An arc crosstalk identification and positioning method based on distributed frequency domain perception

Abstract

A kind of arc crosstalk identification and positioning method based on distributed frequency domain perception, the method includes detecting the power grid node broadcasting arc discovery message of arc;All received broadcast power grid nodes collect and record the arc discovery message;Election master node based on agreed rules;Election winner node broadcasts master node declaration as master node, and the rest nodes enter listening / waiting state;Master node constructs low-frequency mutation matrix;Master node starts turn wave-trap process, constructs wave-trap response matrix;Master node fuses low-frequency mutation matrix and wave-trap response matrix to speculate arc occurrence position and report power grid master station.The method realizes the accurate determination of arc fault source position by the frequency response obtained by multi-node cooperation and multi-parameter fusion perception.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a method for identifying and locating arc crosstalk based on distributed frequency domain sensing. Background Technology

[0002] A fault arc is an abnormal arc discharge phenomenon, usually caused by poor contact, insulation damage, or aging of the circuit. Fault arcs pose the following hazards:

[0003] Fire hazard: The temperature of an electric arc can reach thousands of degrees Celsius, which can easily ignite surrounding insulating materials or flammable materials.

[0004] Equipment damage: The electric arc may burn the contact points, causing the wires to melt or the equipment to fail.

[0005] Electrical safety hazards: could lead to electric shock accidents or instability in the power system.

[0006] The main characteristics of AC fault arcs include:

[0007] 1) Nonlinear waveforms: The current and voltage waveforms are irregular and contain abundant high-order harmonic components.

[0008] 2) High-frequency characteristics: The fault arc signal contains high-frequency components (from several kilohertz to several gigahertz) and has a wide frequency spectrum.

[0009] 3) High randomness: The occurrence of fault arcs is random, with frequent fluctuations and is difficult to predict.

[0010] Electric arc faults are a significant potential hazard in the operation of electrical equipment, potentially leading to equipment damage or fire. Therefore, timely detection of electric arc faults is of great importance.

[0011] Arc fault detection primarily relies on capturing and analyzing abnormal arc signals. The spectrum of an arc fault contains abundant high-frequency components, typically ranging from several kHz to several GHz. The common practice is to extract the high-frequency components of the line current using a specially designed high-frequency transformer; changes in these high-frequency components diagnose whether an arc fault has occurred. However, high-frequency signals are prone to crosstalk, which can lead to multiple detection devices near the fault point detecting the arc signal simultaneously. In such scenarios, this can cause unexpected disconnections in normally operating power branches, preventing accurate fault location and significantly hindering the reliability and market adoption of arc fault detection products.

[0012] Current arc fault detection primarily relies on current sensing. This involves placing a current sensor on a specific phase (L or N) of the power line and analyzing its time-frequency characteristics to diagnose the presence of an arc signal. An alarm signal is then issued, or the line is disconnected to extinguish the arc. However, in practical applications, crosstalk can cause multiple detection devices near the fault location to detect the arc signal. This can lead to unexpected disconnections of normally functioning branches, making it difficult to accurately pinpoint the fault location and significantly hindering the reliability and market adoption of arc fault detection products.

[0013] Patent document CN113009271B, entitled "A Method, Device, and Storage Medium for Fault Arc Detection and Location," proposes a technical solution for fault arc detection and location. This solution involves installing a high-frequency notch filter circuit on the mains power inlet line and measuring the signal amplitude at a specific frequency point downstream of the notch filter. By combining the relative energy change between the notch filter center frequency and its neighboring frequencies, it determines whether the fault or interference signal originates upstream or downstream of the notch filter circuit.

[0014] This technical solution places high demands on the design of the notch filter circuit. Specifically, the method relies on comparing the signal strength at three key frequency points: f0 is the center notch frequency of the notch filter circuit design, and f1 and f2 are located on either side of it, at the edge regions of the notch response curve. Ideally, if the signal enters from upstream of the notch filter circuit, the signal at f0 should be significantly attenuated after passing through the notch filter circuit, while the attenuation at f1 and f2 should be smaller. The energy difference between the three points can reflect whether the signal has passed through the notch filter circuit. However, when the Q value of the notch filter circuit is low and the notch response curve is relatively flat, the difference in signal amplitude between f0, f1, and f2 becomes insignificant, making it difficult to effectively determine the signal source path. In addition, if the fault signal itself has a wide and unstable spectrum, its energy distribution at each frequency point fluctuates greatly under different time windows, further weakening the robustness and accuracy of the three-frequency-point-based judgment. Summary of the Invention

[0015] The technical problem to be solved by this application is to provide a method for arc crosstalk identification and localization that can identify fault arc crosstalk and accurately locate the location of fault arc occurrence.

[0016] According to one aspect of this application, a method for arc crosstalk identification and localization based on distributed frequency domain sensing is provided, comprising the following steps:

[0017] A power grid node broadcasts a message, wherein the power grid node is the one that detected the electric arc, and the message is an arc detection message, which includes an event ID, a detection timestamp, and an arc intensity.

[0018] All grid nodes that receive the broadcast collect and record the arc detection messages from other grid nodes;

[0019] Candidate nodes elect a master node based on agreed-upon rules; the candidate nodes are all power grid nodes that receive the broadcast, and the master node is used for arc crosstalk identification and location. The agreed-upon rules include electing the power grid node that first detects the arc as the master node, or electing the master node based on the arc intensity in the arc detection message.

[0020] The master node broadcasts a master node declaration message, and the other nodes enter a listening / waiting state; the master node declaration message includes the master node's device identifier or communication address, and / or the notch filter configuration parameters to be used;

[0021] The master node extracts the arc intensity information from the arc detection messages of the other nodes and constructs a low-frequency mutation matrix;

[0022] The master node initiates the alternating notch filtering process;

[0023] After the notch filtering process is completed, the notch response indices of all participating nodes are obtained sequentially to construct the notch response matrix;

[0024] The master node integrates the low-frequency mutation matrix and the notch response matrix to infer the location of the electric arc and reports it to the power grid master station.

[0025] According to some embodiments,

[0026] The candidate nodes are determined using an event-aware window mechanism, the steps of which include:

[0027] After the event initiating node detects the abnormal event of an electric arc locally, it starts a timed window and broadcasts an event initiation frame. The event initiating node is the first power grid node to detect the electric arc. The event initiation frame contains an arc discovery message with a unique event ID.

[0028] After receiving the event initiation frame, the other power grid nodes determine whether they have also sensed the abnormal event of the arc. If so, they send an event response frame to the event initiation node within the time window. The event response frame contains an arc detection message indicating that the other power grid nodes have sensed the abnormal event of the arc. The event ID of the event response frame is taken from the event initiation frame.

[0029] The event initiating node collects information on all response nodes that sent event response frames after the timed window ends, forming a candidate node set.

[0030] According to some embodiments,

[0031] The arc detection message also includes: supported notch filter parameters or recommended notch filter parameters;

[0032] After receiving the declaration information from the master node, the remaining nodes extract the notch filter configuration parameters therein and initialize the notch filter circuit according to the extracted notch filter configuration parameters.

[0033] According to some embodiments,

[0034] The power grid node is equipped with an arc sensing unit and a notch filter circuit.

[0035] The arc sensing unit includes: a current sensor, configured on the phase line, for acquiring the low-frequency components of the line current; and a high-frequency coupling circuit, for sensing and extracting the high-frequency noise components in the line voltage.

[0036] The notch circuit is used to deeply attenuate the input signal within a set frequency range, thereby suppressing the signal propagation in that frequency band; the notch circuit includes one or more notch branches, and the notch branch includes a switch and a notch filter.

[0037] According to some embodiments,

[0038] The step of the master node extracting the arc intensity information from the arc detection messages of the other nodes and constructing the low-frequency mutation matrix includes:

[0039] The arc intensity information includes the average signal intensity of the target frequency band and the magnitude of low-frequency abrupt changes before and after the arc occurs;

[0040] A low-frequency mutation matrix is ​​constructed using the magnitude of low-frequency mutations before and after the occurrence of an electric arc, and the power grid node sequence involved in the low-frequency mutation matrix is ​​the same as the power grid node sequence involved in notch filtering.

[0041] The low-frequency mutation size is used to quantify the intensity of low-frequency mutations in response to fault events, providing a basis for determining the structural relationships between nodes; the low-frequency mutation size D i Calculate using the following formula:

[0042]

[0043] in:

[0044] f is the feature number, and p is the total number of features. Let ω represent the low-frequency current characteristic values ​​of the f-th feature of the i-th node before and after the fault. These low-frequency current characteristic values ​​are calculated based on one or more key features, including the cumulative sum of current cycles before and after the arc, the rate of change of current before and after the arc, and the variance of the effective current value. f ε is the weighting coefficient for the f-th feature, and ε is a small constant to avoid division by zero.

[0045] According to some embodiments,

[0046] It also includes a process for filtering the candidate nodes:

[0047] All the candidate nodes mentioned constitute a candidate node set;

[0048] If the number of nodes in the candidate node set is greater than a preset value M, then the candidate nodes are filtered:

[0049] The candidate nodes are sorted in descending order of the detected arc intensity, and the top M candidate nodes are selected as the nodes to participate in the alternating notch filter.

[0050] Alternatively, the candidate nodes using PLC communication can be sorted by PLC network hierarchy, and the candidate nodes that are closer to the master node hierarchy can be selected as the nodes participating in the round-robin notch filtering.

[0051] According to some embodiments,

[0052] The alternating notch filtering process initiated by the master node includes:

[0053] All the candidate nodes mentioned constitute a candidate node set;

[0054] The master node uses unicast messages to control each node in the candidate node set to sequentially connect the notch circuit.

[0055] When a controlled candidate node successfully connects the notch circuit, it broadcasts a message to the candidate node set. The message is used to ensure that the candidate nodes in the candidate set know who has connected the notch circuit.

[0056] After the notch circuit is turned on, the signal strength of multiple frequency points is collected within the set notch bandwidth range, with the notch center frequency as the reference. The average signal strength of the frequency band is calculated as the notch response index, which is used to quantify the channel attenuation between nodes under the influence of the notch.

[0057] The candidate node samples and calculates the notch response index multiple times within the notch circuit turn-on window, and reports the average value within the notch circuit turn-on window period.

[0058] According to some embodiments,

[0059] The notch response index is the relative attenuation rate of the target frequency band signal strength observed by grid node j when the notch circuit is connected at grid node i, denoted as R. ij ,

[0060] Define the average signal strength of grid node j before the notch filter is applied at grid node i as follows: The period of notch filtering is but: R ij Set to R for cases < 0 ij=0, which is considered an abnormal response and should be ignored or handled separately;

[0061] If there are K frequency points {f1, f2, ..., f3} within the target frequency band K The signal strength received by grid node k is S(f) k The average signal strength of the power grid node k in the frequency band. Calculate using the following formula:

[0062]

[0063] According to some embodiments,

[0064] The notch response index R ij for:

[0065]

[0066] In the formula, Let J be the average signal strength in the frequency band of grid node j when notch filtering is applied at grid node i. Let J be the average signal strength of grid node j in the adjacent frequency band when grid node i is connected to notch filtering.

[0067] According to some embodiments,

[0068] The step of the master node fusing the low-frequency mutation matrix and the notch response matrix to infer the location of the electric arc includes:

[0069] Based on the low-frequency mutation matrix, analyze the possible connection types between the arc occurrence location and the power grid nodes;

[0070] Using the notch response matrix, the possible coupling paths of the arc signal are analyzed based on the response of each power grid node before and after the notch.

[0071] By combining the above two methods to eliminate scenarios, analyzing the physical distance and relative position between the arc occurrence location and the power grid node, the location of the arc occurrence can be inferred.

[0072] The beneficial effects of this application are analyzed as follows:

[0073] This application proposes a distributed master node election and notch polling mechanism that does not rely on a concentrator. After a fault event occurs, neighboring meter nodes spontaneously sense the fault signal and participate in the negotiation and election process to select a master node to initiate notch polling and response collection tasks. The response is fast and the communication cost is low: there is no need to send data to the concentrator or master station. The master node election and polling process is completed autonomously by neighboring nodes, which greatly reduces the load on the central node and network communication overhead. Since arc faults have strong local low-frequency and high-frequency disturbance characteristics, false alarms caused by crosstalk also mainly occur in neighboring nodes. Therefore, a local collaborative approach is adopted to carry out notch testing and response analysis, which is more in line with the actual physical characteristics of the power grid. The system only requires the master node selected from the fault neighborhood to report the final fault location result, avoiding the alarm storm caused by all nodes reporting and reducing redundant alarms and master station interference.

[0074] This application also proposes to integrate the low-frequency mutation intensity matrix and the high-frequency notch response matrix for arc location analysis. These two matrices, respectively, analyze the arc fault location of low-voltage lines from two dimensions: the physical structure of the power flow path and the propagation characteristics of high-frequency signals. After constructing the low-frequency mutation matrix and the high-frequency notch response matrix before and after the fault, the possible connection types (series / parallel / common path) between the arc location and the meter node are analyzed based on the low-frequency mutation matrix. This is an advantage of using low-frequency mutation to analyze the arc fault location. However, the low-frequency mutation intensity matrix is ​​easily affected by the line's own load disturbances (such as the start-up of high-power loads, severe harmonic interference in the line itself, etc.). Therefore, after initially analyzing the possible location through low-frequency mutation, the high-frequency notch response matrix is ​​further combined for analysis. By analyzing the response between nodes before and after the notch, the possible coupling paths of the arc signal are analyzed. The combination of both is used to further eliminate scenarios and achieve a more accurate location inference.

[0075] This application also proposes two optional methods for calculating notch response indices to quantify the attenuation of the target frequency band signal strength during notching at other nodes, thereby constructing a notch response matrix. One method compares the average intensity of the frequency band before and after notching. The other method proposes a relative estimation method that does not require a reference value before notching. The advantage of this method is that it does not require snapshot data before notching. Since the measurement time before and after notching is inconsistent, comparing data before and after notching may be affected by factors such as transient disturbances in the power grid, load fluctuations, and changes in background noise. By directly comparing the power difference between the current notched frequency band and its adjacent non-notched frequency band, the deviation caused by the inconsistency of the background in the time period before and after notching is effectively avoided, thus improving the stability and accuracy of the response index.

[0076] This application also provides an implementation scheme for constructing a low-frequency mutation intensity matrix: each meter that detects an electric arc will broadcast an arc detection message to nearby meters; the arc detection message includes, but is not limited to, an event ID, a detection timestamp, and an arc intensity; the arc intensity includes the average signal strength of the target frequency band and the magnitude of the low-frequency mutation before and after the arc occurs, so that the master node can extract the arc intensity information from the "fault detection message" reported by the candidate nodes to construct the low-frequency mutation intensity matrix; there is no need to poll, thus improving the efficiency of information collection.

[0077] This application also provides a scheme for dynamically selecting or adjusting notch filter frequency band parameters: the arc detection message includes supported notch filter parameters and recommended notch filter parameters; such as the number of notch filter branches and corresponding frequency band information, the meter can also recommend suitable notch filter parameters to the master node based on the noise floor; after integrating this information, the master node packages the notch filter parameter information to be used in the "Master Node Declaration" message, and initializes the notch filter configuration parameters to be used by all participating nodes in one batch. In actual deployment, the notch filter integrated in the meter can support multiple preset frequency bands. To adapt to the background noise distribution and signal interference characteristics under different power grid operating environments, this invention allows the master node to dynamically select or adjust notch filter frequency band parameters according to the current power grid noise floor spectrum characteristics, so as to improve the effectiveness of signal recognition and anti-interference capability.

[0078] In summary, the technical solution of this application has the following beneficial effects:

[0079] The technical solution disclosed in this application combines the high-frequency notch response matrix and the low-frequency current mutation characteristic matrix to establish a multi-dimensional perception framework. It performs joint analysis on the location of arc faults in low-voltage lines from two dimensions: the physical structure of the power flow path and the high-frequency signal propagation characteristics, thus overcoming the identification errors of a single response mechanism.

[0080] The distributed master node election and polling mechanism in this application does not rely on the participation of the concentrator. The nodes autonomously elect the master node and initiate notch scheduling and response collection locally. Only the master node selected from the fault neighborhood is responsible for analyzing and reporting the final fault location result, avoiding the alarm storm caused by all nodes reporting and reducing redundant alarms and master station interference.

[0081] Significantly reducing false alarm and false location rates: Most existing fault arc detection methods cause false alarms in adjacent tables due to problems such as arc crosstalk. The technical solution of this application uses distributed frequency sensing, with the master node selected from the fault neighborhood responsible for analyzing and reporting the final fault location results, which significantly reduces the risk of false alarms and improves operation and maintenance efficiency.

[0082] In summary, the arc crosstalk identification and localization method proposed in this application, after detecting an arc, achieves accurate determination of the location of the arc fault source by acquiring the frequency response through multi-node collaboration and multi-parameter fusion sensing. Attached Figure Description

[0083] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0084] Figure 1 A flowchart of an arc crosstalk identification and localization method based on distributed frequency domain sensing, according to an example embodiment, is shown.

[0085] Figure 2 A diagram of a low-voltage power distribution network structure according to an example embodiment is shown. Detailed Implementation

[0086] The embodiments of this application will now be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0087] Those skilled in the art should understand that the following specific embodiments or implementation methods are a series of optimized configurations listed in this application to further explain the specific application content. These configuration methods can be combined or used in conjunction with each other, unless this application explicitly states that some or a specific embodiment or implementation method cannot be associated with or used in conjunction with other embodiments or implementation methods. Furthermore, the following specific embodiments or implementation methods are only considered as optimized configurations and are not intended to limit the scope of protection of this application.

[0088] Example 1

[0089] Figure 1 A flowchart of an arc crosstalk identification and localization method based on distributed frequency domain sensing, according to an example embodiment, is shown.

[0090] like Figure 1 As shown, the following steps are performed for arc crosstalk identification and localization based on distributed frequency domain sensing:

[0091] Step S11: All meters that have detected an electric arc broadcast an arc detection message.

[0092] Figure 2 A diagram of a low-voltage power distribution network structure according to an example embodiment is shown.

[0093] like Figure 2 As shown, the meter includes an arc sensing unit and a notch filter circuit. The arc sensing unit includes a current sensor, positioned on the phase line (L line), for acquiring the low-frequency components of the line current; and a high-frequency coupling circuit for sensing and extracting high-frequency noise components from the line voltage. The notch filter circuit includes components for deeply attenuating the input signal within a set frequency range, thereby suppressing signal propagation in that frequency band. The notch filter circuit includes notch branches, each containing a switching switch and a notch filter. The notch filter can be a passive LC structure or an active band-stop filter; the notch filter circuit can also include multiple notch branches, thus having multiple configurable notch frequency bands. Generally speaking, the arc signal spectrum that the arc sensing unit can sense ranges from several kHz to 30 MHz; the frequency components corresponding to the selected notch band are only a small part of the arc signal spectrum, which is a characteristic subset within the frequency domain that the arc sensing unit can sense; when the notch circuit is turned on, it will cause the impedance change of the line at a specific frequency, thereby changing the propagation characteristics of the arc signal at that frequency and affecting the reception strength of other nodes at that frequency component.

[0094] Due to the influence of arc crosstalk signals causing multiple meters to detect arc events, each meter that detects an arc will broadcast an arc detection message to nearby meters. The arc detection message includes, but is not limited to, the event ID, detection timestamp, arc intensity, etc. If the meter uses PLC carrier communication, it can also include the network layer where the meter is located. For notch filter circuits that support multiple branches, i.e., support multiple notch filter frequency bands, it can also include the supported notch filter parameters, recommended notch filter parameters, etc.

[0095] In step S12, all the meters that received the broadcast collect and record messages from other devices.

[0096] Step S13: Elect a master node based on the agreed rules.

[0097] The broadcast arc detection message uses an "event-aware window mechanism" to determine the set of faulty nodes. There are two types of broadcast arc detection messages: The first meter to detect an arc, after locally sensing the abnormal event, initiates a timed window T1 and broadcasts an arc detection message with a unique event ID (called the "event initiation frame"). Upon receiving the event initiation frame, the remaining nodes determine whether they have also sensed the event and respond with a response information event ("event response frame") within the window time. The event ID of the event response frame is taken from the "event initiation frame." After the window ends, the initiating node collects information from all responding meters, forming a candidate node set, and performs a master node election within this set according to agreed-upon rules.

[0098] Arc detection typically uses the current half-cycle as the counting unit, which is 10ms under a 50Hz power grid. After an arc event occurs on the line, the time difference between when the nearby nodes sense the arc event and when it occurs is generally only a few counting units. Therefore, the first meter to detect the arc starts a timed window T1 after sensing the abnormal event locally. T1 should not be set too long, preferably 100ms. All responding meters within the timed window form a candidate node set, and the master node election is performed within this set according to the agreed rules.

[0099] The election rules can be to elect the earliest detector as the master node; alternatively, the arc intensity in the arc detection message can also be used as the basis; in a specific embodiment, the arc intensity includes the average signal strength of the target frequency band and the magnitude of the low-frequency change before and after the arc occurs; alternatively, in scenarios where PLC is used as the communication method, the network level of the meter itself can also be used as the election basis, and in principle, the lower the level of the meter, the closer it is to the upstream.

[0100] In step S14, the winning node is elected as the master node and broadcasts the master node declaration information, while the remaining nodes enter a listening / waiting state.

[0101] The "Master Node Declaration Message" contains the master node's device identifier or communication address. All meters in the candidate node set can use this address information to communicate with the master node. The "Master Node Declaration Message" may contain the configuration parameters of the notch filter to be used. In specific implementations, the arc detection message includes supported notch filter parameters and recommended notch filter parameters, such as the number of notch filter branches and corresponding frequency band information. Meter nodes can also recommend suitable notch filter parameters to the master node based on noise floor conditions. The master node integrates this information and packages the notch filter parameter information into the "Master Node Declaration Message" message. Upon receiving the "Master Node Declaration Message," the node extracts the notch filter parameter information to initialize the notch filter circuit.

[0102] Step S15: The master node extracts the arc intensity information from the arc discovery messages of the other nodes and constructs a low-frequency mutation matrix.

[0103] The master node extracts arc intensity information from the "arc discovery message" reported by the candidate nodes. As mentioned above, the arc intensity includes the average signal strength of the target frequency band and the magnitude of low-frequency abrupt changes before and after the arc occurs. A low-frequency abrupt change matrix D = [D] is constructed using the magnitude of the low-frequency abrupt changes before and after the arc occurs. i To facilitate subsequent correlation analysis, the meter node sequence involved in the low-frequency mutation matrix is ​​the same as the node sequence involved in notch filtering. Low-frequency current characteristic values ​​are calculated based on one or more key characteristics (e.g., the cumulative sum of current cycles before and after an arc, the rate of change of current before and after an arc, and the variance of the effective current value, etc.); and the normalized change of this characteristic value before and after the fault is calculated, denoted as D. i This value is used to quantify the intensity of low-frequency abrupt changes in a node's response to fault events, providing a basis for determining the structural relationships between nodes. The calculation method is as follows:

[0104]

[0105] in:

[0106] f: Feature number (e.g., current periodic cumulative sum, current change rate, etc.)

[0107] p: Total number of features

[0108] The low-frequency current characteristic value of the f-th feature of the i-th node before and after the fault.

[0109] ω f The weighting coefficient of the f-th feature.

[0110] ε: A tiny constant used to avoid division by zero.

[0111] Step S16: The master node initiates the alternating notch filtering process.

[0112] Due to the unstable nature of electric arcs and their limited duration, if too many nodes participate in the alternating notch filtering, the arc may extinguish prematurely. Therefore, after the master node initiates the alternating notch filtering task, it first obtains the set of electricity meters. If the number of electricity meters in the candidate set is greater than M, the candidate nodes need to be filtered. The following filtering strategies are available: sorting by the detected arc intensity. It can be understood that since the arc intensity includes the average signal strength of the target frequency band, nodes with lower intensity are farther away from the arc occurrence point. Another optional filtering strategy is: if the electricity meters use PLC communication, PLC network hierarchy sorting can be preferred to filter out the electricity meters that are closer to the master node hierarchy as the set of nodes participating in the alternating notch filtering. It can be understood that the farther the hierarchy is from the master node, the greater the physical distance.

[0113] Assume that N meter nodes in the candidate node set, including the master node, all sense the arc event. The master node controls each node in the candidate set to sequentially connect a notch filter circuit, causing attenuation in the target frequency band. The control method uses unicast messages. When a controlled node successfully connects the notch filter circuit, it broadcasts the message to the candidate node set, ensuring that the nodes in the candidate set know which node has connected the notch filter circuit. The notch filter circuit connection window is T2, preferably ranging from 100ms to 200ms. After the notch filter circuit is connected, multiple frequency point signal strengths are collected within the set notch filter bandwidth, using the notch filter center frequency as a reference, and the average signal strength of the frequency band is calculated as the notch filter response index, thereby quantifying the channel attenuation between nodes under the influence of the notch filter. During the notch filter window, the node samples and calculates the notch filter response index multiple times within the notch filter connection window, and reports the average value within the notch filter window (in units of current half-cycle, for example, once every 10ms, for a total of 10 to 20 samples).

[0114] When a meter node detects a fault arc signal, it will activate a frequency band monitoring mechanism to continuously sample and statistically analyze the signal strength within a preset target frequency band. Specifically, the node extracts frequency domain energy characteristics within a fixed time window, calculates the average signal strength of the target frequency band, and uses it as the baseline reference data for notch response analysis. In a specific embodiment, if the notch frequency band is a band centered at frequency f0 with a bandwidth of 2Δf, when the notch circuit is turned on, its equivalent impedance rises sharply within the target frequency band f∈[f0-Δf,f0+Δf]; if there are K frequency points {f1,f2,f...,f...} within the target frequency band... K The signal strength received by a certain node is S(f) k The average signal strength of the frequency band received by the node is...

[0115] During the notch filtering operations performed by other nodes, this node continues to synchronously measure the average intensity of the current frequency band and compare it with the reference value before notching, thereby constructing the relative attenuation index R in the notch response matrix. ij In a specific embodiment, R ij Let be the relative attenuation rate of the target frequency band signal strength observed by meter node j when meter node i is connected to the notch filter circuit; the calculation formula is as follows: Define the average signal strength of meter node j in the frequency band before meter node i is connected to the notch filter as . The period of notch filtering is but: This value can be interpreted as the "relative signal attenuation rate caused by notch filtering"; R ij Set to R for cases < 0 ij =0, which is considered an abnormal response and is ignored or handled separately; therefore, in R ij If R ∈ [0,1],ij A significantly larger value indicates that the signal strength of the target frequency band received by the j-th meter node was drastically attenuated during the notch filtering period at the i-th meter node. This suggests that the notch filtering circuit at the i-th meter node weakened the channel for the arc signal to propagate to the response node, or cut off the main propagation path. Therefore, the magnitude of the response value can serve as an important reference indicator for determining the physical path dependence between nodes and the location of the fault source.

[0116] Step S17: After the notch process is completed, the notch response indices of all participating nodes are obtained sequentially to construct the notch response matrix.

[0117] After the alternating notch filtering task is completed, the data analysis phase begins. The master node polls the notch response indices of each node to construct the response matrix R = [R ij ].

[0118] During the initiation of the alternating notch filtering task, the master node also needs to monitor the arc duration in real time. If the arc signal is detected to disappear prematurely, a partial response matrix is ​​constructed using the existing responses for analysis.

[0119] Step S18: The master node fuses the low-frequency mutation matrix and the notch response matrix to infer the location of the electric arc and reports it to the power grid master station.

[0120] The master node integrates the low-frequency mutation intensity matrix and the high-frequency notch response matrix for analysis. These two matrices provide a joint analysis of low-voltage line arc fault location from two dimensions: the physical structure of the power flow path and the propagation characteristics of high-frequency signals. In a specific embodiment, after constructing the low-frequency mutation matrix and the high-frequency notch response matrix before and after the fault, the possible connection relationship (series / parallel / common path) between the arc location and the meter node is first analyzed based on the low-frequency mutation matrix. Then, the possible coupling path of the arc signal is analyzed based on the response of each meter node before and after the notch. By combining the above two methods, scenario elimination is performed, the physical distance and relative position between the arc location and the meter node are analyzed, the arc location is inferred, and the inference result is reported to the master station system.

[0121] Example 2

[0122] Example 2 provides a specific instance of arc fault location in a specific power grid structure, following the method in Example 1.

[0123] Figure 2 A diagram of a low-voltage power distribution network structure according to an example embodiment is shown.

[0124] like Figure 2As shown, this is a possible low-voltage power distribution network structure in the field. In this network structure, meter 1 is observed to be located upstream of the current structure, with three meter nodes connected downstream. This structure can be considered a local branch of the entire power distribution network. Other meter nodes may still be connected upstream, and the downstream may further extend to form new branches or meter nodes. Therefore, if we assume that an electric arc occurs upstream of meter 1, and the arc signal flows through meter 1 and may be detected by several of meters 1, 2, 3, and 4; in traditional technical solutions, all meters that detect the arc signal will trigger an alarm or even disconnect the line; it would be difficult for maintenance personnel to locate the fault point during troubleshooting.

[0125] According to the method in Example 1, when a fault arc is detected in the line, if the notch filter circuit of control meter 1 is turned on during the arc's duration, the line exhibits a high impedance in the corresponding frequency band of the notch filter. This is manifested in the signal amplitude in the corresponding frequency band sensed by the downstream meter being significantly weakened. However, when the notch filter circuits of the downstream meters are turned on, the signal amplitude in the corresponding frequency band sensed by other nodes does not change significantly, thus inferring that the arc occurrence point is likely located upstream of meter 1.

[0126] In this embodiment, it is assumed that four meter nodes on a local line detect arc signals almost simultaneously. Generally, the arc may occur on the main line, or it may occur in a circuit after a certain meter, while other circuits detect the arc signal due to high-frequency signal coupling. In this situation, it is urgent to pinpoint the exact location, maintain the truly faulty circuit, and prevent normal circuits from being affected. In this example, the low-frequency mutation matrix D is shown in Table 1 below, and the high-frequency notch response matrix R is shown in Table 2.

[0127] Table 1 Low-frequency mutation matrix

[0128] Meter node i <![CDATA[Low-frequency mutant D i > Remark 1 0.41 Located on the main road 2 0.28 Main trunk branches, normal tributaries 3 0.62 Branch where an arc fault occurs 4 0.71 Normal branch, load disturbance

[0129] Table 2 High-frequency notch response matrix

[0130] Table 1 Response Table 2 Response Table 3 Response Table 4 Response Table 1 Notch Wave / 0.26 0.29 0.16 Table 2 Notch Filter 0.32 / 0.37 0.11 Table 3 Notch Wave 0.34 0.48 / 0.13 Table 4 Notch Wave 0.12 0.10 0.11 /

[0131] As shown in the low-frequency mutation matrix D in Table 1, the low-frequency mutations at meter nodes 3 and 4 are significant before and after the arc detection. It is preliminarily inferred that the arc likely occurred in the branch section shared by meters 3 and 4, which is downstream of the meter 2 connection point and upstream of the meter 3 and 4 connection point. Further analysis using the high-frequency notch response matrix R (as shown in Table 2) reveals that meters 3 and 4 show no response when notched to each other, and meter 4 also shows no significant response when notched to other nodes. This indicates that the signal coupling paths of meters 3 and 4 do not overlap, thus ruling out the possibility that the fault is located in the shared section between them. It is speculated that the low-frequency mutation detected by meter 4 is likely caused by current disturbances such as load startup. Combining the conclusions drawn from the analysis of the low-frequency mutation matrix D, it is inferred that the fault most likely occurred in the exclusive path of meter 3 (in the user circuit downstream of the meter).

[0132] Example 3

[0133] Example 3 provides another method for calculating the notch response index in Implementation 1, which is a relative estimation method that does not require a reference value before the notch.

[0134] Because arc signals exhibit broadband characteristics, the signal strength at adjacent frequency points can be approximated as being essentially the same. Therefore, it is not necessary to compare the values ​​before and after notch filtering. Instead, the signal strength within the notch band is compared with the strength difference in adjacent frequency bands to estimate the "relative depression" caused by the notch filtering. The notch filtering response index is defined as follows:

[0135]

[0136] Let J be the average signal strength in the frequency band at meter node j when the notch filter is applied at meter node i. The average signal strength of meter node j in the adjacent frequency band when the notch filter is applied at meter node i (the frequency band without the notch filter can be approximated as the normal propagation strength).

[0137] The advantage of this method is that it does not require snapshot data before notching. Since the measurement time before and after notching is inconsistent, comparing data before and after notching may be affected by factors such as transient disturbances in the power grid, load fluctuations, and changes in background noise. By directly comparing the power difference between the current notched frequency band and its adjacent non-notched frequency band, the deviation caused by the inconsistency of the background in the time period before and after notching is effectively avoided, thus improving the stability and accuracy of the response indicators.

[0138] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for arc crosstalk identification and localization based on distributed frequency domain sensing, characterized in that, Includes the following steps: A power grid node broadcasts a message, wherein the power grid node is the one that detected the electric arc, and the message is an arc detection message, which includes an event ID, a detection timestamp, and an arc intensity. All grid nodes that receive the broadcast collect and record the arc detection messages from other grid nodes; Candidate nodes elect a master node based on agreed-upon rules; The candidate nodes are all power grid nodes that receive the broadcast. The master node is used to identify and locate arc crosstalk. The agreed rules include electing the power grid node that detects the arc earliest as the master node, or electing the master node based on the arc intensity in the arc detection message. The master node broadcasts a master node declaration message, and the other nodes enter a listening / waiting state; the master node declaration message includes the master node's device identifier or communication address, and / or the notch filter configuration parameters to be used; The master node extracts the arc intensity information from the arc detection messages of the other nodes and constructs a low-frequency mutation matrix; The master node initiates the alternating notch filtering process; After the notch filtering process is completed, the notch response indices of all participating nodes are obtained sequentially to construct the notch response matrix; The master node integrates the low-frequency mutation matrix and the notch response matrix to infer the location of the electric arc and reports it to the power grid master station.

2. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The candidate nodes are determined using an event-aware window mechanism, the steps of which include: After the event initiating node detects the abnormal event of an electric arc locally, it starts a timed window and broadcasts an event initiation frame. The event initiating node is the first power grid node to detect the electric arc. The event initiation frame contains an arc discovery message with a unique event ID. After receiving the event initiation frame, the other power grid nodes determine whether they have also sensed the abnormal event of the arc. If so, they send an event response frame to the event initiation node within the time window. The event response frame contains an arc detection message indicating that the other power grid nodes have sensed the abnormal event of the arc. The event ID of the event response frame is taken from the event initiation frame. The event initiating node collects information on all response nodes that sent event response frames after the timed window ends, forming a candidate node set.

3. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The arc detection message also includes: supported notch filter parameters or recommended notch filter parameters; After receiving the declaration information from the master node, the remaining nodes extract the notch filter configuration parameters therein and initialize the notch filter circuit according to the extracted notch filter configuration parameters.

4. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The power grid node is equipped with an arc sensing unit and a notch filter circuit. The arc sensing unit includes: a current sensor, configured on the phase line, for acquiring the low-frequency components of the line current; and a high-frequency coupling circuit, for sensing and extracting the high-frequency noise components in the line voltage. The notch circuit is used to deeply attenuate the input signal within a set frequency range, thereby suppressing the signal propagation in that frequency band; the notch circuit includes one or more notch branches, and the notch branch includes a switch and a notch filter.

5. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The step of the master node extracting the arc intensity information from the arc detection messages of the other nodes and constructing the low-frequency mutation matrix includes: The arc intensity information includes the average signal intensity of the target frequency band and the magnitude of low-frequency abrupt changes before and after the arc occurs; A low-frequency mutation matrix is ​​constructed using the magnitude of low-frequency mutations before and after the occurrence of an electric arc, and the power grid node sequence involved in the low-frequency mutation matrix is ​​the same as the power grid node sequence involved in notch filtering. The low-frequency mutation size is used to quantify the intensity of low-frequency mutations in response to fault events, providing a basis for determining the structural relationships between nodes; the low-frequency mutation size D i Calculate using the following formula: in: f is the feature number, and p is the total number of features. Let ω represent the low-frequency current characteristic values ​​of the f-th feature of the i-th node before and after the fault. These low-frequency current characteristic values ​​are calculated based on one or more key features, including the cumulative sum of current cycles before and after the arc, the rate of change of current before and after the arc, and the variance of the effective current value. f ε is the weighting coefficient for the f-th feature, and ε is a small constant to avoid division by zero.

6. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, It also includes a process for filtering the candidate nodes: All the candidate nodes mentioned constitute a candidate node set; If the number of nodes in the candidate node set is greater than a preset value M, then the candidate nodes are filtered: The candidate nodes are sorted in descending order of the detected arc intensity, and the top M candidate nodes are selected as the nodes to participate in the alternating notch filter. Alternatively, the candidate nodes using PLC communication can be sorted by PLC network hierarchy, and the candidate nodes that are closer to the master node hierarchy can be selected as the nodes participating in the round-robin notch filtering.

7. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The alternating notch filtering process initiated by the master node includes: All the candidate nodes mentioned constitute a candidate node set; The master node uses unicast messages to control each node in the candidate node set to sequentially connect the notch circuit. When a controlled candidate node successfully connects the notch circuit, it broadcasts a message to the candidate node set. The message is used to ensure that the candidate nodes in the candidate set know who has connected the notch circuit. After the notch circuit is turned on, the signal strength of multiple frequency points is collected within the set notch bandwidth range, with the notch center frequency as the reference. The average signal strength of the frequency band is calculated as the notch response index, which is used to quantify the channel attenuation between nodes under the influence of the notch. The candidate node samples and calculates the notch response index multiple times within the notch circuit turn-on window, and reports the average value within the notch circuit turn-on window period.

8. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The notch response index is the relative attenuation rate of the target frequency band signal strength observed by grid node j when the notch circuit is connected at grid node i, denoted as R. ij , Define the average signal strength of grid node j before the notch filter is applied at grid node i as follows: The period of notch filtering is but: R ij Set to R for cases < 0 ij =0, which is considered an abnormal response and should be ignored or handled separately; If there are K frequency points {f1, f2, ..., f3} within the target frequency band K The signal strength received by grid node k is S(f) k The average signal strength of the power grid node k in the frequency band. Calculate using the following formula:

9. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The notch response index R ij for: In the formula, Let J be the average signal strength in the frequency band of grid node j when notch filtering is applied at grid node i. Let J be the average signal strength of grid node j in the adjacent frequency band when grid node i is connected to notch filtering.

10. The arc crosstalk identification and localization method based on distributed frequency domain sensing according to claim 1, characterized in that, The step of the master node fusing the low-frequency mutation matrix and the notch response matrix to infer the location of the electric arc includes: Based on the low-frequency mutation matrix, analyze the possible connection types between the arc occurrence location and the power grid nodes; Using the notch response matrix, the possible coupling paths of the arc signal are analyzed based on the response of each power grid node before and after the notch. By combining the above two methods to eliminate scenarios, analyzing the physical distance and relative position between the arc occurrence location and the power grid node, the location of the arc occurrence can be inferred.

Images