CN116400172A - Cloud-edge cooperative power distribution network fault detection method and system based on random matrix - Google Patents

Cloud-edge cooperative power distribution network fault detection method and system based on random matrix Download PDF

Info

Publication number
CN116400172A
CN116400172A CN202310595273.4A CN202310595273A CN116400172A CN 116400172 A CN116400172 A CN 116400172A CN 202310595273 A CN202310595273 A CN 202310595273A CN 116400172 A CN116400172 A CN 116400172A
Authority
CN
China
Prior art keywords
distribution network
power distribution
data
fault
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310595273.4A
Other languages
Chinese (zh)
Inventor
孙伟
杨建兴
吴应华
刘鑫
杜露露
石情倩
周亚
李奇越
李帷韬
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei University of Technology
State Grid Anhui Electric Power Co Ltd
Chuzhou Power Supply Co of State Grid Anhui Electric Power Co Ltd
Original Assignee
Hefei University of Technology
State Grid Anhui Electric Power Co Ltd
Chuzhou Power Supply Co of State Grid Anhui Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hefei University of Technology, State Grid Anhui Electric Power Co Ltd, Chuzhou Power Supply Co of State Grid Anhui Electric Power Co Ltd filed Critical Hefei University of Technology
Priority to CN202310595273.4A priority Critical patent/CN116400172A/en
Publication of CN116400172A publication Critical patent/CN116400172A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/08Locating faults in cables, transmission lines, or networks
    • G01R31/081Locating faults in cables, transmission lines, or networks according to type of conductors
    • G01R31/086Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/08Locating faults in cables, transmission lines, or networks
    • G01R31/088Aspects of digital computing
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04SSYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
    • Y04S10/52Outage or fault management, e.g. fault detection or location

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Locating Faults (AREA)

Abstract

The invention discloses a cloud edge collaborative power distribution network fault detection method and system based on a random matrix, wherein the method comprises the steps of collecting power distribution network data at the current moment; processing the power distribution network data by using a neural network model, calculating an error between an actual value and a predicted value, and deploying the neural network model in edge equipment; judging whether the power distribution network fails at the current moment based on the error; if not, re-inputting the power distribution network data at the current moment into the neural network model; if yes, uploading the power distribution network data at the current moment to a regional master station; in the regional master station, based on a random matrix, merging historical data of the power distribution network and real-time fault data uploaded by the edge equipment, and carrying out positioning analysis on faults; according to the method, the neural network is used for carrying out the primary fault analysis on the edge side, and the random matrix-based method is used for carrying out the centralized analysis on the cloud side, so that the fault positioning speed and accuracy are improved.

Description

Cloud-edge cooperative power distribution network fault detection method and system based on random matrix
Technical Field
The invention relates to the technical field of power distribution network fault detection, in particular to a cloud edge cooperative power distribution network fault detection method and system based on a random matrix.
Background
The low-voltage distribution network is a key for connecting power users, has a plurality of lines, a complex topological structure and very difficult fault analysis and positioning, and most of power failure accidents are caused by the faults of the distribution network. Therefore, if the fault can be accurately and rapidly subjected to timing positioning analysis, line maintenance can be timely performed, power supply of a user is recovered, and service quality of the user and reliability of power supply are improved.
The prior art mainly transmits the power data to the cloud background in a unified way, and performs fault location by using a matrix algorithm through centralized analysis of the power data, wherein the matrix algorithm is simple and quick, but is easy to judge errors when equivalent measurement information is distorted. With the development of active power distribution network technology, power distribution network lines and network topology are more complex, meanwhile, the development of power technology causes the proliferation of power data, and a centralized fault positioning method can lead to large data volume and high calculation complexity of cloud analysis. These reasons make it difficult to analyze faults only by a centralized analysis method and to accurately and quickly locate faults.
For example, the Chinese patent publication No. CN112098889A describes a single-phase earth fault positioning method based on a neural network and a feature matrix, wherein the scheme is to position the single-phase earth fault of a medium-voltage distribution network instead of analyzing real-time data of the distribution network, and then the scheme is to establish a node feature matrix through a field topological graph.
The Chinese patent document with publication number of CN114137358A describes a power transmission line fault diagnosis method based on a graph-convolution neural network, wherein the scheme adopts a cloud-edge cooperative mode to carry out fault diagnosis, but carries out information fusion operations such as feature extraction and the like on the edge side, and judges on the cloud by using the graph-convolution neural network, so that the centralized fault positioning method can lead to large data volume and high calculation complexity of cloud analysis.
Disclosure of Invention
The invention aims to solve the technical problem of how to realize rapid and accurate positioning of faults of a power distribution network.
The invention solves the technical problems by the following technical means:
the invention provides a cloud edge cooperative power distribution network fault detection method based on a random matrix, which comprises the following steps:
collecting power distribution network data at the current moment;
processing the power distribution network data by using a neural network model, and calculating an error between an output actual value and a predicted value, wherein the neural network model is deployed in edge equipment;
judging whether the power distribution network has faults at the current moment based on the error between the actual value and the predicted value;
if not, the power distribution network data at the next moment is input to the neural network model;
if yes, uploading the power distribution network data at the current moment to a regional master station;
and in the regional master station, based on a random matrix, merging historical data of the power distribution network and real-time fault data uploaded by the edge equipment, and carrying out positioning analysis on faults.
Further, the processing the power distribution network data by using the neural network model, and calculating an error between an output actual value and a predicted value includes:
carrying out normalization pretreatment on the power distribution network data to obtain treated power distribution network data;
processing the processed power distribution network data by utilizing the pre-trained neural network model, and calculating an error between an actual value and a predicted value;
the neural network model adopts a Bi-LSTM neural network, and the output of the Bi-LSTM neural network at the current time t is as follows:
Figure BDA0004246968230000021
Figure BDA0004246968230000022
wherein:
Figure BDA0004246968230000023
outputting a forward LSTM neural network at the time t; f is an activation function; />
Figure BDA0004246968230000024
And->
Figure BDA0004246968230000025
The weight and bias of the forward LSTM neural network are respectively; x is x t The input of the Bi-LSTM neural network at the moment t; h is a t-1 Representing the output of the Bi-LSTM neural network at time (t-1); />
Figure BDA0004246968230000026
For the output of the reverse LSTM neural network at time t, < >>
Figure BDA0004246968230000027
And->
Figure BDA0004246968230000028
The weights and biases of the reverse LSTM neural network; y is t Outputting Bi-LSTM neural network at t moment; g is an activation function; u and c are the weights and biases, respectively, of the bi-directional LSTM neural network.
Further, the determining, based on the error, whether the power distribution network fails at the current moment includes:
based on the error s between the actual value and the predicted value t Calculating an error s t Standard deviation sigma of (2);
error s at current time t t When the power distribution network data is larger than 3 sigma, determining the power distribution network data at the current time t as fault data;
error s at current time t t And when the power distribution network data is smaller than or equal to 3 sigma, determining the power distribution network data at the current time t as normal data.
Further, the re-inputting the power distribution network data at the current moment into the neural network model includes:
judging whether the current time t is an integer multiple of the transmission interval p or not;
if yes, taking the power distribution network data at the current moment t as power distribution network historical data, collecting the power distribution network data at the moment t+1, and re-inputting the power distribution network data into the neural network model;
if not, let t=t+1, gather the distribution network data of t+1 moment and reenter to the neural network model.
Further, the step of fusing, in the area master station, the historical data of the power distribution network and the real-time fault data uploaded by the edge device based on the random matrix to perform positioning analysis on the fault includes:
integrating the historical data of the power distribution network and the real-time fault data, and calculating a centralized sampling matrix
Figure BDA0004246968230000031
Based on the centralized sampling matrix
Figure BDA0004246968230000032
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree;
and positioning and analyzing the faults of the power distribution network, and determining fault points.
Further, the integration of the historical data of the power distribution network and the real-time fault data calculates a centralized sampling matrix
Figure BDA0004246968230000033
Comprising the following steps:
combining the power distribution network historical data with the real-time fault data to obtain N time sequences: { l 1 ,l 2 ,…,l n ,…,l N },l n ={l n,1 ,l n,2 ,…,l n,t ,…,l n,T },n=1,2,…,t,N,l n,t Representing a time sequence l n An element at time t;
in time sequence l n Adding white noise, and performing differential and normalization budget to obtain N preprocessed time sequences
Figure BDA0004246968230000034
The formula is:
l n,t =l n,t +e
Δl n,t =l n,t -l n,t-1
Figure BDA0004246968230000035
wherein: e is white noise meeting standard normal distribution; Δl n,t Representing a time sequence l n The difference between time t and time (t-1);
Figure BDA0004246968230000036
representation l n,t Data after pretreatment, E (Δl) n ) And var (Deltal) n ) Respectively Deltal n Arithmetic mean and variance of the medium elements; Δl n Represents a set, deltal n ={l 1,t -l 1,t-1 ,l 2,t -l 2,t-1 ,…,l n,1 -l n,t-1 ,…,l N,t -l N,t-1 };
N time sequences based on preprocessing
Figure BDA0004246968230000037
Construction of an NxT-dimensional sampling matrix>
Figure BDA0004246968230000038
Figure BDA0004246968230000039
T Transpose the symbol;
based on the sampling matrix
Figure BDA00042469682300000310
Computing a centralized sampling matrix->
Figure BDA00042469682300000311
Figure BDA00042469682300000312
Figure BDA00042469682300000313
Representing the sampling matrix of the kth sample data.
Further, the method is based on the centralized sampling matrix
Figure BDA0004246968230000041
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree, wherein the method comprises the following steps:
method for sampling matrix in the set by utilizing sliding window
Figure BDA0004246968230000042
In selecting a timing analysis matrix->
Figure BDA0004246968230000043
Figure BDA0004246968230000044
Is of size N x W s ,W s Representing the time;
solving the timing analysis matrix
Figure BDA0004246968230000045
Is>
Figure BDA0004246968230000046
The formula is as follows:
Figure BDA0004246968230000047
based on the empirical correlation matrix
Figure BDA0004246968230000048
Calculating the degree of deviation d s The calculation formula is as follows:
Figure BDA0004246968230000049
wherein:
Figure BDA00042469682300000410
respectively->
Figure BDA00042469682300000411
Maximum and minimum eigenvalues of (2); lambda (lambda) max 、λ min The theoretical maximum and minimum eigenvalues of the equi-scale random matrix, respectively, where +.>
Figure BDA00042469682300000412
c is the timing analysis matrix +.>
Figure BDA00042469682300000413
Is a rank ratio of (3);
the degree of deviation d of the spectrum s Threshold of degree of deviation from spectrum
Figure BDA00042469682300000414
Compare, at d s Is greater than->
Figure BDA00042469682300000415
And determining that the power distribution network fails at the current time t.
Further, the calculation formula of the spectrum deviation threshold value is as follows:
Figure BDA00042469682300000416
wherein d h Maximum spectrum deviation degree generated according to history data in a normal state; ρ d For the reserved margin, 0.5 is taken.
Further, the performing positioning analysis on the power distribution network fault to determine a fault point includes:
calculating a positioning analysis matrix of each edge device
Figure BDA00042469682300000417
The method comprises the following steps:
Figure BDA00042469682300000418
Figure BDA00042469682300000419
in the method, in the process of the invention,
Figure BDA00042469682300000420
time series measured for the ith said edge device +.>
Figure BDA00042469682300000421
Corresponding expansion matrix, E is AND +.>
Figure BDA00042469682300000422
Random noise matrices of equal size are used, T transpose the symbol;
positioning analysis matrix based on each edge device
Figure BDA00042469682300000423
Calculating a spectrum deviation improvement index d of the ith edge equipment is
Figure BDA00042469682300000424
Wherein d max D, for the maximum value of the spectrum deviation degree of all edge equipment of the power distribution network dif Is d max Differences, d, from the second largest value of spectral deviation of all nodes i The degree of spectral deviation for the ith edge device; comparing the spectral deviation improvement index d of each edge device is Improving the spectrum deviation degree by an index d is And determining the node where the edge equipment corresponding to the maximum value is located as a fault point.
In addition, the invention also provides a cloud edge collaborative low-voltage distribution network fault detection system based on a random matrix, the system comprises a regional master station and edge equipment arranged in each node in the region, a neural network model is deployed in the edge equipment, wherein:
the edge device includes:
the acquisition module is used for acquiring the power distribution network data at the current moment;
the preliminary fault analysis module is used for processing the power distribution network data by utilizing a neural network model, calculating an error between an actual value and a predicted value, and judging whether the power distribution network has a fault at the current moment based on the error;
the data uploading module is used for inputting the power distribution network data at the next moment into the neural network model when the output result of the preliminary fault analysis module is negative, and uploading the power distribution network data at the current moment to the regional master station when the output result of the preliminary fault analysis module is positive;
and the regional master station is used for fusing the historical data of the power distribution network and the real-time fault data uploaded by the edge equipment based on the random matrix and carrying out positioning analysis on the faults.
The invention has the advantages that:
(1) According to the method, the fault detection is carried out by utilizing a cloud-edge cooperative method, and because the neural network has high complexity and has requirements on equipment computing capacity, the neural network is used for carrying out primary fault analysis on the edge side, the random matrix-based method is utilized for carrying out centralized analysis on the cloud end, whether faults occur or not is judged rapidly on the edge side, and then the cloud end is used for centralized analysis, so that cloud end computing pressure is reduced, computing and load pressure caused by cloud end centralized processing are avoided, the computing speed is high, and the fault positioning speed and accuracy are improved.
(2) And determining a fault analysis matrix on the cloud side based on a random matrix method, positioning and analyzing the fault by taking the spectrum deviation degree as a characteristic index, and fully mining the space-time characteristics of the fault by carrying out high-dimensional matrix analysis on the data so as to improve the judgment accuracy.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
Fig. 1 is a schematic flow chart of a cloud-edge cooperative power distribution network fault detection method based on a random matrix in a first embodiment of the invention;
fig. 2 is a schematic overall flow diagram of a method for detecting faults of a cloud-edge cooperative power distribution network based on a random matrix in a first embodiment of the present invention;
fig. 3 is a schematic structural diagram of a cloud-edge cooperative power distribution network fault detection system based on a random matrix in a second embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions in the embodiments of the present invention will be clearly and completely described in the following in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, a first embodiment of the present invention provides a method for detecting a fault of a cloud-edge cooperative power distribution network based on a random matrix, where the method includes the following steps:
s10, collecting power distribution network data at the current moment;
it should be noted that, in this embodiment, the smart electric meter may be used to collect power distribution network data, where the power distribution network data includes: voltage, current, power, etc.
S20, processing the power distribution network data by using a neural network model, and calculating and outputting an error between an actual value and a predicted value, wherein the neural network model is deployed in edge equipment;
s30, judging whether the power distribution network has faults at the current moment based on the errors, if not, executing the step S40, and if so, executing the step S50;
s40, inputting the power distribution network data at the next moment into the neural network model;
s50, uploading the power distribution network data at the current moment to a regional master station;
s60, fusing historical data of the power distribution network and real-time fault data uploaded by the edge equipment based on a random matrix in the regional master station, and carrying out positioning analysis on faults.
According to the method, the edge equipment is arranged beside each node of the power distribution network, the cloud-edge cooperative method is utilized to detect faults, the neural network is utilized to conduct primary fault analysis on the edge side, the random matrix-based method is utilized to conduct centralized analysis on the cloud, whether faults occur or not is judged rapidly on the edge side, and then centralized analysis is conducted on the cloud, so that cloud computing pressure is reduced, computing and load pressure caused by cloud centralized processing are avoided, and fault positioning speed and accuracy are improved.
In one embodiment, the step S20: and processing the power distribution network data by using a neural network model, and calculating and outputting an error between an actual value and a predicted value, wherein the method specifically comprises the following steps of:
s21, carrying out normalization pretreatment on the power distribution network data to obtain processed power distribution network data;
the power distribution network data is transmitted to the edge equipment according to the transmission interval p, and in the edge equipment, the power distribution network data in the time sequence form is subjected to normalization preprocessing and then subjected to model prediction processing.
S22, processing the processed power distribution network data by utilizing the pre-trained neural network model, and calculating an error between an actual value and a predicted value;
the neural network model adopts a Bi-LSTM neural network, and the output of the Bi-LSTM neural network at the current time t is as follows:
Figure BDA0004246968230000071
Figure BDA0004246968230000072
wherein:
Figure BDA0004246968230000073
outputting a forward LSTM neural network at the time t; f is an activation function; />
Figure BDA0004246968230000074
And->
Figure BDA0004246968230000075
The weight and bias of the forward LSTM neural network are respectively; x is x t The input of the Bi-LSTM neural network at the moment t; h is a t-1 Representing the output of the Bi-LSTM neural network at time (t-1); />
Figure BDA0004246968230000076
For the output of the reverse LSTM neural network at time t, < >>
Figure BDA0004246968230000077
And->
Figure BDA0004246968230000078
The weights and biases of the reverse LSTM neural network; y is t Outputting Bi-LSTM neural network at t moment; g is an activation function; u and c are the weights and biases, respectively, of the bi-directional LSTM neural network.
The neural network model deployed in the edge device is a pre-trained model, and can be divided into a training set and a testing set after normalization preprocessing is performed according to the acquired power distribution network data, and the training set and the testing set are adopted to train and test the neural network model to obtain the network model deployed in the edge device.
In one embodiment, the step S30: based on the error, judging whether the power distribution network has faults at the current moment or not, and comprising the following steps:
s31, based on the error S between the actual value and the predicted value t The mean value mu and standard deviation sigma of the errors are calculated, and the calculation formula is as follows:
Figure BDA0004246968230000079
Figure BDA00042469682300000710
wherein: the error between the actual value and the predicted value at time t is s t The error sequence at time t is denoted as set S, and m is the number of S of the set.
S32, error S at current time t t When the power distribution network data is larger than 3 sigma, determining the power distribution network data at the current time t as fault data;
s33, error S at current time t t When the power distribution network data is smaller than or equal to 3 sigma, determining that the power distribution network data at the current time t is positiveConstant data.
In one embodiment, as shown in fig. 2, the step S40: inputting power distribution network data at the next moment into the neural network model, wherein the method comprises the following steps of:
judging whether the current time t is an integer multiple of the transmission interval p or not;
if yes, taking the power distribution network data at the current moment t as power distribution network historical data, collecting the power distribution network data at the moment t+1, and re-inputting the power distribution network data into the neural network model;
if not, let t=t+1, gather the distribution network data of t+1 moment and reenter to the neural network model.
In this embodiment, if the edge device detects a fault, the edge device uploads real-time data of the fault to the area master station; otherwise, the edge device judges whether to re-input the measurement data at the current moment into the Bi-LSTM neural network as history data according to the transmission interval p, when t=np, the power distribution network data at the current moment t is used as power distribution network history data, otherwise, let t=t+1, acquire the power distribution network data at the moment t+1, and then input the power distribution network data into the edge device.
In one embodiment, the step S60: and based on a random matrix, the regional master station fuses the historical data of the power distribution network and the real-time fault data uploaded by the edge equipment to perform positioning analysis on the faults, and specifically comprises the following steps of:
s61, integrating the historical data of the power distribution network and the real-time fault data, and calculating a centralized sampling matrix
Figure BDA0004246968230000081
S62, based on the centralized sampling matrix
Figure BDA0004246968230000082
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree;
s63, positioning analysis is carried out on the faults of the power distribution network, and fault points are determined.
According to the method, the power distribution network data are processed through the neural network model in the edge equipment, the abnormal data are primarily judged, then the primarily judged abnormal power distribution network data are sent to the cloud, positioning timing analysis of the abnormal data is conducted in the cloud in a concentrated mode, cloud computing pressure is reduced, and fault judgment can be conducted rapidly and accurately.
In one embodiment, the step S61: the historical data of the power distribution network and the real-time fault data are fused, and a centralized sampling matrix is calculated
Figure BDA0004246968230000083
The method comprises the following steps:
combining the power distribution network historical data with the real-time fault data to obtain N time sequences: { l 1 ,l 2 ,…,l n ,…,l N },l n ={l n,1 ,l n,2 ,…,l n,t ,…,l n,T },n=1,2,…,t,N,l n,t Representing a time sequence l n An element at time t;
in time sequence l n Adding white noise, and performing differential and normalization budget to obtain N preprocessed time sequences
Figure BDA0004246968230000084
The formula is:
l n,t =l n,t +e
Δl n,t =l n,t -l n,t-1
Figure BDA0004246968230000091
wherein: e is white noise meeting standard normal distribution; Δl n,t Representing a time sequence l n The difference between the time t and (t-1);
Figure BDA0004246968230000092
representation l n,t Data after pretreatment, E (Δl) n ) And var (Deltal) n ) Respectively Deltal n Arithmetic mean of the elements inValues and variances; Δl n Represents a set, deltal n ={l 1,t -l 1,t-1 ,l 2,t -l 2,t-1 ,…,l n,t -l n,t-1 ,…,l N,t -l N,t-1 -a }; by->
Figure BDA0004246968230000093
Composition of N time sequences after data pretreatment +.>
Figure BDA0004246968230000094
N time sequences based on preprocessing
Figure BDA0004246968230000095
Construction of an NxT-dimensional sampling matrix>
Figure BDA0004246968230000096
Figure BDA0004246968230000097
Based on the sampling matrix
Figure BDA0004246968230000098
Computing a centralized sampling matrix->
Figure BDA0004246968230000099
Figure BDA00042469682300000910
Figure BDA00042469682300000911
Representing the sampling matrix of the kth sample data.
In one embodiment, the step S62: based on the centralized sampling matrix
Figure BDA00042469682300000912
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree, wherein the method comprises the following steps ofThe steps are as follows:
method for sampling matrix in the set by utilizing sliding window
Figure BDA00042469682300000913
In selecting a timing analysis matrix->
Figure BDA00042469682300000914
Figure BDA00042469682300000915
Is of size N x W s ,W s Indicating the time, N<W s <T;
Solving the timing analysis matrix
Figure BDA00042469682300000916
Is>
Figure BDA00042469682300000917
The formula is as follows:
Figure BDA00042469682300000918
based on the empirical correlation matrix
Figure BDA00042469682300000919
Calculating the degree of deviation d s The calculation formula is as follows:
Figure BDA00042469682300000920
wherein:
Figure BDA00042469682300000921
respectively->
Figure BDA00042469682300000922
Maximum and minimum eigenvalues of (2); lambda (lambda) max 、λ min The theoretical maximum of the equi-scale random matrices respectivelyLarge, minimum eigenvalue, wherein->
Figure BDA00042469682300000923
c is the timing analysis matrix +.>
Figure BDA00042469682300000924
Is a rank ratio of (3);
the degree of deviation d of the spectrum s Threshold of degree of deviation from spectrum
Figure BDA00042469682300000925
Compare, at d s Is greater than->
Figure BDA00042469682300000926
And determining that the power distribution network fails at the current time t.
It should be understood that at d s Less than or equal to
Figure BDA00042469682300000927
And determining that the power distribution network does not have faults at the current time t.
In an embodiment, the calculation formula of the spectrum deviation threshold is:
Figure BDA00042469682300000928
wherein d h Maximum spectrum deviation degree generated according to history data in a normal state; ρ d For the reserved margin, 0.5 is taken.
In one embodiment, the step S63: positioning analysis is carried out on faults of the power distribution network, and fault points are determined, wherein the method comprises the following steps:
calculating a positioning analysis matrix of each edge device
Figure BDA0004246968230000101
The method comprises the following steps:
Figure BDA0004246968230000102
Figure BDA0004246968230000103
in the method, in the process of the invention,
Figure BDA0004246968230000104
time series measured for the ith said edge device +.>
Figure BDA0004246968230000105
Corresponding expansion matrix, E is AND +.>
Figure BDA0004246968230000106
Random noise matrices of equal size, wherein the time sequence +.>
Figure BDA0004246968230000107
Is>
Figure BDA0004246968230000108
The calculation process of (2) is the same.
Positioning analysis matrix based on each edge device
Figure BDA0004246968230000109
Calculating a spectrum deviation improvement index d of the ith edge equipment is
Figure BDA00042469682300001010
Wherein d max D, for the maximum value of the spectrum deviation degree of all edge equipment of the power distribution network dif Is d max Differences, d, from the second largest value of spectral deviation of all nodes i The degree of spectral deviation for the ith edge device; d, d s The solution process is the same
Comparing the spectral deviation improvement index d of each edge device is Improving the spectrum deviation degree by an index d is And determining the node where the edge equipment corresponding to the maximum value is located as a fault point.
In the positioning analysis process, the calculation process of the spectrum deviation degree of the edge device is as follows: firstly solving an empirical correlation matrix of a positioning analysis matrix, then calculating the spectral deviation degree of edge equipment according to the empirical correlation matrix, and d s The solving process is the same.
It should be noted that, because when the power distribution network fails,
Figure BDA00042469682300001011
the characteristic value distribution is changed, the maximum characteristic value and the minimum characteristic value are most sensitive to the change of the matrix, the spectrum deviation degree is selected as the characteristic index for judging the fault state, and the rapid and accurate positioning of the fault point can be realized on the basis of realizing the real-time analysis of the fault.
In addition, as shown in fig. 3, the second embodiment of the present invention further provides a cloud-edge collaborative distribution network fault detection system based on a random matrix, where the system includes a regional master station 10 and edge devices 20 disposed at nodes in a region, and a neural network model is deployed in the edge devices 20, where:
the edge device includes:
the acquisition module 21 is used for acquiring power distribution network data at the current moment;
the preliminary fault analysis module 22 is configured to process the power distribution network data by using a neural network model, calculate and output an error between an actual value and a predicted value, and determine whether the power distribution network fails at the current moment based on the error;
the data uploading module 23 is configured to input the power distribution network data at the next moment to the neural network model when the output result of the preliminary fault analysis module is no, and upload the power distribution network data at the current moment to the regional master station when the output result of the preliminary fault analysis module is yes;
the regional master station 10 is configured to fuse the historical data of the power distribution network and the real-time fault data uploaded by the edge device based on a random matrix, and perform positioning analysis on the fault.
According to the method, the cloud-edge cooperative method is utilized to detect faults, the neural network is utilized to conduct primary fault analysis on the edge side, the random matrix-based method is utilized to conduct centralized analysis on the cloud, whether faults occur or not is judged rapidly on the edge side, and then centralized analysis is conducted on the cloud, so that cloud computing pressure is reduced, computing and load pressure caused by cloud centralized processing are avoided, and fault positioning speed and accuracy are improved.
In an embodiment, the neural network model adopts a Bi-LSTM neural network, and the output of the Bi-LSTM neural network at the current time t is:
Figure BDA0004246968230000111
Figure BDA0004246968230000112
wherein:
Figure BDA0004246968230000113
outputting a forward LSTM neural network at the time t; f is an activation function; />
Figure BDA0004246968230000114
And->
Figure BDA0004246968230000115
The weight and bias of the forward LSTM neural network are respectively; x is x t The input of the Bi-LSTM neural network at the moment t; h is a t-1 Representing the output of the Bi-LSTM neural network at time (t-1); />
Figure BDA0004246968230000116
For the output of the reverse LSTM neural network at time t, < >>
Figure BDA0004246968230000117
And->
Figure BDA0004246968230000118
The weights and biases of the reverse LSTM neural network; y is t Outputting Bi-LSTM neural network at t moment; g is an activation function; u and c are the weights and biases, respectively, of the bi-directional LSTM neural network.
In one embodiment, the preliminary fault analysis module includes:
a calculation unit for calculating an error s between the actual value and the predicted value y The mean value mu and standard deviation sigma of the errors are calculated, and the calculation formula is as follows:
Figure BDA0004246968230000119
Figure BDA00042469682300001110
wherein: the error between the actual value and the predicted value at time t is s t The error sequence at time t is denoted as set S, and m is the number of S of the set.
A preliminary fault determination unit for determining an error s at a current time t y When the power distribution network data is larger than 3 sigma, determining the power distribution network data at the current time t as fault data; and an error s for at the current instant t t And when the power distribution network data is smaller than or equal to 3 sigma, determining the power distribution network data at the current time t as normal data.
In an embodiment, the area master station comprises:
the centralized sampling matrix construction module is used for fusing the historical data of the power distribution network with the real-time fault data and calculating a centralized sampling matrix
Figure BDA0004246968230000121
A timing analysis module for based on the centralized sampling matrix
Figure BDA0004246968230000122
Calculating the degree of deviation of the spectrum and the baseDetermining that the power distribution network fails at the current moment t according to the spectrum deviation degree;
and the positioning analysis module is used for positioning and analyzing the faults of the power distribution network and determining fault points.
In an embodiment, the centralized sampling matrix construction module is specifically configured to:
combining the power distribution network historical data with the real-time fault data to obtain N time sequences: { l 1 ,l 2 ,…,l n ,…,l N },l n ={l n,1 ,l n,2 ,…,l n,t ,…,l n,T },n=1,2,…,t,N,l n,t Representing a time sequence l n An element at time t;
in time sequence l n Adding white noise, and performing differential and normalization budget to obtain N preprocessed time sequences
Figure BDA0004246968230000123
The formula is:
l n,t =l n,t +e
Δl n,t =l n,t -l n,t-1
Figure BDA0004246968230000124
wherein: e is white noise meeting standard normal distribution; Δl n,t Representing a time sequence l n The difference between the time t and (t-1);
Figure BDA0004246968230000125
representation l n,t Data after pretreatment, E (Δl) n ) And var (Deltal) n ) Respectively Deltal n Arithmetic mean and variance of the medium elements; Δl n Represents a set, deltal n ={l 1,t -l 1,t-1 ,l 2,t -l 2,t-1 ,…,l n,t -l n,t-1 ,…,l N,t -l N,t-1 -a }; by->
Figure BDA0004246968230000126
Composition of N time sequences after data pretreatment +.>
Figure BDA0004246968230000127
N time sequences based on preprocessing
Figure BDA0004246968230000128
Construction of an NxT-dimensional sampling matrix>
Figure BDA0004246968230000129
Figure BDA00042469682300001210
Based on the sampling matrix
Figure BDA00042469682300001211
Computing a centralized sampling matrix->
Figure BDA00042469682300001212
Figure BDA00042469682300001213
Figure BDA00042469682300001214
Representing the sampling matrix of the kth sample data.
In an embodiment, the timing analysis module is specifically configured to:
method for sampling matrix in the set by utilizing sliding window
Figure BDA00042469682300001215
In selecting a timing analysis matrix->
Figure BDA00042469682300001216
Figure BDA00042469682300001217
Is of the size ofN×E s ,W s Indicating the time, N<W s <T;
Solving the timing analysis matrix
Figure BDA0004246968230000131
Is>
Figure BDA0004246968230000132
The formula is as follows:
Figure BDA0004246968230000133
based on the empirical correlation matrix
Figure BDA0004246968230000134
Calculating the degree of deviation d s The calculation formula is as follows:
Figure BDA0004246968230000135
wherein:
Figure BDA0004246968230000136
respectively->
Figure BDA0004246968230000137
Maximum and minimum eigenvalues of (2); lambda (lambda) max 、λ min The theoretical maximum and minimum eigenvalues of the equi-scale random matrix, respectively, where +.>
Figure BDA0004246968230000138
c is the timing analysis matrix +.>
Figure BDA0004246968230000139
Is a rank ratio of (3); />
The degree of deviation d of the spectrum s Threshold of degree of deviation from spectrum
Figure BDA00042469682300001310
Compare, at d s Is greater than->
Figure BDA00042469682300001311
And determining that the power distribution network fails at the current time t.
In an embodiment, the calculation formula of the spectrum deviation threshold is:
Figure BDA00042469682300001312
wherein d h Maximum spectrum deviation degree generated according to history data in a normal state; pi (II) d For the reserved margin, 0.5 is taken.
In an embodiment, the positioning analysis module is specifically configured to:
calculating a positioning analysis matrix of each edge device
Figure BDA00042469682300001313
The method comprises the following steps:
Figure BDA00042469682300001314
Figure BDA00042469682300001315
in the method, in the process of the invention,
Figure BDA00042469682300001316
time series measured for the ith said edge device +.>
Figure BDA00042469682300001317
Corresponding expansion matrix, E is AND +.>
Figure BDA00042469682300001318
Random noise matrixes with equal sizes;
based on each of the edge arrangementsStandby positioning analysis matrix
Figure BDA00042469682300001319
Calculating a spectrum deviation improvement index d of the ith edge equipment is
Figure BDA00042469682300001320
Wherein d max D, for the maximum value of the spectrum deviation degree of all edge equipment of the power distribution network dif Is d max Differences, d, from the second largest value of spectral deviation of all nodes i The degree of spectral deviation for the ith edge device;
comparing the spectral deviation improvement index d of each edge device is Improving the spectrum deviation degree by an index d is And determining the node where the edge equipment corresponding to the maximum value is located as a fault point. Calculating a positioning analysis matrix of each edge device
Figure BDA00042469682300001321
The method comprises the following steps:
Figure BDA00042469682300001322
Figure BDA00042469682300001323
in the method, in the process of the invention,
Figure BDA00042469682300001324
time series measured for the ith said edge device +.>
Figure BDA00042469682300001325
Corresponding expansion matrix, E is AND +.>
Figure BDA00042469682300001326
Random noise matrixes with equal sizes;
positioning analysis matrix based on each edge device
Figure BDA0004246968230000141
Calculating a spectrum deviation improvement index d of the ith edge equipment is
Figure BDA0004246968230000142
Wherein d max D, for the maximum value of the spectrum deviation degree of all edge equipment of the power distribution network dif Is d max Differences, d, from the second largest value of spectral deviation of all nodes i The degree of spectral deviation for the ith edge device;
comparing the spectral deviation improvement index d of each edge device is Improving the spectrum deviation degree by an index d is And determining the node where the edge equipment corresponding to the maximum value is located as a fault point.
It should be noted that, in other embodiments of the random matrix-based cloud edge collaborative power distribution network fault detection system or the implementation method thereof, reference may be made to the above embodiments of the method, and no redundancy is required here.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.

Claims (10)

1. A cloud edge cooperative power distribution network fault detection method based on a random matrix is characterized by comprising the following steps:
collecting power distribution network data at the current moment;
processing the power distribution network data by using a neural network model, and calculating and outputting an error between an actual value and a predicted value, wherein the neural network model is deployed in edge equipment;
based on the error, judging whether the power distribution network has a fault at the current moment;
if not, the power distribution network data at the next moment is input to the neural network model;
if yes, uploading the power distribution network data at the current moment to a regional master station;
and in the regional master station, based on a random matrix, merging historical data of the power distribution network and real-time fault data uploaded by the edge equipment, and carrying out positioning analysis on faults.
2. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 1, wherein the processing the distribution network data by using a neural network model, calculating and outputting an error between an actual value and a predicted value, comprises:
carrying out normalization pretreatment on the power distribution network data to obtain treated power distribution network data;
processing the processed power distribution network data by utilizing the pre-trained neural network model, and calculating an error between an actual value and a predicted value;
the neural network model adopts a Bi-LSTM neural network, and the output of the Bi-LSTM neural network at the current time t is as follows:
Figure FDA0004246968210000011
Figure FDA0004246968210000012
wherein:
Figure FDA0004246968210000013
outputting a forward LSTM neural network at the time t; f is an activation function; />
Figure FDA0004246968210000014
And->
Figure FDA0004246968210000015
The weight and bias of the forward LSTM neural network are respectively; x is x t The input of the Bi-LSTM neural network at the moment t; h is a t-1 Representing the output of the Bi-LSTM neural network at time (t-1); />
Figure FDA0004246968210000016
For the output of the reverse LSTM neural network at time t, < >>
Figure FDA0004246968210000017
And->
Figure FDA0004246968210000018
The weights and biases of the reverse LSTM neural network; y is t Outputting Bi-LSTM neural network at t moment; g is an activation function; u and c are the weights and biases, respectively, of the bi-directional LSTM neural network.
3. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 1, wherein the determining whether the distribution network has a fault at the current moment based on the error comprises:
based on the error s between the actual value and the predicted value t Calculating an error s t Standard deviation sigma of (2);
error s at current time t t When the power distribution network data is larger than 3 sigma, determining the power distribution network data at the current time t as fault data;
error s at current time t t And when the power distribution network data is smaller than or equal to 3 sigma, determining the power distribution network data at the current time t as normal data.
4. The method for detecting a fault in a cloud-edge collaborative distribution network based on a random matrix according to claim 1, wherein the re-inputting the distribution network data at the next moment into the neural network model comprises:
judging whether the current time t is an integer multiple of the transmission interval p or not;
if yes, taking the power distribution network data at the current moment t as power distribution network historical data, collecting the power distribution network data at the moment t+1, and re-inputting the power distribution network data into the neural network model;
if not, let t=t+1, gather the distribution network data of t+1 moment and reenter to the neural network model.
5. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 1, wherein the step of merging, in the regional master station, the historical data of the distribution network and the real-time fault data uploaded by the edge device based on the random matrix to perform positioning analysis on the fault comprises the following steps:
integrating the historical data of the power distribution network and the real-time fault data, and calculating a centralized sampling matrix
Figure FDA0004246968210000021
Based on the centralized sampling matrix
Figure FDA0004246968210000022
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree;
and positioning and analyzing the faults of the power distribution network, and determining fault points.
6. The method for detecting a fault in a cloud-edge collaborative distribution network based on a random matrix according to claim 5, wherein the integration of the historical data of the distribution network and the real-time fault data calculates a centralized sampling matrix
Figure FDA0004246968210000023
Comprising the following steps:
combining the power distribution network historical data with the real-time fault data to obtain N time sequences: { l 1 ,l 2 ,…,l n ,…,l N },l n ={l n,1 ,l n,2 ,…,l n,t ,…,l n,T N=1, 2, …, N, t=1, 2, …, T is time, l n,t Representing a time sequence l n An element at time t;
in time sequence l n Adding white noise, and performing differential and normalization budget to obtain N preprocessed time sequences
Figure FDA0004246968210000024
The formula is:
l n,t =l n,t +e
Δl n,t =l n,t -l n,t-1
Figure FDA0004246968210000025
wherein: e is white noise meeting standard normal distribution; Δl n,t Representing a time sequence l n The difference between time t and time (t-1);
Figure FDA0004246968210000031
representation l n,t Data after pretreatment, E (Δl) n ) And var (Deltal) n ) Respectively Deltal n Arithmetic mean and variance of the medium elements; Δl n Represents a set, deltal n ={l 1,t -l 1,t-1 ,l 2,t -l 2,t-1 ,…,l n,1 -l n,t-1 ,…,l N,t -l N,t-1 };
N time sequences based on preprocessing
Figure FDA0004246968210000032
Construction of an NxT-dimensional sampling matrix>
Figure FDA0004246968210000033
Figure FDA0004246968210000034
T is a transposed symbol;
based on the sampling matrix
Figure FDA0004246968210000035
Computing a centralized sampling matrix->
Figure FDA0004246968210000036
Figure FDA0004246968210000037
Figure FDA0004246968210000038
Representing the sampling matrix of the kth sample data.
7. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 5, wherein the concentrated sampling matrix is based on
Figure FDA0004246968210000039
Calculating a spectrum deviation degree, and determining that the power distribution network fails at the current time t based on the spectrum deviation degree, wherein the method comprises the following steps:
method for sampling matrix in the set by utilizing sliding window
Figure FDA00042469682100000310
In selecting a timing analysis matrix->
Figure FDA00042469682100000311
Figure FDA00042469682100000312
Is of size n x E s ,W s Representing the time;
solving the timing analysis matrix
Figure FDA00042469682100000313
Is>
Figure FDA00042469682100000314
The formula is as follows:
Figure FDA00042469682100000315
based on the empirical correlation matrix
Figure FDA00042469682100000316
Calculating the degree of deviation d s The calculation formula is as follows:
Figure FDA00042469682100000317
wherein:
Figure FDA00042469682100000318
respectively->
Figure FDA00042469682100000319
Maximum and minimum eigenvalues of (2); lambda (lambda) max 、λ min The theoretical maximum and minimum eigenvalues of the equi-scale random matrix, respectively, where +.>
Figure FDA00042469682100000320
c is the timing analysis matrix +.>
Figure FDA00042469682100000321
Is a rank ratio of (3);
the degree of deviation d of the spectrum s Threshold of degree of deviation from spectrum
Figure FDA00042469682100000322
Compare, at d s Is greater than->
Figure FDA00042469682100000323
And determining that the power distribution network fails at the current time t.
8. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 7, wherein the calculation formula of the spectrum deviation threshold is as follows:
Figure FDA00042469682100000324
wherein d h Maximum spectrum deviation degree generated according to history data in a normal state; ρ d For the reserved margin, 0.5 is taken.
9. The method for detecting the fault of the cloud-edge collaborative distribution network based on the random matrix according to claim 5, wherein the performing positioning analysis on the fault of the distribution network to determine the fault point comprises:
calculating a positioning analysis matrix of each edge device
Figure FDA00042469682100000325
The method comprises the following steps:
Figure FDA0004246968210000041
Figure FDA0004246968210000042
in the method, in the process of the invention,
Figure FDA0004246968210000043
time series measured for the ith said edge device +.>
Figure FDA0004246968210000044
Corresponding expansion matrix, E is AND +.>
Figure FDA0004246968210000045
Random noise matrices of equal size are used, T transpose the symbol;
positioning analysis matrix based on each edge device
Figure FDA0004246968210000046
Calculating a spectrum deviation improvement index d of the ith edge equipment is
Figure FDA0004246968210000047
Wherein d max D, for the maximum value of the spectrum deviation degree of all edge equipment of the power distribution network dif Is d max Differences, d, from the second largest value of spectral deviation of all nodes i The degree of spectral deviation for the ith edge device;
comparing the spectral deviation improvement index d of each edge device is Change the degree of deviation of spectrumIndex of advance d is And determining the node where the edge equipment corresponding to the maximum value is located as a fault point.
10. The utility model provides a cloud limit cooperatees low voltage distribution network fault detection system based on random matrix, its characterized in that, the system includes regional main website and arranges the marginal equipment of each node in the region, be deployed neural network model in the marginal equipment, wherein:
the edge device includes:
the acquisition module is used for acquiring the power distribution network data at the current moment;
the preliminary fault analysis module is used for processing the power distribution network data by utilizing a neural network model, calculating an error between an actual value and a predicted value, and judging whether the power distribution network has a fault at the current moment based on the error;
the data uploading module is used for inputting the power distribution network data at the next moment into the neural network model when the output result of the preliminary fault analysis module is negative, and uploading the power distribution network data at the current moment to the regional master station when the output result of the preliminary fault analysis module is positive;
and the regional master station is used for fusing the historical data of the power distribution network and the real-time fault data uploaded by the edge equipment based on the random matrix and carrying out positioning analysis on the faults.
CN202310595273.4A 2023-05-22 2023-05-22 Cloud-edge cooperative power distribution network fault detection method and system based on random matrix Pending CN116400172A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310595273.4A CN116400172A (en) 2023-05-22 2023-05-22 Cloud-edge cooperative power distribution network fault detection method and system based on random matrix

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310595273.4A CN116400172A (en) 2023-05-22 2023-05-22 Cloud-edge cooperative power distribution network fault detection method and system based on random matrix

Publications (1)

Publication Number Publication Date
CN116400172A true CN116400172A (en) 2023-07-07

Family

ID=87007828

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310595273.4A Pending CN116400172A (en) 2023-05-22 2023-05-22 Cloud-edge cooperative power distribution network fault detection method and system based on random matrix

Country Status (1)

Country Link
CN (1) CN116400172A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117725981A (en) * 2024-02-08 2024-03-19 昆明学院 Power distribution network fault prediction method based on optimal time window mechanism

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117725981A (en) * 2024-02-08 2024-03-19 昆明学院 Power distribution network fault prediction method based on optimal time window mechanism
CN117725981B (en) * 2024-02-08 2024-04-30 昆明学院 Power distribution network fault prediction method based on optimal time window mechanism

Similar Documents

Publication Publication Date Title
CN105846780B (en) A kind of photovoltaic module method for diagnosing faults based on decision-tree model
CN114167214B (en) Low-voltage distribution network edge-cloud cooperative fault detection method based on random matrix theory
CN104617574B (en) A kind of appraisal procedure of power system load region Transient Voltage Stability
CN103616579B (en) Fault diagnosis method used for intelligent substation secondary system
CN108199891B (en) Cps network attack identification method based on artificial neural network multi-angle comprehensive decision
CN107453484B (en) SCADA data calibration method based on WAMS information
CN108520267B (en) Hydrological telemetering data anomaly detection method based on space-time characteristics
CN106557546A (en) A kind of method and system extra-high voltage online monitoring data excavated and is evaluated
CN113239132B (en) Online out-of-tolerance identification method for voltage transformer
CN103103570B (en) Based on the aluminium cell condition diagnostic method of pivot similarity measure
CN116400172A (en) Cloud-edge cooperative power distribution network fault detection method and system based on random matrix
CN105719094A (en) State evaluation method of power transmission equipment
CN114065875A (en) Power grid fault identification system based on big data
CN111856209A (en) Power transmission line fault classification method and device
WO2020177706A1 (en) Method for determining fault section and fault time of power distribution network
CN113220751A (en) Metering system and evaluation method for multi-source data state quantity
CN109378835A (en) Based on the large-scale electrical power system Transient Stability Evaluation system that mutual information redundancy is optimal
CN109901022B (en) Power distribution network area positioning method based on synchronous measurement data
CN114879081A (en) Lightning damage area analysis method based on synchronous dynamic monitoring data of lightning arrester
CN106656603A (en) Cloud service trust evaluation method based on multi-parameter interval number multi-attribute decision-making
CN113644748A (en) System and method for simulating and monitoring substation equipment in area
CN106442830B (en) The detection method and system of gas content in transformer oil warning value
CN109193639B (en) Robust estimation method for power system
CN115575880B (en) Online detection method for metering error state of power transformer
CN109684749B (en) Photovoltaic power station equivalent modeling method considering operating characteristics

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination