length of the cable to invert and calibrate the wave velocity.

[0023] The third level involves environmental correction by overlaying on-site ambient temperature, humidity, and cable insulation aging coefficient.

[0024] The fourth level involves making a slight offset correction to the wave velocity based on the measured error of historical fault location data.

[0025] Furthermore, in the ranging calculation step, the formula for calculating the fault distance is:

[0026] ;

[0027] in, The distance to the fault. This is the corrected traveling wave velocity. The time difference between the incident wave and the reflected wave. This is due to the inherent hardware delay of the system.

[0028] Furthermore, in the coordinate fusion step, a fusion positioning method combining RTK differential positioning and IMU inertial navigation is adopted, and the weighting coefficients are dynamically adjusted according to the number of satellites acquired.

[0029] When the number of satellites searched is greater than or equal to the first threshold, RTK coordinates are used as the primary coordinates.

[0030] When the number of satellites acquired is between the second and third thresholds, IMU inertial navigation is the primary method.

[0031] When the number of satellites searched is less than the third threshold or there is no satellite signal, relative positioning is performed by relying on the pre-stored coordinates of the GIS pipeline turning points and the walking distance, and the accumulated drift error of the IMU is manually eliminated at the cable well.

[0032] Furthermore, in the wavelet decomposition and wavefront identification steps, the Db45 wavelet is used to perform multi-level Mallat tower decomposition, discarding all low-frequency approximation coefficients and retaining only the high-frequency detail coefficients from the first to the kth layers for fault traveling wave determination.

[0033] Furthermore, before the signal acquisition step, there is also a power-on calibration step: the transmitting subsystem injects a fixed frequency reference signal into the cable, and the handheld receiving subsystem acquires the reference waveform at the transmitting point. When the effective value of the received sampling voltage and the signal-to-noise ratio (SNR) exceed the corresponding set values, the calibration is qualified and the coupling compensation parameters are locked.

[0034] A second aspect of the present invention discloses a cable fault detection and location system, comprising:

[0035] The transmitting subsystem is installed on the low-voltage side of the transformer's outgoing phase line and is used to inject a fixed-frequency reference synchronization signal into the cable. This reference signal is used to provide a unified clock reference for both the transmitting and receiving ends and to suppress power frequency harmonic interference in the frequency domain. It is not used as a fault excitation signal. The transmitting subsystem includes a nanocrystalline open-close Rogowski coupling clamp, an LC passive impedance matching network, a DDS digital signal source, a programmable power amplifier circuit, and a time-division main control unit.

[0036] The handheld receiver subsystem is used for mobile inspection along the cable laying path, simultaneously acquiring mixed time-domain signals and geographical location information; the handheld receiver subsystem includes a high-sensitivity sensing coil, a two-stage programmable amplifier circuit, a low-pass filter circuit, an RTK differential positioning module, an IMU inertial navigation unit, and a wireless transmission module;

[0037] The data processing subsystem is used to receive and process data uploaded by the handheld receiving subsystem, and to perform phase mode transformation decoupling, wavelet decomposition and wavefront recognition, ranging calculation and coordinate fusion calculation.

[0038] The electronic system is connected to the transmitting subsystem, the handheld receiving subsystem, and the data processing subsystem to provide operating power.

[0039] Compared with existing technologies, one or more of the above technical solutions have the following beneficial effects:

[0040] By decoupling through phase mode transformation to eliminate spurious pulses generated by electromagnetic coupling between phases of three-phase cables, the fault characteristic components are separated from the coupled interference. Then, Db45 wavelet multi-scale decomposition is used to peel the signal layer by layer according to frequency, effectively separating the high-frequency detail coefficients of the fault traveling wave concentration from low-frequency interference such as power frequency and environmental noise in the frequency domain. Next, high-frequency coefficients are screened step-by-step using three quantitative criteria: energy, amplitude, and steepness. Only when all three conditions are met simultaneously is it determined to be a valid fault wavefront, thus accurately distinguishing the brief, steep, and energy-concentrated real fault traveling wave from various interference pulses with different characteristics. Through this three-level anti-interference architecture of "decoupling-layering-screening," reliable identification and accurate extraction of fault traveling wavefronts are achieved in environments with strong electromagnetic interference. Attached Figure Description

[0041] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0042] Figure 1 This is a schematic diagram of a cable fault detection and location process provided for one or more embodiments of the present invention. Detailed Implementation

[0043] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0044] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0045] As described in the background section, in the traditional 0.4kV cable fault location process, maintenance personnel use induction coils to pick up signals along the cable path. However, due to the parallel laying of multiple cable loops and the confined space, electromagnetic coupling between the three-phase cables generates a large number of false pulses similar to the fault waveform; at the same time, power frequency harmonics and high-frequency noise from the frequency converter completely cover the weak transient traveling wave of the fault in the time domain. Existing equipment directly judges the amplitude or energy threshold of the mixed signal, lacking a means to decouple and separate the true fault characteristics from the strong noise background, resulting in a high wavefront misjudgment rate.

[0046] Therefore, this solution provides a cable fault detection and location method and system. It eliminates interphase electromagnetic coupling interference in three-phase lines through a phase mode transformation decoupling algorithm and removes false pulses caused by mutual inductance. Then, it introduces multi-scale wavelet decomposition technology to peel the signal layer by layer according to frequency and sets three quantitative criteria of energy, amplitude and steepness at each scale to screen and lock the real wavefront that meets the characteristics of the fault traveling wave step by step, thereby achieving highly reliable identification of effective traveling waves and accurate calibration of wavefront time in strong interference environment.

[0047] Example 1:

[0048] A cable fault detection and location system includes a transmitting subsystem, a handheld receiving subsystem, a data processing subsystem, and a power supply subsystem, used for fault location of a 0.4kV cable after a fault trip.

[0049] 1.1 Launch Subsystem

[0050] The transmitting subsystem is snapped onto the 0.4kV outgoing phase line of the low-voltage side of the 10kV transformer. It includes a nanocrystalline openable Rogowski coupling clamp, an LC passive impedance matching network, a DDS digital signal source, a programmable power amplifier circuit, and a time-division main control unit. The transmitting signal is fixed as a 1kHz, 15mV effective value sinusoidal reference signal. The reference signal is only used for all-time clock synchronization and power frequency interference suppression and is not used as fault excitation. The fault traveling wave is generated autonomously by the fault point breakdown arc.

[0051] Among them, the nanocrystalline opening and closing Rogowski coupling clamp is the core coupling device of this component. Its function is to inject a 1kHz reference signal without breaking the wire and with live coupling, so as to realize the reference is conducted along the entire 0.4kV cable without cable stripping or electrical hard wiring. It is the key hardware for the uninterrupted deployment of the reference in this device.

[0052] 1) Opening-closing Rogowski coupling clamp: The magnetic core is made of high-permeability nanocrystalline ribbon wound into shape, with an initial permeability ≥80000. A 0.1mm non-magnetic epoxy gasket is set in the jaw gap to suppress magnetic leakage. The inner diameter of the clamp is divided into three levels: 50mm, 80mm, and 120mm to adapt to the commonly used 0.4kV cables from YJV-3×16 to YJV-3×185. The secondary coil is made of Φ0.12mm high-strength enameled wire with 1200 turns, and is fully epoxy-encapsulated for moisture and shock resistance, making it suitable for cable trenches and outdoor humid and dusty conditions.

[0053] The 1kHz / 15mV alternating electrical signal output by the DDS digital signal source is connected to the secondary winding of the coupling clamp. The alternating current generates an alternating magnetic field in the nanocrystalline magnetic core. Relying on electromagnetic induction, a reference potential of the same frequency is induced on the clamped running cable. The reference is transmitted along the entire cable, which is convenient for the remote receiving coil to pick up in space and complete the synchronization of the transmitting and receiving clocks.

[0054] Formula for calculating monthly calibration coupling efficiency:

[0055] ;

[0056] In the formula: U set U is the transmit reference voltage; rec K represents the sampled voltage at the receiving end. amp To receive the amplification factor.

[0057] 2) LC impedance matching network: A dedicated parallel resonant branch for 1kHz is formed by a C0G high-frequency capacitor and a manganese zinc ferrite inductor with parameters L=1.2mH and C=100μF. This network cancels the inherent leakage inductance loss of the Rogowski coil, compensates for the parasitic parameters of the coil, and improves the reference coupling efficiency at the operating frequency.

[0058] 3) The DDS signal source is AD9833, and the subsequent stage is built with OPA2340 to form a Class A power amplifier. A 10Ω current-limiting fuse resistor is connected in series at the output. The equipment is factory locked to output 1kHz / 15mV and has no frequency or amplitude adaptive adjustment function.

[0059] 4) The time-division main control uses an STM8S003 microcontroller. The control logic is: transmit for 0.5s and sleep for 2s to intermittently output the reference signal, thereby reducing the power consumption of the whole machine.

[0060] 5) Calibration rules: Perform 10s power-on calibration each time the power is turned on, with a fixed output of 1kHz reference. The reference is injected into the cable by the coupling clamp. The calibration is qualified when the received sampling voltage is ≥0.5mV and SNR is ≥15dB, and the on-site clamping status self-check is realized. The coupling efficiency and the inherent delay Δt0 of the system are calibrated monthly using a 100m national standard XLPE0.4kV standard cable.

[0061] 1.2 Handheld Receiver Subsystem

[0062] The handheld receiver subsystem moves along the 0.4kV cable laying path for inspection and includes a high-sensitivity sensing coil, a two-stage programmable amplifier circuit, a 10kHz second-order Butterworth active low-pass filter circuit, an RTK differential positioning module, an MPU6050 six-axis IMU inertial unit, and an EC204G wireless transmission module; the overall gain is fixed at 2000 times, and it is specifically designed for fault transient traveling wave acquisition.

[0063] 1) Induction coil: 3000 turns of Φ0.1mm enameled wire wound with manganese-zinc magnetic winding, no-load noise ≤0.008mV, minimum detection of weak fault traveling wave of 0.01mV;

[0064] 2) Signal conditioning: The first stage uses an INA2128 instrumentation amplifier with a fixed 20x amplification to suppress common-mode interference; the second stage uses a hardware relay with a fixed amplification factor, locking the total amplification factor at 2000x; a TL082 is used to build a 10kHz cutoff frequency filter circuit to filter out RF and switching power supply noise.

[0065] 3) Fusion positioning formula:

[0066] ;

[0067] In open areas, ≥8 satellites are searched, with α=0.9 based on RTK coordinates; for short-term building obstruction, α=0.1 is calculated using IMU integration; in enclosed cable trenches without satellites, relative positioning is achieved by relying on pre-stored GIS pipeline turning points and walking distance; cable wells are manually cleared with one click to eliminate accumulated inertial navigation errors; waveforms and coordinates are uploaded to the data processing subsystem in real time via 4G.

[0068] 1.3 Data Processing Subsystem

[0069] Industrial-grade embedded main controller + local flash memory, with built-in Karrenbauer phase mode transformation, Db45 wavelet 5-level decomposition, multi-parameter wave velocity correction, and fault location algorithm.

[0070] 1) Benchmark theoretical wave speed:

[0071] ;

[0072] XLPE insulation =2.3, PVC insulation =3.2, Rubber insulation =4.0, based on the reference wave velocity selected for 0.4kV cable;

[0073] 2) Level 4 wave velocity correction:

[0074] Level 1: Material selection and loading of theoretical wave velocity;

[0075] Level 2: Input cable total length L real Back-calculation of manual open-circuit test at the end:

[0076] ;

[0077] Level 3: Overlay with temperature and humidity correction, insulation aging correction:

[0078] ;

[0079] Level 4: Fine-tuning wave velocity based on historical fault data;

[0080] 3) Final fault location formula:

[0081] ;

[0082] The time difference between the initial fault wave and the reflected wave. The inherent hardware latency of the system obtained from monthly calibration;

[0083] 4) Wavelet energy criterion:

[0084] ;

[0085] If the three conditions of rising edge ≤ 1ms, amplitude ≥ 0.5mV, and energy E ≥ threshold are met, the traveling wave is determined to be a valid fault; otherwise, it is determined to be environmental interference and the sampled data is discarded.

[0086] It also features an electronic power supply system, using a 12V / 10000mAh ternary lithium battery pack with three independent DC-DC voltage regulators: 5V / 3A supplies the signal source, main control, and 4G module; 3.3V / 2A supplies the IMU, RTK, and ADC; and 12V / 4A is dedicated to the power amplifier. It integrates DW01 overcharge, over-discharge, and reverse connection protection, and can be fully charged in 2 hours with Type-C fast charging. The average power consumption of the whole machine is 15W, and it can work continuously for ≥8 hours.

[0087] The technical principle of system signal source and module division of labor: The fault traveling wave signal detected by the system originates from the continuous electrical activity that still exists within the cable after the faulty branch is de-energized and isolated, rather than the traditional power frequency operating current. Its physical mechanism is as follows:

[0088] 0.4kV power cables are typical distributed parameter elements, with a uniformly distributed capacitance (capacitive energy storage effect) existing between the conductor and the shield / ground along their entire length. During normal cable operation, this distributed capacitance stores electric field energy corresponding to the operating voltage. After a fault trip, although the circuit breaker isolates the faulty cable from the transformer's active busbar, neither end of the cable undergoes discharge grounding to dissipate the load. Therefore:

[0089] (1) The residual charge stored in the cable distributed capacitance cannot be released instantaneously and will maintain a residual voltage in the cable for a period of time;

[0090] (2) The alternating electromagnetic field generated by other normal live cables laid in the same trench will continuously induce a voltage of a certain amplitude (induced voltage) on the core of the faulty cable through spatial electromagnetic coupling.

[0091] The residual voltage and induced voltage superimposed create a sufficiently high potential difference at the point of insulation failure. When this potential difference exceeds the breakdown threshold of the air gap or carbonized channel, an instantaneous breakdown discharge occurs, generating a high-frequency transient traveling wave pulse. After the discharge, the potential difference decreases, and the arc extinguishes; subsequently, residual charge and induced voltage accumulate again, reaching the threshold and causing another breakdown. This cycle repeats, forming an intermittent, continuous micro-breakdown arc discharge sequence that can last for tens of minutes. Each discharge generates a complete high-frequency transient traveling wave that propagates along the cable to both ends.

[0092] Based on the above signal characteristics, the transmitting and receiving modules of this system each perform their respective functions and work together, specifically:

[0093] (1) The transmitting module (including the Rogowski coupling clamp) injects a reference signal;

[0094] While the arcing traveling wave generated at the fault point can be utilized, its signal amplitude is extremely low (millivolt level) and it is an intermittent pulse. If the receiving module is relied upon to directly capture this traveling wave, two major problems will be encountered: ① The transmitting and receiving ends lack a unified clock reference, making it impossible to achieve high-precision time difference measurement; ② The amplitude of the 50Hz power frequency and harmonic interference at the site is much greater than that of the traveling wave signal, resulting in an extremely low signal-to-noise ratio, making it difficult to extract the effective wavefront from the noise.

[0095] To this end, the transmitting module uses Rogowski coupling clamps to non-contactly inject a fixed-frequency sinusoidal reference signal of 1kHz / 15mV into the cable. This reference signal is transmitted along the entire cable and synchronously acquired by the receiving module. Its core functions are: ① to serve as a shared clock synchronization reference for both the transmitting and receiving ends, eliminating the sampling time difference between independent clocks; ② to serve as a reference signal for frequency domain filtering, using signal processing algorithms to lock and suppress the 50Hz power frequency and its harmonic components in the frequency domain, thereby separating weak fault traveling waves from the strong power frequency background.

[0096] The role of the Rogowski coupling clamp in this stage is to inject a weak, stable, and known-frequency "ruler signal," rather than to generate a fault traveling wave. The fault traveling wave originates from the micro-breakdown arc generated at the fault point, and the two are completely independent in terms of hardware and signal path.

[0097] (2) The receiving module simultaneously captures two types of signals;

[0098] The receiving module's induction coil simultaneously captures two types of signals: ① a 1kHz continuous reference sine wave injected by the transmitting module; ② a high-frequency transient traveling wave pulse generated by the micro-breakdown arc at the fault point. These two signals are superimposed in the time domain to form a mixed waveform. The receiving module converts the mixed signal into a digital sequence after hardware filtering (10kHz low-pass) and fixed-gain amplification, and then sends it to the data processing module.

[0099] The receiving module does not distinguish the signal source, but converts the electromagnetic field signal in the cable into a digital waveform. Subsequent signal separation, decoupling, and wavefront identification are completed by the algorithms in the data processing module (S7 phase mode transformation, S8 wavelet decomposition + triple criterion).

[0100] The overall working logic of the system is as follows:

[0101] The fault point (self-generated micro-breakdown arc) generates a high-frequency transient traveling wave (the target signal being detected);

[0102] The transmitting module (Rogowski coupling clamp injection) generates a 1kHz reference signal (a tool signal used for synchronization and anti-interference).

[0103] The receiving module (moving along the path to acquire data) captures both simultaneously, forming a mixed waveform;

[0104] The data processing module separates, decouples, and filters the mixed waveforms, ultimately locking the traveling wave front and calculating the fault distance.

[0105] Example 2:

[0106] A cable fault detection and location method, implemented using a fault location system, requires only isolation of the faulty branch while maintaining power to non-faulty lines. The fault transient traveling wave is naturally generated by the fault point's puncture arc, and the equipment does not apply high-voltage pulse excitation to the cable. The method is divided into eleven major steps: pre-fault isolation preparation, equipment power-on self-test calibration, input of basic line parameters and wave velocity classification correction, continuous operation of reference signals, mobile acquisition of mixed signals along the line, hardware signal preprocessing, phase mode transformation decoupling, wavelet multi-scale decomposition and triple criterion wavefront identification, fault distance quantification, multi-source fusion geographic positioning, on-site verification, and data archiving. Each step details the operating conditions, judgment thresholds, calculation logic, and anomaly handling rules.

[0107] Combination Figure 1 The specific steps are described below.

[0108] S1: Preparations for electrical isolation of the fault branch after fault tripping

[0109] A single-phase grounding and phase-to-phase short-circuit fault occurred in the 0.4kV outgoing cable on the low-voltage side of the 10kV distribution transformer. After the distribution network switch protection tripped, the operation and maintenance personnel checked the fault alarm information and located the outgoing circuit breaker of the branch to which the fault belonged on site.

[0110] 2. Disconnect the circuit breaker of the faulty branch to achieve electrical isolation between the faulty cable and the low-voltage busbar of the transformer, while the transformer body and other non-faulty 0.4kV outgoing lines remain energized.

[0111] 3. Verify that all loads on the faulty line have been disconnected and there is no risk of backfeeding. Confirm that the residual insulation breakdown arc at the fault point can naturally generate transient traveling waves. Only after meeting the on-site data collection conditions can the fault location operation be carried out.

[0112] After the circuit breaker of the faulty branch is tripped, only electrical isolation is achieved on the transformer side; the faulty cable is not grounded for discharge and load removal. The 0.4kV long-distance cable possesses capacitive energy storage characteristics. Other energized outgoing lines laid in the same trench will continuously couple induced voltage to the faulty cable, superimposing residual charge on the cable. This can form a periodic micro-breakdown arc at the fault point, continuously generating high-frequency transient traveling waves to meet the needs of subsequent mobile inspection and data collection.

[0113] 4. Simultaneously record the ambient temperature, air humidity, cable laying age, and laying method (direct burial / conduit / cable trench) as the original parameters for subsequent wave velocity environmental correction.

[0114] Abnormal Handling: When a fault cannot be isolated individually or the entire busbar is shut down, this system can still be used for positioning under the condition of a complete power outage.

[0115] S2: System power-on, hardware self-test, and 10s power-on reference signal calibration

[0116] S2.1 Device power-on hardware self-test

[0117] 1. Secure the transmitting subsystem to any one of the A / B / C phases of the intact 0.4kV outgoing line of the 10kV transformer. During installation, ensure that the jaws of the Rogowski coupling clamp are fully closed and free of debris and cable compression. Power on the handheld receiving subsystem.

[0118] 2. The system automatically performs self-tests in sequence: lithium battery power supply voltage, DDS signal source output port, amplifier chip working status, RTK satellite search module, IMU inertial unit, 4G communication link, and ADC sampling channel; if any hardware fails, the device will issue an audible and visual alarm, stop subsequent calibration, and restart after troubleshooting the hardware failure.

[0119] S2.2 Power-on baseline calibration (fixed 10s timed calibration procedure)

[0120] 1. Transmitter subsystem locked output: 1kHz, 15mV RMS standard sinusoidal reference signal, intermittent mode 0.5s transmit, 2s sleep cycle output;

[0121] 2. The handheld receiving device is brought close to the transmitting point to collect the reference waveform, and the data processing unit calculates the effective value of the sampled voltage in real time. Signal-to-noise ratio (SNR);

[0122] 3. Dual thresholds for calibration pass / fail judgment:

[0123] ;

[0124] The calibration passes if both conditions are met simultaneously.

[0125] 4. Handling of calibration failures: If any parameter fails to meet the standard, the system will automatically pop up a window prompting you to re-tighten the Rogowski coupling clamp. After three consecutive calibration failures, an audible and visual alarm will continue, and the positioning process will be terminated.

[0126] 5. After calibration is successful: the system locks the current coupling compensation parameters, the reference signal continues to be output, and enters the normal fault location standby state.

[0127] S3: Cable foundation parameter input and four-level wave velocity correction calculation

[0128] Maintenance personnel enter line parameters one by one through the human-computer interface of the data processing subsystem. The system automatically calculates the final equivalent wave speed according to the four-level progressive correction model. .

[0129] S3.1 Level 1 Correction: The theoretical reference wave velocity is selected based on the cable insulation material.

[0130] ;

[0131] XLPE insulation =2.3, PVC insulation =3.2, Rubber insulation =4.0, after the selection is completed, the system will automatically load the corresponding theoretical wave velocity.

[0132] S3.2 Second-level correction: Inversion calibration of wave velocity along the actual total length of the line

[0133] 1. Enter the total actual length of the faulty cable. ;

[0134] 2. Temporarily suspend the end of the faulty cable in an open circuit, keep the reference signal of the transmitting unit unchanged, and use a handheld receiving device to collect the reflected wave from the open circuit at the transformer output end;

[0135] 3. The system captures the time interval between the reference signal and the open-circuit reflected wave. Inverse calculation of measured and calibrated wave velocity:

[0136] ;

[0137] 4. If there is no as-built data for the entire line and it is not possible to conduct an open-circuit test at the end, skip the second-level calibration and directly use the theoretical wave velocity of the first-level material to enter the third-level correction.

[0138] S3.3 Level 3 Correction: Correction for ambient temperature and humidity + cable aging coefficient

[0139] ;

[0140] For temperature correction factor, The actual temperature difference between the measured environment and the 20℃ reference temperature. Humidity correction factor This represents the relative humidity offset. For insulation aging correction factor, The values ​​are categorized into three levels based on the cable's service life: 1-3 years, 3-8 years, and over 8 years. The system has built-in calibration data tables for each coefficient, and the entered parameters are automatically substituted to complete the environmental correction.

[0141] S3.4 Level 4 Correction: Minor Adjustments to Historical Fault Samples

[0142] The system retrieves historical fault location data of the same model and laying conditions stored locally, and makes a slight offset correction to the wave velocity based on the historical measured error; if no historical data is available, the fourth-level correction coefficient is set to 0, and the third-level output wave velocity is not changed; after all four levels of calculation are completed, the system is fixed. It serves as the sole constant wave velocity for subsequent ranging.

[0143] S4: Reference synchronization signal for continuous operation of the transmitting subsystem throughout its entire lifecycle.

[0144] 1. The transmitting device is fixed to the 0.4kV low-voltage outgoing phase line of the 10kV distribution transformer throughout the entire process, and continuously outputs a 1kHz / 15mV sine reference from the end of calibration to the end of positioning.

[0145] 2. The reference signal has two main functions: ① to provide a unified clock reference for both the transmitting and receiving ends, eliminating time difference errors caused by asynchronous sampling clocks; ② to suppress the 50Hz power frequency component in the frequency domain, filtering out power frequency harmonic interference from high-power equipment in the distribution area.

[0146] 3. The reference signal does not participate in the fault excitation; the fault traveling wave relies on the continuous transient pulses generated by the breakdown arc of the residual insulation at the fault point.

[0147] S5: Mobile inspection along the cable laying path, synchronous acquisition of mixed waveforms and geographical location.

[0148] 1. Maintenance personnel carry the receiving subsystem and walk at a constant speed along the actual laying path of the faulty 0.4kV cable, controlling the walking speed to 1-3km / h;

[0149] 2. The receiving device synchronously acquires two types of signals: ① a 1kHz reference sine wave coupled from the transformer side; ② a continuous intermittent high-frequency transient fault traveling wave generated by the breakdown arc of the residual insulation at the fault point. The two are superimposed to form a mixed time-domain waveform.

[0150] The continuous intermittent high-frequency transient fault traveling wave is generated by the following mechanism: after the circuit breaker of the faulty branch is disconnected, the residual capacitive charge of the faulty cable and the coupled induced voltage of the live line laid in the same trench work together to form a periodic micro-breakdown arc at the insulation failure point of the fault. This arc can last for tens of minutes and generate multiple detectable transient pulses, which can meet the real-time acquisition needs of mobile inspection along the route.

[0151] 3. Simultaneously, the RTK+IMU positioning unit synchronously acquires the current geographical coordinates at millisecond levels, and the coordinates are bound one-to-one with the waveform sampling point timestamp;

[0152] 4. Differentiated processing of data acquisition scenarios: For open roads, RTK acquires ≥8 satellites, α=0.9, with satellite positioning as the primary method; for buildings and trees that obstruct the view, α=0.1, with IMU inertial navigation as the primary method for position estimation; for enclosed cable trenches without satellite support, the relative positioning is achieved by pre-importing the GIS cable turning point coordinates and walking distance, and the accumulated IMU drift error is manually cleared with one click upon reaching the cable well.

[0153] 5. Each set of waveforms and bound coordinate data is uploaded in real time via 4G to the local storage of the data processing subsystem.

[0154] S6: Receiver hardware-level signal preprocessing

[0155] The input for this step is a superimposed analog signal of a reference sine wave and a fault traveling wave from the induction coil. The output is a three-phase discrete digital sequence after filtering, fixed-gain amplification, and analog-to-digital conversion. , , As the sole input data source for the S7 phase-mode converter; the overall hardware link is fixed and cannot be adjusted, and the signal flow is: induction coil → RC passive pre-filter → second-order Butterworth low-pass filter → two-stage fixed amplification → ADC analog-to-digital sampling.

[0156] 1) RC passive pre-filter: , Initially filters out ultra-high frequency spatial spike interference;

[0157] 2) Second-order 10kHz Butterworth active low-pass filter, transfer function:

[0158] , ;

[0159] The filter retains the effective frequency band of 50Hz to 10kHz and filters out radio frequency and switching power supply noise above 10kHz.

[0160] 3) Two-stage fixed-gain amplification: The first stage uses an INA2128 instrumentation operational amplifier with A1=20x gain to suppress 50Hz common-mode interference; the second stage uses an operational amplifier with A2=100x gain, resulting in a total gain A=A 1× A2=2000, minimum effective input 0.01mV, after amplification meets the ADC input range;

[0161] 4) ADC parameters: sampling frequency 12-bit analog-to-digital conversion, outputting a three-phase time-domain discrete sampling sequence;

[0162] Process connection: The pre-processed three-phase digital sequence is stored in the on-chip buffer in real time and directly sent to the S7 decoupling operation module; if the signal-to-noise ratio (SNR) of the pre-processed waveform is less than 6dB, the noise is judged to be excessive, the data of this group is discarded, and the process is rolled back to S5 for re-acquisition.

[0163] S7: Karrenbauer three-phase phase mode conversion decoupling, separating line mode / zero mode fault components.

[0164] The input for this step is S6, which outputs a three-phase discrete digital waveform containing two types of superimposed signals:

[0165] First, there is the global synchronous common-mode electromagnetic interference generated by the spatial radiation of surrounding live cables, with the amplitude and phase of the three-phase induced waveforms being basically the same.

[0166] Second, the asymmetrical arc transient traveling wave generated by the residual induced voltage breakdown at the insulation failure point only exists in the single-phase / two-phase fault, while the three-phase pulse waveform has obvious amplitude and timing differences.

[0167] By separating the two types of signals through phase-mode transformation, the three-phase uniform common-mode interference is eliminated, and the effective fault differential-mode characteristics are extracted, which is not affected by the unsteady power frequency load current of the fault branch.

[0168] 1) A Karrenbauer decoupling matrix adapted to 0.4kV three-phase cables is used to convert the three-phase phase domain signals into zero-mode and two line-mode components;

[0169] 2) Signal differentiation logic: All common-mode interference in the entire space is mapped to the zero-mode component, and the asymmetrical differential-mode pulse of the fault arc is mapped to the two line-mode components; the system directly discards the zero-mode channel containing only background interference and selects only the line-mode components as the input signal for subsequent wavelet decomposition.

[0170] 3) Adaptive selection rules for fault components: Single-phase grounding faults extract the corresponding single-line mode component, and phase-to-phase short-circuit faults extract the two-phase coupled-line mode component; the system automatically determines the fault type and filters the effective modes by the instantaneous amplitude difference of the three-phase waveforms;

[0171] 4) Hardware noise reduction support: The common-mode rejection ratio of the front-end INA2128 instrumentation amplifier is >120dB, which greatly attenuates the global common-mode spatial interference and ensures that the line-mode fault characteristics have an effective signal-to-noise ratio; the phase-to-phase coupling interference suppression rate is ≥90%, which filters out the false pulses caused by mutual inductance.

[0172] The decoupled pure line-mode fault components are stored in a buffer and used as the input data source for the S8 Db45 wavelet five-level decomposition.

[0173] S8: Db45 wavelet 5-level multi-scale decomposition + triple quantitative criteria for screening effective fault wavefronts

[0174] This step takes as input the single-channel pure fault mode feature component S from the phase-mode decoupling output of S7. Building upon the processing results of S7 ("separating common-mode interference and extracting fault differential-mode signals"), it addresses the pain points of weak residual micro-arc signals, mixed noise, and the ease of misjudgment / missed detection due to indiscriminate determination of massive wavelet coefficients under power outage conditions. It completes the entire process of multi-scale wavelet decomposition, interference stripping, candidate wavefront screening, quantization threshold verification, and multi-scenario wavefront pairing. This step ultimately outputs the compliant incident wave and reflected wave times and the precise round-trip time difference Δt, which is sent to S9 for fault location calculation. If no valid paired wavefront is found, the data set is discarded, and the process jumps to S5 to re-acquire waveforms, forming a closed-loop iterative detection logic. This chapter comprehensively covers the wavelet parameter selection mechanism, low-frequency component selection criteria, massive data screening algorithms, threshold quantization sources, and multi-wavefront pairing rules, completely filling all the logical gaps raised by experts.

[0175] S8.1 Selection Basis and Technical Principles of Wavelet Basis and Decomposition Level

[0176] This solution consistently employs Db45 orthogonal wavelets and a 5-layer Mallat tower decomposition. The selection is not based on experience, but rather on the signal characteristics of residual micro-arcs under 0.4kV power outage isolation conditions, the distribution of on-site interference spectrum, hardware sampling parameters, and the computational constraints of embedded devices. Specific technical advantages and compatibility are as follows:

[0177] 1. Core basis for selecting the Db45 (45th order Daubechies) wavelet basis;

[0178] The fault signal in this solution is a microsecond-level narrow pulse discharge arc generated by the residual capacitive charge in the cable and the coupled induced voltage of adjacent lines after the faulty branch is de-energized and isolated. The pulse rise time is ≤1ms, the signal is weak and easily submerged by low-frequency periodic interference. Compared with Db low-order wavelets, Sym wavelets, Coif wavelets and Mexican hat wavelets, Db45 has unique adaptability:

[0179] (1) Ultra-high time domain positioning accuracy: Db45 has 89 orthogonal tight support intervals, which can capture steep wavefronts within 1ms without distortion. It will not produce a smoothing and passivation effect on narrow pulses, accurately lock the wavefront start time, ensure the time difference Δt extraction accuracy, and solve the problems of blurred wavefronts and large ranging errors in low-voltage weak fault pulses. The short support interval of low-order Db wavelets will smooth out steep wavefronts and cause time sequence extraction deviation.

[0180] (2) Strong low-frequency interference suppression capability: Db45 has a 45th order vanishing moment, which can deeply attenuate continuous periodic interference such as 50Hz power frequency, 1kHz synchronous reference sine wave, and 3rd / 5th / 7th low-frequency harmonics of frequency converter, so that the low-frequency interference energy is almost not leaked to the high-frequency detail coefficient, thus avoiding the fault judgment of periodic interference from the root, and is suitable for multiple cables in the same trench and strong electromagnetic interference conditions in the transformer area.

[0181] (3) Low pseudo-oscillation characteristics: Db45 has high high regularity. For the signal change point where the reference signal and the fault pulse are superimposed, there is no obvious Gibbs pseudo-oscillation after decomposition, and no false glitch pulse will be generated, thus eliminating the misjudgment of no fault.

[0182] (4) Embedded computing power adaptability: The Db series is a standard orthogonal wavelet, which supports Mallat fast discrete tower decomposition, without complex number operations and signal energy aliasing. It is adapted to real-time computing of handheld low-power embedded terminals. Compared with Coif wavelet and complex wavelet, the computing power overhead is greatly reduced, which meets the real-time positioning requirements of walking and measuring.

[0183] Quantitative Basis for Selecting the Number of Layers in a 2.5-Layer Mallat Tower Decomposition Structure

[0184] The ADC sampling frequency of this device is f=20kHz. The wavelet decomposition has the characteristic of frequency band bisection. The number of layers is uniquely determined by the boundary relationship between the effective frequency band of the fault signal and the interference frequency band: after 5 layers of decomposition, the low-frequency approximate component A corresponds to the frequency band 0~625Hz, and the five high-frequency detail components D~D correspond to the frequency band 625Hz~10kHz.

[0185] (1) Fault signal frequency band characteristics: In this scheme, more than 99% of the energy of the residual micro-arc narrow pulse is concentrated in the high frequency range of 625Hz to 10kHz, and there are no effective fault characteristics in the low frequency range of 0 to 625Hz.

[0186] (2) Interference frequency band characteristics: 50Hz power frequency harmonics, 1kHz reference sine wave, and hardware temperature drift noise are all concentrated in the low frequency range of 0 to 625Hz.

[0187] (3) Optimal logic for the number of layers: Decomposition of 3 layers or less cannot completely isolate the 1kHz reference interference, and the interference is easily mixed into the high-frequency layer, causing misjudgment; Decomposition of 7 layers or more will over-split the weak fault pulse energy, causing the actual fault energy to be lower than the judgment threshold, resulting in missed judgment, and multiplying the computation delay; 5 layers is the only optimal number of layers that can completely isolate high and low frequencies and take into account the weak signal recognition sensitivity and embedded real-time computing power.

[0188] This scheme uses a 5-level Mallat pyramid decomposition standard iterative formula to complete the hierarchical decomposition of the Db45 wavelet discrete signal. The specific formula is as follows:

[0189] ;

[0190] In the formula: This is the Db45 wavelet low-pass filter coefficient set. This is the Db45 wavelet high-pass filter coefficient set; For the first Low-frequency approximation coefficient sequence of the layer, For the first High-frequency detail coefficient sequence of layer; initial input Fault characteristic components output by S7 After completing 5 levels of iterative decomposition, the final result is... Five-layer low-frequency approximation components and Five layers of high-frequency detail components.

[0191] S8.2 Low-frequency component selection mechanism and rationale

[0192] The conventional judgment logic of this algorithm completely abandons the five-layer low-frequency approximation component A and only uses the five-layer high-frequency detail coefficients D to D to carry out fault judgment. This choice has strict physical and engineering rationality. At the same time, a low-frequency auxiliary verification fallback mechanism is set up to prevent the problem of missed judgment caused by pulse broadening of ultra-long cables and adapt to the weak differential mode fault signal characteristics of S7 output.

[0193] 1. The core basis for discarding low-frequency components

[0194] (1) Operating Spectrum Constraints: This scheme is a residual micro-arc discharge after power failure, with extremely narrow time-domain pulses. According to the Fourier time-domain-frequency domain duality, the signal energy is highly concentrated in the high-frequency range. The fault energy in the low-frequency band of 0-625Hz accounts for less than 1%, which has no effective identification value. The rich low-frequency characteristics of conventional high-current grounding faults under load are not applicable to this power failure isolation operation scenario.

[0195] (2) Interference suppression requirements: The low frequency range of 0-625Hz only includes three types of interference: power frequency harmonics, low frequency sidelobe of 1kHz reference signal, and hardware zero-point temperature drift, and has no fault transient characteristics; if included in the judgment, the large-amplitude low-frequency periodic interference will continuously trigger energy and amplitude thresholds, generate a large number of false wavefronts, and completely deteriorate the signal-to-noise ratio.

[0196] (3) Optimization requirements for computing power: Discarding low-frequency components can reduce the amount of computation by nearly half, reduce the algorithm latency of handheld terminals, and ensure the real-time solution capability of mobile inspection.

[0197] 2. Low-frequency auxiliary verification fallback mechanism under extreme operating conditions

[0198] To avoid the risk of missed detection due to pulse transmission broadening in ultra-long cables (>800m) and the infiltration of a small amount of fault energy into the low-frequency range, the algorithm buffers the low-frequency approximate component A throughout the process, does not participate in the regular judgment, and only initiates auxiliary verification under weak signal conditions:

[0199] (1) Triggering condition: The total high-frequency energy of the five layers meets the requirements. (High-frequency energy is close to the threshold but does not meet the standard);

[0200] (2) Verification rules: Check whether there is a transient abrupt spike in A that is completely synchronized with the timing of the high-frequency pulses in the fifth layer;

[0201] (3) Energy correction: If the timing is synchronized, the total energy is corrected by superimposing the low-frequency mutation energy. If the total energy reaches the standard after correction, it can be judged as a valid fault wavefront;

[0202] (4) On-site backup: In extreme weak signal scenarios, the fault circuit breaker can be opened and closed for a short time to generate a high-intensity narrow pulse, so that the fault energy is completely concentrated in the high-frequency range and the fault can be stably identified.

[0203] S8.3 Five-layer high-frequency coefficient three-level candidate wavefront screening process

[0204] The five layers of high-frequency detail coefficients (D to D) contain a massive amount of sampled data points. Directly traversing and judging all of them would result in computational redundancy, noise-induced false triggers, and an inability to meet the requirements for identifying weak fault signals. This solution, building upon the high-frequency signal characteristics described earlier, employs a three-level progressive screening mechanism: hierarchical extreme value coarse screening, cross-layer time-series clustering, and sliding window deduplication. This mechanism accurately extracts a limited number of candidate wavefronts from the massive data, with only candidate wavefronts participating in the subsequent triple criterion verification. This significantly reduces computational load and filters out random noise and residual interference.

[0205] 1. Layered Local Extreme Value Coarse Screening: The five layers of high-frequency coefficients are traversed independently layer by layer. A 5-point sliding window is used to extract local maxima, and a pre-amplitude threshold is set for each layer. Only extreme points where the amplitude exceeds the threshold are retained, and uniform time sequence and amplitude parameters are recorded. More than 90% of minor hardware noise and glitches are removed to complete the coarse data compression.

[0206] 2. Cross-layer timing synchronization clustering fine screening: Based on the broadband synchronization characteristics of real fault pulses (the same discharge wavefront will synchronously show peak values ​​in multiple high-frequency coefficients, while random noise only exists in isolation in a single layer), the following settings are configured: Temporal clustering deviation threshold: If the temporal difference between at least 3 extreme points is ≤0.2ms, they are merged into a group of candidate wavefronts. Isolated single-layer and double-layer extreme points without synchronous peaks are directly judged as noise and discarded. In extreme weak signal conditions, the synchronization matching can be flexibly reduced to 2 layers, and the energy threshold is synchronously increased to prevent misjudgment.

[0207] 3. Time-domain sliding window deduplication: Settings In the determination window, multiple adjacent candidate wavefronts within the same window are determined to be multiple peaks derived from the same arc pulse. Only the unique independent candidate wavefront with the largest total energy is retained, and duplicate candidate points are eliminated.

[0208] S8.4 Triple Synchronization Quantization Judgment Criteria and Threshold Value Basis

[0209] For the independent candidate wavefronts output by the three-level screening, a triple synchronous judgment is performed on energy, amplitude, and rise edge steepness. All three conditions must be met simultaneously to be recognized as a valid fault wavefront. All thresholds are quantitatively calibrated based on hardware noise basis, arc physical characteristics, and hundreds of sets of field working condition tests, and are not empirical values.

[0210] 1. Energy threshold: Calculation formula and judgment criteria for total high-frequency energy of five layers;

[0211] The total energy of the five high-frequency details is the sum of the high-frequency coefficient energies of each layer, calculated using the following formula:

[0212] ;

[0213] In the formula: For the first Layer One Db45 wavelet high-frequency detail coefficient For the first Total length of the high-frequency coefficient sequence of the layer.

[0214] Energy threshold determination formula:

[0215] ;

[0216] Value selection criteria: The maximum energy of total hardware and electromagnetic noise under no-load conditions is 0.06mV², and the minimum residual arc fault pulse energy that can be stably identified on-site is 0.15mV²; the intermediate value is taken and calibrated with a margin for engineering tolerance as 0.10mV²; the equipment can be adaptively fine-tuned and fixed according to the noise level of each unit before leaving the factory to ensure equipment consistency. Weak signal fallback judgment conditions: This interval automatically triggers a low-frequency auxiliary verification mechanism.

[0217] 2. Amplitude Threshold: Formula for determining the peak amplitude of high frequencies in a single frame:

[0218] ;

[0219] The values ​​are based on the following criteria: the maximum amplitude of the inherent noise of the equipment under no-load conditions is ≤0.25mV; the upper limit of the amplitude of the common-mode interference spikes in the entire field space is 0.4mV; the high-frequency peak values ​​of all measured effective residual arc discharge pulses are ≥0.55mV; and 0.5mV is the critical threshold for accurately distinguishing noise from real fault signals.

[0220] 3. Steepness threshold: Formula for determining the rising edge duration of the wavefront:

[0221] ;

[0222] Exemption conditions for flexibility of extra-long cables: Cables over 800m meet the requirements. And simultaneously satisfy When the peak amplitude is ≥0.8mV, it is determined to be a valid fault wavefront.

[0223] Value selection criteria: The instantaneous breakdown arc of residual capacitive charge is a steep and narrow pulse, and the measured effective fault pulse rise edge is concentrated in 0.1 to 0.8 ms; interference such as power frequency, 1 kHz reference sine wave, and inverter harmonics are slowly varying waveforms, and the rise edge is > 2.5 ms; the 1 ms threshold can completely distinguish fault pulses from periodic interference, while reserving a 0.2 ms margin to accommodate the slight pulse widening caused by long-distance transmission.

[0224] S8.5 Effective wave head identification, pairing, and multi-scenario processing logic

[0225] After obtaining the effective wavefront set through triple determination, it is sorted in ascending order of time sequence. Combining this with the physical characteristics of cable traveling wave propagation, incident and reflected waves are distinguished. Matching of fault-related wave pairs is completed for three scenarios: single wavefront, dual wavefront, and multi-wavefront, completely resolving the technical problems of multi-wavefront confusion, mismatched pairing, and lack of effective time difference. The core physical principle is that the fault discharge incident wave arrives at the first-end acquisition device first, with an earlier time sequence and higher energy; the reflected wave, after being folded back at the impedance discontinuity point at the far end of the cable, has a later time sequence, and its energy after line attenuation is 0.3 to 0.7 times that of the incident wave; the system pre-stores the total cable length... With wave speed The maximum legal round-trip time difference is calculated using the following formula:

[0226] ;

[0227] This value serves as the core time difference boundary constraint for wavefront pairing.

[0228] 1. Scenario 1: Number of effective wavefronts n=1 (single wavefront only)

[0229] If only the incident wave of the fault is captured and there is no corresponding reflected wave, the round-trip time difference cannot be calculated and the ranging operation is not performed; the system automatically extends the acquisition window and continues to acquire subsequent waveforms while waiting for the reflected pulse; when there are only single wavefronts for 3 consecutive frames, the system prompts the maintenance personnel to move the inspection point to the far end of the cable to increase the coupling strength of the reflected wave.

[0230] 2. Scenario 2: Number of effective wavefronts n=2 (paired wavefronts)

[0231] Assuming a time series t < t, four constraints must be satisfied simultaneously for the incident-reflected wave pair to be identified as belonging to the same fault:

[0232] (1) Energy decay constraint: ;

[0233] (2) Time difference boundary constraints: ;

[0234] (3) Waveform shape constraint: The difference in the rising edge duration between the two sets of wavefronts ≤ ;

[0235] (4) Timing logic constraint: the incident wave is always earlier than the reflected wave.

[0236] Pairing is successful: (Incident wave moment) (Reflection wave time), calculate the pulse round-trip time using the following formula:

[0237] ;

[0238] If any constraint is not met, it is determined to be two independent and unrelated pulses, and the process is switched to the multi-wavehead processing logic.

[0239] 3. Scenario 3: Number of effective wavefronts n≥3 (multiple wavefronts redundant screening)

[0240] (1) Pair all permutations together and apply the four-term constraint of the double wavefront to screen all legal incident-reflected wave pairs;

[0241] (2) Eliminate illegal pairing groups that exceed the time difference boundary or have abnormal energy decay;

[0242] (3) Select the main fault: Select the legal pair with the largest total energy as the core fault wave pair for ranging calculation; mark the remaining legal pairs as multi-point fault locations and record and output them synchronously.

[0243] (4) Excessive redundancy processing: When there are more than 3 valid pairs, the inspection movement speed is automatically reduced, the acquisition window is extended, the waveforms are cross-verified in the second acquisition, and false pairs are filtered out.

[0244] 4. Two-way locking rule for incident / reflected waves;

[0245] Within the same pairing group, pulses with earlier timing and higher energy are fixed as incident waves, while pulses with later timing and lower energy are fixed as reflected waves. Timing and energy are interlocked to prevent the waveheads from being reversed or confused.

[0246] S8.6 Process connection logic;

[0247] After successfully matching a valid incident and reflected wave pair, the round-trip time difference Δt and the corresponding wavefront timestamp T are locked and buffered and sent to S9 to complete the fault distance quantification calculation; if there is no valid paired wave pair or no valid time difference, the buffered data of this group is cleared and the process jumps to S5 to reacquire the waveform, realizing the full closed-loop link of "acquisition-decoupling-decomposition-identification-iterative retesting".

[0248] S9: Traveling wave time difference extraction, system delay compensation, and accurate quantitative solution for fault distance;

[0249] Input for this step: S8 output round-trip time difference S3 curing four-level corrected wave velocity Monthly calibration and solidification system inherent delay ; Distance of the output fault point from the transformer outgoing line And bind wavehead timestamp Input into S10 coordinate fusion.

[0250] 1) Parameter description:

[0251] Time difference between incident and reflected waves of the fault traveling wave The output of the wavefront pairing logic of S8 specifically includes: three-level candidate wavefront screening after five-level decomposition of Db45 wavelets, triple criterion verification, and multi-scenario pairing processing based on four judgment conditions: energy attenuation constraint, time difference boundary constraint, waveform morphology constraint, and timing logic constraint. Finally, it outputs a compliant incident-reflected wave pair time difference. .

[0252] The fixed hardware delay of the entire device, including the coil, amplifier circuit, and ADC transmission, is stored in Flash after monthly calibration and cannot be changed.

[0253] 2) Compensated distance measurement calculation formula:

[0254] ;

[0255] Dividing by 2 in the formula represents the round trip distance of the traveling wave from the transformer end to the fault point and then back.

[0256] The parameters input in this step have undergone validity verification through the multi-scenario pairing logic of S8.5: when the number of valid wavefronts is 2, the pairing can be confirmed only if four constraints are met simultaneously: energy attenuation of 0.3-0.7 times, time difference ≤ 2L / V, rise time difference ≤ 0.3ms, and incident wave timing earlier than reflected wave. When the number of valid wavefronts is ≥ 3, the legal pairing group with the largest total energy is selected as the main fault wave pair for ranging calculation. This ensures that only legal time differences that have undergone complete physical logic verification are entered into the ranging formula, guaranteeing ranging accuracy from the source.

[0257] Process connection: After completing the distance calculation, mark the timestamp with the wavefront. To retrieve keywords, cached location data at the same time is retrieved and passed to the S10 multi-source coordinate fusion module.

[0258] S10: RTK+IMU Multi-Source Fusion Geographic Positioning

[0259] This step requires: S9 synchronization timestamp Corresponding RTK original coordinates IMU coordinate calculation Output the actual geographic coordinates of the fault. , with fault length The combination forms the final positioning result.

[0260] 1) Coordinate weighted fusion formula:

[0261] ;

[0262] The weighting coefficients are dynamically adjusted and automatically changed based on the number of satellites acquired.

[0263] In open areas, satellite search exceeds the first threshold (≥8 satellites): =0.9, RTK-dominated positioning;

[0264] Building / tree shade obstruction, search for 3-7 satellites (between the second and third thresholds): =0.1, IMU inertial navigation is the primary method;

[0265] No satellite in the enclosed cable trench (satellite search < third threshold, third threshold = 3): =0, relying on GIS inflection points + walking mileage to correct inertial navigation coordinates, and clear the accumulated IMU drift at the cable well with one click.

[0266] 2) Timing matching rules: Only AND operations are used. Millisecond-aligned positional data and time-series misaligned data are directly discarded;

[0267] Process integration: After the coordinate fusion is completed, the system synchronously displays the fault distance, actual coordinates and fault waveform curve, and enters the S11 maintenance archiving process.

[0268] S11: On-site fault diagnosis and verification, full data archiving and storage

[0269] 1. Maintenance personnel use distance data and actual coordinates to excavate at the target location to locate the insulation breakdown fault point;

[0270] 2. Measured Fault Locations and Calculations For error comparison, the measured data can be stored in the device as a historical sample for subsequent fine-tuning of the fourth-level wave velocity under the same working conditions.

[0271] 3. After the fault is cleared and the line is powered back on, the parameters of the entire process are automatically stored in the non-volatile flash memory of the data processing subsystem, completing the closed-loop archiving of a single fault location.

[0272] The original analog signal undergoes unified noise reduction, amplification, and analog-to-digital conversion via hardware preprocessing in S6, outputting a regular three-phase digital waveform. S7 decouples phase-mode signals to eliminate inter-phase coupling noise and selects the best single-path fault characteristic components. S8 extracts high-frequency fault features through Db45 five-layer wavelet decomposition, filters out interference pulses using energy, amplitude, and rising edge criteria, retaining only the true fault traveling wave and extracting the round-trip time difference. S9 combines the previously corrected wave velocity and the inherent delay of the hardware with the ranging formula to quantify the fault cable length. S10 uses the fault wavefront timestamp as an index to match synchronous point data, and obtains the actual fault coordinates through dynamic weighted fusion, realizing continuous operation across the entire link from on-site analog signal acquisition → electrical feature analysis → fault length calculation → geographical location.

[0273] This solution constructs a three-tiered anti-interference architecture of "decoupling-layering-screening". The first layer achieves decoupling through Karrenbauer phase-mode transformation. For three-phase parallel cables, inter-phase electromagnetic coupling induces spurious pulses similar to the real traveling wave waveform on each phase line, directly interfering with wavefront identification. Phase-mode transformation recombines the three-phase sampling sequence into zero-mode and line-mode components. The coupling interference has different distribution characteristics in each component, while the real fault signal has concentrated energy in specific components. By adaptively selecting the corresponding component according to the fault type (single-phase grounding or inter-phase short circuit), the real fault characteristic component is separated from the coupling interference.

[0274] In the original signal, interference signals such as power frequency components and environmental noise change slowly and last for a long time, and their energy is mainly distributed in the low-frequency approximation coefficients of wavelet decomposition. In contrast, the fault traveling wave is a transient signal with a short duration and drastic changes, and its energy is concentrated in the high-frequency detail coefficients. Therefore, the second stage achieves layering through Db45 wavelet multi-scale decomposition. Wavelet decomposition uses filter banks of different scales to peel the signal layer by layer according to frequency. By discarding the low-frequency approximation coefficients and retaining only the high-frequency detail coefficients, effective separation of the fault traveling wave and environmental interference in the frequency domain is achieved.

[0275] Even after phase-mode transformation and wavelet decomposition, the remaining high-frequency detail coefficients may still contain a small number of interference pulses. Therefore, the third stage uses a triple quantitative criterion for filtering, setting three independent criteria: energy criterion (total high-frequency energy exceeds a preset threshold), amplitude criterion (wavefront amplitude ≥ 0.5mV), and steepness criterion (rise time ≤ 1ms). Real fault traveling waves are generated by arc breakdown and have the physical characteristics of concentrated energy, relatively large amplitude, and extremely steep rise; while residual interference pulses usually have dispersed energy, small amplitude, or slow changes, making it difficult to simultaneously satisfy all three criteria. Through the "AND" logic filtering of the triple criteria, real traveling waves can be accurately distinguished from various types of interference.

[0276] The aforementioned three-layer anti-interference architecture achieves highly reliable identification of fault traveling wave fronts in environments with strong electromagnetic interference. The first layer, decoupling, eliminates the strongest source of interference—phase-to-phase coupling—transforming the signal from "three-wire mixed" to "single-path clear." The second layer, layering, separates the fault traveling wave from environmental noise in the frequency domain, further improving the signal-to-noise ratio. The third layer, filtering based on the inherent physical characteristics of the fault traveling wave, achieves precise locking of the true wavefront. The synergistic effect of these three layers enables reliable extraction of the traveling wave front from a strong noise background, even in harsh electromagnetic environments with multiple cables in the same trench and densely packed frequency converters, laying the foundation for subsequent accurate ranging.

[0277] A 0.4kV XLPE cable with a total length of 410m was output from the low-voltage side of a 10kV distribution transformer. When a phase-to-phase short circuit fault occurred, the protection tripped. The maintenance personnel disconnected the faulty branch switch to isolate the fault, while the remaining output lines remained energized.

[0278] 1) The system's power-on hardware self-test is complete, the power-on benchmark calibration is qualified, and XLPE insulation has been selected. =2.3, the total length of the cable is 410m, the ambient temperature is +5℃, and it has been laid for 3 years with slight aging. The system automatically completes the fourth level of wave velocity correction.

[0279] 2) The transmitting coupling clamp is fixed to phase A of the 0.4kV output line of the transformer, and the equipment continuously outputs a 1kHz / 15mV synchronization reference;

[0280] 3) Maintenance personnel use handheld receivers to move along the cable path to collect the mixed signal of continuous intermittent transient traveling waves generated by the fault arc;

[0281] 4) After three-phase decoupling, Db45 wavelet decomposition, and triple threshold screening, the effective fault wavefront was selected and substituted into the ranging formula to calculate the fault distance as 203.2m;

[0282] 5) Based on the RTK+IMU fusion, the geographical coordinates of the fault were obtained. The actual measured fault location during on-site excavation was 202.9m with a positioning error of 0.3m, thus completing the accurate location of the fault.

[0283] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting and locating cable faults, characterized in that, Includes the following steps: After a fault trip occurs in the low-voltage side outgoing cable of the transformer, disconnect the circuit breaker of the faulty branch and keep the transformer body and other non-faulty outgoing circuits energized and operating normally. The system moves along the cable laying path for inspection, collecting mixed time-domain signals from the cable and geographical location information of the current location; the mixed time-domain signals carry transient traveling wave signals generated by fault point puncture arc; The collected three-phase time-domain sampling sequence was subjected to Karrenbauer phase-mode transformation, and zero-mode and line-mode components were obtained after decoupling. A fault characteristic component was selected according to the fault type to eliminate inter-phase electromagnetic coupling interference between the three-phase cables. The fault feature components are decomposed by Db45 wavelet multi-scale decomposition to extract high-frequency detail coefficients. The high-frequency detail coefficients are then screened by a triple quantitative criterion of energy threshold, amplitude threshold and steepness threshold. Effective fault traveling wave fronts are identified from the background of strong electromagnetic interference, and the time difference between incident wave and reflected wave is extracted. Based on the time difference, the traveling wave velocity after multi-dimensional correction, and the inherent hardware delay of the system, the distance from the fault point to the transformer output terminal is calculated. By performing multi-source fusion with geographic location information, the actual geographic coordinates of the fault point are generated.

2. The cable fault detection and location method as described in claim 1, characterized in that, During the acquisition of signals from the cable, a fixed-frequency reference synchronization signal injected into the cable by the transmitting subsystem is also acquired. The reference synchronization signal is used to provide a unified clock reference for both the transmitting and receiving ends and to suppress power frequency harmonic interference in the frequency domain.

3. The cable fault detection and location method as described in claim 1, characterized in that, The fault type selection rules are as follows: During the phase-mode transformation decoupling period, the three-phase sampling sequence is transformed by Karrenbauer to obtain zero-mode components and line-mode components; among them, the global spatial induced common-mode interference is mapped to the zero-mode component, and the fault arc differential-mode pulse is mapped to the line-mode component; the zero-mode component containing only spatial interference is discarded, and the line-mode component is selected as the fault characteristic component; the corresponding single-phase line-mode component is extracted for single-phase grounding faults, and the two-phase coupled line-mode component is extracted for phase-to-phase short-circuit faults.

4. The cable fault detection and location method as described in claim 1, characterized in that, The triple quantitative criteria include: The energy criterion is as follows: the total energy of the high-frequency detail coefficients exceeds a preset energy threshold. The amplitude criterion is as follows: the wavefront amplitude is not less than the set value; The steepness criterion is as follows: the duration of the rising edge of the wave head does not exceed a set value. When all three criteria are met simultaneously, the wave is determined to be a valid faulty traveling wave head.

5. The cable fault detection and location method as described in claim 1, characterized in that, The traveling wave velocity was obtained through a four-level progressive correction model: The first level involves selecting the theoretical reference wave velocity based on the type of cable insulation material. The second stage involves conducting an open-circuit reflection test based on the actual total length of the cable to invert and calibrate the wave velocity. The third level involves environmental correction by overlaying on-site ambient temperature, humidity, and cable insulation aging coefficient. The fourth level involves making a slight offset correction to the wave velocity based on the measured error of historical fault location data.

6. The cable fault detection and location method as described in claim 1, characterized in that, In the ranging calculation step, the formula for calculating the fault distance is: ； in, The distance to the fault. This is the corrected traveling wave velocity. The time difference between the incident wave and the reflected wave. This is due to the inherent hardware delay of the system.

7. The cable fault detection and location method as described in claim 1, characterized in that, In the coordinate fusion step, a fusion positioning method combining RTK differential positioning and IMU inertial navigation is adopted, and the weighting coefficients are dynamically adjusted according to the number of satellites acquired. When the number of satellites searched is greater than or equal to the first threshold, RTK coordinates are used as the primary coordinates. When the number of satellites acquired is between the second and third thresholds, IMU inertial navigation is the primary method. When the number of satellites searched is less than the third threshold or there is no satellite signal, relative positioning is performed by relying on the pre-stored coordinates of the GIS pipeline turning points and the walking distance, and the accumulated drift error of the IMU is manually eliminated at the cable well.

8. The cable fault detection and location method as described in claim 1, characterized in that, In the wavelet decomposition and wavefront identification steps, the Db45 wavelet is used for multi-level Mallat tower decomposition. All low-frequency approximation coefficients are discarded, and only the high-frequency detail coefficients from the first to the kth layer are retained for fault traveling wave determination.

9. The cable fault detection and location method as described in claim 1, characterized in that, Before the signal acquisition step, there is also a power-on calibration step: the transmitting subsystem injects a fixed frequency reference signal into the cable, and the handheld receiving subsystem acquires the reference waveform at the transmitting point. When the effective value of the received sampling voltage and the signal-to-noise ratio (SNR) exceed the corresponding set values, the calibration is qualified and the coupling compensation parameters are locked.

10. A cable fault detection and location system, used to implement the cable fault detection and location method as described in any one of claims 1-9, characterized in that, include: The transmitting subsystem is installed on the low-voltage side of the transformer's outgoing phase line and is used to inject a fixed-frequency reference synchronization signal into the cable. This reference signal is used to provide a unified clock reference for both the transmitting and receiving ends and to suppress power frequency harmonic interference in the frequency domain. It is not used as a fault excitation signal. The transmitting subsystem includes a nanocrystalline open-close Rogowski coupling clamp, an LC passive impedance matching network, a DDS digital signal source, a programmable power amplifier circuit, and a time-division main control unit. The handheld receiving subsystem is used for mobile inspection along the cable laying path, simultaneously acquiring mixed time-domain signals and geographical location information; The handheld receiver subsystem includes a high-sensitivity sensing coil, a two-stage programmable amplifier circuit, a low-pass filter circuit, an RTK differential positioning module, an IMU inertial navigation unit, and a wireless transmission module; The data processing subsystem is used to receive and process data uploaded by the handheld receiving subsystem, and to perform phase mode transformation decoupling, wavelet decomposition and wavefront recognition, ranging calculation and coordinate fusion calculation. The electronic power supply system is connected to the transmitting subsystem, the handheld receiving subsystem, and the data processing subsystem to provide operating power.

Images