A power distribution network monitoring system and a power distribution network

Abstract

This invention relates to the field of power distribution technology, specifically disclosing a power distribution network monitoring system and a power distribution network. It addresses the problems of existing systems where weak wavefronts, dispersion, multi-branch reflections, and timing jitter coexist, making unified tuning and consistent time scales impossible. These problems easily lead to missed detections, false peak preemption, arrival time offsets, unstable cross-end fusion, and large positioning errors. The system includes a synchronous transient sampling unit, a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time scale fusion unit, a data storage unit, and a master station processor. This invention uses kurtosis K, entropy H, and kurtosis-entropy ratio as unified parameter sources, integrating phase mode decoupling, MRSVD, VMD, SDEO, and time scale fusion to construct a cross-unit collaborative tuning and traceable distribution mechanism. This enables the linkage between line mode selection, singular value truncation, mode number, and center frequency, forming a closed loop with topology and timing.

Description

Technical Field

[0001] This invention relates to the field of power distribution technology, and in particular to a power distribution network monitoring system and a power distribution network. Background Technology

[0002] In the traveling wave front calibration process for fault location in power distribution networks, existing technologies typically first perform phase-mode transformation on the three-phase fault voltage traveling wave at each measurement point to extract line mode 1. Then, they use MRSVD (Multi-Resolution Singular Value Decomposition) to decompose the wave into detail and approximate components. The kurtosis is used to determine the validity of the detail components. If the condition is not met, VMD (Variational Mode Decomposition) is applied to the approximate components, and the kurtosis ratio is used to screen for valid IMF (Intrinsic Mode Function) components. Finally, the calibration is combined with SDEO (Symmetric Differencing Energy) calibration. The energy spectrum peak value of the operator (symmetric differential energy operator) determines the arrival time of the initial wavefront; however, the kurtosis ratio and kurtosis statistics K and entropy measure H, which are the core thresholds for VMD screening, are designed to only perform a single selection function. They are not used to constrain or calibrate the linear mode purity of phase mode transformation, MRSVD singular value truncation and layer selection, VMD mode number self-adaptation, SDEO window length and threshold linkage, and the weighting and anomaly removal in multi-terminal time-scale fusion. As a result, when the fault presents complex characteristics such as weak wavefront, dispersion broadening, multi-branch reflection, inverter noise superposition, and time jitter, the shape carried by K and H in the prior is not properly considered. The status and uncertainty information are not transformed into cross-step parameter adjustment signals and quality control indicators, making it difficult to suppress screening omissions, false peak preemption and arrival time offsets, thus limiting the robustness under real distribution network conditions. At the same time, this single-purpose setting ignores K and H as prior and subsequent verification evidence for cross-layer iteration, which are used to dynamically constrain the modal frequency search range, suppress modal aliasing, guide the energy allocation of details and approximate components, correct spectral leakage caused by clipping and quantization noise, provide weights for the discrimination of zero crossover density and energy abrupt changes, and serve as the basis for time scale consistency and abnormal end removal when multiple terminals cooperate. Therefore, it is difficult to form a closed loop in the method chain. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide a power distribution network monitoring system and power distribution network. By using kurtosis K, entropy H and kurtosis-entropy ratio as unified parameter sources, and integrating phase mode decoupling, MRSVD, VMD, SDEO and time scale fusion, a cross-unit collaborative tuning and traceable distribution mechanism is constructed, which enables the linkage of line mode selection, singular value truncation, mode number and center frequency, and forms a closed loop with topology and time synchronization.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A power distribution network monitoring system includes a synchronous transient sampling unit, a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time-scale fusion unit, a data storage unit, and a master station processor. The phase mode decoupling unit performs phase mode decoupling on the traveling wave and outputs line mode 1. The MRSVD decomposition unit generates detail components and approximate components for line mode 1. The VMD decomposition unit generates an IMF for the approximate components. The feature statistics unit calculates kurtosis K, entropy H, and kurtosis-entropy ratio for the detail components and IMF and outputs statistics. The parameter management unit receives statistics and timing quality indicators, uses kurtosis, entropy, and kurtosis-entropy ratio as a unified parameter source, and sends them to the phase mode decomposition unit. The coupling unit distributes the decoupling matrix version identifier, line mode selection, and matrix correction coefficients; the MRSVD decomposition unit distributes the singular value truncation threshold, decomposition layer number, and segmented window length; the VMD decomposition unit distributes the mode number, penalty factor, and initial center frequency constraint; the SDEO calibration unit distributes short and long windows, energy threshold, and consistency tolerance; the time-scaled fusion unit distributes measurement point weights, anomaly removal criteria, and topology index; and the data storage unit distributes the archiving strategy and index key. The SDEO calibration unit outputs arrival time candidates on the received components, the time-scaled fusion unit outputs the measurement point-time series, and the data storage unit accesses the original traveling wave, decomposition results, statistics, control variables, and timestamps.

[0006] As a further technical solution of the present invention, the synchronous transient sampling unit includes a micro PMU, a primary sampler, an anti-aliasing filter, a digital-to-analog converter, a GPS timing module, and a communication interface. The timing module provides a unified time base and provides timing for the micro PMU. The micro PMU sends out sampling trigger parameters. The primary sampler outputs a three-phase secondary signal, which is sent to the digital-to-analog converter after being filtered by the anti-aliasing filter. The signal is then sampled synchronously according to a unified time scale. The micro PMU adds node ID, sampling rate, time scale, and timing quality to the sampling sequence and generates a data frame. The communication interface sends the data frame to the main station processor and the data storage unit according to the protocol.

[0007] As a further technical solution of the present invention, the phase mode decoupling unit stores a configurable phase mode decoupling matrix and receives the decoupling matrix version identifier, line mode selection parameters and matrix correction coefficients issued by the parameter management unit. The three-phase transient traveling wave output by the synchronous transient sampling unit is divided into frequency domains according to a preset segment window length and sampling rate. After phase reference alignment, the corresponding decoupling matrix is ​​called according to the version identifier, and the element amplitude and phase of the matrix entries are corrected according to the matrix correction coefficients. The voltage side and current side mapping relationship is established according to the topology search and the device list. The matrix operation is performed to obtain candidate modes, and line mode 1 is selected from the candidate modes according to the line mode selection parameters. The line mode 1 with the version of the matrix used and the timestamp identifier is output, and the version used, correction coefficients and segment index are returned to the parameter management unit and the data storage unit.

[0008] As a further technical solution of the present invention, the parameters distributed by the parameter management unit to the phase mode decoupling unit include: a decoupling matrix version identifier composed of topology code, transformer connection group code, grounding method code, frequency band index and calibration batch number; a line mode number for specifying the output line mode, a purity threshold index for determining the line mode selection and a phase reference identifier; and three-phase amplitude correction coefficients and three-phase phase correction coefficients for item-level correction of the three-phase amplitude and phase.

[0009] The parameters distributed by the parameter management unit to the multi-resolution singular value decomposition unit include: a singular value truncation threshold with a value between zero and one, a decomposition layer number with a value from one to a preset maximum layer number, a segmented window length with a length that is an integer power of two, and an identifier corresponding to the sampling rate and an identifier for the window alignment method.

[0010] As a further technical solution of the present invention, the parameters distributed by the parameter management unit to the variational mode decomposition unit include: the number of modes with values ​​ranging from one to a preset upper limit; positive real coefficients for adjusting decomposition convergence and bandwidth allocation; an initial center frequency constraint interval given for each mode, wherein the lower limit of the initial center frequency constraint interval is greater than zero, the upper limit is less than half of the sampling frequency, and a preset minimum interval is maintained between adjacent intervals; and a bandwidth upper limit, frequency quantization step size, and a random seed for initialization are given for each mode.

[0011] As a further technical solution of the present invention, the parameters distributed by the parameter management unit to the symmetric differential energy operator calibration unit include: a short window length and a long window length determined by positive integer multiples of the sampling period, wherein the long window length is greater than the short window length; an energy threshold given in the form of quantile index, with a value between zero and one; a consistency tolerance time expressed as an integer multiple of the sampling period; an identifier for specifying the window function type; an identifier for specifying the standardization processing method; and a processing batch number for process traceability.

[0012] As a further technical solution of the present invention, the parameters distributed by the parameter management unit to the time-scale fusion unit include: non-negative weights corresponding to each measurement point and the sum of the weights is agreed to be one; anomaly elimination and pairing rules consisting of limit codes, continuous frame counts and pairing tolerances; topology index identifiers for indicating network structure and clock source identifiers for indicating time synchronization sources;

[0013] The parameters distributed by the parameter management unit to the data management unit include: an archiving strategy consisting of archiving period, scrolling window size, number of versions retained, and compression level; an index key consisting of node identifier, version identifier, start time, end time, decomposition level identifier, and modality number; and a naming rule identifier used to unify the generation of files and directories.

[0014] A power distribution network, employing the aforementioned power distribution network monitoring system, includes an electrical topology comprising substation busbars, outgoing feeders, branch switches, ring network units, distribution transformers, grounding devices, and a line layer. The line layer includes overhead sections and cable sections. Measurement points are located at nodes within the line layer, and a synchronous transient sampling unit is installed at each measurement point. The synchronous transient sampling unit acquires a unified time scale and sends three-phase transient traveling wave and time scale information to a master station processor via a communication network. The master station processor carries a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time scale fusion unit, and a data storage unit, and maintains topology indexes corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points. The data management unit is connected to the master station processor to access raw traveling waves, decomposition and statistical records, and parameter indexes. Distributed power supply grid connection points and energy storage grid connection points are connected to the line layer via nodes and branch switches.

[0015] As a further technical solution of the present invention, the connection relationship of the electrical topology is as follows: the outgoing end of the substation busbar is electrically connected to the beginning of each outgoing feeder; each outgoing feeder is composed of an overhead section and a cable section connected in sequence, and the two are electrically connected at the node; a branch switch is set at the node, so that the branch circuit is electrically connected to the main outgoing feeder at the node; the ring network unit is set between adjacent outgoing feeders or adjacent sections of the same outgoing feeder, and each of its interfaces is electrically connected to the corresponding feeder section at the node; the high-voltage side of the distribution transformer is electrically connected to the corresponding outgoing feeder at the node through a branch circuit, and its grounding end is connected through a grounding device; the grounding device is electrically connected to the feeder circuit and the equipment casing at the node; each node is equipped with a measurement point, and at the measurement point, the primary sampling interface of the synchronous transient sampling unit is electrically connected to the three-phase conductor and connected to the grounding device; the communication interface of the synchronous transient sampling unit is connected to the main station processor through a communication network.

[0016] As a further technical solution of the present invention, the following structures and fields are set at the line layer and node side: An event ring buffer, pre-trigger and post-trigger segment identifiers, sampling rate identifiers, and timing quality identifiers are set in the synchronous transient sampling unit of each measurement point, and a node identifier and topology index are carried in the message header; Parameter channel identifiers and parameter encoding tables are set for the phase mode decoupling unit, MRSVD decomposition unit, VMD decomposition unit, SDEO calibration unit, time scale fusion unit, and data storage unit respectively on the master station processor side; Subject identifiers and sequence number fields are set for parameter distribution on the communication network side; Measurement boxes at nodes are set with... Fiber optic or power line carrier coupling units are configured with node codes and phase sequence identification markers; timing distribution nodes and fiber optic aggregation points are set up on the substation bus side; synchronous transient sampling unit installation positions, power supply interfaces, timing interfaces, and communication interfaces are set up at branch switch cabinets and ring network units; primary sampling leads and grounding terminals are set up on the primary side of the distribution transformer, and synchronous transient sampling unit installation positions are set up in the secondary side cabinet; the master station processor maintains the topology index and parameter mapping table corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points, which is used to address-distribute and record parameter codes generated based on kurtosis, entropy, and kurtosis-entropy ratio.

[0017] The technical advantages of the power distribution network monitoring system and power distribution network of the present invention are as follows:

[0018] This invention introduces a cross-unit collaborative tuning mechanism with kurtosis K, entropy H, and kurtosis-entropy ratio as unified parameter sources into the processing links of phase mode decoupling, MRSVD, VMD, SDEO, and time-scale fusion. This mechanism achieves unified driving of key control variables such as line mode selection, singular value truncation and layer number, mode number and penalty factor and center frequency constraints, short and long windows and thresholds and consistency tolerance, measurement point weights and anomaly removal. Furthermore, it incorporates robust preprocessing and parameterized distribution and recording of dual-window consistent wavefront confirmation, topology indexing, and time synchronization quality identification. This enables arrival time calibration and multi-terminal time series construction under conditions of weak wavefronts, dispersion broadening, multi-branch reflections, and strong noise to achieve more... Strong consistency, anti-interference and traceability support the rapid and accurate location and engineering deployment of fault sections in the power distribution network. At the same time, the parameter mapping and versioning records based on nodes and timestamps connect the closed-loop data chain of acquisition, decoupling, decomposition, calibration and fusion. It is adaptable to different wiring groups and grounding methods, overhead and cable mixed lines, and distributed power and energy storage grid connection scenarios. It provides stable parameter transmission and tracking for online topology changes, timing jitter and sampling drift. It enables the system to maintain unified configuration and consistent output under large-scale multi-measurement point deployment and complex communication conditions. It also supports historical replay and audit trace, which facilitates cross-site collaboration and long-term maintenance. Attached Figure Description

[0019] Figure 1 This is a block diagram of a power distribution network monitoring system according to the present invention;

[0020] Figure 2 This is a schematic diagram of the power distribution feeder structure in Embodiment 1 of the present invention;

[0021] Figure 3 This is the original waveform diagram of the three measurement points of the present invention;

[0022] Figure 4 This is a schematic diagram showing the SDEO energy and its arrival time according to the present invention;

[0023] Figure 5 This is a bar chart showing the statistical characteristics of kurtosis K, entropy H, and kurtosis-entropy ratio in this invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1

[0026] like Figure 1 As shown, the present invention proposes a power distribution network monitoring system, comprising a synchronous transient sampling unit, a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time-scale fusion unit, a data storage unit, and a master station processor. The phase mode decoupling unit performs phase mode decoupling on the traveling wave and outputs line mode 1. The MRSVD decomposition unit generates detail components and approximate components for line mode 1. The VMD decomposition unit generates an IMF for the approximate components. The feature statistics unit calculates kurtosis K, entropy H, and kurtosis-entropy ratio for the detail components and IMF and outputs statistics. The parameter management unit receives the statistics and timing quality indicators, using kurtosis, entropy, and kurtosis-entropy ratio as a unified parameter source. The decoupling matrix version identifier, line mode selection, and matrix correction coefficients are distributed to the phase mode decoupling unit; the singular value truncation threshold, decomposition layer number, and segmented window length are distributed to the MRSVD decomposition unit; the mode number, penalty factor, and initial center frequency constraint are distributed to the VMD decomposition unit; short and long windows, energy threshold, and consistency tolerance are distributed to the SDEO calibration unit; measurement point weights, anomaly removal criteria, and topology index are distributed to the time-scaled fusion unit; and archiving strategy and index key are distributed to the data storage unit. The SDEO calibration unit outputs arrival time candidates on the received components, the time-scaled fusion unit outputs measurement point-time series, and the data storage unit accesses the original traveling wave, decomposition results, statistics, control variables, and timestamps.

[0027] The technical problem of this invention is that existing processes only use kurtosis K, entropy H, and kurtosis-entropy ratio as VMD screening thresholds, without involving them in key stages such as phase mode decoupling, MRSVD layer number and truncation, SDEO window parameters, and multi-terminal fusion. This results in parameters being independent of each other across steps, easily leading to candidate conflicts and arrival time offsets when weak wavefronts, dispersion broadening, and multi-branch reflections are superimposed. To address this, the system uses kurtosis K, entropy H, and kurtosis-entropy ratio as a unified parameter source in the acquisition, decoupling, decomposition, statistics, tuning, calibration, fusion, and storage links. The parameter management unit synchronously distributes decoupling matrix version, line mode selection, singular value threshold and layer number, mode number and penalty factor, center frequency constraints, SDEO short / long window and threshold, fusion weight, and pairing rules, ensuring that each module adopts a consistent prior for the same signal morphology and uncertainty. Furthermore, with topology index and timing quality as constraints, a mapping is established between parameters and nodes, timestamps, levels, and modes, forming a collaborative configuration and consistent output for complex distribution networks.

[0028] It should be noted that the synchronous transient sampling unit includes a micro PMU, a primary sampler, an anti-aliasing filter, a digital-to-analog converter, a GPS timing module, and a communication interface. The timing module provides a unified time base and provides timing for the micro PMU. The micro PMU sends out sampling trigger parameters. The primary sampler outputs a three-phase secondary signal, which is sent to the digital-to-analog converter after being filtered by the anti-aliasing filter. The signal is then sampled synchronously according to a unified time scale. The micro PMU adds node ID, sampling rate, time scale, and timing quality to the sampling sequence and generates a data frame. The communication interface sends the data frame to the main station processor and data storage unit according to the protocol.

[0029] By providing a unified time base to the micro PMU through the timing module in the synchronous transient sampling unit, uniformly issuing sampling trigger parameters by the micro PMU, forming a stable secondary signal through a primary sampler and anti-aliasing filter, and synchronously sampling by the digital-to-analog converter according to a unified time scale, and attaching node ID, sampling rate, time scale, and timing quality to the data frame and transmitting it through the communication interface, the transient traveling waves across measurement points have consistent alignment and source identification on the time axis. This provides reliable input and parameter linkage basis for phase mode decoupling, MRSVD and VMD decomposition, SDEO wavefront calibration, and time scale fusion, thereby reducing parameter mismatch and candidate conflicts caused by timing jitter, sampling drift, and link differences. It also enables the kurtosis, entropy, and kurtosis-entropy ratio calculated by the feature statistics unit under a unified time reference to correspond with the node and frequency band index. The parameter management unit can then generate codes such as decoupling matrix version, line mode selection, singular value truncation and decomposition layer number, mode number and penalty factor, short and long windows and thresholds, and consistency tolerance, and distribute them synchronously to each unit to ensure the consistency and traceability of subsequent processing link configurations.

[0030] It should be noted that the phase mode decoupling unit stores configurable phase mode decoupling matrices and receives decoupling matrix version identifiers, line mode selection parameters, and matrix correction coefficients from the parameter management unit. It divides the three-phase transient traveling wave output by the synchronous transient sampling unit into frequency domains according to the preset segmented window length and sampling rate. After phase reference alignment, it calls the corresponding decoupling matrix according to the version identifier and performs element amplitude and phase correction on the matrix entries according to the matrix correction coefficients. It establishes the voltage side and current side mapping relationship based on topology search and equipment list, performs matrix operations to obtain candidate modes, selects line mode 1 from the candidate modes according to the line mode selection parameters, outputs line mode 1 with the version of the matrix used and the timestamp identifier, and returns the version used, correction coefficients, and segment index to the parameter management unit and data storage unit.

[0031] In the phase mode decoupling unit, the receiving parameter management unit generates a parameter set based on kurtosis K, entropy H, and kurtosis-entropy ratio and their statistical index. The parameter set includes: decoupling matrix version identifier (including topology code, wiring group code, grounding method code and calibration batch number), frequency band selection index and segment window length identifier (given by entropy and its index), line mode selection parameters (including line mode number and purity threshold index, given by kurtosis), matrix correction coefficient number (amplitude and phase correction table mapped by kurtosis-entropy ratio), phase reference identifier and window alignment method identifier. The phase-mode decoupling unit first determines the segmentation window based on the frequency band selection index and sampling rate, and segments and divides the three-phase transient traveling wave output by the synchronous transient sampling unit into frequency domains. Then, it completes phase alignment using a phase reference identifier and loads the phase-mode decoupling matrix from the matrix library according to the decoupling matrix version identifier. It calls the amplitude correction vector and phase correction vector according to the matrix correction coefficient number, and performs element-level amplitude and phase correction on the matrix entries. It establishes a mapping relationship between the voltage side and the current side based on the topology search results and the device list. It performs matrix operations on each segment to obtain a candidate mode set. Based on the line mode number and purity threshold index specified by the line mode selection parameters, it selects a line mode from the candidate mode set and appends the matrix version and timestamp to form the output. Simultaneously, it sends the matrix version identifier, matrix correction coefficient number, frequency band selection index, segmentation window length identifier, segmentation index, and node identifier back to the parameter management unit and writes them into the data storage unit.

[0032] It should be noted that the parameters distributed by the parameter management unit to the phase mode decoupling unit include: a decoupling matrix version identifier consisting of topology code, transformer connection group code, grounding method code, frequency band index, and calibration batch number; a line mode number for specifying the output line mode, a purity threshold index for determining the line mode selection, and a phase reference identifier; and three-phase amplitude correction coefficients and three-phase phase correction coefficients for item-level correction of the three-phase amplitude and phase.

[0033] The parameters distributed by the parameter management unit to the multi-resolution singular value decomposition unit include: a singular value truncation threshold with a value between zero and one, a decomposition layer number with a value from one to a preset maximum layer number, a segmented window length with a length that is an integer power of two, and an identifier corresponding to the sampling rate and an identifier for the window alignment method.

[0034] Compared to traditional methods that only use kurtosis, entropy, and kurtosis-entropy ratio for VMD component selection, this invention uses a parameter management unit as a hub. It uniformly distributes the decoupling matrix version identifier (composed of topology coding, wiring group, grounding method, frequency band index, and calibration batch), along with the connected line mode number, purity threshold index, phase reference identifier, and three-phase amplitude / phase correction coefficients, to the phase-mode decoupling unit. Furthermore, it uniformly distributes the singular value truncation threshold, decomposition level, and power-of-two window length, along with the sampling rate and alignment method identifier, to the MRDVD decomposition unit, achieving cross-mode decomposition. The same source data tuning and configuration of the blocks are traceable and parameter consistent, reducing decoupling mismatch and decomposition drift, improving time-frequency consistency and comparability and deployment efficiency in reproducible scenarios. At the same time, K, H and their ratio are used as a unified parameter source to generate parameter codes for decoupling and decomposition, and bind them with node ID, timestamp, and frequency band index to form records. This ensures that line mode selection, matrix correction, window configuration and hierarchical setting are consistent and the process is smooth under different measurement points and different sampling frequencies, avoiding repeated tuning and conflicting configurations, and facilitating cross-site reuse and backtracking.

[0035] It should be noted that the parameters distributed by the parameter management unit to the variational mode decomposition unit include: the number of modes with values ​​ranging from one to a preset upper limit; positive real coefficients used to adjust decomposition convergence and bandwidth allocation; an initial center frequency constraint interval given for each mode, wherein the lower limit of the initial center frequency constraint interval is greater than zero and the upper limit is less than half of the sampling frequency, and a preset minimum interval is maintained between adjacent intervals; and a bandwidth upper limit, frequency quantization step size, and a random seed for initialization are given for each mode.

[0036] By uniformly distributing a set of parameters—including the number of modes, penalty coefficient, initial center frequency constraint interval for each mode (located between 0 and half the sampling frequency with minimum intervals between intervals), upper bound of bandwidth, frequency quantization step size, and random seed—to the VMD decomposition unit from the parameter management unit, the modal search space, bandwidth, center frequency range, and step size can be standardized and the initialization can be kept repeatable under given sampling and topological constraints. This reduces the decomposition's sensitivity to noise and initial values, reduces modal aliasing and drift, and ensures the consistency and comparability of time-frequency characterization and cross-measurement results.

[0037] It should be noted that the parameters distributed by the parameter management unit to the symmetric differential energy operator calibration unit include: the short window length and the long window length, which are determined by positive integer multiples of the sampling period, wherein the long window length is greater than the short window length; the energy threshold given in the form of quantile index, with values ​​between zero and one; the consistency tolerance time expressed as an integer multiple of the sampling period; the identifier used to specify the window function type; the identifier used to specify the standardization processing method; and the processing batch number used for process traceability.

[0038] Unlike existing technologies that often use fixed single windows and empirical thresholds without establishing consistent calibration with the sampling period, this technology defines short and long windows as integer multiples of the sampling period and limits the long window to be greater than the short window. It uses quantile indexes to unify energy thresholds, sets consistency tolerances as integer multiples of the sampling period, and encodes operator configurations with window function type identifiers, standardized processing identifiers, and processing batch numbers. This ensures that window parameters, thresholds, and consistency criteria under different sampling rates, different terminals, and different operating conditions have the same caliber, are portable, and traceable, reducing configuration drift and false triggering. It also enables unified parameter distribution and historical replay comparison in multi-terminal collaborative positioning.

[0039] It should be noted that the parameters distributed by the parameter management unit to the time-scale fusion unit include: the non-negative weights corresponding to each measurement point and the agreement that the sum of the weights is one; the anomaly elimination and pairing rules consisting of limit codes, continuous frame counts and pairing tolerances; the topology index identifier used to indicate the network structure; and the clock source identifier used to indicate the time synchronization source.

[0040] The parameters distributed by the parameter management unit to the data management unit include: an archiving strategy consisting of archiving period, scrolling window size, number of versions retained, and compression level; an index key consisting of node identifier, version identifier, start time, end time, decomposition level identifier, and modality number; and a naming rule identifier used to unify the generation of files and directories.

[0041] Compared to existing technologies that rely on experience to configure and maintain time-stamp fusion and data storage separately, the present invention employs a distributable, machine-readable parameter set: time-stamp fusion is uniformly encoded with non-negative and normalized measurement point weights, anomalies consisting of limit codes, continuous frame counts, and pairing tolerances, as well as pairing rules, topology index identifiers, and timing source identifiers; data management uses archiving cycles, rolling windows, version retention, and compression levels to form an archiving strategy, and uses index keys containing nodes, versions, start and end times, decomposition levels, and modality numbers, along with naming rules, to form standardized metadata. This enables consistent configuration and traceable management across sites, sampling rates, and topology changes, and supports parameter versioning, batch distribution, historical playback, and comparison. Furthermore, it ensures message order and verifiability through topic identifiers and sequence number fields, and binds to node IDs and timestamps to form configuration link records, supporting cross-domain collaborative maintenance and reducing the risk of human misconfiguration.

[0042] For example, such as Figure 2 As shown, a 10kV hybrid feeder with a 6.5km overhead busbar and a 3.5km cable length is used. There is a branch switch at 2km of the cable, and an inverter DG is connected to the cable section. Synchronous transient sampling units are installed at three locations to measure the data at the busbar end P1, the overhead / cable transition node P2, and the cable end P3, respectively. High-resistance grounding occurs at the cable branch (weak wavefront, broadened dispersion, and dense near-end reflections). The timing quality is good, but there is a ±10 microsecond jitter upper limit. Using the above-mentioned technical means proposed in this invention, the phase mode decoupling unit simultaneously sends out the frequency band index and window length obtained from H, the line mode selection parameters obtained from K, and the matrix version and correction coefficients obtained from the kurtosis ratio; the MRSVD decomposition unit sends out the singular value truncation threshold from K and the layer number encoding from the kurtosis ratio, and sends out the segmented window and alignment method from H; the VMD decomposition unit sends out the number of modes from K, the penalty factor range and bandwidth upper bound from H, and the initialization center frequency constraint and quantization step size from the kurtosis ratio; and the SDEO calibration unit sends out the short window from K / H / kurtosis ratio. Long windows Energy threshold and consistency tolerance The nodes are bound to timestamps, and the time stamp fusion unit and data storage unit distribute measurement point weights, pairing tolerances, index keys, and archiving strategies. For example... Figure 3 As shown, after the above processing: After MRSVD / VMD screening, the SDEO dual-window consistency yields arrival time candidates at P1 / P2 / P3 respectively, P1=72.1µs, P2=44.8µs, P3=27.3µs (all three window consistency checks pass, and the discrepancy between terminals is <1µs). Converted to the equivalent propagation speed of 0.20km / µs for cables and 0.30km / µs for overhead lines, the location calculation falls on the upstream side of the main trunk near the cable branch switch, with a distance error of <0.2km from the branch point; the three terminal calculations are consistent. If the conventional approach in the literature is followed (each stage has independent parameter settings, K / H / kurtosis ratio is only used for VMD selection), the phase mode decoupling matrix is ​​fixed, and no frequency band / topology correction is performed; the MRSVD threshold and number of layers, and window length are set independently; VMD only uses kurtosis ratio to screen IMFs; and SDEO uses a single window with a fixed length. For example... Figure 4 As shown, the corresponding results (using the same data) are as follows: At point P1, the SDEO peak preempts the DG switch spike under a single window, giving an arrival time of approximately 95µs; at point P2, due to the low number of VMD modes leading to mode aliasing, the SDEO peak time is approximately 63µs; at point P3, due to dispersion broadening causing insufficient kurtosis of line mode 1, it is mistakenly discarded, and only the IMF candidate gives an arrival time of approximately 54µs. The inter-terminal dispersion reaches 20–40µs; based on a cable speed of 0.20km / µs, the equivalent positioning error is 4–8km; the results are not reproducible under different sampling rates or window lengths and are sensitive to topology switching. Figure 5As shown in the statistical characteristic histogram, the K (kurtosis) histogram reflects the strength of signal abrupt changes / impulsivity at each measurement point, the H (entropy) histogram depicts the dispersion / complexity of energy distribution, and the K / H histogram provides a comprehensive selectivity of both. In this embodiment, P1 exhibits a higher K (significant DG switch spikes), and P3 exhibits a higher H (stronger cable dispersion). Based on this, this technical solution supports the parameter management unit using K, H, and K / H as unified parameter sources to drive the selection of line modes and matrix versions for phase mode decoupling, the truncation threshold and number of layers for the MRSVD decomposition unit, the mode and frequency band constraints for the VMD decomposition unit, and the SDEO calibration unit, respectively. The encoding and distribution of weights are integrated with thresholds and time stamps, and are linked to segment window length, alignment, and consistency tolerance. And data archiving strategies / index keys to achieve consistent configuration of sampling rates across measurement points.

[0043] Example 2

[0044] The difference between Embodiment 2 and Embodiment 1 is that this embodiment introduces a power distribution network.

[0045] A power distribution network, using the power distribution network monitoring system described in Embodiment 1, includes an electrical topology. The electrical topology includes substation busbars, outgoing feeders, branch switches, ring network units, distribution transformers, grounding devices, and a line layer. The line layer includes overhead sections and cable sections. Measurement points are located at nodes in the line layer, and a synchronous transient sampling unit is installed at each measurement point. The synchronous transient sampling unit acquires a unified time scale and sends three-phase transient traveling waves and time scale information to the master station processor via a communication network. The master station processor carries a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time scale fusion unit, and a data storage unit, and maintains topology indexes corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points. The data management unit is connected to the master station processor to access raw traveling waves, decomposition and statistical records, and parameter indexes. Distributed power grid connection points and energy storage grid connection points are connected to the line layer via nodes and branch switches.

[0046] To adapt to the monitoring system described in Embodiment 1, the distribution network proposed in this invention adds a time synchronization distribution node and an optical fiber aggregation point on the bus side. Each node's measurement box is equipped with a synchronous transient sampling unit mounting position, sampling lead-out and shielding connectors, phase sequence identification, and a unique node identifier. On the communication side, optical fiber / carrier coupling and message subject and sequence number fields are set. On the master station processor side, parameter channels and encoding tables for phase mode decoupling, MRSVD, VMD, SDEO, fusion, and storage are established. Measurement bypass and isolation terminals are set at the DG / energy storage grid connection point. The data management unit uses an index key containing node / version / start and end time / level / mode. The archiving strategy includes a built-in event ring buffer, pre-trigger / post-trigger segment identifiers, and sampling rate and timing quality tags for each synchronous transient sampling unit; measurement grounding terminals and equipotential bonding rows are added to the node side; the master station maintains the topology index and matrix version mapping, supporting the cross-unit distribution and backtracking of codes generated by the parameter management unit according to kurtosis, entropy, and kurtosis-entropy ratio; the above changes make the cross-measurement traveling wave timescale consistent, the parameter source distribution traceable, the candidate consistency improved in weak wavefront and dispersion scenarios, the suppression of multi-branch reflection pseudo-peaks, the improvement of multi-terminal fusion consistency and the facilitation of historical replay auditing, and the improvement of configuration alignment capability and engineering maintainability under complex topology switching.

[0047] It should be noted that the electrical topology connections are as follows: the outgoing ends of the substation busbar are electrically connected to the beginning of each outgoing feeder; each outgoing feeder is composed of an overhead section and a cable section electrically connected sequentially, and the two are electrically connected at the nodes; branch switches are set at the nodes, so that the branch circuits are electrically connected to the main outgoing feeders at these nodes; ring network units are set between adjacent outgoing feeders or adjacent sections of the same outgoing feeder, and each of their interfaces is electrically connected to the corresponding feeder section at the nodes; the high-voltage side of the distribution transformer is electrically connected to the corresponding outgoing feeder at the nodes through branch circuits, and its grounding end is connected through a grounding device; the grounding device is electrically connected to the feeder circuit and the equipment casing at the nodes; each node has a measurement point, and at the measurement point, the primary sampling interface of the synchronous transient sampling unit is electrically connected to the three-phase conductors and connected to the grounding device; the communication interface of the synchronous transient sampling unit is connected to the main station processor through the communication network.

[0048] The aforementioned connections enable the busbar, feeder, branch, ring network, transformer, and grounding to form a closed and traceable conductive path. Overhead and cable are connected under a unified node definition. All nodes are equipped with measurement points and connected to the three-phase primary side and grounding with synchronous transient sampling units according to a unified time scale. The communication interface directly reaches the master station processor, thereby obtaining spatially continuous, temporally consistent, and source-identifiable transient traveling wave and equipment status data within the same topological coordinate system. This supports the collaborative processing of phase mode decoupling, decomposition, wavefront calibration, and time scale fusion, reduces cross-node mismatch and grounding loop interference, and provides a stable interface and metadata anchor point for unified parameter distribution, historical replay, and cross-site comparison.

[0049] It should be noted that the following structures and fields are set at the line layer and node side: In the synchronous transient sampling unit of each measurement point, an event ring buffer, pre-trigger and post-trigger segment identifiers, sampling rate identifiers, and timing quality identifiers are set, and the node identifier and topology index are carried in the message header; on the master station processor side, parameter channel identifiers and parameter encoding tables are set for the phase mode decoupling unit, MRSVD decomposition unit, VMD decomposition unit, SDEO calibration unit, time scale fusion unit, and data storage unit, respectively; on the communication network side, a topic identifier and sequence number field are set for parameter distribution; and optical fibers or electrical cables are installed in the measurement box at the node. The system includes a force carrier coupling unit, configured with node codes and phase sequence identification markers; a time distribution node and fiber optic aggregation point are set up on the substation bus side; synchronous transient sampling unit installation positions, power supply interfaces, time distribution interfaces, and communication interfaces are set up at branch switch cabinets and ring network units; a primary sampling lead-out terminal and a grounding terminal are set up on the primary side of the distribution transformer, and a synchronous transient sampling unit installation position is set up in the secondary side cabinet; the main station processor maintains the topology index and parameter mapping table corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points, which is used to address-distribute and record parameter codes generated based on kurtosis, entropy, and kurtosis-entropy ratio.

[0050] By synchronously configuring event ring buffers, pre / post-trigger segment identifiers, sampling rate and timing quality identifiers, node identifiers and topology indexes, parameter channel identifiers and encoding tables, topic identifiers and sequence numbers, node encoding and phase sequence identification, timing allocation nodes and fiber optic aggregation points, standardized installation positions / power supply / timing / communication interfaces, and primary sampling leads and grounding terminals on three sides—measurement points, communication networks, and the main station processor—and having the main station maintain the topology index and parameter mapping table, a metadata link with unified time stamps, ordered messages, addressable parameters, and traceable versions is constructed. This enables cross-unit parameters generated based on kurtosis, entropy, and kurtosis-entropy ratio to be consistently distributed according to addressing rules under conditions of multiple nodes, multiple sampling rates, multiple topologies, and multiple timing sources. These parameters are bidirectionally bound to the data stream and can be historically replayed and audited, reducing the risk of manual configuration discrepancies and message out-of-order delivery.

[0051] In summary, this invention introduces a cross-unit collaborative tuning mechanism with kurtosis K, entropy H, and kurtosis-entropy ratio as unified parameter sources into the processing links of phase mode decoupling, MRSVD, VMD, SDEO, and time-scale fusion. This mechanism achieves unified driving of key control variables such as line mode selection, singular value truncation and layer number, mode number and penalty factor and center frequency constraints, short and long windows and thresholds and consistency tolerance, measurement point weights and anomaly removal. Furthermore, it incorporates robust preprocessing and parameterized distribution and recording of dual-window consistent wavefront confirmation, topology indexing, and time synchronization quality identification. This enables arrival time calibration and multi-terminal time series construction under conditions of weak wavefronts, dispersion broadening, multi-branch reflections, and strong noise. It possesses stronger consistency, anti-interference capabilities, and traceability, supporting rapid and accurate location and engineering deployment of faulty sections in power distribution networks. Simultaneously, based on node and timestamp-based parameter mapping and versioned recording, it connects a closed-loop data chain of acquisition, decoupling, decomposition, calibration, and fusion, adapting to different wiring groups and grounding methods, overhead and cable mixed lines, and distributed power sources and energy storage grid-connected scenarios. It provides stable parameter transmission and tracking for online topology changes, timing jitter, and sampling drift, enabling the system to maintain unified configuration and consistent output under large-scale multi-measurement point deployment and complex communication conditions. It also supports historical replay and audit tracing, facilitating cross-site collaboration and long-term maintenance.

[0052] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0053] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power distribution network monitoring system, characterized in that, The system includes a synchronous transient sampling unit, a phase mode decoupling unit, an MRSVD decomposition unit, a VMD decomposition unit, a feature statistics unit, a parameter management unit, an SDEO calibration unit, a time-scaled fusion unit, a data storage unit, and a master station processor. The phase mode decoupling unit performs phase mode decoupling on the traveling wave and outputs linear mode 1. The MRSVD decomposition unit generates detail components and approximate components for linear mode 1. The VMD decomposition unit generates an IMF for the approximate components. The feature statistics unit calculates kurtosis K, entropy H, and kurtosis-entropy ratio for the detail components and IMF and outputs the statistics. The parameter management unit receives the statistics and timing quality indicators, uses kurtosis, entropy, and kurtosis-entropy ratio as a unified parameter source, and distributes them to the phase mode decoupling unit. The decoupling matrix version identifier, line mode selection, and matrix correction coefficients are used to distribute singular value truncation thresholds, decomposition layer numbers, and segmented window lengths to the MRSVD decomposition unit; the number of modes, penalty factors, and initial center frequency constraints are distributed to the VMD decomposition unit; short and long windows, energy thresholds, and consistency tolerances are distributed to the SDEO calibration unit; measurement point weights, anomaly removal criteria, and topology indexes are distributed to the time-scaled fusion unit; and archiving strategies and index keys are distributed to the data storage unit. The SDEO calibration unit outputs arrival time candidates on the received components, the time-scaled fusion unit outputs measurement point-time series, and the data storage unit accesses the original traveling wave, decomposition results, statistics, control variables, and timestamps. The parameters distributed by the parameter management unit to the phase mode decoupling unit include: a decoupling matrix version identifier consisting of topology code, transformer connection group code, grounding method code, frequency band index, and calibration batch number; a line mode number for specifying the output line mode, a purity threshold index for determining the line mode selection, and a phase reference identifier; and three-phase amplitude correction coefficients and three-phase phase correction coefficients for item-level correction of the three-phase amplitude and phase. The parameters distributed by the parameter management unit to the multi-resolution singular value decomposition unit include: a singular value truncation threshold with a value between zero and one, a decomposition layer number with a value from one to a preset maximum layer number, a segmented window length with a length that is an integer power of two, and an identifier corresponding to the sampling rate and an identifier for the window alignment method. The parameters distributed by the parameter management unit to the variational mode decomposition unit include: the number of modes with values ​​ranging from one to a preset upper limit; positive real coefficients used to adjust decomposition convergence and bandwidth allocation; an initial center frequency constraint interval given for each mode, wherein the lower limit of the initial center frequency constraint interval is greater than zero, the upper limit is less than half of the sampling frequency, and a preset minimum interval is maintained between adjacent intervals; and a bandwidth upper limit, frequency quantization step size, and random seed for initialization are given for each mode.

2. The power distribution network monitoring system according to claim 1, characterized in that, The synchronous transient sampling unit includes a micro PMU, a primary sampler, an anti-aliasing filter, a digital-to-analog converter, a GPS timing module, and a communication interface. The timing module provides a unified time base and provides timing for the micro PMU. The micro PMU sends out sampling trigger parameters. The primary sampler outputs a three-phase secondary signal, which is sent to the digital-to-analog converter after being filtered by the anti-aliasing filter. The signal is then sampled synchronously according to a unified time scale. The micro PMU adds node ID, sampling rate, time scale, and timing quality to the sampling sequence and generates a data frame. The communication interface sends the data frame to the main station processor and data storage unit according to the protocol.

3. The power distribution network monitoring system according to claim 1, characterized in that, The phase mode decoupling unit stores configurable phase mode decoupling matrices and receives decoupling matrix version identifiers, line mode selection parameters, and matrix correction coefficients from the parameter management unit. It divides the three-phase transient traveling wave output by the synchronous transient sampling unit into frequency domains according to preset segment window lengths and sampling rates. After phase reference alignment, it calls the corresponding decoupling matrix according to the version identifier and performs element amplitude and phase correction on the matrix entries according to the matrix correction coefficients. It establishes voltage-side and current-side mapping relationships based on topology search and equipment list, performs matrix operations to obtain candidate modes, selects line mode 1 from the candidate modes according to the line mode selection parameters, outputs line mode 1 with the used matrix version and timestamp identifier, and returns the used version, correction coefficients, and segment index to the parameter management unit and data storage unit.

4. The power distribution network monitoring system according to claim 1, characterized in that, The parameters distributed by the parameter management unit to the symmetric differential energy operator calibration unit include: a short window length and a long window length determined by positive integer multiples of the sampling period, wherein the long window length is greater than the short window length; an energy threshold given in quantile index form, with values ​​between zero and one; a consistency tolerance time expressed as an integer multiple of the sampling period; an identifier for specifying the window function type; an identifier for specifying the standardization processing method; and a processing batch number for process traceability.

5. A power distribution network monitoring system according to claim 2, characterized in that, The parameters distributed by the parameter management unit to the time-scale fusion unit include: non-negative weights corresponding to each measurement point, with the sum of the weights agreed to be one; anomaly elimination and pairing rules consisting of limit codes, continuous frame counts, and pairing tolerances; topology index identifiers used to indicate network structure; and clock source identifiers used to indicate the time synchronization source. The parameters distributed by the parameter management unit to the data storage unit include: an archiving strategy consisting of archiving period, rolling window size, number of versions retained, and compression level; an index key consisting of node identifier, version identifier, start time, end time, decomposition level identifier, and modality number; and a naming rule identifier used to unify the generation of files and directories.

6. A power distribution network, employing the power distribution network monitoring system according to any one of claims 1-5, characterized in that, The system includes an electrical topology, comprising substation busbars, outgoing feeders, branch switches, ring network units, distribution transformers, grounding devices, and a line layer. The line layer includes overhead sections and cable sections. Measurement points are located at nodes within the line layer, and a synchronous transient sampling unit is installed at each measurement point. The synchronous transient sampling unit acquires a unified time scale and sends three-phase transient traveling wave and time scale information to the master station processor via a communication network. The master station processor houses the phase mode decoupling unit, MRSVD decomposition unit, VMD decomposition unit, characteristic statistics unit, parameter management unit, SDEO calibration unit, time scale fusion unit, and data storage unit. It also maintains topology indexes corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points. The data storage unit is connected to the master station processor to access raw traveling waves, decomposition and statistical records, and parameter indexes. Distributed power grid connection points and energy storage grid connection points are connected to the line layer via nodes and branch switches.

7. A power distribution network according to claim 6, characterized in that, The electrical topology is as follows: the outgoing ends of the substation busbar are electrically connected to the beginning of each outgoing feeder; each outgoing feeder is composed of an overhead section and a cable section connected sequentially, and the two are electrically connected at the nodes; branch switches are set at the nodes, so that the branch circuits are electrically connected to the main outgoing feeders at the nodes; ring network units are set between adjacent outgoing feeders or adjacent sections of the same outgoing feeder, and each of their interfaces is electrically connected to the corresponding feeder section at the nodes; the high-voltage side of the distribution transformer is electrically connected to the corresponding outgoing feeder at the nodes through branch circuits, and its grounding end is connected through a grounding device; the grounding device is electrically connected to the feeder circuit and the equipment casing at the nodes; each node is equipped with a measurement point, and at the measurement point, the primary sampling interface of the synchronous transient sampling unit is electrically connected to the three-phase conductors and connected to the grounding device; the communication interface of the synchronous transient sampling unit is connected to the main station processor through the communication network.

8. A power distribution network according to claim 6, characterized in that, The following structures and fields are set at the line layer and node side: An event ring buffer, pre-trigger and post-trigger segment identifiers, sampling rate identifiers, and timing quality identifiers are set in the synchronous transient sampling unit of each measurement point, and the node identifier and topology index are carried in the message header; Parameter channel identifiers and parameter encoding tables are set for the phase mode decoupling unit, MRSVD decomposition unit, VMD decomposition unit, SDEO calibration unit, time scale fusion unit, and data storage unit, respectively, on the master station processor side; a topic identifier and sequence number field are set for parameter distribution on the communication network side; and fiber optic or power line carrier is installed in the measurement box at the node. The system includes a coupling unit, configured with node codes and phase sequence identification markers; a time distribution node and fiber optic aggregation point are set up on the substation bus side; synchronous transient sampling unit installation positions, power supply interfaces, time distribution interfaces, and communication interfaces are set up at branch switch cabinets and ring network units; a primary sampling lead-out terminal and a grounding terminal are set up on the primary side of the distribution transformer, and a synchronous transient sampling unit installation position is set up in the secondary side cabinet; the master station processor maintains the topology index and parameter mapping table corresponding to nodes, branch switches, ring network units, distribution transformers, and measurement points, which is used to address-distribute and record parameter codes generated based on kurtosis, entropy, and kurtosis-entropy ratio.

Images