Power distribution network fault protection and positioning method and system based on fast switch

Abstract

According to the power distribution network fault protection and positioning method and system based on the fast switch, an edge calculation node adopts noise self-adaptive improved Morlet wavelet transform, and a current change rate and a voltage break variable are extracted from a transient signal; constructing a three-dimensional decision model taking the current change rate, the voltage break variable and the line characteristic impedance as variables; based on a three-dimensional decision-making model, when it is judged that an internal fault occurs by using the extracted current change rate and voltage break variable, the distributed detection unit sends the action state quantity of the equipment where the distributed detection unit is located to an edge computing node; after the edge computing node utilizes the action state matrix of the equipment, a state reasoning method based on a fuzzy Petri network is adopted to determine a fault section; verifying the fault section according to the matching degree of the fault current weighting and the characteristic impedance of the equipment in the fault section; after the fault section is verified, fault isolation is executed; and after fault isolation, fault positioning is optimized, and it is ensured that the cooperative breaking time sequence of the multi-stage switch is accurately matched with a fault diffusion path.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network fault protection, and in particular, relates to a distribution network fault protection and positioning method and system based on fast switches. Background Art

[0002] With the large-scale integration of new energy sources and the increasing complexity of urban distribution networks, traditional fault protection technologies face severe challenges. As the terminal link of the power system, the probability of failure of the distribution network accounts for a large proportion of the total power system failures.

[0003] Existing technologies for fault protection and location in new distribution networks with high penetration of distributed power generation (DGs) suffer from the following common issues: A conflict between speed and selectivity. Traditional overcurrent protection relies on detecting the steady-state value of the fault current, requiring a 20-40ms wait for the current to stabilize before triggering. This makes it unable to adapt to the rapid isolation requirements of microsecond-level fault propagation in scenarios with high penetration of renewable energy. Statistics show that for every 10ms delay in fault removal, arc energy increases by 35%, leading to a 23% increase in equipment damage, making it difficult to meet the rapid isolation requirements of high-penetration DGs. Furthermore, there is a conflict between sensitivity and reliability: the accuracy of single-phase ground fault detection in low-current grounding systems is less than 75%, especially when the transition resistance is greater than 1kΩ, where the traditional zero-sequence current direction method has a misjudgment rate of up to 40%. Harmonic injection from new energy inverters further degrades the fault signature signal-to-noise ratio by 6-8dB, resulting in frequent protection failures or malfunctions. There's a conflict between positioning accuracy and cost control: existing traveling wave positioning devices cost over 50,000 yuan per point, making dense network-wide deployment difficult. There's also a conflict between positioning accuracy and cost: existing traveling wave positioning devices rely on dedicated measurement points, costing over 50,000 yuan per point, while full network coverage would require tens of millions of yuan in investment. Simple impedance positioning methods can yield positioning errors of 300-500 meters, and in mixed cable and overhead line networks, errors can exceed 1km, increasing troubleshooting time by 2-3 times. Summary of the Invention

[0004] To address the deficiencies in the prior art, the present invention provides a distribution network fault protection and positioning method and system based on fast switches, enabling adaptive complex fault detection and high-precision fault-tolerant positioning for rapid fault isolation and arc suppression. To address the issues of insufficient speed, low positioning accuracy, and difficulty in detecting high-resistance faults in traditional solutions, a multi-level fast switch network is constructed, with solid-state circuit breakers deployed every 300±50 meters along the feeder. High-frequency sampling modules and transient traveling wave detection units are integrated to achieve a 1MHz sampling rate and 200ns-level fault feature capture. An improved Morlet wavelet algorithm is used to adaptively extract fault frequency bands, combining a three-dimensional decision model with fuzzy Petri net reasoning to form a millisecond-level collaborative triggering mechanism. A triple-redundant positioning strategy is designed, integrating dynamic compensation for traveling wave time differences, impedance spectrum matching, and switch action sequence verification to improve positioning reliability in complex branch networks. 5G / LoRa dual-mode communication and TSN time-sensitive networks are used to ensure μs-level synchronization and reliable data transmission.

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

[0006] The present invention proposes a distribution network fault protection and location method based on fast switches, comprising:

[0007] Utilize the distributed detection units integrated on the electrical equipment in the distribution network to obtain transient signals of distribution network faults;

[0008] Each group of edge computing nodes uses a noise-adaptive improved Morlet wavelet transform to extract the current change rate and voltage mutation amount from the transient signal. A three-dimensional decision model is constructed with the current change rate, voltage mutation amount, and line characteristic impedance as variables. Based on the three-dimensional decision model, the edge computing node uses the extracted current change rate and voltage mutation amount to determine whether an in-zone fault has occurred. Each set of distributed detection units is configured with a group of edge computing nodes.

[0009] When a fault is detected within a zone, the distributed detection unit sends the device's action state information to the edge computing node. The edge computing node uses the calculated device action state matrix and a state reasoning method based on fuzzy Petri nets to determine the fault zone.

[0010] The determined fault section is verified based on the fault current weighting and characteristic impedance matching of the equipment within the fault section; after the fault section is verified, fault isolation is performed; after fault isolation, fault location is optimized.

[0011] Solid-state circuit breakers are arranged at equal intervals on the distribution network feeders, and magnetically controlled disconnectors and bidirectional current detection intelligent switches are set at the feeder branch nodes; a distributed detection unit is integrated in each solid-state circuit breaker, magnetically controlled disconnector and bidirectional current detection intelligent switch.

[0012] A wavelet transform processor is deployed in the edge computing node. The wavelet transform processor uses a noise-adaptive improved Morlet wavelet transform to process transient signals.

[0013] A transient energy spectrum analysis module is deployed in the edge computing node. The module uses the processed transient signal to calculate the energy entropy value in the [1Hz, 100kHz] frequency band. If the sudden increase in the energy entropy value in any frequency band is greater than three times the energy entropy threshold, the current and voltage in that frequency band are extracted.

[0014] The current change rate is calculated by differential calculation, and the voltage mutation is measured by the difference between the effective values ​​of the half cycle before and after the fault;

[0015] The weighted sum of current change rate, voltage mutation and line characteristic impedance is used as a three-dimensional decision model.

[0016] In the noise-adaptive improved Morlet wavelet algorithm, the scale parameter is dynamically adjusted according to the real-time noise level of the sampled data. The basis function of the noise-adaptive improved Morlet wavelet transform is:

[0017]

[0018] Where Ψ(t) is the basis function of the noise-adaptive improved Morlet wavelet transform, t is the sampling time, and σ is the scale parameter adaptively adjusted in the range of [0.1, 2.0] according to the real-time noise level.

[0019] A Morlet wavelet transform with σ=0.5 is performed on the data window [-10ms, 10ms], and the energy entropy value in the characteristic frequency band of [35Hz, 80kHz] is extracted as the energy entropy threshold.

[0020] In the period before the current detection moment, the sliding window is used to obtain the average voltage and current, and the average voltage and current are substituted into the three-dimensional decision-making model, with 85% of the calculated values ​​as the dynamic fault threshold. The current change rate and voltage mutation amount extracted by the edge computing node are substituted into the three-dimensional decision-making model. If the calculated real-time value is greater than the dynamic fault threshold and the duration is greater than 1 / 4 cycle, it is determined that an internal fault has occurred; otherwise, it is determined that no internal fault has occurred.

[0021] The edge computing node uses the product of the device's action state quantity obtained by the distributed detection unit and the collection time vector of the distributed detection unit as the device's action state matrix; among them, the device's action state quantity obtained by the j-th set of distributed detection units is S j ∈{-1,0,1}, a value of 1 indicates that the device is in forward motion, a value of -1 indicates that the device is in reverse motion, and a value of 0 indicates that the device is not in motion.

[0022] The weight coefficient of the device is determined according to its location in the distribution network, and the distributed detection unit in the fault section obtains the fault current of the device;

[0023] When the weighted sum of the fault currents of multiple devices within the fault section is greater than 1.25 times the current threshold, the two solid-state circuit breakers at the boundary of the fault section are driven to perform a tripping operation, and a blocking instruction is sent to the non-fault area. At the same time, multiple devices corresponding to the weighted sum of the fault currents greater than 1.25 times the current threshold are driven to perform a coordinated tripping operation;

[0024] Inject a 150Hz third harmonic at the device with the largest weight coefficient within the fault section and test the matching degree of the characteristic impedance within a 2ms time window. If the matching degree is greater than 90%, the fault section is verified and the blocking instruction of the non-fault section is released.

[0025] After the fault section is verified, the solid-state circuit breaker is driven to open in the transient priority mode. When the transient priority mode fails, the magnetic control disconnector is driven to open in the steady-state backup mode to perform fault isolation.

[0026] Set the weight coefficient of the trunk equipment to 1.2, the weight coefficient of the branch equipment to 0.8, the weight coefficient of the segment equipment to 1.0, and the weight coefficient of the communication equipment to 0.9.

[0027] Driving the solid-state circuit breaker to open in transient priority mode includes:

[0028] When the voltage mutation rate is greater than 5kV / μs, the wave front is marked and the wave front arrival time is calculated using wave front velocity interpolation. When the wave front amplitude is greater than 0.15pu and the polarity is correct, the solid-state circuit breaker is driven to open.

[0029] The matching is as follows:

[0030]

[0031] In the formula, ζ is the matching degree, Z measured is the impedance modulus measured at 150Hz, Z0 is the line characteristic impedance, and Z min is the lower impedance threshold.

[0032] Driving the magnetically controlled disconnector to open in steady-state backup mode includes:

[0033] The dynamic impedance sweep method is used to obtain the system impedance, calculate the inverse time parameter, and determine the real-time current value based on the inverse time characteristic curve. When the real-time current value is greater than 1.2 times the rated current, the magnetically controlled isolating switch is triggered.

[0034] After the fault is isolated, the traveling wave time difference of arrival positioning method, impedance amplitude and angle characteristic analysis method and equipment action sequence matching method are used simultaneously to optimize the fault location coordinates;

[0035] Among them, the traveling wave time difference of arrival location method uses the corrected traveling wave propagation velocity, the coordinates of each monitoring point, and the arrival time of each monitoring point to determine the candidate fault location. In the impedance amplitude characteristic analysis method, the cosine similarity of the amplitude angle measurement value and the reference value at each frequency point in the [1Hz, 100kHz] frequency band is calculated, and the location with a confidence level greater than 0.95 is selected as the candidate fault location. In the equipment action sequence matching method, the Hamming distance between the equipment action sequence and the expected fault mode is calculated.

[0036] When the minimum Hamming distance is no more than 2, the candidate fault location determined by the traveling wave time difference of arrival location method is used as the optimized fault location. If the cosine similarity is greater than 0.9, the optimized fault location is obtained by dividing the sum of 0.5 times the candidate fault location determined by the traveling wave time difference of arrival location method and 0.3 times the candidate fault location determined by the impedance amplitude and angle characteristic analysis method by 0.8. If the minimum Hamming distance is greater than 2 and / or the cosine similarity is no more than 0.9, manual fault location is required.

[0037] The present invention also proposes a distribution network fault protection and positioning system based on fast switches, comprising: a solid-state circuit breaker, a magnetically controlled isolating switch, a bidirectional current detection intelligent switch, a distributed detection unit, and an edge computing node;

[0038] Solid-state circuit breakers are arranged at equal intervals on the distribution network feeders, and magnetically controlled disconnectors and bidirectional current detection intelligent switches are installed at the feeder branch nodes. A distributed detection unit is integrated into each solid-state circuit breaker, magnetically controlled disconnector, and bidirectional current detection intelligent switch.

[0039] Distributed detection unit, used to obtain transient signals of distribution network faults; when it is determined that a fault has occurred in the area, the action status of the device where it is located is sent to the edge computing node;

[0040] Each group of edge computing nodes is used to extract the current change rate and voltage mutation amount from the transient signal using a noise-adaptive improved Morlet wavelet transform; a three-dimensional decision model is constructed with the current change rate, voltage mutation amount, and line characteristic impedance as variables; based on the three-dimensional decision model, the edge computing nodes use the extracted current change rate and voltage mutation amount to determine whether an in-zone fault has occurred, wherein a group of edge computing nodes is configured for each set of distributed detection units; using the calculated action state matrix of the equipment, a state reasoning method based on fuzzy Petri nets is used to determine the fault section; the determined fault section is verified based on the fault current weighting and characteristic impedance matching of the equipment in the fault section; after the fault section is verified, fault isolation is performed; and after fault isolation, the fault location is optimized.

[0041] The present invention also provides a terminal, comprising a processor and a storage medium; the storage medium is used to store instructions; and the processor is used to operate according to the instructions to execute steps of the method.

[0042] The present invention also relates to a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method when the program is executed by a processor.

[0043] The beneficial effects of the present invention, compared with the prior art, include at least the following: the method proposed in the present invention constructs a multi-stage switch network by densely deploying fast solid-state circuit breakers along the feeder, and combines transient traveling wave detection with high-frequency sampling technology to capture the microsecond-level current mutation characteristics of the initial fault in real time. When an abnormal polarity of the traveling wave head is detected, the coordinated triggering logic immediately drives the adjacent circuit breakers to operate synchronously, forming an electrical isolation barrier for the fault section. The edge computing node achieves μs-level synchronization of switching actions through a time-sensitive network, ensuring that the coordinated disconnection timing of the multi-stage switches accurately matches the fault diffusion path, significantly shortening the fault removal time, effectively suppressing arc development, and avoiding the expansion of the fault range and equipment damage.

[0044] A triple redundancy check mechanism employs traveling wave time difference of arrival, impedance spectrum analysis, and switching action sequence matching. The traveling wave detection module dynamically compensates for the impact of line parameter fluctuations on wave velocity. The impedance spectrum matching algorithm compares pre-stored line characteristics across a wide frequency band. The switching action logic verifies topological action consistency using Hamming distance. Edge computing nodes perform a majority vote on the three-way positioning results, automatically eliminating anomalous data. This enables stable positioning even in complex branch networks and high-noise environments, accurately pinpointing fault points and significantly reducing manual troubleshooting time.

[0045] An adaptive algorithm based on an improved wavelet transform dynamically adjusts the feature extraction frequency band based on real-time noise levels, effectively separating fault signals from renewable energy harmonic interference. A built-in neural network online learns operating condition characteristics such as different grounding resistances and initial fault phase angles, and fuses multi-source information through fuzzy Petri nets for state inference. The dual-mode communication architecture seamlessly switches to the LoRa link in the event of a 5G outage, ensuring continuous execution of protection logic. This overcomes the detection bottleneck of high-resistance grounding and low-current faults, maintains high reliability in scenarios with high renewable energy penetration, and adapts to extreme environments and fluctuating communication conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of a distribution network fault protection and positioning method based on fast switches proposed by the present invention. DETAILED DESCRIPTION

[0047] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0048] The invention proposes a distribution network fault protection and location method based on fast switching, such as Figure 1 Shown, including:

[0049] Step 1: Arrange solid-state circuit breakers at equal intervals on the distribution network feeder, and set magnetically controlled disconnectors and bidirectional current detection intelligent switches at feeder branch nodes.

[0050] The three-level coordinated architecture for distribution network fault protection and location consists of a multi-level fast-switch network. This network comprises a core protection layer consisting of fast solid-state circuit breakers distributed at intervals of 300 ± 50 meters along the feeder lines. These breakers utilize silicon carbide (SiC) devices to achieve a response time of ≤ 2ms. Magnetic fast disconnectors are added at branch nodes as a secondary protection layer, and bidirectional current sensing intelligent switches are deployed at access points to form the terminal control layer.

[0051] In this embodiment, fast solid-state circuit breakers are installed every 300±50 meters along the main channel of a 10kV feeder. The main circuit is constructed using silicon carbide (SiC) MOSFET devices, with an internal integrated SiC MOSFET power module and IGBT drive circuit. A parallel RC snubber circuit (capacitor 4μF / resistor 10Ω) is used to suppress overvoltage during disconnection. A magnetically controlled fast disconnector (operation time 4.8ms) is deployed at the branch node. Its operating mechanism is pre-installed with a pressure sensor to monitor the contact pressure of the contacts in real time (standard value 20±2N). When the pressure deviates by more than 10%, the self-test program is triggered. All access points are equipped with a bidirectional current detection intelligent switch with a built-in Rogowski coil and a capacitive voltage divider. It uses a LEM ITC 2000 sensor with a range of 0-20kA and an accuracy of ±0.5%. After installation, the switch operation timing is tested using the pulse current injection method to ensure that the delay difference of the coordinated disconnection of adjacent circuit breakers is ≤50μs.

[0052] Step 2: Integrate a distributed detection unit in each solid-state circuit breaker, magnetically controlled disconnector, and bidirectional current detection intelligent switch; and use the distributed detection unit to obtain transient signals of distribution network faults.

[0053] Transient signals of distribution network faults include, but are not limited to, voltage and current. In the three-level collaborative architecture for distribution network fault protection and location, the second level is a distributed detection unit. Each switch integrates a high-frequency sampling module with a 1MHz sampling rate, a transient traveling wave detection circuit with a response time of ≤200ns, and an intelligent decision-making module that integrates ground fault direction determination and 5G / LoRa dual-mode communication to achieve localized processing of fault signals. The high-frequency sampling module continuously caches four cycles (80ms) of waveform data at a 1MHz rate. When the transient detection circuit is triggered, it locks in the data for the two cycles before and after the fault.

[0054] In the embodiment, a high-frequency sampling board is integrated in the switch cabinet corresponding to each solid-state circuit breaker, magnetically controlled disconnect switch and bidirectional current detection intelligent switch, and an ADS8588S 14-bit ADC chip is used to synchronously collect three-phase voltage and current at a sampling rate of 1MHz. The transient traveling wave detection circuit is composed of a high-speed comparator (response time ≤ 5ns) and an FPGA logic module: when du / dt>5kV / μs or di / dt>10A / μs is detected, a pulse signal with a width of 200ns is immediately started to latch the wave head polarity, where u and i are transient signals of distribution network faults. The ground fault direction judgment unit is based on the zero-sequence power direction algorithm. When the zero-sequence voltage U0>7%U n When the phase of zero-sequence current I0 lags behind U0 by 85°-95°, a direction flag is generated, where U nThe 5G / LoRa dual-mode communication module has preset channel switching logic: it automatically switches to LoRa spread spectrum mode when the signal strength is less than -90dBm, ensuring a transmission success rate of more than 99.9%.

[0055] In step 3, multiple sets of distributed detection units are configured with a group of edge computing nodes. Each group of edge computing nodes uses a noise-adaptive improved Morlet wavelet transform to extract the current change rate and voltage mutation amount from the transient signal. A three-dimensional decision model is constructed with the current change rate, voltage mutation amount, and line characteristic impedance as variables. Based on the three-dimensional decision model, the edge computing node uses the extracted current change rate and voltage mutation amount to determine whether an in-zone fault has occurred.

[0056] In the three-level collaborative architecture for distribution network fault protection and location, the third level is the edge computing node. When the distributed detection units are electrically connected to each other, one edge computing node with wavelet transform and transient energy spectrum analysis capabilities is deployed for every five distributed detection units. Time-Sensitive Networking (TSN) is used to establish a coordinated control loop with μs-level synchronization between these edge computing nodes, with synchronization errors controlled within ±1 μs. The edge computing nodes use Xilinx Zynq UltraScale+ MPSoC chips and deploy TSN switches that comply with the IEEE 802.1AS protocol. Fiber optic interfaces enable μs-level clock synchronization between nodes.

[0057] Specifically, step 3 includes:

[0058] Step 3.1: Deploy a wavelet transform processor in the edge computing node. The wavelet transform processor uses a noise-adaptive improved Morlet wavelet transform to process transient signals.

[0059] In the embodiment, in the noise-adaptive improved Morlet wavelet algorithm, the scale parameter is dynamically adjusted according to the real-time noise level of the sampling data in the previous second. The basis function of the noise-adaptive improved Morlet wavelet transform is:

[0060]

[0061] Where Ψ(t) is the basis function of the noise-adaptive improved Morlet wavelet transform, t is the sampling time, and σ is the scale parameter adaptively adjusted in the range of [0.1, 2.0] according to the real-time noise level;

[0062] In the embodiment, the wavelet transform processor loads the kernel function of Morlet wavelet transform when σ=0.5, and performs 128-scale decomposition on the current signal sampled at 1MHz. When the noise variance exceeds 0.1V 2When σ is used, the scale parameter σ is automatically expanded from 0.5 to 2.0 in the range of [0.1, 2.0] with a step size of 0.1, and 1024-point signal transformation is completed in each cycle; the present invention adopts a noise-adaptive improved Morlet wavelet algorithm, which can realize automatic adjustment of the processing signal amount according to the interference situation to ensure sufficient data.

[0063] In step 3.2, a transient energy spectrum analysis module is deployed in the edge computing node. The transient energy spectrum analysis module uses the processed transient signal to calculate the energy entropy value in the [1 Hz, 100 kHz] frequency band. If the sudden increase in the energy entropy value of any frequency band is greater than 3 times the energy entropy threshold, the current and voltage in the frequency band are extracted.

[0064] Among them, Morlet wavelet transform with σ=0.5 is implemented in the data window [-10ms, 10ms], and the energy entropy value in the characteristic frequency band of [35Hz, 80kHz] is extracted as the energy entropy threshold.

[0065] Step 3.3, calculate the current change rate by differential Wherein, Δt = 1μs; the voltage mutation ΔU is the difference between the effective values ​​of the half cycle before and after the fault;

[0066] In step 3.4, the weighted sum of the current change rate, voltage mutation, and line characteristic impedance is used as the three-dimensional decision model, as follows:

[0067]

[0068] Where F(x,y,z) is a three-dimensional decision model, is the current change rate, ΔU is the voltage mutation amount, Z0 is the line characteristic impedance, and the pre-stored values ​​in the line parameter library are used. α, β, and γ are decision parameters, α+β+γ=1. In the embodiment, α=0.6, β=0.3, and γ=0.1.

[0069] In step 3.5, during the period before the current detection moment, the mean voltage and mean current are obtained using a sliding window, and the mean voltage and mean current are substituted into the three-dimensional decision model, with 85% of the calculated values ​​being used as the dynamic fault threshold. The current change rate and voltage mutation amount extracted by the edge computing node are substituted into the three-dimensional decision model. If the calculated real-time value is greater than the dynamic fault threshold and the duration is greater than 1 / 4 cycle, it is determined that an in-zone fault has occurred; otherwise, it is determined that no in-zone fault has occurred.

[0070] The width of the sliding window is set to 10 periods.

[0071] In step 4, when it is determined that a fault has occurred within the area, the distributed detection unit sends the action state quantity of the device to the edge computing node; the edge computing node uses the calculated action state matrix of the device and adopts a state reasoning method based on fuzzy Petri nets to determine the fault section.

[0072] Specifically, the action state matrix of the device calculated by the edge computing node is:

[0073] L(k)=[S1,S2,…,S n ]×[t1,t2,…,t n ] T

[0074] Where L(k) is the action state matrix of the device calculated by the kth group of edge computing nodes, S j ∈{-1,0,1} represents the action state of the device obtained by the jth set of distributed detection units, S j A value of 1 indicates that the device is in positive motion. j A value of -1 indicates that the device moves in the reverse direction. j A value of 0 indicates that the device is not in action. j is the collection time of the jth set of distributed detection units, j = 1, 2, ..., n, n is the number of distributed detection units;

[0075] The forward action, reverse action and no action of the device are defined based on the logical state of the fault current direction, specifically:

[0076] 1) The physical operation corresponding to the positive action is: when the device is tripped, the fault point is located downstream of the device and the fault current is detected to be flowing from the power side to the load side. In this embodiment, if the traveling wave head is detected to be positive polarity, it is determined that the fault current is flowing from the power side to the load side;

[0077] 2) The physical operation corresponding to the reverse action is: when the device is locked and opened and sends a lockout message, the fault point is located upstream of the device and the fault current is detected to be flowing from the load side to the power side. In this embodiment, if the traveling wave head is detected to be negative polarity, it is determined that the fault current is flowing from the load side to the power side;

[0078] 3) The physical operation corresponding to no action is: the device remains in its original state, the fault is not within the protection range of the device and the fault characteristics do not reach the action threshold, or a locking instruction is received from an adjacent device.

[0079] Specifically, a collaborative logic controller is deployed in the edge computing node. The collaborative logic controller is used to adopt a state reasoning method based on fuzzy Petri nets to determine the fault section by reading the action state matrix of the device. For example, when the fault point is between switches S2 and S3, the action state vector is [1, 0, -1, 0, ...], which triggers the isolation of the corresponding section.

[0080] In the present invention, the communication architecture of each group of edge computing nodes adopts a layered fusion design. Among them, the bottom switch nodes use LoRa to build an ad hoc network, achieving 10km coverage and 50kbps effective transmission in the 470MHz frequency band, and supporting neighbor node relay forwarding; the edge layer deploys a 5G NR-U dedicated channel, using the 3.5GHz frequency band 200MHz bandwidth to ensure end-to-end delay ≤5ms, and a single-node peak rate of 500Mbps; the time synchronization system integrates GPS pulse-second timing and IEEE 1588 precision protocol, automatically switching to network clock synchronization when the satellite signal fails, maintaining a time base of ±1μs for the entire system. Among them, LoRa communication uses the SX1276 chip (470MHz frequency band), configured with a spreading factor SF=10, a bandwidth of 125kHz, and a coding rate of 4 / 5, to achieve 50kbps transmission within a range of 10km. The 5G NR-U module (Quectel RG500Q) allocates a 20MHz dedicated channel in the 3.5GHz band, employing LDPC coding and 256QAM modulation, achieving a peak rate of 800Mbps. The time synchronization system uses the ublox ZED-F9T dual-mode receiver, outputting a 1PPS signal with an accuracy of ±30ns. The TSN switch distributes clocks using the PTPv2 protocol, ensuring slave node clock deviation of ≤800ns.

[0081] Edge computing nodes broadcast action commands via TSN. The target circuit breaker opens within 2 milliseconds of receiving the command, and switches within the faulty section are locked or linked. Edge computing nodes are interconnected via TSN switches and clock synchronization is achieved using the gPTP protocol. A synchronization error compensation algorithm is calibrated every 10 milliseconds.

[0082] Step 5: Verify the determined fault section based on the fault current weighting of the equipment in the fault section and the matching degree of the characteristic impedance; after the fault section is verified, perform fault isolation.

[0083] Specifically, step 5 includes:

[0084] Step 5.1: Determine the weight coefficient of the device according to its location in the distribution network, and the distributed detection units in the fault section obtain the fault current of the device;

[0085] The weight coefficient of the trunk line equipment is set to 1.2, and the trunk line equipment uses solid-state circuit breakers; the weight coefficient of the branch equipment is set to 0.8, and the branch equipment uses magnetically controlled disconnectors; the weight coefficient of the section equipment is set to 1.0, and the section equipment uses magnetically controlled disconnectors with isolation operation only; the weight coefficient of the contact equipment is set to 0.9, and the contact equipment uses solid-state circuit breakers.

[0086] Step 5.2: When the weighted sum of the fault currents of multiple devices within the fault section is greater than 1.25 times the current threshold, the two solid-state circuit breakers at the boundary of the fault section are driven to perform a tripping operation, and a blocking instruction is sent to the non-fault section. At the same time, multiple devices corresponding to the weighted sum of the fault currents greater than 1.25 times the current threshold are driven to perform a coordinated tripping operation;

[0087] The current threshold is 1.25 times the rated current of the device with the largest weight coefficient in the fault section;

[0088] When determining the fault section, rather than tripping all devices within the fault section, which would expand the power outage, the system selects the device range for tripping based on the relationship between the weighted sum of the fault current and the threshold. Edge computing nodes are fully utilized to quickly determine the device range corresponding to the weighted sum of the fault current being greater than 1.25 times the current threshold, thereby properly controlling the power outage range and facilitating fault removal. The tripping of the two solid-state circuit breakers is completed within 2ms, and a locking command is sent to the upstream devices of each solid-state circuit breaker to ensure that power is maintained in non-fault areas.

[0089] In the embodiment, after the fault section is determined, the edge computing node quickly determines through multiple calculations: the fault current detected by the main line switch S1 is 8kA, the rated current is 12kA, and the corresponding weight coefficient is 1.2; the fault current detected by the section switch S2 that receives the trigger pulse is 6kA, and the corresponding weight coefficient is 1.0; the fault current detected by the branch switch S3 that receives the trigger pulse is 4kA, and the corresponding weight coefficient is 0.8. At this time, the weighted sum of the fault currents is 18.8kA, and the current threshold is 15kA, which meets the conditions for triggering the linked tripping; therefore, the main control FPGA sends a trigger pulse to the main line switch S1, the section switch S2, and the branch switch S3 to perform the linked tripping operation.

[0090] In step 5.3, inject the 150 Hz third harmonic at the device with the largest weight coefficient in the fault section and detect the matching degree of the characteristic impedance within the 2 ms time window, as shown below:

[0091]

[0092] In the formula, ζ is the matching degree, Z measuredZ0 is the characteristic impedance of the line, the pre-stored value of the line parameter library is used, Z min Zmin is the lower threshold of the impedance, the lower threshold of the impedance in the embodiment is 0.1 Ω to prevent zero error;

[0093] Step 5.4, if the matching degree is greater than 90%, the fault section passes the verification, and the locking instruction of the non-fault area is released;

[0094] When the matching degree is greater than 90%, it means that the fault point positioning is accurate, that is, the fault section determined in the foregoing is confirmed, so that the locking instruction of the non-fault area can be released, thereby realizing the multi-stage cooperative fault verification; if the confirmed fault section is located downstream of the distributed power supply, island protection is started.

[0095] Step 5.5, after the fault section verification, the solid-state circuit breaker is driven to trip in the transient priority mode, and when the transient priority mode fails, the magnetic control isolating switch is driven to trip in the steady-state backup mode, so as to realize fault isolation;

[0096] When the multi-stage verification of the fault section is completed, the trigger mechanism in the transient priority mode is preferentially executed, and when the transient priority mode fails (for example, the traveling wave detection fails), the trigger mechanism in the steady-state backup mode is automatically switched to be executed; the trigger mechanism proposed in the application adopts a dual-mode complementary design, including:

[0097] 1) Trigger mechanism in the transient priority mode:

[0098] Depend on the polarity and amplitude of the wave head to identify, complete the wave head arrival time interpolation calculation within 200 μs, start the microsecond level breaking of the solid-state circuit breaker, and the execution process is specifically:

[0099] First, when the voltage mutation rate du / dt> 5 kV / μs is detected, the wave head is marked, and then the wave head arrival time t arrival = argmin t (|V(t)-V th |)+0.12·d 2 V / dt 2 , V(t) is the wave head speed at time t, V th is the wave head threshold, and the calibration value of the second derivative compensation coefficient in the formula is 0.12; when the wave head amplitude is greater than 0.15 p.u. and the polarity meets the preset direction, the solid-state circuit breaker is driven to trip.

[0100] 2) Trigger mechanism in the steady-state backup mode:

[0101] Based on dynamic impedance sweep technology, the system impedance is obtained and an inverse time characteristic curve is generated as secondary protection, ensuring that fault isolation can be completed within 20ms even if the transient trigger fails. In transient trigger mode, the AD8610 high-speed comparator (response time 3ns) is used for traveling wave head detection. When du / dt>5kV / μs is detected, the ADF4158 phase-locked loop is activated to measure the time difference. In steady-state backup mode, the system impedance Z is measured every 5 minutes using the sweep method (1-100kHz sine injection). sys , dynamic calculation of inverse time parameter τ=0.8Z sys , then based on the inverse time characteristic curve t(I)=K / (I 2.5 -1)+0.01, t(I) is the inverse time characteristic parameter, I is the real-time value of current, when I>1.2I n And when the duration is t, I n The rated current triggers the magnetic isolation switch to operate.

[0102] In addition, when communication is interrupted, it automatically switches to the local inverse time protection mode, where the action time equation used is: Z sys The system impedance sweep frequency measurement value for the last 5 minutes is used, and the delay curve parameters of the magnetically controlled isolating switch are dynamically adjusted based on the real-time impedance measurement value.

[0103] Step 6: After the fault is isolated, the fault location coordinates and location reliability are optimized.

[0104] The present invention realizes fault isolation by adopting a trigger mechanism of dual-mode complementary design, activates a fault-tolerant positioning strategy, and improves reliability through triple redundancy verification.

[0105] The three-level redundancy check fault-tolerant positioning strategy adopted by the present invention includes:

[0106] 1) Traveling wave time difference of arrival positioning, time resolution ≤ 100ns;

[0107] 2) Impedance amplitude and angle characteristic analysis, frequency scanning range [1Hz, 100kHz];

[0108] 3) Switch action sequence matching, using Hamming distance verification;

[0109] The first verification layer adopts the wave velocity dynamic compensation algorithm to correct the traveling wave propagation velocity in real time according to the ambient temperature and humidity, compressing the time difference positioning error to ±0.3%; the wave velocity correction formula is v=0.97c×(1+α T ΔT)×(1+a H ΔH), where v is the corrected wave velocity, c is the speed of light, and α T , α Hare the temperature coefficient and humidity coefficient respectively, ΔT is the temperature change, and ΔH is the humidity change; the positioning calculation formula is (x,y) is the fault location coordinate, (x i ,y i ) is the coordinate of the monitoring point, t i -t0 is the arrival time of each monitoring point, i=1,2,3, t0 is the starting time, and the candidate fault location is finally output.

[0110] The second verification layer matches the pre-stored 128 line impedance feature database and performs amplitude and angle similarity analysis in the [1Hz, 100kHz] frequency band. The cosine similarity calculation formula is:

[0111]

[0112] Where, M = 128, is the number of pre-stored line impedances, f k There are 128 characteristic frequency points from 1kHz to 100kHz, θ meas ,θ db is the angle measurement value and the reference value; the candidate fault location output based on the cosine similarity sim is {P i |sim(P i )>0.95}, P i is the probability of the candidate fault location.

[0113] The third verification layer uses the Hamming distance algorithm to check the deviation between the equipment action sequence and the expected failure mode;

[0114] Finally, the positioning result is output by the majority voting mechanism to realize the screening matching mode. When at least two layers of verification are consistent, the topology coloring instruction is activated. In the embodiment, the topology coloring is performed after the fault is isolated, including: generating a red highlight display of the fault current path; marking the adjacent healthy section in green; displaying the suspicious transition section (impedance change rate ≥ 20% / ms) in yellow flashing; and pushing the colored map to the operation and maintenance terminal in real time through the human-computer interface.

[0115] The final decision is:

[0116]

[0117] Where, P TW is the candidate fault location determined by the traveling wave time difference of arrival location method, P Z is the candidate fault location determined by the impedance amplitude characteristic analysis method, P final is the optimized fault location, d H,min is the minimum value of the Hamming distance in the device action sequence matching method, and sim is the cosine similarity in the impedance amplitude characteristic analysis method;

[0118] When the cosine similarity is greater than 0.9, the calculation method is actually to define P TW The weight is 50%, P Z The weight is 30%; the weight of the device action sequence matching is 20%. This is essentially because when the traveling wave and impedance results are significantly different, the action sequence verification result has the lowest weight and is excluded, and only the first two data items are integrated.

[0119] In this embodiment, traveling wave time difference positioning uses a TDC-GPX2 time-to-digital converter chip (50ps resolution). The arrival times measured at three monitoring points are t1 = 100.2μs, t2 = 101.5μs, and t3 = 103.0μs. The fault distance is calculated based on the corrected wave velocity v = 294m / μs. Impedance spectrum analysis uses a pre-stored database (covering 128 line types, including YJV22-300 and LGJ-240). When the measured impedance at a frequency of 50kHz with an amplitude of 85° matches the characteristics of the YJV22-300 cable (standard value of 84.5°), the similarity score is 0.98. During switch operation sequence verification, the Hamming distance between the actual sequence "1010" and the expected pattern "1001" is 2, meeting the fault tolerance threshold.

[0120] The final positioning result is determined by a majority voting mechanism, and the fault location coordinates, location reliability, and abnormal flag are output.

[0121] In the embodiment, the deployed LSTM neural network constructs an adaptive learning framework to implement the various steps of the method proposed in the present invention, including: the input layer receives a 12-dimensional feature vector including the voltage mutation rate and the third harmonic distortion rate, and extracts spatiotemporal correlation features through a 3-layer 128-node gated cyclic unit; the output layer generates 6 types of fault probability distributions, updates the network weights every 5 minutes through online incremental learning, and maintains a recognition accuracy of >97% in a scenario where the photovoltaic output fluctuates by ±30%; the fault type library covers complex working conditions such as high-resistance grounding and arc short circuit, and supports dynamic expansion of classification dimensions based on historical data. The deployed LSTM neural network consists of the following: an input layer with a 12-dimensional feature vector, including ΔU / Δt (calculated in a 1μs time window), third harmonic content (150Hz component ratio calculated using FFT), and zero-sequence power direction (-1 / 0 / +1 encoding); a hidden layer with three layers of gated recurrent units (GRUs), each with 128 neurons. The hidden layer uses GRUs (128 nodes), a 128×128 gated weight matrix, and a tanh activation function; and an output layer with a 6-dimensional vector generated by softmax to represent the probability distribution of fault types, with an online update period of ≤5 minutes. For example, the output vector [0.02, 0.91, 0.03, 0, 0.04, 0] indicates a 91% probability of a phase B ground fault. The online learning module extracts 300 samples from the ring buffer every 5 minutes and performs SGD optimization to update the network parameters.

[0122] In the fault protection method, after fault isolation, visual topology reconstruction is performed. The master station system generates a dynamic coloring map based on an SVG vector graphics engine. The fault path is highlighted in red with a line width 3 times the original width, while the adjacent healthy segments are displayed in semi-transparent green. A yellow flashing warning is activated for transition segments with an impedance mutation rate greater than 20% / ms, with the flashing frequency positively correlated with the residual energy of the fault. The map data is compressed and encrypted and pushed to mobile operation and maintenance terminals via the 5G network, supporting millisecond-level refresh and multi-level zooming for multi-person collaboration. The topology coloring engine parses the CIM / E model file and generates an SVG vector image. The fault path is rendered in RGB(255,0,0) with a line width of 6px, while the healthy segments are rendered in RGBA(0,255,0,0.4) with a line width of 2px. HSL animation is enabled for transition segments, with the hue value H periodically varying at a 2Hz frequency between 50° and 70°, saturation S = 100%, and brightness L = 50%. The atlas data is compressed with zlib (compression rate > 60%) and published to the mobile terminal through the MQTT protocol. The terminal implements 60fps dynamic rendering based on the WebGL engine and supports multi-touch zooming (zoom level 1%-1000%).

[0123] The beneficial effect of the present invention is that, compared with the prior art,

[0124] 1. Rapid fault isolation and arc suppression

[0125] By densely deploying fast solid-state circuit breakers along the feeder lines to create a multi-stage switching network, combined with transient traveling wave detection and high-frequency sampling technology, the microsecond-level current mutation characteristics of the initial fault are captured in real time. When an abnormal polarity of the traveling wave head is detected, the coordinated triggering logic immediately drives the adjacent circuit breakers to operate synchronously, forming an electrical isolation barrier for the fault section. Edge computing nodes achieve μs-level synchronization of switching operations through a time-sensitive network, ensuring that the coordinated disconnection timing of the multi-stage switches precisely matches the fault propagation path, significantly shortening fault clearing time and effectively suppressing arc propagation, preventing the fault from expanding and damaging equipment.

[0126] 2. High-precision fault-tolerant positioning

[0127] A triple redundancy check mechanism employs traveling wave time difference of arrival, impedance spectrum analysis, and switching action sequence matching. The traveling wave detection module dynamically compensates for the impact of line parameter fluctuations on wave velocity. The impedance spectrum matching algorithm compares pre-stored line characteristics across a wide frequency band. The switching action logic verifies topological action consistency using Hamming distance. Edge nodes perform a majority vote on the three-way positioning results, automatically eliminating anomalous data. This ensures stable positioning even in complex branch networks and high-noise environments, accurately pinpointing fault points and significantly reducing manual troubleshooting time.

[0128] 3. Adaptive complex fault detection

[0129] Based on the improved wavelet transform adaptive algorithm, the feature extraction frequency band is dynamically adjusted according to the real-time noise level, and the fault signal and new energy harmonic interference are effectively separated. The built-in neural network learns the working condition characteristics such as different grounding resistances and fault initial phase angles online, and the multi-source information is fused through the fuzzy Petri net to carry out state reasoning. The dual-mode communication architecture seamlessly switches to the LoRa link when the 5G is interrupted, guarantees the continuous execution of the protection logic, breaks through the detection bottleneck of high-resistance grounding and small-current faults, maintains high reliability in the new energy high-penetration scene, and adapts to extreme environments and communication fluctuation conditions.

[0130] The present disclosure can be a system, a method, and / or a computer program product. The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present disclosure.

[0131] The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium can be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or punched-tape, a holographic storage medium, or any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.

[0132] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network can comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0133] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions. The electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A distribution network fault protection and location method based on fast switching, characterized in that: include: Utilize the distributed detection units integrated on the electrical equipment in the distribution network to obtain transient signals of distribution network faults; Each group of edge computing nodes uses a noise-adaptive improved Morlet wavelet transform to extract the current change rate and voltage mutation amount from the transient signal. A three-dimensional decision model is constructed with the current change rate, voltage mutation amount, and line characteristic impedance as variables. Based on the three-dimensional decision model, the edge computing node uses the extracted current change rate and voltage mutation amount to determine whether an in-zone fault has occurred. Each set of distributed detection units is configured with a group of edge computing nodes. When a fault is detected within the area, the distributed detection unit sends the action state of the device to the edge computing node. The edge computing node uses the calculated action state matrix of the device and adopts a state reasoning method based on fuzzy Petri nets to determine the fault section and perform fault protection.

2. The method for distribution network fault protection and location based on fast switches according to claim 1, characterized in that: The determined fault section is verified based on the fault current weighting and characteristic impedance matching of the equipment within the fault section; after the fault section is verified, fault isolation is performed; after fault isolation, fault location is optimized.

3. The method for distribution network fault protection and location based on fast switches according to claim 1, characterized in that: Solid-state circuit breakers are arranged at equal intervals on the distribution network feeders, and magnetically controlled disconnectors and bidirectional current detection intelligent switches are set at the feeder branch nodes; a distributed detection unit is integrated in each solid-state circuit breaker, magnetically controlled disconnector and bidirectional current detection intelligent switch.

4. The method for distribution network fault protection and location based on fast switches according to claim 1, characterized in that: A wavelet transform processor is deployed in the edge computing node. The wavelet transform processor uses a noise-adaptive improved Morlet wavelet transform to process transient signals. A transient energy spectrum analysis module is deployed in the edge computing node. The module uses the processed transient signal to calculate the energy entropy value in the [1Hz, 100kHz] frequency band. If the sudden increase in the energy entropy value in any frequency band is greater than three times the energy entropy threshold, the current and voltage in that frequency band are extracted. The current change rate is calculated by differential calculation, and the voltage mutation is measured by the difference between the effective values ​​of the half cycle before and after the fault; The weighted sum of current change rate, voltage mutation and line characteristic impedance is used as a three-dimensional decision model.

5. The method for distribution network fault protection and location based on fast switches according to claim 4, characterized in that: In the noise-adaptive improved Morlet wavelet algorithm, the scale parameter is dynamically adjusted according to the real-time noise level of the sampled data. The basis function of the noise-adaptive improved Morlet wavelet transform is: Where Ψ(t) is the basis function of the noise-adaptive improved Morlet wavelet transform, t is the sampling time, and σ is the scale parameter adaptively adjusted in the range of [0.1, 2.0] according to the real-time noise level.

6. The method for distribution network fault protection and location based on fast switches according to claim 5, characterized in that: A Morlet wavelet transform with σ=0.5 is performed on the data window [-10ms, 10ms], and the energy entropy value in the characteristic frequency band of [35Hz, 80kHz] is extracted as the energy entropy threshold.

7. The method for distribution network fault protection and location based on fast switches according to claim 3, characterized in that: In the period before the current detection moment, the sliding window is used to obtain the average voltage and current, and the average voltage and current are substituted into the three-dimensional decision-making model, with 85% of the calculated values ​​as the dynamic fault threshold. The current change rate and voltage mutation amount extracted by the edge computing node are substituted into the three-dimensional decision-making model. If the calculated real-time value is greater than the dynamic fault threshold and the duration is greater than 1 / 4 cycle, it is determined that an internal fault has occurred; otherwise, it is determined that no internal fault has occurred.

8. The method for distribution network fault protection and location based on fast switches according to claim 1, characterized in that: The edge computing node uses the product of the device's action state quantity obtained by the distributed detection unit and the collection time vector of the distributed detection unit as the device's action state matrix; among them, the device's action state quantity obtained by the j-th set of distributed detection units is S j ∈{-1,0,1}, a value of 1 indicates that the device is in forward motion, a value of -1 indicates that the device is in reverse motion, and a value of 0 indicates that the device is not in motion.

9. The method for distribution network fault protection and location based on fast switches according to claim 1, characterized in that: The weight coefficient of the device is determined according to its location in the distribution network, and the distributed detection unit in the fault section obtains the fault current of the device; When the weighted sum of the fault currents of multiple devices within the fault section is greater than 1.25 times the current threshold, the two solid-state circuit breakers at the boundary of the fault section are driven to perform a tripping operation, and a blocking instruction is sent to the non-fault area. At the same time, multiple devices corresponding to the weighted sum of the fault currents greater than 1.25 times the current threshold are driven to perform a coordinated tripping operation; Inject a 150Hz third harmonic at the device with the largest weight coefficient within the fault section and test the matching degree of the characteristic impedance within a 2ms time window. If the matching degree is greater than 90%, the fault section is verified and the blocking instruction of the non-fault section is released. After the fault section is verified, the solid-state circuit breaker is driven to open in the transient priority mode. When the transient priority mode fails, the magnetic control disconnector is driven to open in the steady-state backup mode to perform fault isolation.

10. The method for distribution network fault protection and location based on fast switches according to claim 9, characterized in that: Set the weight coefficient of the trunk equipment to 1.2, the weight coefficient of the branch equipment to 0.8, the weight coefficient of the segment equipment to 1.0, and the weight coefficient of the communication equipment to 0.

9.

11. The method for distribution network fault protection and location based on fast switches according to claim 10, characterized in that: Driving the solid-state circuit breaker to open in transient priority mode includes: When the voltage mutation rate is greater than 5kV / μs, the wave front is marked and the wave front arrival time is calculated using wave front velocity interpolation. When the wave front amplitude is greater than 0.15pu and the polarity is correct, the solid-state circuit breaker is driven to open.

12. The method for distribution network fault protection and location based on fast switches according to claim 9, characterized in that: The matching is as follows: In the formula, ζ is the matching degree, Z measured is the impedance modulus measured at 150Hz, Z0 is the line characteristic impedance, and Z min is the lower impedance threshold.

13. The method for distribution network fault protection and location based on fast switches according to claim 9, characterized in that: Driving the magnetically controlled disconnector to open in steady-state backup mode includes: The dynamic impedance sweep method is used to obtain the system impedance, calculate the inverse time parameter, and determine the real-time current value based on the inverse time characteristic curve. When the real-time current value is greater than 1.2 times the rated current, the magnetically controlled isolating switch is triggered.

14. The method for distribution network fault protection and location based on fast switches according to claim 2, characterized in that: After the fault is isolated, the traveling wave time difference of arrival positioning method, impedance amplitude and angle characteristic analysis method and equipment action sequence matching method are used simultaneously to optimize the fault location coordinates; Among them, the traveling wave time difference of arrival location method uses the corrected traveling wave propagation velocity, the coordinates of each monitoring point, and the arrival time of each monitoring point to determine the candidate fault location. In the impedance amplitude characteristic analysis method, the cosine similarity of the amplitude angle measurement value and the reference value at each frequency point in the [1Hz, 100kHz] frequency band is calculated, and the location with a confidence level greater than 0.95 is selected as the candidate fault location. In the equipment action sequence matching method, the Hamming distance between the equipment action sequence and the expected fault mode is calculated. When the minimum Hamming distance is no more than 2, the candidate fault location determined by the traveling wave time difference of arrival location method is used as the optimized fault location. If the cosine similarity is greater than 0.9, the optimized fault location is obtained by dividing the sum of 0.5 times the candidate fault location determined by the traveling wave time difference of arrival location method and 0.3 times the candidate fault location determined by the impedance amplitude and angle characteristic analysis method by 0.

8. If the minimum Hamming distance is greater than 2 and / or the cosine similarity is no more than 0.9, manual fault location is required.

15. A distribution network fault protection and location system based on a fast switch, used to implement the distribution network fault protection and location method based on a fast switch according to any one of claims 1 to 14, characterized in that: include: Solid-state circuit breakers, magnetically controlled disconnect switches, bidirectional current detection intelligent switches, distributed detection units, and edge computing nodes; Solid-state circuit breakers are arranged at equal intervals on the distribution network feeders, and magnetically controlled disconnectors and bidirectional current detection intelligent switches are installed at the feeder branch nodes. A distributed detection unit is integrated into each solid-state circuit breaker, magnetically controlled disconnector, and bidirectional current detection intelligent switch. Distributed detection unit, used to obtain transient signals of distribution network faults; when it is determined that a fault has occurred in the area, the action status of the device where it is located is sent to the edge computing node; Each group of edge computing nodes uses a noise-adaptive improved Morlet wavelet transform to extract the current change rate and voltage mutation amount from the transient signal. A three-dimensional decision model is constructed with the current change rate, voltage mutation amount, and line characteristic impedance as variables. Based on the three-dimensional decision model, the edge computing node uses the extracted current change rate and voltage mutation to determine whether a fault has occurred in the area. Among them, each set of distributed detection units is configured with a group of edge computing nodes; After using the calculated action state matrix of the equipment, the state reasoning method based on fuzzy Petri net is adopted to determine the fault section to perform fault protection.

16. The distribution network fault protection and positioning system based on fast switches according to claim 15, characterized in that: Each group of edge computing nodes is also used to verify the determined fault section based on the fault current weighting of the equipment in the fault section and the matching degree of the characteristic impedance; after the fault section is verified, fault isolation is performed; after fault isolation, the fault location is optimized.

17. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 14.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 14 are implemented.

Images