CN112000923A - Power grid fault diagnosis method, system and equipment - Google Patents
Power grid fault diagnosis method, system and equipment Download PDFInfo
- Publication number
- CN112000923A CN112000923A CN202010676093.5A CN202010676093A CN112000923A CN 112000923 A CN112000923 A CN 112000923A CN 202010676093 A CN202010676093 A CN 202010676093A CN 112000923 A CN112000923 A CN 112000923A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- dcvpfi
- power
- cnn
- 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
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 47
- 239000011159 matrix material Substances 0.000 claims abstract description 22
- 238000013507 mapping Methods 0.000 claims abstract description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 70
- 238000006243 chemical reaction Methods 0.000 claims description 23
- 230000005283 ground state Effects 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 13
- 210000002569 neuron Anatomy 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 7
- 239000002131 composite material Substances 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 238000012800 visualization Methods 0.000 abstract description 4
- 238000004088 simulation Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 12
- 238000012545 processing Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000003860 storage Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000001012 protector Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
-
- Y—GENERAL 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS 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
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
-
- Y—GENERAL 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS 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
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
Abstract
The invention provides a power grid fault diagnosis method based on dynamic computer visualization load flow pictures, which comprises the steps of forming load flow data into a two-dimensional computer visualization load flow data matrix CVPFM; mapping the CVPFM to an HSV color space to obtain a computer visualization flow picture CVPFI; and combining the CVPFIs before and after the fault into the DCVPFI. The DCVPFI is used for replacing the power grid load flow data in the numerical form as the input of the CNN, so that the space and time information contained in the load flow can be better extracted, and the fault position of the system can be judged; compared with the traditional mode of acquiring the tidal current data on line from the SCADA as a data source, the method can simulate the infrequent fault condition of the simulation power grid, and strengthens the generalization capability of the CNN to the fault state of the system.
Description
Technical Field
The invention relates to the field of dispatching control of power systems, in particular to a power grid fault diagnosis method, system and equipment.
Background
The power system comprises a plurality of devices such as generators, power transmission lines and buses, and operators can monitor the operation state of the system through a relay protector, a circuit breaker and a communication device. When the power system receives disturbance, an operator can perform fault diagnosis and analysis through Data provided by a Supervisory Control and Data Acquisition (SCADA) system. However, when a fault occurs, the SCADA transmits a large amount of alarm information to the control center operator in a short time, which burdens the dispatcher and makes fault diagnosis difficult. In addition, the grid measurement device is likely to be subjected to power outage or noise interference in an actual operating environment. Under the first condition, the operator cannot judge whether the line is broken or not, and cannot find a fault point. In the second case, even if the line has been disconnected, the measurement device still displays a false measurement value that appears "real", instead of the ideal zero value that follows the electrical law, which makes the fault point data obtained by the operator completely erroneous. Both of these situations can mislead the operating personnel to a large extent, making fault diagnosis difficult. Therefore, there is a need to develop a method for fault diagnosis of a power grid operating in a severe interference environment.
In the field of fault diagnosis research, researchers at home and abroad propose a plurality of methods based on artificial intelligence theory, wherein the methods comprise a Petri network, a Bayesian network, an optimization method, a causal network, an expert system and the like. Among them, the application of Artificial Neural Network (ANN) is relatively wide, but because of the self-defect of the Network, it has the problems of "dimension disaster" and poor portability in the practical application. In recent years, deep learning has been rapidly developed, and related researchers have proposed deep learning models such as a stacked automatic encoder, a recurrent Neural Network, a deep belief Network, and a Convolutional Neural Network (CNN). Among these models, CNN is a new technology with wide application and maturity, which can automatically extract effective feature information from raw data (especially from images). Meanwhile, compared with the traditional ANN method, the sparse connection and weight sharing technology can greatly simplify the network and accelerate the training speed of the neural network. Furthermore, the CNN can capture the features of the image in a more macroscopic manner, and can still recognize the input well even when it is disturbed by noise or is incomplete. How to use the CNN technology to rapidly and effectively judge the power grid fault becomes a problem which needs to be solved urgently in power grid maintenance.
Disclosure of Invention
In order to achieve the purpose, the invention provides a power grid fault diagnosis method, a system and equipment, which are reasonable in design, simple in structure and convenient to use, and carry out fault diagnosis on a power grid by using a convolutional neural network based on dynamic computer visualization tide pictures and comprehensively considering active power changes before and after a fault.
The invention provides a power grid fault diagnosis method, which comprises the following steps:
s1, when the power grid fails, correspondingly converting the power flow data before the failure and the power flow data after the failure into DCVPFI data according to a preset conversion rule;
s2, inputting the DCVPFI data into a Convolutional Neural Network (CNN) model pre-trained by massive DCVPFI data samples to obtain fault probabilities of different fault states; and determining a power grid fault diagnosis result according to the fault state with the maximum fault probability.
Further, the step S2 includes the following steps:
s201, two groups of convolution-activation-pooling structures are adopted in the CNN model, and fault characteristics of different fault states are extracted;
s202, inputting the extracted fault features into a full-connection layer containing a plurality of neurons, and synthesizing the fault features to obtain a classification decision;
s203, the neurons of the output layer respectively correspond to different fault states of the power grid, and the classification decision output by the neurons is converted into the fault probability of the corresponding position given by the CNN.
Further, the convolutional neural network CNN model is obtained by the following steps:
generating a mass DCVPFI data source in an off-line manner on the basis of the ground state data of the power grid system;
and constructing by using the generated mass DCVPFI data source to obtain mass DCVPFI data samples, and training the CNN model to obtain a pre-trained convolutional neural network CNN model.
Still further, the method for generating the massive DCVPFI data source in an off-line mode on the basis of the ground state data of the power grid system comprises the following steps:
a. endowing each load with random active power and reactive power by the following formula, and simulating the fluctuation of the load power in a real environment;
in the formula (I), the compound is shown in the specification,andgenerating random active and reactive power; pLiAnd QLiThe base state active power and reactive power of the load; k is a radical of1And k2Two parameters generated randomly; the function rand represents the random number in a range of a subsequent interval;generating a set random active power for the generated set; pGiThe ground state active power of the unit; k is a randomly generated parameter;
b. carrying out load flow calculation on the topological structure of the fault-free system by using the random power of the generator and the load to obtain the active power of each branch of the system before fault;
c. randomly acquiring faults from a preset fault set, and performing load flow calculation by adopting a system topology after the faults to acquire active power of each branch of the system after the faults;
d. summarizing the active power of the branch circuits before and after the fault and the power of each generator set and load in the network at the moment to obtain a DCVPFI data source;
e. and repeating the steps a to d to obtain a massive DCVPFI data source which is generated by the ground state data of the power grid system and covers all fault states.
Further, when the load flow data before the fault is converted into the DCVPFI data according to the conversion rule, the method includes the following steps:
s101, assigning values to the corresponding pixel blocks by the power flow data before the fault;
s102, simulating a topological structure of a network, and arranging pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (composite video frequency domain) which is used as CNN (common channel network) input and is in one-to-one correspondence with load flow distribution before grid faults;
s103, mapping the CVPFM before the fault to an HSV color space to obtain the CVPFI before the fault.
Further, when the power flow data after the fault is converted into the DCVPFI data according to the conversion rule, the method includes the following steps:
s111, assigning values to the corresponding pixel blocks of the power flow data after the fault;
s112, simulating a topological structure of a network to arrange pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (composite video frequency domain) which is used as CNN (CNN) input and is in one-to-one correspondence with load flow distribution after power grid faults;
s113, mapping the CVPFM after the fault to an HSV color space to obtain a CVPFI after the fault; during mapping, setting pixel values corresponding to fault lines in pixel blocks as random values according to the following formula;
Pl=rand(-1,1)
in the formula, PlIs the random active power generated.
Further, after the power flow data before the fault and the power flow data after the fault are converted into corresponding CVPFIs, the original two-dimensional CVPFM is expanded into a three-dimensional matrix, and the third dimension of the three-dimensional matrix is the frame number of the dynamic CVPFI; the CVPFI before the fault forms a first frame, the CVPFI after the fault forms a second frame, and the first frame and the second frame are combined to form DCVPFI data or DCVPFI data samples corresponding to a DCVPFI data source; and forming a mass of DCVPFI data samples corresponding to the mass of DCVPFI data sources.
A power grid fault diagnosis system comprises a data conversion module and a fault diagnosis module;
the data conversion module is used for correspondingly converting the power flow data before the fault and the power flow data after the fault into DCVPFI data according to a preset conversion rule when the power grid has the fault; inputting DCVPFI data into a CNN model;
the fault diagnosis module is used for inputting the DCVPFI data into a Convolutional Neural Network (CNN) model which is trained by massive DCVPFI data samples in advance, outputting fault probabilities of different fault states, and determining a power grid fault diagnosis result according to the fault state with the maximum fault probability.
Preferably, the system also comprises a mass sample module which is used for generating a mass DCVPFI data source in an off-line manner on the basis of the ground state data of the power grid system; and constructing and obtaining a mass of DCVPFI data samples by using the generated mass of DCVPFI data sources.
A power grid fault diagnosis device comprises a power grid fault diagnosis module,
a memory for storing a computer program;
a processor for implementing the steps of the grid fault diagnosis method as described above when executing the computer program.
Compared with the prior art, the Power grid fault diagnosis method, the system, the equipment and the storage medium provided by the invention have the advantages that the network topology of the active Power Flow before and after the fault combined with the Power grid structure is converted into a Dynamic Computer Visual Power Flow Image (DCVPFI), and then the DCVPFI is input into a Convolutional Neural Network (CNN) for fault diagnosis; the DCVPFI is used for replacing the power grid load flow data in the numerical form as the input of the CNN, so that the spatial and time information contained in the load flow can be better extracted, and the fault position of the system can be judged.
Furthermore, a mass DCVPFI data source is generated offline on the basis of the basic state data of the system, compared with the traditional mode that tidal current data is acquired online from an SCADA as a data source, the fault condition which is not frequently generated in a simulation power grid can be simulated, and the generalization capability of the CNN to the fault state of the system is enhanced.
Further, forming a two-dimensional computer-visualized power flow matrix (CVPFM) by the power flow data; mapping the CVPFM to an HSV color space to obtain a Computer-visualized Power Flow Image (CVPFI); and combining the CVPFIs before and after the fault into the DCVPFI.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
fig. 1 is a flowchart of a power grid fault diagnosis method according to an embodiment of the present invention;
fig. 2 is a flowchart of acquiring a DCVPFI data source in an embodiment of a power grid fault diagnosis method provided by the present invention;
fig. 3(a) is a diagram of an active power distribution of a network before a fault in an embodiment of a grid fault diagnosis method provided by the present invention;
fig. 3(b) is a diagram of an active power distribution of a network after a fault in an embodiment of a grid fault diagnosis method provided by the present invention;
FIG. 4 is a CVPFM before failure in an embodiment of the grid failure diagnosis method provided by the present invention;
FIG. 5 is a CVPFI before failure in one embodiment of the grid fault diagnosis method provided by the present invention;
fig. 6(a) is a CVPFI when the power of the fault point after the fault is an ideal value in an embodiment of the grid fault diagnosis method provided by the present invention;
fig. 6(b) is a CVPFI when the power of the fault point after the fault is a random value in an embodiment of the grid fault diagnosis method provided by the present invention;
fig. 7 is a DCVPFI screenshot in an embodiment of the power grid fault diagnosis method provided by the present invention;
fig. 8 is a process diagram of CNN identifying dynamic DCVPFI in an embodiment of the power grid fault diagnosis method provided by the present invention;
fig. 9 is a schematic structural diagram of a power grid fault diagnosis system provided by the present invention.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further details of the invention. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
As shown in fig. 1, a method for diagnosing a grid fault includes the following steps:
s1, when the power grid fails, correspondingly converting the power flow data before the failure and the power flow data after the failure into DCVPFI data according to a preset conversion rule;
s2, inputting the DCVPFI data into a Convolutional Neural Network (CNN) model pre-trained by massive DCVPFI data samples to obtain fault probabilities of different fault states; and determining the result of the power grid fault diagnosis according to the fault state with the maximum fault probability.
As shown in fig. 2, fig. 3(a) and fig. 3(b), the method for generating a massive DCVPFI data source offline based on the system ground state data comprises the following steps;
a. endowing each load with random active power and reactive power by the following formula, and simulating the fluctuation of the load power in a real environment;
in the formula (I), the compound is shown in the specification,andgenerating random active and reactive power; pLiAnd QLiThe base state active power and reactive power of the load; k is a radical of1And k2Two parameters generated randomly; the function rand represents the random number in a range of a subsequent interval;generating a set random active power for the generated set; pGiThe ground state active power of the unit; k is a randomly generated parameter;
b. carrying out load flow calculation on the topological structure of the fault-free system by using the random power of the generator and the load to obtain the active power of each branch of the system before fault;
c. randomly acquiring faults from the fault set, and performing load flow calculation by adopting a system topology after the faults to acquire active power of each branch of the system after the faults;
d. the active power of the branch before and after the fault and the power of each generator set and the load in the network at the moment are gathered and packed into a DCVPFI data source;
e. and (d) continuously repeating the steps a-d, and generating a mass DCVPFI data source covering all fault conditions by the network ground state parameters.
Wherein fig. 3(a) shows the active power of the network generator, load, transformer and transmission line before the failure of the IEEE9 system, and the active power of the whole network is redistributed after the line L4-5 is disconnected, as shown in fig. 3 (b). Storing both in a structured array, a source of DCVPFI data is obtained. And the massive DCVPFI data source can be obtained by continuously changing the power of the unit, the load and the fault type of the network.
As shown in fig. 4-7, when each DCVPFI data source is converted into DCVPFI data, the structured array is converted into a picture; when the grid fails, and when DCVPFI data samples are formed in the previous period, the same conversion rule is adopted, which includes the following two parts.
The first part comprises the following steps when the power flow data before the fault is converted into DCVPFI data according to the conversion rule:
s101, assigning values to the corresponding pixel blocks by the power flow data before the fault;
s102, simulating a topological structure of a network to arrange pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (continuous variable frequency-domain data matrix) which can be used as CNN (common channel network) input and is in one-to-one correspondence with load flow distribution before grid faults;
s103, mapping the CVPFM before the fault to the HSV color space to obtain the CVPFI before the fault.
The second part is that when the power flow data after the fault is converted into DCVPFI data according to the conversion rule, the former two steps are the same as the processing method of the power flow data before the fault,
s111, assigning values to the corresponding pixel blocks of the power flow data after the fault;
s112, simulating a topological structure of a network to arrange pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (composite video frequency domain) which is used as CNN (CNN) input and is in one-to-one correspondence with load flow distribution after power grid faults;
the difference is that special processing is carried out on the pixel block corresponding to the fault; when a fault occurs, the power of the fault line is zero, so that the corresponding pixel block does not exist. Theoretically, after a CNN is trained using a large number of such CVPFI, the CNN can obtain good fault identification capability. However, because the missing of the faulty device pixel block is a very obvious feature, in the iterative process, the CNN continuously increases the weight connected to the location of the faulty device pixel block, and the identification of the feature is strengthened but neglected. Then, in an actual environment, once a problem occurs in the fault point measurement (zero value is not displayed), the CNN cannot give an accurate identification result; therefore, the method further includes step S113 of mapping the failed CVPFM to the HSV color space to obtain the failed CVPFI; during mapping, the pixel value corresponding to the fault line is set as a random value according to the following formula:
Pl=rand(-1,1)
in the formula, Pl is the random active power generated. The specific conversion rules are shown in the following table:
table 1 is a table of conversion rules from the electrical device to the pixel block.
After converting the power flow data before the fault and the power flow data after the fault into corresponding CVPFI data, expanding the original two-dimensional CVPFM into a three-dimensional matrix, wherein the third dimension of the three-dimensional matrix is the frame number of the dynamic CVPFI; the first frame formed by the CVPFI before the fault and the second frame formed by the CVPFI after the fault are combined to form a complete DCVPFI data or a DCVPFI data sample corresponding to a DCVPFI data source; and forming a mass of DCVPFI data samples corresponding to the mass of DCVPFI data sources.
The calculation results of the power before and after the failure shown in fig. 3(a) and 3(b) are taken as an example. As shown in fig. 4, before a fault occurs, each device in the power grid and its active power are converted into a pixel block according to a preset rule, and the generated CVPFM (28 × 28 pixels) is finally converted into a CVPFI before the fault (as shown in fig. 5).
As shown in fig. 6(a) and 6(b), the CVPFI after failure has an ideal zero value for the failed pixel block in fig. 6(a), and a point a in fig. 6(b) indicates that the value of the failed pixel block is a random value. Combining the pre-fault CVPFI and post-fault CVPFI results in the DCVPFI shown in fig. 6, whose mathematical nature is a three-dimensional matrix of 28 × 2. And obtaining the DCVPFI sample set required by the CNN training by using the mass DCVPFI data source obtained previously according to the conversion process.
As shown in fig. 7, a CNN structure model suitable for DCVPFI is proposed on the basis of the LeNet-5 network, and the following steps are adopted in step S3:
s201, two groups of convolution-activation-pooling structures are adopted in the CNN model, and fault characteristics are extracted;
s202, inputting the extracted fault features into a full-connection layer containing a plurality of neurons, and synthesizing the fault features to obtain a classification decision;
s203, the neurons of the output layer respectively correspond to different fault states of the network, and the classification decision output by the neurons is converted into the fault probability of the position given by the CNN.
As shown in fig. 8, the present invention employs a classical CNN structure. The neurons of the input layer are arranged in a cube of 28 × 2 (one corresponds to 28 × 2 pixel points of the DCVPFI), then the spatial structure characteristics of the DCVPFI are extracted by using two convolution-activation-pooling combinations, and finally, 7 longitudinally arranged neurons of the network output layer respectively correspond to seven fault types including normal conditions of the network. Specific structural parameters of CNN are shown in the following table.
Table 2 is a structural parameter table of CNN.
The procedure for CNN to identify DCVPFI is as follows: after a DCVPFI sample is input into the CNN, the CNN extracts the shallow space-time structural features of CVPFI through the 1 st "convolution-activation-Pooling" and expresses them into 8 feature maps with 12 × 12 size, as shown in the Pooling1 layer in fig. 8. After 2 nd "convolution-activation-Pooling", the CNN extracted the deep space-time structural features of CVPFI, as shown by the Pooling2 layer in fig. 8, the size of the feature map was 4 x 4, for a total of 16. And finally, converting the 16 characteristic graphs obtained after the 2 nd pooling into column vectors and inputting the column vectors into a full-connection layer. A large number of DCVPFI and their labels (fault types, including normal conditions) are input into the CNN, and the CNN will continuously adjust the parameters of the network by itself, so that when one DCVPFI is input, the output of the network is continuously close to the label of the DCVPFI. When the identification rate of the network is high enough, the CNN has the overall sensing capability of the DCVPFI, and further can perform fault diagnosis on the power grid by using measurement data provided by the SCADA under the actual condition.
As shown in fig. 9, the present invention further provides a grid fault diagnosis system, which includes a data conversion module 1 and a fault diagnosis module 2;
the data conversion module 1 is used for correspondingly converting the power flow data before the fault and the power flow data after the fault into DCVPFI data according to a preset conversion rule when the power grid has the fault; inputting DCVPFI data into a CNN model;
and the fault diagnosis module 2 is used for inputting the DCVPFI data into a Convolutional Neural Network (CNN) model which is trained by massive DCVPFI data samples in advance, outputting fault probabilities of different fault states, and obtaining the fault state with the maximum fault probability as a power grid fault diagnosis result.
The system comprises a mass sample module and a data processing module, wherein the mass sample module is used for generating a mass DCVPFI data source in an off-line mode on the basis of the ground state data of the power grid system; and constructing and obtaining a mass of DCVPFI data samples by using the generated mass of DCVPFI data sources. And further, the generated massive DCVPFI data samples can be used for training a Convolutional Neural Network (CNN) model.
In practice, the operations a to e in the above method are performed in the mass sample module, the data conversion module performs the operations S101 to S103 in the above method, and the CNN model performs the operations S111 to S113.
The invention also provides a power grid fault diagnosis device, which comprises a memory, a control unit and a power grid, wherein the memory is used for storing a computer program; a processor for implementing the steps of a grid fault diagnosis method as in any of the above embodiments when executing the computer program.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be appreciated by those skilled in the art that the invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The embodiments disclosed above are therefore to be considered in all respects as illustrative and not restrictive. All changes which come within the scope of or equivalence to the invention are intended to be embraced therein.
Claims (10)
1. A power grid fault diagnosis method is characterized by comprising the following steps:
s1, when the power grid fails, correspondingly converting the power flow data before the failure and the power flow data after the failure into DCVPFI data according to a preset conversion rule;
s2, inputting the DCVPFI data into a Convolutional Neural Network (CNN) model pre-trained by massive DCVPFI data samples to obtain fault probabilities of different fault states; and determining a power grid fault diagnosis result according to the fault state with the maximum fault probability.
2. The grid fault diagnosis method according to claim 1, wherein step S2 includes the following steps:
s201, two groups of convolution-activation-pooling structures are adopted in the CNN model, and fault characteristics of different fault states are extracted;
s202, inputting the extracted fault features into a full-connection layer containing a plurality of neurons, and synthesizing the fault features to obtain a classification decision;
s203, the neurons of the output layer respectively correspond to different fault states of the power grid, and the classification decision output by the neurons is converted into the fault probability of the corresponding position given by the CNN.
3. The grid fault diagnosis method according to claim 1, wherein the Convolutional Neural Network (CNN) model is obtained by:
generating a mass DCVPFI data source in an off-line manner on the basis of the ground state data of the power grid system;
and constructing by using the generated mass DCVPFI data source to obtain mass DCVPFI data samples, and training the CNN model to obtain a pre-trained convolutional neural network CNN model.
4. The power grid fault diagnosis method according to claim 3, wherein the step of generating massive DCVPFI data sources offline on the basis of the ground state data of the power grid system comprises the following steps:
a. endowing each load with random active power and reactive power by the following formula, and simulating the fluctuation of the load power in a real environment;
in the formula (I), the compound is shown in the specification,andgenerating random active and reactive power; pLiAnd QLiThe base state active power and reactive power of the load; k is a radical of1And k2Two parameters generated randomly; the function rand represents the random number in a range of a subsequent interval;generating a set random active power for the generated set; pGiThe ground state active power of the unit; k is a randomly generated parameter;
b. carrying out load flow calculation on the topological structure of the fault-free system by using the random power of the generator and the load to obtain the active power of each branch of the system before fault;
c. randomly acquiring faults from a preset fault set, and performing load flow calculation by adopting a system topology after the faults to acquire active power of each branch of the system after the faults;
d. summarizing the active power of the branch circuits before and after the fault and the power of each generator set and load in the network at the moment to obtain a DCVPFI data source;
e. and repeating the steps a to d to obtain a massive DCVPFI data source which is generated by the ground state data of the power grid system and covers all fault states.
5. The grid fault diagnosis method according to claim 1 or 4, wherein the step of converting the pre-fault power flow data into DCVPFI data according to the conversion rule comprises the following steps:
s101, assigning values to the corresponding pixel blocks by the power flow data before the fault;
s102, simulating a topological structure of a network, and arranging pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (composite video frequency domain) which is used as CNN (common channel network) input and is in one-to-one correspondence with load flow distribution before grid faults;
s103, mapping the CVPFM before the fault to an HSV color space to obtain the CVPFI before the fault.
6. The grid fault diagnosis method according to claim 1 or 4, wherein the step of converting the fault load flow data into DCVPFI data according to the conversion rule comprises the following steps:
s111, assigning values to the corresponding pixel blocks of the power flow data after the fault;
s112, simulating a topological structure of a network to arrange pixel blocks on a two-dimensional zero-value pixel matrix to form a two-dimensional computer vision load flow data matrix CVPFM (composite video frequency domain) which is used as CNN (CNN) input and is in one-to-one correspondence with load flow distribution after power grid faults;
s113, mapping the CVPFM after the fault to an HSV color space to obtain a CVPFI after the fault; during mapping, setting pixel values corresponding to fault lines in pixel blocks as random values according to the following formula;
Pl=rand(-1,1)
in the formula, PlIs the random active power generated.
7. The grid fault diagnosis method according to claim 6,
after converting the power flow data before the fault and the power flow data after the fault into corresponding CVPFI, expanding the original two-dimensional CVPFM into a three-dimensional matrix, wherein the third dimension of the three-dimensional matrix is the frame number of the dynamic CVPFI; the CVPFI before the fault forms a first frame, the CVPFI after the fault forms a second frame, and the first frame and the second frame are combined to form DCVPFI data or DCVPFI data samples corresponding to a DCVPFI data source; and forming a mass of DCVPFI data samples corresponding to the mass of DCVPFI data sources.
8. The power grid fault diagnosis system is characterized by comprising a data conversion module and a fault diagnosis module;
the data conversion module is used for correspondingly converting the power flow data before the fault and the power flow data after the fault into DCVPFI data according to a preset conversion rule when the power grid has the fault; inputting DCVPFI data into a CNN model;
the fault diagnosis module is used for inputting the DCVPFI data into a Convolutional Neural Network (CNN) model which is trained by massive DCVPFI data samples in advance, outputting fault probabilities of different fault states, and determining a power grid fault diagnosis result according to the fault state with the maximum fault probability.
9. The power grid fault diagnosis system of claim 8, further comprising a mass sample module for generating a mass DCVPFI data source offline based on the grid system ground state data; and constructing and obtaining a mass of DCVPFI data samples by using the generated mass of DCVPFI data sources.
10. A power grid fault diagnosis device is characterized by comprising,
a memory for storing a computer program;
a processor for implementing the steps of a grid fault diagnosis method as claimed in any one of claims 1 to 7 when executing the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010676093.5A CN112000923A (en) | 2020-07-14 | 2020-07-14 | Power grid fault diagnosis method, system and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010676093.5A CN112000923A (en) | 2020-07-14 | 2020-07-14 | Power grid fault diagnosis method, system and equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112000923A true CN112000923A (en) | 2020-11-27 |
Family
ID=73467957
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010676093.5A Pending CN112000923A (en) | 2020-07-14 | 2020-07-14 | Power grid fault diagnosis method, system and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112000923A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113609765A (en) * | 2021-07-29 | 2021-11-05 | 国网河北省电力有限公司邯郸供电分公司 | Overvoltage prediction method |
CN113917285A (en) * | 2021-10-21 | 2022-01-11 | 国网山西省电力公司运城供电公司 | Fault intelligent positioning method and system based on power distribution network power flow distribution identification |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109061391A (en) * | 2018-09-07 | 2018-12-21 | 中国电力科学研究院有限公司 | A kind of electric network failure diagnosis method and system based on computer vision tidal current chart |
CN109768877A (en) * | 2018-11-23 | 2019-05-17 | 国网上海市电力公司 | A kind of electric network failure diagnosis method based on space optimum code collection and DHNN error correction |
CN109814523A (en) * | 2018-12-04 | 2019-05-28 | 合肥工业大学 | Method for diagnosing faults based on CNN-LSTM deep learning method and more attribute time series datas |
CN110109015A (en) * | 2019-05-31 | 2019-08-09 | 中南大学 | A kind of asynchronous motor Fault monitoring and diagnosis method based on deep learning |
CN110826632A (en) * | 2019-11-11 | 2020-02-21 | 深圳前海微众银行股份有限公司 | Image change detection method, device, equipment and computer readable storage medium |
CN111161229A (en) * | 2019-12-23 | 2020-05-15 | 大连理工大学 | Change detection method based on geometric active contour model and sparse self-coding |
CN111323228A (en) * | 2020-03-20 | 2020-06-23 | 广东技术师范大学 | Fault diagnosis method based on time series mapping and convolutional neural network |
-
2020
- 2020-07-14 CN CN202010676093.5A patent/CN112000923A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109061391A (en) * | 2018-09-07 | 2018-12-21 | 中国电力科学研究院有限公司 | A kind of electric network failure diagnosis method and system based on computer vision tidal current chart |
CN109768877A (en) * | 2018-11-23 | 2019-05-17 | 国网上海市电力公司 | A kind of electric network failure diagnosis method based on space optimum code collection and DHNN error correction |
CN109814523A (en) * | 2018-12-04 | 2019-05-28 | 合肥工业大学 | Method for diagnosing faults based on CNN-LSTM deep learning method and more attribute time series datas |
CN110109015A (en) * | 2019-05-31 | 2019-08-09 | 中南大学 | A kind of asynchronous motor Fault monitoring and diagnosis method based on deep learning |
CN110826632A (en) * | 2019-11-11 | 2020-02-21 | 深圳前海微众银行股份有限公司 | Image change detection method, device, equipment and computer readable storage medium |
CN111161229A (en) * | 2019-12-23 | 2020-05-15 | 大连理工大学 | Change detection method based on geometric active contour model and sparse self-coding |
CN111323228A (en) * | 2020-03-20 | 2020-06-23 | 广东技术师范大学 | Fault diagnosis method based on time series mapping and convolutional neural network |
Non-Patent Citations (1)
Title |
---|
SONGYAN WANG等: "《Deep‐learning based fault diagnosis using computer‐visualised power flow》", 《ET GENERATION, TRANSMISSION & DISTRIBUTION》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113609765A (en) * | 2021-07-29 | 2021-11-05 | 国网河北省电力有限公司邯郸供电分公司 | Overvoltage prediction method |
CN113609765B (en) * | 2021-07-29 | 2023-12-26 | 国网河北省电力有限公司邯郸供电分公司 | Overvoltage prediction method |
CN113917285A (en) * | 2021-10-21 | 2022-01-11 | 国网山西省电力公司运城供电公司 | Fault intelligent positioning method and system based on power distribution network power flow distribution identification |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wilson et al. | Deep learning-aided cyber-attack detection in power transmission systems | |
CN111596604B (en) | Intelligent fault diagnosis and self-healing control system and method for engineering equipment based on digital twinning | |
CN114429153B (en) | Gear box increment fault diagnosis method and system based on life learning | |
CN107730017A (en) | Intelligent metallurgical equipment operational diagnostics system | |
CN109061391B (en) | Power grid fault diagnosis method and system based on computer vision tidal current diagram | |
CN108022039B (en) | Transformer substation simulation training system based on augmented reality | |
CN107066759A (en) | A kind of Vibration Fault Diagnosis of Turbine Rotor method and device | |
CN114297947B (en) | Data-driven wind power system twinning method and system based on deep learning network | |
CN110703057A (en) | Power equipment partial discharge diagnosis method based on data enhancement and neural network | |
CN112000923A (en) | Power grid fault diagnosis method, system and equipment | |
CN113344875B (en) | Abnormal image detection method based on self-supervision learning | |
CN104615121B (en) | A kind of train fault diagnostic method and system | |
CN111580479A (en) | Intelligent manufacturing industry parameter optimization method based on machine learning and industrial Internet of things | |
CN114676121B (en) | Missing monitoring data filling method based on multi-scale space-time memory sharing network | |
CN114492150B (en) | Digital twin-body-based power distribution network typical service scene early warning method | |
CN110361609A (en) | Extra-high voltage equipment monitors system and method | |
CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
CN113935237A (en) | Power transmission line fault type distinguishing method and system based on capsule network | |
CN117639228A (en) | Power distribution network running state prediction method and system based on digital twin | |
CN113406439B (en) | Fault positioning method for power distribution network | |
CN110826201A (en) | Three-dimensional visualization platform for operation and maintenance management system of data center | |
CN117523934A (en) | Distribution network operation and maintenance simulation training system based on big data | |
CN110674240B (en) | GIS-based distributed multistage intelligent fault diagnosis system for power equipment | |
CN116415925A (en) | Distribution network fault studying and judging method based on artificial intelligence | |
CN115560962A (en) | Fan abnormal state detection method based on multi-level space-time diagram neural network |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201127 |