Low-voltage power distribution network fault active alarm system and method thereof

Abstract

The application discloses a low-voltage power distribution network fault active alarm system and method thereof, the system comprising an edge computing terminal and four levels of sensing devices deployed at the outgoing line side of a distribution transformer, the outgoing line side of a branch box, the incoming line side of a meter box and the user side; each sensing device is internally provided with a backup power supply, a power failure holding circuit and a data buffer area, and when voltage sag is detected, the power failure holding circuit is triggered, and the voltage and current waveform data of the last N cycles before power failure is actively uploaded to the edge computing terminal by the backup power supply; the active alarm module generates structured alarm information according to a fault event, selects a corresponding alarm channel according to a preset alarm priority, actively sends the alarm information to a cloud-based master station and / or an operation and maintenance terminal, and realizes hierarchical and multi-channel fault active alarm. The application realizes accurate positioning and active alarm of low-voltage power distribution network faults, and can be widely applied to the fields of fault repair and power supply reliability management of low-voltage power distribution networks.

Description

Technical Field

[0001] This invention relates to the field of power distribution automation and alarm technology, specifically to fault monitoring and active alarm technology for low-voltage power distribution networks, and particularly to an active fault alarm system and method for low-voltage power distribution networks. Background Technology

[0002] As the end-user segment of the power system, the reliability of low-voltage distribution networks directly affects users' electricity experience and the quality of power supply services. For a long time, emergency repairs of low-voltage distribution networks have mainly faced the following technical challenges:

[0003] First, fault alarm methods are passive and response is delayed. Currently, fault detection in low-voltage distribution networks mainly relies on user telephone reports or maintenance personnel inspections, resulting in a significant time blind spot between fault occurrence and alarm information generation. Although some solutions attempt to report power outages through smart meters or concentrators, these devices lose their power supply after a power outage and cannot proactively send complete alarm information containing fault characteristics at the moment of the outage. Consequently, the main station can only know that "a certain transformer area is out of power" but cannot determine the outage range and specific location. The alarm information is coarse-grained and cannot directly guide emergency repairs.

[0004] Second, the lack of accurate topology references in alarm information leads to false alarms and missed alarms. Frequent changes in the topology of low-voltage distribution networks cause discrepancies between the PMS system records and the actual situation. Existing alarm solutions either do not rely on topology (judging solely based on electrical quantity thresholds), resulting in poor alarm location accuracy; or they statically call outdated records, causing significant deviations in the fault location in the alarm information. Maintenance personnel still need to investigate segment by segment upon arrival at the site, greatly diminishing the practical value of the alarms.

[0005] Third, fault signals are difficult to trace effectively in hierarchical networks, and alarm information is contradictory. Low-voltage distribution networks pass through multiple levels from distribution transformers to users, and the electrical quantities of faults are attenuated and distorted as they flow through different levels. Existing technologies lack the ability to perform correlation analysis on alarm signals reported by devices at different levels, making it impossible to determine whether a power outage originates from a fault at this level or is caused by the spread of a fault at a higher level. This often results in a chaotic situation where "all outages are reported and all alarms are responded to," with alarm platforms receiving massive amounts of redundant alarms but unable to extract effective information.

[0006] Fourth, the alarm information is coarse-grained and cannot directly guide emergency repairs. Existing solutions often only provide a general description of "voltage loss in a certain phase of a certain transformer area," without specifying the fault location, the affected user range, alarm priority, or alarm channel. The power supply service command platform still requires manual analysis before dispatching a repair order, essentially failing to achieve "proactive alarm and precise push notifications." Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a proactive, hierarchical, multi-channel alarm system and method for low-voltage distribution network faults, which shortens the response time for fault repair.

[0008] To achieve the above objectives, the present invention employs the following solution:

[0009] A low-voltage distribution network fault active alarm system is characterized in that: the system includes an edge computing terminal, multiple levels of sensing devices, a topology identification module, a fault event determination module, and an active alarm module.

[0010] The edge computing terminal is deployed on the low-voltage side of the distribution transformer area and includes a dual-core heterogeneous processor, an AC sampling module, a multi-standard communication module, and a storage module. The edge computing terminal has a built-in containerized operating environment and deploys a topology identification app, a fault event determination app, and an active alarm app in a software-defined manner. The edge computing terminal communicates with the cloud-based main station via a remote communication network and with multiple levels of sensing devices via a local communication network. This architecture allows computational tasks such as topology identification, fault event determination, and alarm generation to be completed on the edge side, avoiding the communication and computational pressure caused by uploading massive amounts of raw data to the main station. Simultaneously, the containerized operating environment supports independent upgrades and canary releases of each functional app, ensuring the system's maintainability and continuous evolution capabilities.

[0011] The multiple levels of sensing devices are deployed at various nodes of the low-voltage distribution network, including: First-level sensing devices: low-voltage fault sensors deployed on the low-voltage outgoing line side of the distribution transformer, with built-in backup power supply, used to collect voltage, current and fault signals of the distribution transformer outgoing line; Second-level sensing devices: low-voltage fault sensors deployed at the outgoing line of the branch box, with built-in backup power supply, used to collect voltage, current and fault signals of the branch line; Third-level sensing devices: end monitoring units deployed on the incoming line side of the meter box, with built-in backup power supply, used to collect voltage and current signals and user power outage events at the meter box level; Fourth-level sensing devices: user smart meters, which interact with edge computing terminals through concentrators.

[0012] The aforementioned four levels of sensing devices form a complete sensing chain from the distribution transformer to the user. If a fault occurs at any level, all downstream sensing devices can detect the change in electrical quantities, providing full-path cross-sectional data from the power supply side to the user side for fault event determination. This achieves full coverage and no blind spots in low-voltage distribution network fault sensing. Each level of sensing device has a built-in backup power supply, enabling short-term operation even when external power is interrupted, ensuring the reliability of data reporting and alarm triggering during power outages.

[0013] Each level of sensing device has a built-in power outage retention circuit and a data buffer for the moment of power failure. When the device detects that the voltage has dropped below a threshold, it triggers the power outage retention circuit, which is powered by a backup power supply. The voltage and current waveform data of the last N cycles before the power outage, along with the power outage time marker, are transmitted to the edge computing terminal via the local communication network, where N is a preset positive integer. This mechanism enables the edge computing terminal to obtain complete electrical cross-sectional data at the moment of the fault—that is, the device actively transmits "what was the voltage and current at the last moment before the power outage"—instead of the traditional binary state of "I lost power." Even if the sensing device subsequently loses power completely, the key fault characteristic information has already been reported beforehand, thus providing a data-driven decision-making basis for the fault event determination module, thereby triggering the generation of alarm information. This fundamentally solves the defect in traditional solutions where the device loses its alarm triggering capability upon power failure.

[0014] The topology identification module, located within the edge computing terminal, initiates low-voltage distribution area topology identification during equipment updates or at set intervals. It identifies the relationship between households and transformers through voltage distortion signal injection and the feeder hierarchy through pulse current characteristic signals, generating and storing a real-time topology file of the low-voltage distribution network. This module enables the edge computing terminal to autonomously establish and update the real-time power network topology of low-voltage distribution areas without relying on PMS ledgers. The voltage distortion signal injection method solves the attribution problem of "which distribution area and phase does the equipment belong to," while the pulse current signal method solves the hierarchical relationship problem of "which equipment is upstream and which is downstream." The fusion of these two methods forms a complete tree-like topology structure, providing accurate power network references for fault location information and fundamentally eliminating the negative impact of ledger distortion on alarm location accuracy.

[0015] The fault event determination module is located within the edge computing terminal. It receives fault signals and pre-outage section data reported by multiple levels of sensing devices. Combining this with the real-time topology file, it locates the fault section and determines the fault type and a list of affected downstream sensing devices by tracing back from the user side to the distribution transformer side. This tracing-back logic simulates the thought process of engineers checking from the end upwards during on-site troubleshooting and formalizes it into an algorithm that can be automatically executed at the edge. Using the voltage data of each level of equipment at the time of the fault as an objective criterion, it compares the data level by level according to the topology, accurately determining whether the power outage originated from a fault in the current level of equipment or an interruption in power supply from an upstream level, providing accurate event source information to the alarm module.

[0016] The proactive alarm module, located within the edge computing terminal, generates structured alarm information based on the fault segment, fault type, and list of affected downstream sensing devices after the fault event determination module identifies a fault event. It then proactively sends the structured alarm information to the cloud-based main station and / or a pre-defined maintenance terminal, selecting the appropriate alarm channel according to a preset alarm priority, to trigger fault alarms and emergency repair notifications. This structured alarm information includes the fault device ID, the fault topology level, a description of the fault segment location, the fault type, the fault time, a list of affected downstream sensing devices, and the alarm priority. Upon receiving this information, the power supply service command platform can directly generate an emergency repair work order without manual analysis, achieving full-chain automation from "sensing → determination → alarm → work order dispatch," thus possessing genuine practical engineering value for proactive emergency repair.

[0017] Furthermore, the local communication network adopts a hybrid communication method combining LoRa, broadband power line carrier, and RS485: the edge computing terminal is connected to the near-end sensing devices within its installation cabinet via an RS485 bus or CAN bus, ensuring high reliability and real-time performance of communication within the cabinet; it is connected to the remote sensing devices via LoRa wireless communication or broadband power line carrier, solving the problems of wireless coverage and power line communication coverage for remote devices. These three communication methods complement each other, achieving wide coverage and low cost for low-voltage IoT communication.

[0018] Furthermore, the specific method by which the topology identification module identifies the relationship between households and transformers is as follows: the edge computing terminal applies a specifically coded modulation distortion signal near the zero-crossing point of the voltage of the three-phase (A, B, C) lines of the distribution transformer in the distribution area. This distortion signal cannot pass through the transformer to the 10kV busbar due to the filtering effect of the transformer winding impedance, and therefore can only be detected within the current distribution area. Each level of sensing device has a built-in zero-crossing detection circuit. After detecting the modulation distortion signal, it parses its code and feeds it back to the edge computing terminal. The edge computing terminal determines the distribution area and phase to which each sensing device belongs based on the feedback signal. This identification method utilizes the electrical characteristics of the distribution transformer itself as a natural signal isolation means, achieving signal isolation between distribution areas without additional equipment, and has the technical advantages of low implementation cost and reliable identification.

[0019] Furthermore, the specific method by which the topology identification module identifies the feeder hierarchy is as follows: the edge computing terminal sends a pulse current generation command to each of the currently identified sensing devices in sequence according to the current list of identified sensing devices. The sensing device receiving the command injects a pulse current signal with specific time-domain characteristics (such as a pulse width of 500μs and a peak current of 5A) into its access phase line. After detecting the pulse current signal, the sensing devices at each level upstream report to the edge computing terminal. The edge computing terminal constructs a signal propagation tree based on the signal propagation relationship, and the hierarchical relationship of this signal propagation tree corresponds to the feeder hierarchy of the low-voltage line. This identification method directly utilizes the physical topology characteristics of the low-voltage line, and the identification result is not affected by the line load condition or the change of transformer area load, thus having the technical advantage of high identification accuracy.

[0020] Furthermore, the judgment logic of the fault event determination module is as follows: When a fault occurs, the fault event determination module first collects a list of all sensing devices that reported cross-sectional data and fault signals before the power outage; then, starting from the highest-level reporting device, i.e., the one closest to the user, it traces upstream along the topology level by level, judging whether each level of device still has voltage data at the time of the fault, until the nearest upstream device that still has voltage at the time of the fault is found. The fault section is then the line section between that device and its next-level device without voltage. If a sensing device itself has no voltage at the time of the fault but did not report cross-sectional data before the power outage, then the device itself or its internal components are determined to be the fault point. This step-by-step backtracking logic uses the "presence or absence" of voltage data as the basic criterion, comparing step by step along the topology. It has the technical advantages of low computational load, clear judgment rules, and simple engineering implementation, and is suitable for real-time operation on edge computing terminals with limited computing resources.

[0021] Furthermore, the active alarm module includes an alarm information encapsulation unit, an alarm priority determination unit, and a multi-channel transmission unit; the alarm information encapsulation unit encapsulates the fault segment, fault type, fault time, and affected device information into a structured alarm message according to a preset data structure; the alarm priority determination unit determines the alarm priority based on the fault type, the fault topology level, the number of affected devices, and the fault impact range; the multi-channel transmission unit selects one or more of the following for alarm transmission: remote communication network, SMS platform, mobile work terminal, and / or instant messaging platform, based on the alarm priority.

[0022] A method for proactive fault alarm in low-voltage distribution networks based on the above system, characterized by the following steps: Step S1, Topology Establishment Step: Obtain the distribution area and phase affiliation and feeder hierarchy of sensing devices through voltage distortion signal injection and pulse current characteristic signal methods respectively, and establish a real-time low-voltage distribution network topology; Step S2, Fault Detection Step: Real-time collection of voltage, current and fault status information by sensing devices at each level, and triggering power outage retention when the voltage drops below a preset threshold, and proactively reporting the voltage and current waveform data of the last N cycles before the power outage and the power outage time marker; Step S3, Fault Judgment Step. The edge computing terminal determines the fault section, fault type, and affected area by tracing back level by level based on the pre-outage section data and fault status information reported by the sensing devices at each level and the real-time low-voltage distribution network topology. Step S4: Alarm generation step: Generate structured alarm information based on the fault section, fault type, and affected area, and determine the corresponding alarm priority. Step S5: Active alarm step: Select one or more alarm channels according to the alarm priority, and actively send the structured alarm information to the cloud master station, the distribution network mobile operation terminal, and / or the notification terminal corresponding to the affected user to realize fault alarm and emergency repair prompt.

[0023] The above method links topology identification, data acquisition, fault determination, and alarm push into a complete closed-loop process. Topology identification provides alarm location reference (step S1), hierarchical perception provides alarm trigger data source (step S2), proactive reporting of power outage sections solves the problem of missing alarm information (step S3), step-by-step back-tracking judgment achieves accurate event source identification (step S4), and structured alarm and hierarchical multi-channel transmission achieve effective information transmission (step S5). Each step is interconnected and together constitutes an end-to-end automated solution from fault occurrence to alarm push.

[0024] Furthermore, in step S4, the alarm priority is determined according to the fault type, the topology level of the fault, the number of downstream sensing devices affected, and the expected power outage range; wherein, the alarm priority is increased when the fault type is a short circuit fault, and the alarm priority is increased when the topology level of the fault is closer to the transformer side and the number of downstream sensing devices affected is greater.

[0025] Furthermore, in step S5, the active alarm module receives an acknowledgment from the alarm channel after sending an alarm message; if no acknowledgment is received within a preset time, the alarm message is resent according to a preset number of resends, and the alarm is switched to another alarm channel according to the alarm priority; when at least one alarm channel returns an acknowledgment, the alarm sending time, alarm channel, acknowledgment status, and number of resends are recorded to form an alarm sending record.

[0026] Furthermore, in step S5, after receiving the structured alarm information, the cloud-based master station determines the scope of affected users based on the fault section and the list of affected downstream sensing devices, and generates a corresponding proactive repair work order; at the same time, it sends the fault location, fault type, fault time and expected power outage range to the distribution network mobile operation terminal according to the alarm priority, and sends a power outage notification containing the expected power outage range and fault status to the affected users.

[0027] Compared with existing technologies, this invention has the following advantages: Each level of sensing device in this invention is equipped with a backup power supply and a power outage retention circuit. Before the equipment loses power, it automatically reports the electrical cross-sectional data at the moment of the fault. The edge computing terminal then actively generates an alarm based on this data, transforming fault detection from "verification after user report" to "active alarm at the moment of equipment power failure," eliminating the time blind spot in alarm sensing. Through topology identification methods based on two independent physical principles—voltage distortion signal injection and pulse current signal—the edge computing terminal can autonomously establish a real-time electrical network topology, ensuring that the fault location in the alarm information is always based on an accurate electrical network reference, fundamentally eliminating the impact of ledger distortion on alarm location accuracy. The fault determination result is converted into a data structure including hierarchical coding and device ID columns. Structured alarm information, including elements such as the fault table and the scope of impact, allows the power supply service command platform to directly dispatch orders without manual analysis, achieving full-chain automation of "perception → judgment → alarm → order dispatch". Alarm priorities are dynamically determined based on the severity of the fault and the scope of impact, and one or more channels are intelligently selected from remote communication networks, SMS platforms, mobile work terminals, and instant messaging platforms to send alarms, ensuring real-time delivery of critical alarm information while avoiding excessive channel resource consumption by low-priority alarms. Topology identification and fault determination calculations are all completed on the edge computing terminal; only structured location result alarm information, rather than raw waveform data, is sent to the main station, effectively saving remote communication network bandwidth resources and ensuring the real-time transmission of alarm information. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall architecture of the low-voltage distribution network fault active alarm system of the present invention.

[0029] Figure 2 This is a schematic diagram showing the deployment locations of the sensing devices at each level in the low-voltage power distribution network of the present invention.

[0030] Figure 3 This is a block diagram of the module structure of the edge computing terminal in the system of this invention.

[0031] Figure 4 This is a flowchart illustrating the active alarm method for low-voltage distribution network faults according to the present invention.

[0032] Figure 5 This is a flowchart of the fault location step-by-step backtracking analysis logic of the present invention.

[0033] Figure 6 This is a schematic diagram illustrating the principle of identifying the relationship between households and transformers using the voltage distortion signal injection method in the topology identification of this invention.

[0034] Figure 7 This is a schematic diagram illustrating the principle of identifying feeder hierarchical relationships using the pulse current characteristic signal method in topology identification of this invention.

[0035] Figure 8 This is a timing diagram illustrating the proactive reporting of power outage section data by the sensing device in this invention.

[0036] Figure 9 This is a schematic diagram of the data structure of the structured alarm information of the present invention. Detailed Implementation

[0037] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. These embodiments are for illustrative purposes only and do not constitute a limitation on the scope of the invention.

[0038] Example 1: System Overall Architecture and Device-Level Deployment

[0039] Please see Figure 1 As shown, the present invention provides a low-voltage distribution network fault active alarm system, which includes an edge computing terminal (i.e., a smart distribution transformer terminal) and multiple sensing devices deployed at various levels of the distribution area.

[0040] Edge computing terminals are deployed on the low-voltage side of distribution transformer areas (such as within JP cabinets). They include a dual-core heterogeneous processor (one core for management and non-real-time computing, and the other for building a hard real-time operating environment), an AC sampling module (for collecting voltage and current data from the distribution transformer side), a multi-standard communication module (supporting RS485, broadband carrier, LoRa, 4G / fiber optic, and other communication methods), and a storage module (for storing topology files and fault records). The edge computing terminal communicates with the cloud-based master station via a remote communication network (4G / fiber optic) and with multiple levels of sensing devices via a local communication network.

[0041] like Figure 2 As shown, multiple layers of sensing devices are deployed according to the actual physical hierarchy of the low-voltage distribution network:

[0042] The first level of sensing equipment—the low-voltage fault sensor—is installed at the outgoing line of the universal circuit breaker on the low-voltage outgoing line side of the distribution transformer. Each outgoing circuit breaker is equipped with one set of low-voltage fault sensors, which are mounted on the outgoing cable using an open CT method. Power is obtained through puncture, and a backup battery (such as a 1000mAh lithium battery) is built-in. This level of low-voltage fault sensors collects the three-phase voltage, three-phase current, active power, reactive power, power factor, and fault alarm signals of each outgoing line of the distribution transformer.

[0043] The second-level sensing device—the low-voltage fault sensor—is installed at each outgoing switch in the branch box. It is also installed using an open-type CT kit with puncture-based power extraction, and has a built-in backup power supply. It collects voltage, current, and fault signals from the branch lines and communicates with the edge computing terminal via LoRa wireless communication (operating frequency 470MHz, transmission rate 5.4kbps).

[0044] The third-level sensing device—the end-point monitoring unit—is installed on the incoming line side of each meter box. It has a built-in backup power supply (designed life of 8 to 10 years) and communicates with the edge computing terminal via broadband power line carrier. The end-point monitoring unit collects the voltage and current of the meter box's incoming line and monitors power outage events for each user within the box, featuring data buffering and proactive reporting functions during power outages.

[0045] The fourth level of sensing devices is the user's smart meter. After the data is aggregated via an RS485 bus through a Type I concentrator, the user's smart meter interacts with the edge computing terminal and reports the voltage, current, power, and power outage / restoration events of each user.

[0046] The aforementioned four levels of sensing devices form a complete sensing chain from the distribution transformer to the user. If a fault occurs at any level, the sensing devices at all levels below it can detect the change in electrical quantities, providing sufficient cross-sectional data for subsequent fault location.

[0047] In this embodiment, the local communication network adopts a hybrid communication method: devices within the JP cabinet (near-end devices such as smart capacitors and low-voltage smart switches) are connected to the edge computing terminal via RS485 bus using the Modbus protocol; low-voltage fault sensors (remote devices) in the branch box are connected wirelessly via LoRa; and end-of-line monitoring units (remote devices) in the meter box are connected via broadband power line carrier. These three communication methods complement each other: LoRa solves the problems of wireless coverage and low-cost access, broadband carrier solves the problem of power line communication coverage, and RS485 solves the problem of high-reliability communication within the cabinet.

[0048] The beneficial effects of this embodiment are as follows: by deploying hierarchical sensing devices at various nodes of the low-voltage distribution network, full-path electrical quantity monitoring from distribution transformers to users is realized, providing a complete data foundation for fault location; through the combined use of hybrid communication methods, wide coverage and low cost are achieved while ensuring communication reliability.

[0049] Example 2: Dynamic Topology Identification – Household-Transformer Relationship Identification

[0050] This embodiment details how the present invention achieves dynamic identification of the relationship between households and transformers through a voltage distortion signal injection method.

[0051] like Figure 6As shown, the edge computing terminal is electrically connected to the three phases (A, B, C) of the low-voltage side of the distribution transformer in the area. During the topology identification process, the edge computing terminal injects voltage distortion signals according to the following steps:

[0052] Step 1: The edge computing terminal detects the zero-crossing points of the three-phase voltages on the low-voltage side of the distribution transformer and marks the zero-crossing times of phases A, B, and C respectively.

[0053] Step 2: 1ms after the zero-crossing point of phase A voltage, a modulation load is connected between phase A and the neutral line through a power electronic switch. The connection of the load causes a voltage dip of 200μs to be generated near the zero-crossing point of phase A voltage. This dip is used as the modulation distortion signal.

[0054] Step 3: The modulated distortion signal is modulated on and off according to a preset coding sequence. The coding sequence is in Manchester coding format and includes station identification information and phase identification information.

[0055] Step 4: Since the lines of the same phase in the same distribution area are electrically directly connected, the distorted signal is transmitted along the A-phase conductor to each branch box and meter box within the distribution area. Simultaneously, due to the low-pass filtering effect of the distribution transformer's winding impedance on the high-frequency distorted signal (the transformer leakage reactance exhibits high impedance for non-power frequency signals), the distorted signal is significantly attenuated when passing through the transformer and cannot be transmitted to the 10kV busbar. Therefore, adjacent distribution areas do not receive this signal.

[0056] Step 5: Each level of sensing device has a built-in zero-crossing detection circuit (implemented based on a low-cost comparator, such as the LM339 comparator chip). After detecting a voltage zero-crossing point, the zero-crossing detection circuit continuously monitors for any abnormal voltage dips near the zero-crossing point. When the coded sequence is detected, the device reports the decoded area identifier and phase identifier to the edge computing terminal via the local communication network;

[0057] Step 6: The edge computing terminal summarizes all reported decoding results to determine the station area and phase to which each sensing device belongs.

[0058] Taking a specific transformer substation as an example: The A-phase outgoing line of substation #3 is connected to branch boxes 1 (K111, K112, K113) and branch boxes 2 (K121, K122). When the edge computing terminal injects a modulated distorted signal encoded as "3#-A" into phase A, the second-level sensing devices at K111, K112, K113, K121, and K122 all detect the signal and report the decoding result "3#-A". However, the sensing devices on phases B and C of the same substation do not detect the signal; nor do any sensing devices in the adjacent #4 substation. Based on this, the edge computing terminal confirms that the outgoing lines of the above five branch boxes all belong to phase A of substation #3. The actual ledger records are completely consistent with this, verifying the accuracy of the identification.

[0059] The advantages of this embodiment are: it utilizes the electrical characteristics of the distribution transformer itself as a natural signal isolation boundary, achieving signal isolation between stations without additional equipment; each sensing device only requires a low-cost zero-crossing detection circuit for reliable identification, avoiding the cost increase brought by high-precision ADCs; this scheme can be combined with a carrier system to achieve 100% reliable identification of stations and phases.

[0060] Example 3: Dynamic Topology Identification – Feeder Hierarchy Relationship Identification

[0061] This embodiment details how the present invention achieves dynamic identification of feeder hierarchy using the pulse current characteristic signal method.

[0062] like Figure 7 As shown, after completing the identification of household-transformer relationships, the edge computing terminal has a list of all subordinate sensing devices. Based on this, it performs feeder hierarchy relationship identification, with the following specific steps:

[0063] Step 1: The edge computing terminal obtains a list of all currently identified sensing devices and sorts them by device type (first level → second level → third level).

[0064] Step 2: The edge computing terminal sends a "pulse generation command" to each sensing device in the list in sequence. The command includes the amplitude requirement of the pulse current and the time domain characteristic parameters (pulse width 500μs, peak current 5A).

[0065] Step 3: Upon receiving the instruction, the sensing device injects a 500μs wide, 5A peak pulse current signal into its connected phase line through its internal pulse current generation circuit. This signal propagates along the line in two directions: upstream (towards the power supply side) and downstream (towards the user side).

[0066] Step 4: Since the pulse current signal has clear time-domain characteristics (500μs pulse width, specific rise edge steepness), other sensing devices installed on the same line will report the pulse current signal to the edge computing terminal after detecting it. The report content includes "a pulse signal from a certain device has been detected".

[0067] Step 5: The edge computing terminal aggregates the reported information from all sensing devices and constructs a signal propagation tree starting from the transmitting device. All devices that detect the pulse signal are downstream nodes of the transmitting device. The structure of the signal propagation tree reflects the actual feeder hierarchy of the low-voltage line.

[0068] Step 6: The edge computing terminal repeats steps 2 to 5 for each sensing device in the list in turn. After completing the pulse injection and detection feedback of all devices, it summarizes all signal propagation trees to form a complete feeder hierarchical topology structure for the transformer area.

[0069] Taking a specific transformer substation as an example: The edge computing terminal first sends a pulse generation command to the first-level sensing device at D11. After D11 injects the pulse signal, the second-level sensing devices K111, K112, and K113, as well as the third-level sensing devices B1111 and B1112, all detect the signal and report it. Subsequently, the edge computing terminal sends a pulse generation command to the second-level sensing device at K111. After K111 injects the pulse signal, the third-level sensing devices B1111 and B1112 detect the signal and report it, but K112 and K113 do not detect the signal. Therefore, the edge computing terminal determines that B1111 and B1112 are downstream of K111, while K111, K112, and K113 are downstream of D11 but connected in parallel. Through device-by-device injection and feedback collection, a complete tree-like topology is ultimately constructed.

[0070] The beneficial effects of this embodiment are as follows: the pulse current signal propagates directly along the physical path of the line, unaffected by the load conditions of the transformer area and changes in line impedance, and the identification result accurately reflects the actual connection relationship of the line; by injecting and detecting each device in sequence, the branch hierarchical structure of the line can be accurately identified, overcoming the problem of weak branch structure identification capability of the single carrier scheme.

[0071] Example 4: Dual-signal fusion verification

[0072] This embodiment illustrates how the present invention fuses and verifies the identification results of the voltage distortion signal injection method and the pulse current characteristic signal method.

[0073] In a certain operating transformer area, the voltage distortion signal injection method showed that the decoding result reported by device S001 (the second-level sensing device) was "Transformer Area 3 - Phase A", while the pulse current signal method showed that device S001 was located in a branch of the Phase B line. The two methods showed inconsistencies in phase assignment.

[0074] The topology identification module of the edge computing terminal processes data according to the following procedure:

[0075] Step 1: Mark device S001 as "topology anomaly pending confirmation" and do not include it in the formal topology structure;

[0076] Step 2: In the next round of timed topology identification (3:00 AM the next day), the device will be subject to in-depth verification—while simultaneously increasing the injection power of voltage distortion signals and the detection sensitivity of pulse current signals;

[0077] Step 3: The second round of identification results shows that the voltage distortion method still identifies it as "3# transformer area - phase A", and the pulse current method still identifies it as "phase B line branch".

[0078] Step 4: The edge computing terminal retrieves the installation record and wiring file of device S001 and finds that the device is actually installed on phase A line. It confirms that the result of the voltage distortion method is correct and the result of the pulse current method is a misjudgment (possibly due to line induction).

[0079] Step 5: Based on the A-phase assignment of the voltage distortion signal method, correct the result of the pulse current signal method, formally assign device S001 to the A-phase topology, and generate a topology change log and send it to the master station.

[0080] In another scenario, the two identification methods for device S002 yielded consistent results—the voltage distortion method identified it as "3# transformer area - phase C", while the pulse current method identified it as "phase C line branch". The edge computing terminal directly confirmed that the device belonged to phase C of 3# transformer area and incorporated it into the formal topology.

[0081] The beneficial effects of this embodiment are as follows: the dual-signal fusion verification mechanism can effectively identify and correct misjudgments by a single method; the rule of confirming only when the results of two consecutive rounds of identification are consistent reduces the misjudgment rate; and the mechanism of automatically marking anomalies and focusing on review when the results of the two methods are inconsistent ensures the reliability of topology identification. Field tests have shown that this fusion verification method increases the accuracy of topology identification from approximately 95% with a single method to over 99%.

[0082] Example 5: Active Reporting of Power Outage Section Data

[0083] This embodiment takes a single-phase short-circuit fault in a branch line as an example to explain in detail the specific implementation of the active reporting of power outage section data by the sensing device in this invention.

[0084] Scenario setting: A phase A metallic short circuit fault occurs on the line section between phase A outgoing line D1 of transformer substation #3 and branch box 1 K111 outgoing line to meter box B1111. Both circuit breakers D1 and K111 trip within 40ms after the fault.

[0085] The specific responses of each level of sensing equipment after a fault occurs are as follows:

[0086] (a) Response of the first-level sensing device (at point D1):

[0087] The low-voltage fault sensor at D1 has a real-time sampling rate of 80 points per cycle (sampling frequency 4kHz), and its internal loop buffer stores the sampling data of the most recent 20 cycles (400ms). At t0+40ms, the device detects that the phase A voltage has dropped from 220V to 62V (approximately 28% of the rated value), triggering the power failure holding circuit. The device performs the following operations:

[0088] Stop writing new data to the circular buffer and lock the data of the first 10 cycles (200ms) before time t0 in the buffer as the cross-sectional data before the power outage;

[0089] A backup battery (1000mAh lithium battery) provides immediate power to maintain the operation of the MCU and communication module;

[0090] Package the cross-sectional data—including the device ID "D1_Sensor" and the power outage time marker. (Unix timestamp), Phase A voltage sequence (200 sampling points, 16-bit precision per point), Phase A current sequence (200 sampling points);

[0091] Data packets are sent to the edge computing terminal via a broadband carrier communication module; the data packet size is approximately 2.4KB.

[0092] Wait for the ACK confirmation from the edge computing terminal. Once received, enter sleep mode. If not received within 50ms, retransmit once, up to a maximum of 3 times.

[0093] (ii) Response of the second-level sensing device (at K111):

[0094] The low-voltage fault sensor at K111 performs the same operation as at D1, but the cross-sectional data characteristics are different—the voltage of phase A drops to 0 (due to the tripping of the K111 circuit breaker), and the current of phase A is recorded at the peak short-circuit current (approximately 8.2 times the rated current). The device transmits the cross-sectional data to the edge computing terminal via LoRa wireless communication.

[0095] (iii) Response of the third-level sensing device (end-point monitoring unit at B1111):

[0096] The end monitoring unit at B1111 detected a voltage drop to 77V (35% of the rated value), with no overcurrent record. The device transmits the cross-sectional data via broadband power line carrier, and the data characteristics are "voltage drops sharply but is not zero, and the current decays to 0r after returning to normal".

[0097] (iv) Collection status of edge computing terminals:

[0098] from Starting at +45ms, the edge computing terminal successively receives cross-sectional data reports from various sensing devices:

[0099] +45ms: Received report from D1 (wideband carrier, delay approximately 5ms);

[0100] +120ms: Received report from K111 (LoRa wireless, latency approximately 80ms).

[0101] +180ms: Received report from B1111 (wideband carrier, delay of about 5ms, but device response is slightly slower).

[0102] +500ms: Received reports from all 21 end-of-cell monitoring units and 3 outgoing sensors.

[0103] The beneficial effect of this embodiment is that the mechanism by which sensing devices at each level actively send buffered waveform data using backup power at the moment of power failure ensures that the edge computing terminal can obtain complete fault section information—including the precise voltage and current values ​​of each level of device at the moment of fault occurrence—rather than only receiving indirect information such as "a certain device is offline." This provides direct data support for subsequent accurate positioning and overcomes the fundamental defect of traditional solutions where devices lose their reporting capability upon power failure.

[0104] Example 6: Complete Location Process of Single-Phase Short-Circuit Fault

[0105] This embodiment uses the short-circuit fault described in Embodiment 5 as a background example to explain in detail the complete judgment process of the edge computing terminal fault location module.

[0106] The cross-sectional data collected by the fault location module after the fault occurred are summarized as follows:

[0107]

[0108] The fault location module performs a step-by-step backtracking:

[0109] The first round – starting from the user side (fourth layer):

[0110] The user's electricity meter reports "no voltage, no current," but this is true for all user meters. Because the user's electricity meter does not have the ability to buffer power outages (it reports power outage events from the concentrator), these reports are paused, awaiting data from the upstream equipment.

[0111] Second round – ascending to the third level:

[0112] The cross-sectional data of the B1111 terminal monitoring unit shows: voltage 77V (non-zero) and current normal. The cross-sectional data of the B1112 and B1113 terminal monitoring units show: voltage 0V and current 0A.

[0113] The fault location module discovered a key difference: B1111 had residual voltage (77V), while B1112 and B1113 had no voltage. This difference indicates that the upstream of B1111 (K111) still had voltage output at the time of the fault (although it dropped, it did not reach 0), while the upstream of B1112 and B1113 (K112, K113) had no voltage at the time of the fault. It is speculated that an anomaly occurred in the K111 branch, while the power loss in the K112 and K113 branches was caused by the tripping of D1 (upstream power outage).

[0114] Third round – Ascending to the second level:

[0115] The cross-sectional data of the K111 sensing device shows: voltage 0V (but note that the 0V of K111 is because its circuit breaker has tripped, not because there is no power upstream), current 0A, and no overcurrent record.

[0116] The cross-sectional data of the K112 and K113 sensing devices show: voltage 0V, current 0A, and no overcurrent record.

[0117] However, the fault location module retrieves K111 in The buffered data from the previous cycle was used to find its voltage at the time point. It is still 215V at -20ms (normal). The current suddenly dropped to 0, and a short-circuit impact was recorded. This indicates that the "0V" of K111 was due to its own tripping after a short circuit, rather than a power outage upstream.

[0118] Fourth round – Verifying upstream D1:

[0119] The cross-sectional data from the D1 sensing device shows: voltage 62V (non-zero, although there was a drop, the power supply is still working), and the current recorded is the short-circuit current (8.2 times the rated value).

[0120] Final positioning conclusion:

[0121] The fault location module determined that D1 has voltage (although the voltage drop occurred, the power supply was not interrupted), its downstream K111 has no voltage and a short-circuit impact was recorded, and B1111 has residual voltage (77V, the value after attenuation by the line impedance). The short circuit point is located between the K111 outgoing line and the B1111 incoming line (on the line segment from K111 to B1111), for the following reasons:

[0122] A short-circuit impact was recorded at K111 (indicating that the short-circuit point is downstream of it).

[0123] There is residual voltage at B1111 but no short-circuit current (indicating that B1111 is downstream of the short-circuit point, experiencing a voltage drop but the short-circuit current does not flow through the measurement point of B1111, which is consistent with the electrical characteristic that "the fault point is located between K111 and B1111").

[0124] There is voltage at D1 and the short-circuit current was recorded (confirming it was a short-circuit fault rather than a broken wire).

[0125] The location result output is: "The fault section is "the A-phase line section between the K111 outgoing line of branch box 1 and the B1111 incoming line of meter box", and the fault type is "A-phase metallic short circuit".

[0126] The beneficial effects of this embodiment are as follows: by tracing back step by step and comparing data across levels, the fault location module can infer the precise location of the fault point from the independent cross-sectional data of each level of equipment. This method does not rely on complex electrical quantity feature extraction algorithms, but only uses the presence or absence of voltage data and the presence or absence of short-circuit impact as the criteria for judgment. It has a small amount of calculation, clear judgment rules, and accurate location results.

[0127] Example 7: Differentiating and Locating Open-Circuit Faults and Short-Circuit Faults

[0128] This embodiment illustrates how the present invention distinguishes between two different types of faults, namely open-circuit faults and short-circuit faults, based on the key feature of "whether an overcurrent is recorded".

[0129] Scenario A – Open Circuit Fault (Compared with Example 5):

[0130] Assume that the connection terminal between the A-phase busbar inside branch box 1 and the outgoing switch K111 is burned and melted, causing the A-phase outgoing line of K111 to lose power, but no short-circuit current is generated.

[0131] The cross-sectional data characteristics of each level of sensing equipment are as follows:

[0132]

[0133] The fault location module performs the analysis:

[0134] K111 has no voltage (voltage drops to 0) and no overcurrent record (unlike the case of short-circuit impact in Example 5).

[0135] D1 has voltage (normal) and no overcurrent record (unlike in Example 5 where there is short-circuit current).

[0136] The voltage drop of K111 is "a direct drop from normal to 0", rather than "a sudden drop to some intermediate value".

[0137] Location conclusion: The faulty section is "the K111 outgoing switch inside branch box 1 or its upstream connection terminal", and the fault type is "phase A open circuit / poor contact".

[0138] Scenario B – User-side device failure:

[0139] Suppose that the A-phase metering chip inside user Zhang San's smart meter is damaged.

[0140] The characteristics of the cross-sectional data are:

[0141] The voltages of D1, K111, and B1111 are all normal (220V).

[0142] Zhang San's meter reported a voltage loss in phase A, but there was no cross-sectional data (the equipment was not powered off).

[0143] The voltage of other users' meters in the same meter box is normal.

[0144] Location conclusion: User-side equipment malfunction, not a power supply line fault.

[0145] The beneficial effect of this embodiment is that the present invention can accurately distinguish three different types of faults based on the electrical characteristic quantity of "whether there is overcurrent", namely, short circuit fault (with overcurrent), open circuit fault (no overcurrent but power failure) and user-side equipment abnormality (upper level is normal but the end is abnormal), thus avoiding invalid emergency repairs caused by misreporting user-side equipment as line faults.

[0146] Example 8: Structured Alarm Information and Automatic Work Order Dispatch

[0147] This embodiment details the data structure of the structured alarm information generated by the present invention and its parsing application on the power supply service command platform side.

[0148] like Figure 9 As shown, the structured alarm information is encapsulated in JSON format and contains the following key fields:

[0149] fault_id: A unique identifier for the fault;

[0150] fault_type: Fault type code (e.g., "001" represents a metallic short circuit);

[0151] fault_location: contains a hierarchy description and a list of devices;

[0152] affected_devices: A list of downstream sensing device IDs that have been affected;

[0153] priority: Alarm priority (levels 1-3, with level 1 being the highest);

[0154] timestamp: The time when the fault occurred.

[0155] After generating the JSON message, the active alarm module executes the following multi-channel sending strategy:

[0156] When the priority is level 1 (such as a short circuit at the distribution transformer outlet with a large impact range), the remote communication network (4G / fiber) is selected to send the data to the main station, the SMS platform is used to notify the emergency repair team leader, and the mobile operation terminal is used to push the work order.

[0157] When the priority is level 2 (such as a branch box outgoing line failure affecting some users), select remote communication network to send to the main station and mobile operation terminal for push;

[0158] When the priority is level 3 (such as a single meter box failure), the data is sent to the main station only via remote communication network, and the corresponding grid worker is notified at the same time via instant messaging platform.

[0159] After receiving the structured alarm information, the power supply service command platform parses the JSON and associates it with user data from the marketing system, automatically generating a repair work order. Simultaneously, the SMS platform sends power outage notifications to affected users. The entire process, from the occurrence of the fault to the alarm reaching all terminals, takes approximately 2-3 minutes, more than 10 times more efficient than the traditional manual fault reporting method.

[0160] The processing procedure after the power supply service command platform receives this structured alarm information is as follows:

[0161] Step 1: Parse the JSON data, extract the level_description from the fault_location, and obtain the directly readable text description of the fault section: "3# transformer area - No. 1 distribution transformer outgoing line - No. 111 branch box outgoing line to the lower level meter box";

[0162] Step 2: Extract the affected_devices list, query it in conjunction with the user profile database of the marketing system, and obtain the names, contact numbers and addresses of 12 affected users;

[0163] Step 3: Based on the type_code ("001") in fault_type, match the preset emergency repair resource template - "Metallic short circuit" matches "Emergency repair team type A+ requires tool kit X (insulation tester, multimeter, cable fault locator)";

[0164] Step 4: Automatically generate an emergency repair work order. The work order includes the fault location, fault type, list of affected users, and recommended emergency repair resources. It is then pushed to the distribution network mobile operation terminal (PDA) of the emergency repair team closest to the fault location.

[0165] Step 5: At the same time, based on the list of affected users, a power outage notification was sent to 12 users via SMS and WeChat. The notification read: "Dear power users, your area (between K111 and B1111) is currently under repair due to a short circuit fault in phase A. The estimated restoration time is yet to be determined. We apologize for any inconvenience caused."

[0166] The entire process described above, from the occurrence of a fault to the dispatch of a work order and notification to the user, takes approximately 2 to 3 minutes. In contrast, the traditional solution typically takes more than 30 minutes from user report of a fault to manual assessment and work order dispatch, and users can only check the progress by phone.

[0167] The beneficial effects of this embodiment are as follows: the structured alarm information contains all the key elements for fault location, and the power supply service command platform can directly dispatch orders after parsing it without the need for manual analysis; the accurate provision of the list of affected users enables power outage notices to be accurately pushed to the truly affected users, avoiding false and missed notifications.

[0168] Example 9: Adaptive Topology Identification After Adding a New Device

[0169] This embodiment illustrates the adaptive update capability of the topology identification module of the present invention when equipment changes (addition of meter boxes) occur in a low-voltage distribution network.

[0170] Scenario: In a residential community, a new meter box B1114 (containing 8 households) was added at the end of the K111 outgoing line of branch box 1 in the No. 3 transformer substation of a certain residential community, based on the existing topology. After the construction was completed, the maintenance personnel marked the "equipment change" event in the system.

[0171] After the edge computing terminal detects a device change marker, it executes the following process in the next timed identification cycle (or immediately):

[0172] Step 1: The edge computing terminal executes the voltage distortion signal injection method—injecting coded distortion signals into the three phases ABC and collecting feedback from all sensing devices;

[0173] Step 2: A new device, terminal monitoring unit B1114, was reported in the feedback list (device ID "B1114_EMU", decoded result "3# area - phase A"). This device did not exist in the previous topology.

[0174] Step 3: The edge computing terminal executes the pulse current characteristic signal method—sending pulse generation commands to each sensing device in sequence;

[0175] Step 4: When a pulse command is sent to the second-level sensing device at K111, the new device B1114 detects the pulse signal and reports it; however, when a pulse command is sent to K112, B1114 does not detect the signal.

[0176] Step 5: Edge computing terminal determination - B1114 is located downstream of K111 and belongs to the same branch as B1111, B1112, and B1113;

[0177] Step 6: The new topology is “3# transformer area → D1 → K111 → {B1111, B1112, B1113, B1114}”. The edge computing terminal stores the updated topology file and generates a topology change log, which is then sent to the cloud-based main station.

[0178] Step 7: The cloud-based main site compares and verifies the change log with the PMS2.0 ledger. If there is no record of B1114 in the ledger, the ledger correction process is initiated.

[0179] The entire adaptive identification process, from device change marking to topology update completion, takes about 5 minutes (including communication and computing time), without the need for manual on-site surveying and record entry.

[0180] The beneficial effects of this embodiment are as follows: the topology identification module can automatically detect the access of new devices and update the topology structure without manual intervention; the edge computing terminal always maintains the latest topology structure, providing an accurate reference for subsequent fault location; the automatic uploading of topology change logs also promotes the synchronous update of PMS ledgers, forming a virtuous cycle of "edge-side perception → topology update → ledger synchronization".

[0181] Example 10: Local Independent Location When Communication Link is Abnormal

[0182] This embodiment illustrates the robust advantage of the present invention, which allows the edge computing terminal to independently locate faults even when a remote communication link is interrupted.

[0183] The scenario is the same as in Example 6, but the 4G link is interrupted. After the fault occurs, the edge computing terminal performs the judgment locally and generates structured alarm information, which is then stored in the queue to be sent. The link is restored after 45 minutes, and the active alarm module immediately sends the alarm according to priority. The main station can still dispatch repair orders upon receiving the alarm. Traditional solutions cannot generate alarms at all when the link is interrupted; this invention achieves reliable alarms that do not depend on the remote link.

[0184] Scenario Setting: One day, the 4G remote communication link between transformer substation #3 and the cloud-based main station was interrupted due to a network failure of the operator. During this period, a short-circuit fault, similar to that in Example 5, occurred within the transformer substation.

[0185] The troubleshooting process after a malfunction occurs:

[0186] Step 1: Sensing devices at each level collect and report section data according to the normal process (link interruption does not affect the local communication network - LoRa, broadband carrier, and RS485 are all local communication in the transformer area, independent of the remote 4G link).

[0187] Step 2: The edge computing terminal collects all cross-sectional data within approximately 500ms;

[0188] Step 3: The fault location module completes the step-by-step back-tracking analysis and outputs the location result: "The A-phase line segment between the K111 outgoing line of branch box 1 and the B1111 incoming line of meter box 1 is short-circuited in phase A".

[0189] Step 4: The active alarm module generates structured alarm information and attempts to transmit it via 4G remote communication network, but finds that the link is interrupted;

[0190] Step 5: The active alarm module stores the alarm information in the queue to be sent in the local storage module, marks it with a timestamp, and continuously monitors the status of the remote link;

[0191] Step 6: After about 45 minutes, the 4G link is restored, and the active alarm module immediately sends the alarm information in the queue to the cloud main station.

[0192] Step 7: The power supply service command platform receives the alarm information. Although there is a 45-minute delay, the repair personnel can still obtain accurate fault location information and go to the site for repairs.

[0193] In contrast, if a traditional solution relying on centralized computing by a master station is adopted, when the remote link is interrupted, the data from the sensing device cannot be uploaded to the master station, and the master station cannot detect the fault at all. After the link is restored, there is also a lack of locally cached historical data, resulting in the complete failure to report the fault.

[0194] The advantages of this embodiment are as follows: the fault location calculation of the present invention is completed entirely locally on the edge computing terminal, without relying on the real-time connectivity of the remote communication link; the remote link is only used for uploading alarm information, and even if the link is interrupted, the location function can still be executed normally, and the alarm information can be re-sent after the link is restored. This design significantly improves the reliability and robustness of the system under harsh communication conditions.

[0195] Example 11: Hardware Implementation and Performance Metrics

[0196] This embodiment illustrates the specific hardware implementation scheme and measured performance indicators of the system of the present invention.

[0197] (a) Edge computing terminal hardware specifications

[0198] The edge computing terminal uses a dual-core ARM Cortex-A53 processor (1.2GHz), is equipped with 2GB of DDR4 memory and 8GB of eMMC storage. The terminal has the following built-in communication interfaces:

[0199] Two RS485 serial ports (for communication with devices inside the JP cabinet);

[0200] One broadband power line carrier communication module (used to communicate with the end monitoring unit of the meter box).

[0201] One-channel LoRa wireless communication module (SX1278 chip, operating frequency 470MHz, transmission rate 5.4kbps, communication distance 1km in open environment).

[0202] One-channel 4G full-network communication module (used for remote communication with the cloud-based main station).

[0203] One 100M Ethernet port (backup remote communication channel).

[0204] The terminal uses a Linux operating system and deploys a Docker container runtime environment. The topology identification app, fault location app, and proactive alarm app are all deployed in a containerized manner, supporting remote upgrades and canary releases. The apps interact with each other via an internal message queue—the topology structure file generated by the topology identification app is stored in a shared storage area, the fault location app reads it from the storage and combines it with cross-sectional data for analysis, and the proactive alarm app receives the location results from the fault location app and generates alarm information—this loosely coupled design between modules ensures that each functional module can be upgraded and maintained independently.

[0205] (II) Hardware Specifications of Low-Voltage Fault Sensor

[0206] The low-voltage fault sensor uses an STM32F103 series MCU as the main control chip (72MHz) and is equipped with an ADS131M04 quad-channel 24-bit microcontroller. The ADC is used for voltage and current sampling (sampling rate 4kHz, 80 sampling points per cycle), with a built-in pulse current generation circuit (peak current adjustable 5A, pulse width programmable from 200μs to 1ms), and a built-in 1000mAh lithium battery as a backup power source (under normal operating conditions, it can maintain the device for 8 hours, and can meet 10 complete reports when only used for power outage reporting).

[0207] The sensor uses an open CT type kit (CT inner diameter is suitable for cables with less than 35mm), and the puncture needle for power extraction extends 15mm (suitable for common low-voltage cables with insulation thickness of 5 to 8mm).

[0208] (III) Measured performance indicators

[0209] Topology recognition accuracy The number of distribution stations is ≤5 and the number of devices is ≤100. ≥99% Topology recognition time Same as above ≤120 seconds Delay in reporting power outage section data From power outage to data collection at the edge ≤500ms Fault location accuracy Single-phase short circuit, open circuit, and equipment failure are the three types. ≥97% Fault location accuracy — Branch box - Meter box level (positioning range ≤ 3 meter boxes) Alarm generation and transmission delay From location completion to main station reception ≤700ms

[0210] Among them, the fault location accuracy of "branch box-meter box level" means that the location conclusion can narrow down the fault range to a specific outgoing line of a specific branch box to one or several adjacent meter boxes connected to that outgoing line (involving about 15-20 users), rather than the traditional "area level" (involving hundreds of users) or "feeder level" (involving dozens of users), providing accurate on-site guidance for emergency repair personnel.

[0211] The beneficial effects of this embodiment are as follows: through standardized hardware selection and modular software architecture, the system of the present invention has engineering practical value with controllable cost, reliable performance and strong maintainability; the measured data verified the excellent performance of the system in core indicators such as topology identification accuracy, fault location accuracy and response latency.

[0212] Example 12: System adaptability under different transformer substation sizes

[0213] This embodiment illustrates the adaptability of the system of the present invention in low-voltage distribution substations of different sizes.

[0214] Scenario 1 – Small transformer substation (single outgoing line, no branch box):

[0215] A rural distribution area has only one outgoing line from the transformer, directly supplying power to three centralized meter boxes, each serving 8 to 12 households, with no branch boxes. The equipment deployment is as follows: one set of first-level sensing devices (outgoing line side of the transformer), three sets of third-level sensing devices (incoming line side of each meter box), and several fourth-level sensing devices (user smart meters).

[0216] System performance:

[0217] The topology is “TTU→D1→{B1, B2, B3}”, with two levels (transformer outgoing line → meter box).

[0218] When a fault occurs, the step-by-step backtracking only requires comparison of two levels, and the positioning speed is fast (positioning is completed in about 200ms after cross-section data collection).

[0219] The fault location accuracy is "meter box level" (accurate to the specific meter box that lost power).

[0220] Scenario 2 – Medium-sized distribution area (multiple outgoing lines, multiple branch boxes):

[0221] A residential transformer substation in a certain city has 3 outgoing transformer lines, each with 2 to 3 branch boxes, and each branch box has 4 to 6 outgoing lines, totaling 21 meter boxes. The equipment deployment is as follows: 3 sets of first-level sensing devices, 21 sets of second-level sensing devices (outgoing lines from each branch box), and 21 sets of third-level sensing devices (incoming lines from each meter box).

[0222] System performance:

[0223] The topology is “TTU→{D1, D2, D3}→{K111...K326}→{B1111...B3321}”, with 3 levels (distribution transformer outgoing line → branch box outgoing line → meter box).

[0224] The step-by-step backtracking process requires traversing three levels, and the positioning is completed approximately 600ms after the cross-sectional data is collected.

[0225] The fault location accuracy is "branch box outgoing line - meter box level" (accurate to a specific outgoing line of a specific branch box and its downstream meter box).

[0226] Scenario 3 – Large transformer substations (including distributed power supply access):

[0227] In addition to conventional loads, a commercial complex's distribution area is also connected to rooftop photovoltaic (distributed power) systems and a cluster of electric vehicle charging stations. The equipment is deployed at the conventional level, with additional plug-and-play communication units serving as sensing nodes at the photovoltaic grid connection points and charging stations.

[0228] System performance:

[0229] The topology has been expanded to include branches for "TTU → Photovoltaic Inverter" and "TTU → Charging Pile Cluster".

[0230] The topology identification module can identify newly added distributed power nodes and incorporate them into the topology.

[0231] When a fault occurs (such as a fault in the photovoltaic grid-connected line), tracing back level by level can also locate the photovoltaic grid-connected point.

[0232] The power outage data from sensing devices are also applicable to scenarios with distributed power sources (the fault current may contain reverse power flow characteristics, but voltage drop detection is still effective).

[0233] This embodiment demonstrates that the system of the present invention has good scalability—from small distribution areas (level 2 topology, number of devices <10) to large distribution areas (level 3 to 4 topology, number of devices >100), the system's topology identification and fault location functions can operate effectively; the hierarchical deployment architecture of sensing devices is naturally adapted to low-voltage distribution networks of different sizes and complexities, and has broad application value.

[0234] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should be included in the scope of the present invention.

Claims

1. A low-voltage distribution network fault active alarm system, characterized in that, The system includes an edge computing terminal, multiple layers of sensing devices, a topology identification module, a fault event determination module, and an active alarm module. The multiple layers of sensing devices are deployed hierarchically along the low-voltage distribution network from the low-voltage outgoing line side of the distribution transformer, the outgoing line of the branch box, the incoming line side of the meter box, to the user side. Each layer of sensing device collects voltage, current, and fault status information of its corresponding node and is equipped with a backup power supply, a power outage protection circuit, and a data buffer for instantaneous power failure. The edge computing terminal communicates with the multiple layers of sensing devices and is equipped with the topology identification module, fault event determination module, and active alarm module. The topology identification module identifies the transformer area and phase affiliation of the sensing devices through voltage distortion signal injection and identifies the feeder hierarchy relationship between the sensing devices through pulse current characteristic signals, forming a real-time low-voltage distribution network topology. The fault event determination module performs a step-by-step retrospective determination of fault events based on the pre-outage section data, fault status information, and the real-time low-voltage distribution network topology reported by the multiple layers of sensing devices, identifying the fault section, fault type, and affected downstream sensing devices. The active alarm module is used to generate structured alarm information based on the fault section, fault type and affected downstream sensing devices after a fault event is determined, and to actively send the structured alarm information to the cloud main station and / or the preset operation and maintenance terminal according to the corresponding alarm channel selected according to the preset alarm priority, so as to trigger fault alarm and emergency repair prompt.

2. The low-voltage distribution network fault active alarm system according to claim 1, characterized in that, The structured alarm information includes at least the faulty device identifier, the topology level of the fault, the location of the faulty section, the fault type, the time of the fault occurrence, a list of affected downstream sensing devices, and the alarm priority; the active alarm module generates alarm information of different levels according to the alarm priority.

3. The low-voltage distribution network fault active alarm system according to claim 1, characterized in that, When the sensing devices at each level detect a voltage drop below a preset threshold, they trigger the power outage holding circuit. The backup power supply maintains the processing and communication functions of the sensing devices and actively sends the voltage and current waveform data of the last N cycles cached before the power outage, as well as the power outage time marker, to the edge computing terminal as the triggering and judgment basis for the fault event judgment module to generate alarm information.

4. The low-voltage distribution network fault active alarm system according to claim 1, characterized in that, The active alarm module includes an alarm information encapsulation unit, an alarm priority determination unit, and a multi-channel transmission unit. The alarm information encapsulation unit encapsulates the fault segment, fault type, fault time, and affected device information into a structured alarm message according to a preset data structure. The alarm priority determination unit determines the alarm priority based on the fault type, the fault topology level, the number of affected devices, and the fault impact range. The multi-channel transmission unit selects one or more of the following for alarm transmission: remote communication network, SMS platform, mobile work terminal, and / or instant messaging platform, based on the alarm priority.

5. The low-voltage distribution network fault active alarm system according to claim 1, characterized in that, When the topology identification module identifies the relationship between the household and the transformer, it applies a modulation distortion signal with a preset code near the zero-crossing point of the three-phase line voltage on the low-voltage side of the distribution transformer. After each level of sensing device detects and parses the modulation distortion signal, it feeds back the substation identification and phase identification to the edge computing terminal. When identifying the relationship between the feeders, a designated sensing device injects a pulse current signal with a preset pulse width and peak value into the line. Other sensing devices detect the pulse current signal and feed back the detection result. The topology identification module establishes the relationship between the feeders based on the feedback result.

6. The low-voltage distribution network fault active alarm system according to claim 1, characterized in that, The fault event determination module uses the downstream sensing device that experienced a power outage and actively reported the cross-sectional data before the power outage as the starting point for backtracking. Following the real-time low-voltage distribution network topology, it compares the voltage status of each sensing device at the time of the fault with the voltage status of the sensing devices at each level towards the power source. When the upstream sensing device has voltage data but the downstream sensing device does not, the line segment between them is determined as the fault segment. When neither the upstream nor downstream sensing device has voltage data, the backtracking object is moved towards the power source. When the downstream sensing device has no voltage data but its upstream device has voltage data, and the cross-sectional data before the power outage of the downstream sensing device contains fault current characteristics, the line segment corresponding to that downstream sensing device is determined as the fault segment, and the fault type is determined to be a short-circuit fault.

7. A method for proactive fault alarm in low-voltage distribution networks based on the system described in any one of claims 1 to 6, characterized in that, include: S1. Topology establishment steps: Obtain the distribution area and phase affiliation and feeder hierarchy of the sensing equipment by voltage distortion signal injection and pulse current characteristic signal respectively, and establish the real-time low-voltage distribution network topology. S2. Fault detection steps: The sensing devices at each level collect voltage, current and fault status information in real time, and trigger power outage retention when the voltage drops below the preset threshold. The voltage and current waveform data of the last N cycles before the power outage and the power outage time mark are actively reported. S3. Fault Determination Step: The edge computing terminal determines the fault section, fault type, and affected area by tracing back level by level based on the pre-outage section data and fault status information reported by the sensing devices at each level and the real-time low-voltage distribution network topology. S4. Alarm Generation Step: Structured alarm information is generated based on the fault section, fault type, and affected area, and the corresponding alarm priority is determined. S5. Active Alarm Step: One or more alarm channels are selected according to the alarm priority, and the structured alarm information is actively sent to the cloud master station, the distribution network mobile operation terminal, and / or the notification terminal corresponding to the affected user to realize fault alarm and emergency repair prompts.

8. The method for active fault alarm in low-voltage distribution networks according to claim 7, characterized in that, In S4, the alarm priority is determined according to the fault type, the topology level of the fault, the number of downstream sensing devices affected, and the expected power outage range. Specifically, the alarm priority is increased when the fault type is a short circuit fault, and the alarm priority is increased when the topology level of the fault is closer to the transformer side and the number of downstream sensing devices affected is greater.

9. The method for active fault alarm in low-voltage distribution networks according to claim 7, characterized in that, In step S5, the active alarm module receives an acknowledgment from the alarm channel after sending an alarm message; if no acknowledgment is received within a preset time, the alarm message is resent according to a preset number of resentments, and the alarm is switched to another alarm channel according to the alarm priority; when at least one alarm channel returns an acknowledgment, the alarm sending time, alarm channel, acknowledgment status and number of resentments are recorded to form an alarm sending record.

10. The method for active fault alarm in low-voltage distribution networks according to claim 7, characterized in that, In step S5, after receiving the structured alarm information, the cloud-based master station determines the scope of affected users based on the fault section and the list of affected downstream sensing devices, and generates a corresponding proactive repair work order. At the same time, it sends the fault location, fault type, fault time and expected power outage range to the distribution network mobile operation terminal according to the alarm priority, and sends a power outage notification containing the expected power outage range and fault status to the affected users.

Images