CN-122017665-A - Low-voltage distribution network state monitoring method and system

Abstract

The application discloses a state monitoring method of a low-voltage power distribution network, and belongs to the technical field of power system monitoring. The method comprises the steps of disposing edge computing nodes on a power distribution site, synchronously collecting multiphase current time sequence data, multiphase voltage time sequence data and temperature data of a target line in real time, carrying out parallel monitoring on the time sequence data based on a plurality of trigger criteria such as a current mutation rate criterion, a current effective value out-of-limit criterion, a voltage effective value out-of-limit criterion and a temperature change rate criterion, triggering and storing waveform data sets in a time window before and after a trigger point when any trigger criterion is met, processing the waveform data sets and associated steady-state data, extracting waveform feature vectors, impedance feature vectors and temperature associated feature vectors, inputting the extracted feature vectors into a multi-criterion fusion diagnosis model disposed on the edge computing nodes, and carrying out fusion analysis on the multi-dimensional feature vectors to obtain diagnosis results containing fault type classification. The method can realize real-time state monitoring and fault diagnosis of the low-voltage distribution network and improve the safety and reliability of a distribution system.

Inventors

• GAO CHENGBO
• ZHANG JINGQIAO
• MI GENG
• WANG QIYUAN
• Dou Jiahao
• Liu Haoruo
• LIU SHIYU
• LIU CHAOYANG
• YU JIAFENG
• WANG WEIQI
• SHANG KUN
• WANG ZHI
• ZHANG YAN
• WANG HAO
• ZHAO WENLONG
• NIU CHENGLIN
• WANG YUCHEN
• GU YUNPENG

Assignees

• 国网陕西省电力有限公司
• 国网陕西省电力有限公司铜川供电公司

Dates

Publication Date

20260512

Application Date

20260416

Claims (10)

1. 1. A method of low voltage power distribution network condition monitoring, the method performed at an edge computing node deployed at a power distribution site, comprising: synchronously acquiring multiphase current time sequence data, multiphase voltage time sequence data and temperature data of a target line in real time; The method comprises the steps of carrying out parallel monitoring on time sequence data based on a plurality of preset trigger criteria, wherein the plurality of trigger criteria comprise at least one of a current mutation rate criterion and a temperature change rate criterion, and triggering and storing a waveform data set containing waveform data in a time window before and after a trigger point when any trigger criterion is met; Processing the waveform data set and the associated steady state data at the edge computing node to extract a waveform feature vector, an impedance feature vector and a temperature associated feature vector; and inputting the waveform characteristic vector, the impedance characteristic vector and the temperature-related characteristic vector into a multi-criterion fusion diagnosis model deployed at the edge computing node to obtain a diagnosis result containing fault type classification.
2. 2. The method of claim 1, wherein the plurality of trigger criteria includes at least two of a current slew rate criterion, a current effective value out-of-limit criterion, a voltage effective value out-of-limit criterion, and a temperature rate of change criterion.
3. 3. The method of claim 1, wherein the impedance feature vector comprises calculating a transient loop impedance during a fault based on the waveform dataset and comparing the transient loop impedance to a normal state impedance.
4. 4. The method for monitoring the state of the low-voltage distribution network according to claim 3, wherein the method is characterized in that the transient loop impedance during the fault is calculated based on the waveform data set, specifically, the method comprises the steps of selecting preset cycle data after the fault occurs, extracting fundamental wave components, and calculating the voltage and the current of a fault loop to obtain the transient loop impedance; The method comprises the steps of calculating transient loop impedance during faults based on the waveform data set, extracting harmonic voltage and harmonic current in a preset characteristic frequency band, calculating harmonic transient loop impedance, and taking the ratio of fundamental transient loop impedance to harmonic transient loop impedance as a part of the impedance characteristic vector.
5. 5. The method of claim 1, further comprising, after obtaining the diagnostic result, evaluating an operational performance of an associated circuit breaker based on the waveform dataset if the diagnostic result includes a short circuit type fault.
6. 6. The method of claim 5, wherein evaluating the operational performance of the associated circuit breaker based on the waveform dataset includes extracting a fault current profile and a circuit breaker operational time point from the waveform dataset, and comparing the fault current profile to a standard operational characteristic to generate a performance evaluation result.
7. 7. The method for monitoring the state of the low-voltage distribution network according to claim 1, wherein the waveform characteristic vector comprises at least one of an amplitude mutation rate, a phase deviation angle and a waveform distortion rate of each phase of current voltage extracted based on cycle comparison before and after a fault; the waveform distortion rate is the total harmonic distortion rate, and the calculation formula is as follows: Wherein the method comprises the steps of As the effective value of the fundamental wave, Is the effective value of the h harmonic.
8. 8. The method for monitoring the state of a low-voltage distribution network according to claim 7, wherein the waveform feature vector comprises a total harmonic distortion rate, and the method further comprises identifying that fault properties are arc faults or metallic faults according to the value of the total harmonic distortion rate and the change trend of the total harmonic distortion rate with time; The temperature-related characteristic vector comprises a temperature-current coupling coefficient, and the temperature-current coupling coefficient is used for representing the degree of correlation between temperature change and current change so as to distinguish overload heating faults from poor-contact heating faults of the equipment body.
9. 9. The method of claim 1, further comprising a model optimization step of forming a training data set from diagnostic reports and corresponding validation results from a plurality of edge computing nodes, for retraining and optimizing the multi-criteria fusion diagnostic model, and issuing optimized model parameters to the edge computing nodes.
10. 10. A low voltage power distribution network condition monitoring system deployed in an edge computing node of a power distribution site, comprising: The data acquisition and triggering module is used for synchronously acquiring multiphase current time sequence data, multiphase voltage time sequence data and temperature data of a target line in real time, carrying out parallel monitoring on the time sequence data based on a plurality of preset triggering criteria, and triggering and storing a waveform data set containing waveform data in a time window before and after a triggering point when any triggering criterion is met; The characteristic extraction module is used for processing the waveform data set and the associated steady-state data on the edge computing node and extracting waveform characteristic vectors, impedance characteristic vectors and temperature associated characteristic vectors; and the intelligent diagnosis module is embedded with a multi-criterion fusion diagnosis model and is used for receiving the waveform characteristic vector, the impedance characteristic vector and the temperature correlation characteristic vector and outputting a diagnosis result containing fault type classification.

Description

Low-voltage distribution network state monitoring method and system Technical Field The application relates to a method for monitoring the state of a power distribution network, in particular to a state monitoring method for realizing multi-criterion fusion fault diagnosis of a low-voltage power distribution network based on edge computing nodes. Background In the power system, a low-voltage distribution network is used as a key link for connecting a user side, and the running state of the low-voltage distribution network is directly related to the power supply reliability and the power quality. At present, state monitoring for a low-voltage power distribution network mainly depends on various protection devices and monitoring terminals deployed on site. In the conventional method, fault detection means based on a single electric quantity threshold value are widely adopted, for example, whether the line current exceeds a set value or not is monitored to judge an overcurrent fault, or voltage drop is detected to judge an undervoltage event. In addition, when detecting the electric quantity mutation, the monitoring device with the wave recording function can trigger the wave recording, and the collected original wave data is uploaded to a background analysis system at the main station side through a network, and the main station is used for completing further processing and fault judgment of the data. At present, the power distribution network fault monitoring technology mainly adopts a distributed monitoring method based on edge calculation. The Chinese patent application with publication number of CN120632538A discloses a distribution network line fault analysis method based on edge calculation, wherein sensors distributed on all nodes are used for collecting distribution network physical facilities and operation monitoring information in real time, and distribution network line layout anomaly degree and electric power deviation degree are extracted based on the edge calculation method. The Chinese patent application with publication number of CN120214478A provides a power distribution network fault monitoring system, which adopts edge calculation to perform primary processing and feature extraction on multi-source data and utilizes a multi-source data fusion technology to generate fault feature vectors. The Chinese patent application with publication number CN120539535A proposes a fault identification method based on low-voltage distribution network topology modeling and intelligent algorithm, and data such as voltage, current and the like are collected in real time through an Internet of things terminal, and a data set is constructed by combining an edge computing technology. In addition, chinese patent application with publication number CN110646677a describes a method for identifying topology and line impedance of a low-voltage distribution network in a transformer area, and performs waveform data sampling and topology identification by using an edge computing terminal. The Chinese patent application with publication number CN117454234A discloses a county power grid fault identification method based on cloud-edge cooperation, and fault identification pretreatment is carried out through edge computing nodes. However, existing low-voltage (below 400V) power distribution network fault monitoring methods still have significant shortcomings in coping with increasingly complex power distribution network operating environments. Firstly, the traditional fault capturing mechanism usually depends on a single current or voltage threshold value as a trigger condition, and the judging mode is difficult to comprehensively capture the complete transient process before and after the fault occurs, so that a plurality of transient faults or high-resistance faults are missed due to unobvious characteristics, and tripping events with unknown on-site frequent reasons are the direct manifestation of the limitation. Secondly, in the data processing mode, the method of uploading massive original waveform data to the master station occupies a large amount of communication bandwidth resources, the analysis and processing of the data are completely dependent on the master station side, obvious time delay exists in the whole diagnosis flow, and the requirement of real-time quick response to faults is difficult to meet. Moreover, the existing method focuses on simple threshold judgment of single electric quantity, lacks the capability of deep mining on abundant characteristics contained in waveform data, and particularly fails to effectively fuse non-electric quantity information such as temperature and the like which can reflect the state of the equipment body, so that the judgment on fault properties is rough. In addition, the diagnostic logic of most monitoring systems is relatively solidified at present, and depending on preset fixed values and rules, when the line characteristics change or a novel fault mode occurs, the system