CN116937820B - High-voltage circuit state monitoring method based on deep learning algorithm - Google Patents
High-voltage circuit state monitoring method based on deep learning algorithm Download PDFInfo
- Publication number
- CN116937820B CN116937820B CN202311207814.8A CN202311207814A CN116937820B CN 116937820 B CN116937820 B CN 116937820B CN 202311207814 A CN202311207814 A CN 202311207814A CN 116937820 B CN116937820 B CN 116937820B
- Authority
- CN
- China
- Prior art keywords
- data
- optimal solution
- particle
- monitoring
- line
- 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.)
- Active
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013135 deep learning Methods 0.000 title claims abstract description 24
- 239000002245 particle Substances 0.000 claims abstract description 76
- 230000002159 abnormal effect Effects 0.000 claims abstract description 68
- 238000012545 processing Methods 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims abstract description 38
- 238000003062 neural network model Methods 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims abstract description 18
- 238000005457 optimization Methods 0.000 claims abstract description 12
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 230000009467 reduction Effects 0.000 claims abstract description 11
- 238000004140 cleaning Methods 0.000 claims abstract description 8
- 230000003044 adaptive effect Effects 0.000 claims description 39
- 239000011159 matrix material Substances 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 18
- 238000005259 measurement Methods 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 238000007689 inspection Methods 0.000 abstract description 5
- 230000005540 biological transmission Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 239000006185 dispersion Substances 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00002—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by monitoring
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R19/00—Arrangements for measuring currents or voltages or for indicating presence or sign thereof
- G01R19/0092—Arrangements for measuring currents or voltages or for indicating presence or sign thereof measuring current only
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00001—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by the display of information or by user interaction, e.g. supervisory control and data acquisition systems [SCADA] or graphical user interfaces [GUI]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00032—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Power Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Computer Interaction (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention relates to the field of high-voltage circuit monitoring, and discloses a high-voltage circuit state monitoring method based on a deep learning algorithm, which comprises the following steps: acquiring a line data acquisition instruction, and receiving node line data output by each node position of the high-voltage line according to the line data acquisition instruction; performing data cleaning processing on the node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set; performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting a training set into the optimized neural network model for training to obtain a state monitoring model; detecting the test set by using a state monitoring model to obtain a line state monitoring result; when an abnormal state exists in the line state detection result, generating abnormal state information and outputting the abnormal state information; the invention improves the working efficiency, reduces line inspection personnel, and effectively improves the power supply reliability.
Description
Technical Field
The invention relates to the field of high-voltage circuit monitoring, in particular to a high-voltage circuit state monitoring method based on a deep learning algorithm.
Background
The high-voltage power supply is an electric service wire with the alternating-current voltage of an assigned electric wire line being more than 1000V or the direct-current voltage being more than 1500V; power transmission and distribution lines in an electrical power system, such as high voltage towers, or substations. Or a large amount of electricity users; trains, high-speed rails and express trains using electric power as main power, and high-voltage cables and collecting brushes or high-voltage rails are used; in a power distribution network operation system, the circuit branches are more, the operation mode is extremely complex, particularly, the high-voltage transmission lines are long in transmission distance and are generally distributed in unmanned zones, the management and maintenance workload of the lines is extremely high, the power supply reliability is low, therefore, the operation state of the high-voltage circuit lines is determined in advance, the electric energy transmission accidents caused by sudden faults of the high-voltage circuit lines can be avoided, and therefore, how to determine the operation state of the high-voltage circuit lines is a problem to be solved urgently at present.
Disclosure of Invention
The invention aims to solve the problems, and designs a high-voltage circuit state monitoring method based on a deep learning algorithm.
The first aspect of the invention provides a high-voltage circuit state monitoring method based on a deep learning algorithm, which comprises the following steps:
acquiring a line data acquisition instruction, and receiving node line data output by each node position of a high-voltage line according to the line data acquisition instruction;
performing data cleaning processing on the node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set;
performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting the training set into the optimized neural network model for training to obtain a state monitoring model;
detecting the test set by using the state monitoring model to obtain a line state monitoring result;
and when the abnormal state exists in the line state detection result, generating abnormal state information and outputting the abnormal state information.
Optionally, in a first implementation manner of the first aspect of the present invention, the obtaining a line data acquisition instruction, receiving node line data output by each node position of the high-voltage line according to the line data acquisition instruction, includes:
receiving a line data acquisition instruction sent by a monitoring center, and sending the line data acquisition instruction to each node position sensor of a high-voltage line;
acquiring time domain information in the line data acquisition instruction, and acquiring initial line data collected by the sensor according to the time domain information;
acquiring a data sequence in the initial line data, splitting the data sequence into integer data and floating point data, and performing run-length coding on the integer data to obtain variable data and repeated data;
and sorting the floating point data, the variable data and the repeated data to obtain node line data, and outputting the node line data.
Optionally, in a second implementation manner of the first aspect of the present invention, a node network is formed between the monitoring center and the sensors, and the sensors are distributed on the high-voltage line according to a preset interval.
Optionally, in a third implementation manner of the first aspect of the present invention, the performing data cleaning processing on the node line data to obtain monitoring index data includes:
the node line data is received, and the node line data is marked according to the node position on the high-voltage line, so that node marking data is obtained;
normalizing the node labeling data to obtain first processing data, and determining characteristic data in the first processing data;
performing redundancy processing on the first processing data through a covariance matrix according to the characteristic data to obtain second processing data, wherein the covariance matrix is composed of the characteristic data and a covariance coefficient;
and carrying out variance screening treatment on the second treatment data to obtain monitoring index data.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set includes:
acquiring the monitoring index data, generating a first monitoring data set, and marking a plurality of sample points in the first monitoring data set;
performing decentralization processing on a plurality of sample points in the first monitoring data set, circularly traversing the sample points, and determining linear representation of the sample points;
reconstructing the sample points to obtain a reconstruction coefficient, establishing a sample matrix according to the reconstruction coefficient, determining eigenvalue eigenvectors of the sample matrix, and taking a preset number of eigenvectors from large to small to form an eigenvector;
and obtaining a second monitoring data set after dimension reduction according to the feature matrix, and dividing the second monitoring data set into a training set and a testing set according to a certain proportion.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the performing parameter optimization on the neural network model using an adaptive particle swarm algorithm includes:
initializing parameters of the self-adaptive particle swarm algorithm, and setting iteration times and initial position speed;
calculating the adaptive value of each particle, and updating the parameters based on the adaptive value of each particle to obtain an optimal solution parameter set;
and carrying out parameter optimization on the neural network model by adopting the optimal solution parameter set.
Optionally, in a sixth implementation manner of the first aspect of the present invention, the calculating an adaptive value of each particle, updating a parameter based on the adaptive value of each particle, to obtain an optimal solution parameter set includes:
acquiring an initial optimal solution and a group initial optimal solution of each particle, and updating according to the initial optimal solution and the group initial optimal solution of each particle to obtain a current optimal solution and a group current optimal solution of each particle;
updating the initial position speed according to the current optimal solution of each particle and the current optimal solution of the group to obtain the current position speed;
acquiring adaptive dynamic inertia factors, and updating the inertia factors according to the adaptive values of each particle;
updating variation probability according to the adaptive value of each particle, judging whether the adaptive particle swarm algorithm reaches an end condition, outputting a parameter combination if the adaptive particle swarm algorithm reaches the end condition, and re-calculating the adaptive value of each particle and updating the iteration times if the adaptive particle swarm algorithm does not reach the end condition;
and obtaining an optimal solution parameter set according to the current optimal solution of each particle, the current optimal solution of the group, the current position speed, the inertia factor and the variation probability.
Optionally, in a seventh implementation manner of the first aspect of the present invention, the obtaining an initial optimal solution and a group initial optimal solution of each particle, and updating the initial optimal solution and the group initial optimal solution according to the initial optimal solution and the group initial optimal solution of each particle, to obtain a current optimal solution and a group current optimal solution of each particle includes:
comparing the adaptive value of each particle with the initial optimal solution of each particle to obtain a first comparison result, and updating the initial optimal solution of each particle according to the first comparison result to obtain a current optimal solution of each particle;
and comparing the adaptive value of each particle with the initial optimal solution of the group to obtain a second comparison result, and updating the initial optimal solution of the group according to the second comparison result to obtain the current optimal solution of the group.
Optionally, in an eighth implementation manner of the first aspect of the present invention, the training set input to the optimized neural network model to obtain a state monitoring model includes:
randomly selecting training data from the training set, and constructing a plurality of meta-learning tasks based on the training data;
inputting a plurality of element learning tasks into the optimized neural network model, and classifying, updating and optimizing the loss function of the neural network model to obtain a first loss function;
acquiring a target domain and a source domain in the training set, inputting the target domain into an optimized neural network model, calculating the maximum mean difference between the target domain and the source domain, and updating a loss function of the optimized neural network model to obtain a second loss function, wherein the target domain is high-voltage line data in a normal state;
and taking a model when the first loss function and the second loss function are converged as the state monitoring model.
Optionally, in a ninth implementation manner of the first aspect of the present invention, when an abnormal state exists in the line state detection result, generating abnormal state information, and outputting the abnormal state information includes:
obtaining an abnormal current value from the line state detection result, and obtaining a current abnormal component by comparing the abnormal current value with normal current data in a normal state;
and judging an abnormal section based on the current abnormal component, sending out measurement quality in the abnormal section, measuring the traveling wave speed according to the measurement quality, carrying out abnormal positioning to obtain abnormal state information, and outputting the abnormal state information, wherein the abnormal state information comprises an abnormal type and an abnormal positioning result.
According to the technical scheme provided by the invention, the node line data output by each node position of the high-voltage line is received according to the line data acquisition instruction by acquiring the line data acquisition instruction; performing data cleaning processing on the node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set; performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting the training set into the optimized neural network model for training to obtain a state monitoring model; detecting the test set by using the state monitoring model to obtain a line state monitoring result; generating abnormal state information when an abnormal state exists in the line state detection result, and outputting the abnormal state information; the invention can rapidly monitor the state of the high-voltage circuit, determine the abnormal state, and correspondingly process the abnormal state in time, thereby avoiding electric energy transmission accidents caused by sudden faults of the high-voltage circuit, improving the working efficiency, reducing the number of circuit inspection staff, and effectively improving the power supply reliability.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention.
Fig. 1 is a schematic diagram of a first embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a second embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a third embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a fourth embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a fifth embodiment of a high-voltage circuit state monitoring method based on a deep learning algorithm according to an embodiment of the present invention.
Detailed Description
The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, the following describes a specific flow of an embodiment of the present invention, please refer to fig. 1 for a first embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm, which specifically includes the following steps:
step 101, acquiring a line data acquisition instruction, and receiving node line data output by each node position of a high-voltage line according to the line data acquisition instruction;
102, performing data cleaning processing on node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set;
step 103, performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting a training set into the optimized neural network model for training to obtain a state monitoring model;
in this embodiment, training data is randomly selected from a training set, and a plurality of meta-learning tasks are constructed based on the training data; inputting a plurality of element learning tasks into the optimized neural network model, and classifying, updating and optimizing the loss function of the neural network model to obtain a first loss function; acquiring a target domain and a source domain in a training set, inputting the target domain into an optimized neural network model, calculating the maximum mean difference between the target domain and the source domain, and updating a loss function of the optimized neural network model to obtain a second loss function, wherein the target domain is high-voltage line data in a normal state; and taking the model when the first loss function and the second loss function are converged as a state monitoring model.
104, detecting the test set by using a state monitoring model to obtain a line state monitoring result;
step 105, when an abnormal state exists in the line state detection result, generating abnormal state information and outputting the abnormal state information.
In the embodiment, an abnormal current value is obtained from a line state detection result, and a current abnormal component is obtained by comparing the abnormal current value with normal current data in a normal state; and judging an abnormal section based on the current abnormal component, sending out measurement quality in the abnormal section, measuring the traveling wave velocity according to the measurement quality, carrying out abnormal positioning to obtain abnormal state information, and outputting the abnormal state information, wherein the abnormal state information comprises an abnormal type and an abnormal positioning result.
In the embodiment of the invention, the node line data output by each node position of the high-voltage line is received according to the line data acquisition instruction by acquiring the line data acquisition instruction; performing data cleaning processing on the node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set; performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting a training set into the optimized neural network model for training to obtain a state monitoring model; detecting the test set by using a state monitoring model to obtain a line state monitoring result; when an abnormal state exists in the line state detection result, generating abnormal state information and outputting the abnormal state information; the invention can rapidly monitor the state of the high-voltage circuit, determine the abnormal state, and correspondingly process the abnormal state in time, thereby avoiding electric energy transmission accidents caused by sudden faults of the high-voltage circuit, improving the working efficiency, reducing the number of circuit inspection staff, and effectively improving the power supply reliability.
Referring to fig. 2, a second embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention is shown, where the method includes:
step 201, receiving a line data acquisition instruction sent by a monitoring center, and sending the line data acquisition instruction to each node position sensor of a high-voltage line;
in this embodiment, a node network is formed between the monitoring center and the sensors, and the sensors are distributed on the high-voltage line according to a preset interval.
Step 202, acquiring time domain information in a line data acquisition instruction, and acquiring initial line data collected by a sensor according to the time domain information;
step 203, acquiring a data sequence in the initial line data, splitting the data sequence into integer data and floating point data, and performing run-length coding on the integer data to obtain variable data and repeated data;
and 204, sorting the floating point data, the variable data and the repeated data to obtain node line data, and outputting the node line data.
In the embodiment of the invention, a line data acquisition instruction is sent to each node position sensor of a high-voltage line by receiving the line data acquisition instruction sent from a monitoring center; acquiring time domain information in a line data acquisition instruction, and acquiring initial line data collected by a sensor according to the time domain information; acquiring a data sequence in initial line data, splitting the data sequence into integer data and floating point data, and performing run-length coding on the integer data to obtain variable data and repeated data; sorting floating point data, variable data and repeated data to obtain node line data, and outputting the node line data; the invention can achieve higher space saving, is beneficial to reducing network communication time delay and communication load.
Referring to fig. 3, a third embodiment of a high-voltage circuit state monitoring method based on a deep learning algorithm according to an embodiment of the present invention is shown, where the method includes:
step 301, receiving node line data, marking the node line data according to the node position on the high-voltage line, and obtaining node marking data;
step 302, carrying out normalization processing on the node labeling data to obtain first processing data, and determining characteristic data in the first processing data;
in this embodiment, the normalization process maps data to the range of [0,1] or [1,1] interval, the dimensions of different features are different, the value ranges are different, and singular values exist, which has an influence on training.
Step 303, performing redundancy processing on the first processing data through a covariance matrix according to the characteristic data to obtain second processing data, wherein the covariance matrix is composed of the characteristic data and a covariance coefficient;
in this embodiment, the mathematical expectation of the square of the dispersion of the random variable is called the variance of the random variable, which is always a non-negative number, and is smaller when the possible values of the random variable are concentrated in the vicinity of the mathematical expectation; on the contrary, the variance is larger, so that the dispersion degree of the random variable distribution can be deduced according to the variance, and the variance can reflect the dispersion degree of all possible values of the random variable around the mathematical expectation; diagonal elements in the covariance matrix C represent variances and non-diagonal elements represent covariance between different components of the random vector X. It can be seen from (1, 10) that the covariance shows a correlation to a certain extent, so that C can be used as a criterion for characterizing the correlation between different components, and if the correlation between different components is smaller, the value of the non-diagonal element of C is smaller, and in particular, if the different components are not correlated with each other, C becomes a diagonal matrix.
And 304, performing variance screening processing on the second processing data to obtain monitoring index data.
In the embodiment of the invention, node line data is marked according to the node position on the high-voltage line by receiving the node line data, so as to obtain node marking data; carrying out normalization processing on the node labeling data to obtain first processing data, and determining characteristic data in the first processing data; performing redundancy processing on the first processing data through a covariance matrix according to the characteristic data to obtain second processing data, wherein the covariance matrix consists of the characteristic data and a covariance coefficient; performing variance screening treatment on the second treatment data to obtain monitoring index data; the invention can effectively improve the data cleaning efficiency and accuracy and the working efficiency.
Referring to fig. 4, a fourth embodiment of a method for monitoring a state of a high-voltage circuit based on a deep learning algorithm according to an embodiment of the present invention is shown, where the method includes:
step 401, acquiring monitoring index data, generating a first monitoring data set, and marking a plurality of sample points in the first monitoring data set;
step 402, performing decentralization processing on a plurality of sample points in the first monitoring data set, circularly traversing the sample points, and determining linear representation of the sample points;
in this embodiment, the decentration process is to make the data satisfy the mean value of 0, but there is no requirement for the standard deviation.
Step 403, reconstructing the sample points to obtain a reconstruction coefficient, establishing a sample matrix according to the reconstruction coefficient, determining eigenvalue eigenvectors of the sample matrix, and taking a preset number of eigenvectors from large to small to form an eigenvector;
and 404, obtaining a second monitoring data set after dimension reduction according to the feature matrix, and dividing the second monitoring data set into a training set and a testing set according to a certain proportion.
In the embodiment of the invention, a first monitoring data set is generated by acquiring monitoring index data, and a plurality of sample points in the first monitoring data set are marked; performing decentralization processing on a plurality of sample points in the first monitoring data set, circularly traversing the sample points, and determining linear representation of the sample points; reconstructing the sample points to obtain a reconstruction coefficient, establishing a sample matrix according to the reconstruction coefficient, determining eigenvalue eigenvectors of the sample matrix, and taking a preset number of eigenvectors from large to small to form the eigenvector; obtaining a second monitoring data set after dimension reduction according to the feature matrix, and dividing the second monitoring data set into a training set and a testing set according to a certain proportion; according to the invention, the training set and the testing set are obtained to carry out deep learning on the neural network model, so that the state of the high-voltage circuit line can be rapidly monitored, the abnormal state is determined, the abnormal state is correspondingly processed in time, the electric energy transmission accident caused by sudden failure of the high-voltage circuit line is avoided, the working efficiency is improved, the number of line inspection personnel is reduced, and the power supply reliability is effectively improved.
Referring to fig. 5, a fifth embodiment of a high-voltage circuit state monitoring method based on a deep learning algorithm according to an embodiment of the present invention is shown, where the method includes:
step 501, initializing parameters of a self-adaptive particle swarm algorithm, and setting iteration times and initial position speed;
step 502, calculating an adaptive value of each particle, and updating parameters based on the adaptive value of each particle to obtain an optimal solution parameter set;
in the embodiment, an initial optimal solution and a group initial optimal solution of each particle are obtained, and a current optimal solution and a group current optimal solution of each particle are obtained according to the initial optimal solution and the group initial optimal solution of each particle;
updating the initial position speed according to the current optimal solution of each particle and the current optimal solution of the group to obtain the current position speed; acquiring an adaptive dynamic inertia factor, and updating the inertia factor according to the adaptive value of each particle; updating variation probability according to the adaptive value of each particle, judging whether the adaptive particle swarm algorithm reaches an end condition, if so, outputting parameter combinations, if not, recalculating the adaptive value of each particle, and updating iteration times; and obtaining an optimal solution parameter set according to the current optimal solution of each particle, the current optimal solution of the group, the current position speed, the inertia factor and the variation probability.
In this embodiment, the adaptive value of each particle is compared with the initial optimal solution of each particle to obtain a first comparison result, and the initial optimal solution of each particle is updated according to the first comparison result to obtain the current optimal solution of each particle; and comparing the adaptive value of each particle with the initial optimal solution of the population to obtain a second comparison result, and updating the initial optimal solution of the population according to the second comparison result to obtain the current optimal solution of the population.
And 503, carrying out parameter optimization on the neural network model by adopting an optimal solution parameter set.
In the embodiment of the invention, the iteration times and the initial position speed are set by initializing parameters of the self-adaptive particle swarm algorithm; calculating the adaptive value of each particle, and updating the parameters based on the adaptive value of each particle to obtain an optimal solution parameter set; performing parameter optimization on the neural network model by adopting an optimal solution parameter set; the invention improves the accuracy of the neural network model through the deep learning algorithm, can rapidly monitor the state of the high-voltage circuit, determine the abnormal state, and correspondingly process the abnormal state in time, thereby avoiding the electric energy transmission accident caused by the sudden failure of the high-voltage circuit, improving the working efficiency, reducing the number of circuit inspection staff and effectively improving the power supply reliability.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (6)
1. The high-voltage circuit state monitoring method based on the deep learning algorithm is characterized by comprising the following steps of:
acquiring a line data acquisition instruction, and receiving node line data output by each node position of a high-voltage line according to the line data acquisition instruction;
receiving a line data acquisition instruction sent by a monitoring center, and sending the line data acquisition instruction to each node position sensor of a high-voltage line;
acquiring time domain information in the line data acquisition instruction, and acquiring initial line data collected by the sensor according to the time domain information;
acquiring a data sequence in the initial line data, splitting the data sequence into integer data and floating point data, and performing run-length coding on the integer data to obtain variable data and repeated data;
sorting the floating point data, the variable data and the repeated data to obtain node line data, and outputting the node line data;
performing data cleaning processing on the node line data to obtain monitoring index data, and performing data dimension reduction processing on the monitoring index data to obtain a training set and a testing set;
the node line data is received, and the node line data is marked according to the node position on the high-voltage line, so that node marking data is obtained;
normalizing the node labeling data to obtain first processing data, and determining characteristic data in the first processing data;
performing redundancy processing on the first processing data through a covariance matrix according to the characteristic data to obtain second processing data, wherein the covariance matrix is composed of the characteristic data and a covariance coefficient;
performing variance screening treatment on the second treatment data to obtain monitoring index data;
acquiring the monitoring index data, generating a first monitoring data set, and marking a plurality of sample points in the first monitoring data set;
performing decentralization processing on a plurality of sample points in the first monitoring data set, circularly traversing the sample points, and determining linear representation of the sample points;
reconstructing the sample points to obtain a reconstruction coefficient, establishing a sample matrix according to the reconstruction coefficient, determining eigenvalue eigenvectors of the sample matrix, and taking a preset number of eigenvectors from large to small to form an eigenvector;
obtaining a second monitoring data set after dimension reduction according to the feature matrix, and dividing the second monitoring data set into a training set and a testing set according to a certain proportion;
performing parameter optimization on the neural network model by adopting a self-adaptive particle swarm algorithm, and inputting the training set into the optimized neural network model for training to obtain a state monitoring model;
randomly selecting training data from a training set, and constructing a plurality of meta-learning tasks based on the training data;
inputting a plurality of element learning tasks into the optimized neural network model, and classifying, updating and optimizing the loss function of the neural network model to obtain a first loss function;
acquiring a target domain and a source domain in a training set, inputting the target domain into an optimized neural network model, calculating the maximum mean difference between the target domain and the source domain, and updating a loss function of the optimized neural network model to obtain a second loss function, wherein the target domain is high-voltage line data in a normal state;
taking a model when the first loss function and the second loss function are converged as a state monitoring model;
detecting the test set by using the state monitoring model to obtain a line state monitoring result;
and when the abnormal state exists in the line state detection result, generating abnormal state information and outputting the abnormal state information.
2. The method for monitoring the state of a high-voltage circuit based on a deep learning algorithm according to claim 1, wherein a node network is formed between the monitoring center and the sensors, and the sensors are distributed on the high-voltage circuit according to a preset interval.
3. The method for monitoring the state of a high-voltage circuit based on a deep learning algorithm as claimed in claim 1, wherein the parameter optimization of the neural network model by adopting the adaptive particle swarm algorithm comprises the following steps:
initializing parameters of the self-adaptive particle swarm algorithm, and setting iteration times and initial position speed;
calculating the adaptive value of each particle, and updating the parameters based on the adaptive value of each particle to obtain an optimal solution parameter set;
and carrying out parameter optimization on the neural network model by adopting the optimal solution parameter set.
4. A method for monitoring a state of a high-voltage circuit based on a deep learning algorithm as claimed in claim 3, wherein said calculating an adaptive value of each particle, updating parameters based on said adaptive value of each particle, and obtaining an optimal solution parameter set includes:
acquiring an initial optimal solution and a group initial optimal solution of each particle, and updating according to the initial optimal solution and the group initial optimal solution of each particle to obtain a current optimal solution and a group current optimal solution of each particle;
updating the initial position speed according to the current optimal solution of each particle and the current optimal solution of the group to obtain the current position speed;
acquiring adaptive dynamic inertia factors, and updating the inertia factors according to the adaptive values of each particle;
updating variation probability according to the adaptive value of each particle, judging whether the adaptive particle swarm algorithm reaches an end condition, outputting a parameter combination if the adaptive particle swarm algorithm reaches the end condition, and re-calculating the adaptive value of each particle and updating the iteration times if the adaptive particle swarm algorithm does not reach the end condition;
and obtaining an optimal solution parameter set according to the current optimal solution of each particle, the current optimal solution of the group, the current position speed, the inertia factor and the variation probability.
5. The method for monitoring the state of a high-voltage circuit based on a deep learning algorithm as claimed in claim 4, wherein the obtaining the initial optimal solution and the initial optimal solution of the population for each particle, and updating the initial optimal solution and the initial optimal solution of the population for each particle to obtain the current optimal solution and the current optimal solution of the population for each particle, comprises:
comparing the adaptive value of each particle with the initial optimal solution of each particle to obtain a first comparison result, and updating the initial optimal solution of each particle according to the first comparison result to obtain a current optimal solution of each particle;
and comparing the adaptive value of each particle with the initial optimal solution of the group to obtain a second comparison result, and updating the initial optimal solution of the group according to the second comparison result to obtain the current optimal solution of the group.
6. The method for monitoring the state of a high-voltage circuit based on a deep learning algorithm according to claim 1, wherein when an abnormal state exists in the circuit state detection result, generating abnormal state information and outputting the abnormal state information comprises:
obtaining an abnormal current value from the line state detection result, and obtaining a current abnormal component by comparing the abnormal current value with normal current data in a normal state;
and judging an abnormal section based on the current abnormal component, sending a measurement instruction in the abnormal section, measuring the traveling wave speed according to the measurement instruction, carrying out abnormal positioning to obtain abnormal state information, and outputting the abnormal state information, wherein the abnormal state information comprises an abnormal type and an abnormal positioning result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311207814.8A CN116937820B (en) | 2023-09-19 | 2023-09-19 | High-voltage circuit state monitoring method based on deep learning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311207814.8A CN116937820B (en) | 2023-09-19 | 2023-09-19 | High-voltage circuit state monitoring method based on deep learning algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116937820A CN116937820A (en) | 2023-10-24 |
| CN116937820B true CN116937820B (en) | 2024-01-05 |
Family
ID=88384755
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311207814.8A Active CN116937820B (en) | 2023-09-19 | 2023-09-19 | High-voltage circuit state monitoring method based on deep learning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116937820B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916369A (en) * | 2010-08-20 | 2010-12-15 | 西安电子科技大学 | Face recognition method based on kernel nearest subspace |
| CN108089099A (en) * | 2017-12-18 | 2018-05-29 | 广东电网有限责任公司佛山供电局 | The diagnostic method of distribution network failure based on depth confidence network |
| CN112504682A (en) * | 2020-12-21 | 2021-03-16 | 中国人民解放军63963部队 | Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm |
| CN112557034A (en) * | 2020-12-25 | 2021-03-26 | 南京理工大学 | Bearing fault diagnosis method based on PCA _ CNNS |
| WO2021159815A1 (en) * | 2020-09-04 | 2021-08-19 | 平安科技（深圳）有限公司 | Method and apparatus for training facial recognition model, and computer device |
| CN113379528A (en) * | 2021-05-25 | 2021-09-10 | 杭州搜车数据科技有限公司 | Wind control model establishing method and device and risk control method |
| CN113675947A (en) * | 2021-07-27 | 2021-11-19 | 北京智芯微电子科技有限公司 | Power transmission side equipment state monitoring method of power transmission gateway and power transmission gateway |
| CN114869298A (en) * | 2022-06-15 | 2022-08-09 | 浙大宁波理工学院 | Depression detection method and system based on electroencephalogram signals and storage medium |
| CN116559593A (en) * | 2023-05-17 | 2023-08-08 | 南方电网能源发展研究院有限责任公司 | Transmission line fault processing method, device, computer equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190219994A1 (en) * | 2018-01-18 | 2019-07-18 | General Electric Company | Feature extractions to model large-scale complex control systems |
-
2023
- 2023-09-19 CN CN202311207814.8A patent/CN116937820B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916369A (en) * | 2010-08-20 | 2010-12-15 | 西安电子科技大学 | Face recognition method based on kernel nearest subspace |
| CN108089099A (en) * | 2017-12-18 | 2018-05-29 | 广东电网有限责任公司佛山供电局 | The diagnostic method of distribution network failure based on depth confidence network |
| WO2021159815A1 (en) * | 2020-09-04 | 2021-08-19 | 平安科技（深圳）有限公司 | Method and apparatus for training facial recognition model, and computer device |
| CN112504682A (en) * | 2020-12-21 | 2021-03-16 | 中国人民解放军63963部队 | Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm |
| CN112557034A (en) * | 2020-12-25 | 2021-03-26 | 南京理工大学 | Bearing fault diagnosis method based on PCA _ CNNS |
| CN113379528A (en) * | 2021-05-25 | 2021-09-10 | 杭州搜车数据科技有限公司 | Wind control model establishing method and device and risk control method |
| CN113675947A (en) * | 2021-07-27 | 2021-11-19 | 北京智芯微电子科技有限公司 | Power transmission side equipment state monitoring method of power transmission gateway and power transmission gateway |
| CN114869298A (en) * | 2022-06-15 | 2022-08-09 | 浙大宁波理工学院 | Depression detection method and system based on electroencephalogram signals and storage medium |
| CN116559593A (en) * | 2023-05-17 | 2023-08-08 | 南方电网能源发展研究院有限责任公司 | Transmission line fault processing method, device, computer equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于深度学***台资金流预测技术研究;沈桂竹;《中国优秀硕士学位论文全文数据库》(第12期);第1-49页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116937820A (en) | 2023-10-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107219457B (en) | Frame-type circuit breaker fault diagnosis and degree assessment method based on operation attachment electric current | |
| CN111598166B (en) | Single-phase earth fault classification method and system based on principal component analysis and Softmax function | |
| CN109444667B (en) | Power distribution network early fault classification method and device based on convolutional neural network | |
| CN112327100B (en) | Power failure detection method and system based on Internet of things | |
| CN111080105A (en) | Transformer area user-to-user relationship identification method and system based on voltage time sequence data | |
| CN113340353B (en) | Monitoring method, equipment and medium for power transmission line | |
| CN113496440B (en) | User abnormal electricity consumption detection method and system | |
| El-Hag | Application of machine learning in outdoor insulators condition monitoring and diagnostics | |
| Jacob et al. | Fault diagnostics in shipboard power systems using graph neural networks | |
| CN112381351A (en) | Power utilization behavior change detection method and system based on singular spectrum analysis | |
| CN113689307A (en) | Power supply system level fault association enhanced cognition and diagnosis based on association graph network | |
| CN116739829A (en) | Big data-based power data analysis method, system and medium | |
| CN113092044B (en) | Rotary machine fault diagnosis method based on weighted level visible graph | |
| CN116937820B (en) | High-voltage circuit state monitoring method based on deep learning algorithm | |
| CN110673024A (en) | High-voltage circuit breaker fault early warning method based on power internet of things | |
| CN113514743A (en) | Construction method of GIS partial discharge pattern recognition system based on multi-dimensional features | |
| CN116070140B (en) | Power distribution substation safe operation state monitoring system and method | |
| CN112241595A (en) | Airborne power supply reliability increase performance degradation analysis method based on logistic regression | |
| CN112946369B (en) | Cable aging identification method using power line communication device | |
| Akbari et al. | Fault Detection, Classification, and Location in VSC-HVDC Systems Using Metaheuristic Optimized ANFIS and Hilbert Huang Transform Method | |
| CN113591909A (en) | Abnormality detection method, abnormality detection device, and storage medium for power system | |
| CN112329628A (en) | Method and device for extracting multi-feature fusion features of converter transformer vibration signal | |
| JPH07270467A (en) | Failure state discrimination method | |
| CN115684835B (en) | Power distribution network fault diagnosis method, device, equipment and storage medium | |
| CN117713099A (en) | Short-time impulse current quick response method of flexible voltage treatment device |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |