CN117216485A - Objective weighting-based power transmission wave-recording bird damage fault judging method and system - Google Patents
Objective weighting-based power transmission wave-recording bird damage fault judging method and system Download PDFInfo
- Publication number
- CN117216485A CN117216485A CN202311481469.7A CN202311481469A CN117216485A CN 117216485 A CN117216485 A CN 117216485A CN 202311481469 A CN202311481469 A CN 202311481469A CN 117216485 A CN117216485 A CN 117216485A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- recording
- features
- power transmission
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000005540 biological transmission Effects 0.000 title claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000010801 machine learning Methods 0.000 claims abstract description 10
- 238000013528 artificial neural network Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000012850 discrimination method Methods 0.000 claims 5
- 230000003247 decreasing effect Effects 0.000 claims 1
- 230000005611 electricity Effects 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000000354 decomposition reaction Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000001052 transient effect Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000001131 transforming effect Effects 0.000 description 3
- 241000607479 Yersinia pestis Species 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 235000005770 birds nest Nutrition 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 210000003608 fece Anatomy 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 235000005765 wild carrot Nutrition 0.000 description 2
- 208000025274 Lightning injury Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000016507 interphase Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- 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
Abstract
The invention discloses a method and a system for discriminating a wave-recording bird trouble fault of power transmission based on objective weighting, and relates to the technical field of power transmission network fault monitoring. The method comprises the following steps: acquiring fault recording data to obtain real data in the recording data; extracting fault interval characteristics according to a set threshold value; adopting a synthetic minority class oversampling algorithm to enhance the fault interval characteristics; performing weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model; learning the continuous features by utilizing a self-adaptive learning mode to obtain a second fault discrimination model; and combining the discrimination results of the first fault discrimination model and the second fault discrimination model on the data to be detected to serve as the final result of bird damage fault discrimination. According to the invention, the discrete characteristic weight is learned in a self-adaptive learning mode, and the continuity characteristic is learned by utilizing an automatic machine learning framework, so that the accurate discrimination of the wave-recording bird damage fault of the power transmission fault is finally realized.
Description
Technical Field
The invention relates to the technical field of transmission network fault monitoring, in particular to a method and a system for discriminating transmission wave-recording bird damage faults based on objective weighting.
Background
The transmission line wave recording technology is a technical means widely applied to the fault detection and analysis of a power system. By arranging the wave recording device on the power transmission line, various fault waveform data generated by the power system can be captured. In the discrimination of bird trouble faults, the wave recording data can provide information about the current and voltage changes, fault waveform characteristics and the like when faults occur so as to discriminate and locate fault types. When birds, bird droppings, bird nests or the like form a series path on the transmission line, current abnormality and current oscillation are caused; when birds, bird droppings or bird nests and the like form a short circuit path between the wires, excessive current is caused to cause tripping, and accordingly the reliability of power supply and the operation safety of a power system are threatened. At present, the traditional method for judging the wave-recording bird damage fault of the power transmission fault mainly depends on manual experience and professional knowledge, and has the problems of strong subjectivity, low judging accuracy, low processing efficiency and the like. The existing power transmission line fault judging model cannot provide multi-dimensional and multi-angle characteristic information in the characteristic extraction process, and ignores continuity between time domain characteristics and frequency domains, so that the judging result of bird damage is not accurate enough, instantaneity is poor, and risks of delaying fault processing and increasing potential line safety hazards are possibly caused.
Disclosure of Invention
Aiming at the defects existing in the prior art, the invention aims to provide the method and the system for judging the power transmission wave-recording bird damage faults based on objective weighting, which learn discrete characteristic weights in a self-adaptive learning mode and learn continuity characteristics by utilizing an automatic machine learning framework so as to finally realize the accurate judgment of the power transmission wave-recording bird damage faults.
In order to achieve the above object, the present invention is realized by the following technical scheme:
the invention provides a method for discriminating power transmission wave-recording bird damage faults based on objective weighting, which comprises the following steps:
acquiring fault recording data, and preprocessing the recording data to obtain real data in the recording data;
extracting fault interval features according to a set threshold value, wherein continuity features are extracted from two aspects of a time domain and a frequency domain, and discreteness features are extracted from multiple dimensions;
adopting a synthetic minority class oversampling algorithm to enhance the fault interval characteristics;
performing weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model;
learning the continuous features by utilizing a self-adaptive learning mode to obtain a second fault discrimination model;
and combining the discrimination results of the first fault discrimination model and the second fault discrimination model on the data to be detected to serve as the final result of bird damage fault discrimination.
Further, the specific steps of preprocessing the recording data are as follows:
and performing irrelevant data rejection operation according to a file format obtained by the recording data to obtain current and voltage data of each phase as real data.
Further, the continuity features include a direct current content of the zero sequence current, a zero sequence current amplitude variation, a zero sequence current 3 rd harmonic content, a fault resistance average value, and a zero sequence current wavelet packet transformation energy feature.
Further, discrete features are extracted through four dimensions of fault phase type, reclosing condition, time period and weather.
Further, the specific steps of adopting a synthetic minority class oversampling algorithm to enhance the fault interval characteristics are as follows:
constructing a new minority class sample by adopting a K neighbor algorithm;
and combining the new minority class samples with the original fault interval characteristic data set to generate a new training set.
Furthermore, the specific steps of constructing a new minority class sample by adopting the K nearest neighbor algorithm are as follows:
determining a minority class sample;
calculating K neighbors of each minority class sample by adopting a nearest neighbor algorithm;
randomly selecting N samples from K neighbors to perform random linear interpolation;
a new minority class sample is obtained.
Further, the discrete features are subjected to weight training by using an objective weighting method, and the specific steps for obtaining the first fault discrimination model are as follows:
determining initial weights by using an objective weighting method aiming at discrete features in a data set;
and optimizing the initial weight of the discrete feature determined by the objective weighting method by utilizing the self-adaptive neural network to obtain a first fault discrimination model.
Further, the specific steps of optimizing the initial weight of the discrete feature determined by the objective weighting method by utilizing the adaptive neural network are as follows:
the structure and the connection weight of the neural network are continuously adjusted, and the model and the parameters are automatically adjusted according to the change of the data and the change of the network structure, so that the learning effect and the adaptability of the model are improved, and the optimal discrete feature weight data are obtained.
Further, the specific steps of continuously adjusting the structure and the connection weight of the neural network include:
initializing connection weights and bias items of the neural network at the beginning of training;
forward propagation is carried out through a neural network by using input data, and a prediction result of the network is calculated;
comparing the predicted result of the network with the actual label, and calculating the error of the predicted result, namely the loss function;
and gradually reducing the value of the loss function through multiple iterations, so as to obtain the optimal discrete feature weight data.
Further, the specific steps of gradually reducing the value of the loss function through multiple iterations are:
calculating the contribution degree of each connection weight and the bias term to the loss through a back propagation algorithm by using the value of the loss function;
the connection weights and bias terms are updated in the opposite direction of the gradient using a gradient descent optimization algorithm, thereby reducing the value of the loss function.
Further, the continuous features are learned using an automatic machine learning framework AutoGluon.
The second aspect of the invention provides a power transmission wave-recording bird damage fault discrimination system based on objective weighting, which comprises the following components:
the data acquisition module is configured to acquire fault recording data, and preprocess the recording data to obtain real data in the recording data;
the feature extraction module is configured to extract fault interval features according to a set threshold, wherein the continuity features are extracted from the time domain and the frequency domain, and the discreteness features are extracted from the multiple dimensions;
the feature enhancement module is configured to enhance the fault interval features by adopting a synthetic minority class oversampling algorithm;
the first model training module is configured to perform weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model;
the second model training module is configured to learn the continuous features by utilizing an adaptive learning mode to obtain a second fault discrimination model;
the fault judging module is configured to combine judging results of the first fault judging model and the second fault judging model on the data to be tested as final judging results of bird damage faults.
The one or more of the above technical solutions have the following beneficial effects:
the invention discloses a method and a system for judging a bird damage fault of a power transmission record based on objective weighting. The method saves labor cost, and provides a more accurate, efficient and real-time bird pest tripping fault distinguishing method by using a modern machine learning technology, so that the bird pest fault distinguishing is realized more accurately and efficiently, and the power supply reliability and the operation safety of a power system are improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention.
FIG. 1 is a flowchart of a method for discriminating a fault of a wave-recording bird trouble of power transmission based on objective weighting in a first embodiment of the present invention;
FIG. 2 is a schematic diagram of real data waveforms of a wave-recording zero-sequence current of a bird fault in a first embodiment of the invention;
fig. 3 is a schematic diagram of a real data waveform of zero sequence voltage of a wave-recording bus for bird trouble in the first embodiment of the invention.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the invention. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that 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 present invention. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof;
embodiment one:
the first embodiment of the invention provides a method for discriminating a power transmission wave-recording bird trouble based on objective weighting, which is shown in fig. 1 and comprises the following steps:
step 1, acquiring fault recording data, and preprocessing the recording data to obtain real data in the recording data.
And 2, extracting fault interval characteristics according to a set threshold value, wherein the continuity characteristics are extracted from the time domain and the frequency domain, and the discreteness characteristics are extracted from the multi-dimension.
And 3, adopting a synthetic minority class oversampling algorithm (Synthetic Minority Oversampling Technique, SMOTE) to enhance the fault interval characteristics.
And step 4, carrying out weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model.
And step 5, learning the continuous features by utilizing a self-adaptive learning mode to obtain a second fault discrimination model.
And 6, combining the discrimination results of the first fault discrimination model and the second fault discrimination model on the data to be detected to serve as the final discrimination result of the bird damage faults.
In step 1, the specific steps of preprocessing the recording data are as follows:
and performing irrelevant data rejection operation according to a file format obtained by the recording data to obtain current and voltage data of each phase as real data.
In a specific embodiment, COMTRADE (Common format for transient data exchange) is an IEEE standard power system transient data exchange universal format, which defines a format for the files of transient waveforms and accident data collected by the power system model. Each COMTRADE record has a set of up to 4 files associated with it, each of the 4 files having a different information level, the 4 files comprising: title file (· hdr), configuration file (· cfg), data file (· dat), information file (· inf).
In order to obtain the real values of the current and voltage data of each phase, the embodiment extracts the cfg file and the dat file in the COMTRADE file format, and finally derives the csv format data and waveform pictures of the current and voltage of each phase obtained according to data sampling, as shown in fig. 2 and 3, data sampling pictures of the zero sequence current and the bus zero sequence voltage extracted by the real values of the fault recording data of the bird damage mill line are respectively given, and the data fluctuation interval in fig. 2 is the discrete data point in the fault time of the power transmission line.
In step 2, after extracting the true value of the wave recording file, extracting the fault interval data in the wave recording through setting a certain threshold value, and obtaining a final training data set after determining the fault data characteristics. The extracted fault data features in this embodiment include a continuity feature and a discreteness feature. The threshold is set to the average of the first 3 fundamental waves. To prevent erroneous judgment, it is set that when the zero sequence current interval is greater than the threshold value 3 times in succession, it is judged as a fault interval.
In a specific embodiment, the continuity features include a direct current content of the zero sequence current, a zero sequence current amplitude variation, a zero sequence current 3 rd harmonic content, a fault resistance mean value, and a zero sequence current wavelet packet transformed energy feature. And the discrete characteristics are extracted through four dimensions of fault phase type, reclosing condition, time period and weather.
(1) Continuity features
By extracting the continuity characteristics of the fault record data from two angles of the time domain and the frequency domain, the waveform characteristics in the fault record can be represented abundantly, and intelligent diagnosis of the fault cause is realized. Time domain features are direct descriptions of signals over time, such as maximum and minimum, mean, amplitude, etc., that have intuitive physical meaning that can provide basic information about the signal. The frequency domain features can provide information of the signal in the frequency domain, including spectral energy, spectral peaks, etc. Through frequency domain analysis, the frequency components and energy distribution of the signals can be known, so that beneficial information is provided for fault types or fault frequencies. The comprehensive use of the time domain and frequency domain features can fully utilize the information of fault wave recording data, provide multi-dimensional and multi-angle feature representation, and better perform fault diagnosis and classification, thereby improving the accuracy and robustness of fault discrimination. The main extracted continuity features in this embodiment are:
1. dc content of zero sequence current: the direct current content of the zero sequence current is the average value of the zero sequence current data when all faults are extracted;
2. zero sequence current amplitude variation: the zero sequence current amplitude variation characteristic can be described as a multiple of the zero sequence current amplitude and the fundamental wave amplitude.
3. Zero sequence current 3 rd harmonic content: the 3 rd harmonic content of the zero sequence current is obtained by discrete Fourier transform and can be described as follows:
。
in the method, in the process of the invention,for discrete fault data sequences->In the present invention, k is the harmonic order>N is the number of fault data samples for a cycle and N is the frequency of the wave.
4. Failure resistance mean: taking all zero-sequence current and zero-sequence voltage data points at fault positions, wherein the zero-sequence current is as followsZero sequence current data I of the barrier line, and zero sequence voltage is bus zero sequence voltage data U, namely fault resistance average valueThe method comprises the following steps:
。
where Means () is the mean function.
5. Zero sequence current wavelet packet transformation energy characteristics: the wavelet packet transformation energy characteristic of the zero sequence current consists of three characteristics of wavelet packet transformation energy mean value, energy variance and energy entropy. Extracting wavelet packet transformation energy mean value characteristics, wherein the higher the value is, the higher the energy level of the fault cause is; extracting wavelet packet transformation energy variance characteristics, wherein the higher the variance is, the greater the energy fluctuation degree along with the time in the frequency band is; and extracting wavelet packet transformation energy entropy characteristics, wherein entropy is used for describing the complexity degree of transient signal energy distribution in a probability space. The wavelet packet decomposition algorithm can be described as:
。
in the method, in the process of the invention,the decomposition result of the wavelet packet of the upper layer is obtained; />And->The next stage of decomposition result; j is a scale index; l is a position index; n is the frequency of the wave; k is a variable; h is a 0 And h 1 The multi-resolution filter coefficients employed for the decomposition. After j layers of wavelet packet decomposition is carried out on the discrete signals, j layers of wavelet packet coefficients +.>M is the wavelet packet space position identification.
From the above, the wavelet packet individual band decomposition energy E j,k Can be expressed as:
。
wavelet packet transformation energy mean, energy variance and energy entropy can be achieved through E j,k The description is as follows:
。
wherein E is me Transforming energy mean features for wavelet packets; e (E) var Transforming energy variance characteristics for the wavelet packet; e (E) ent Transforming energy entropy characteristics for the wavelet packet; means () is a mean function; variance () is a mean square error function; entropy () is an Entropy function.
(2) Discrete features
The fault phase characteristics, reclosing characteristics, time period characteristics and weather characteristics are discrete characteristics. The fault phase characteristics can be used for distinguishing different types of faults, such as short circuit, grounding faults and the like; the reclosing characteristics can be associated with the fault occurrence and power supply restoration processes, and the state before the fault occurrence and the influence after the fault restoration can be determined according to the reclosing characteristics, so that more comprehensive information is provided for fault diagnosis and treatment; the time period characteristics can be used for analyzing time periods of different fault types, for example, the time period of bird damage fault occurrence is generally early morning, and line faults caused by construction generally occur in daytime; weather features can be used for analyzing influences of different fault types under different weather conditions, for example, in thunderstorm weather conditions, line tripping can be caused by lightning stroke, in ice and snow weather, line galloping can be caused by line icing, line tripping can be caused, and the background and environmental factors of fault occurrence can be better understood through analyzing the weather features. Discrete features are described in this disclosure as:
1. fault phase characteristics: dividing fault phase characteristics into single-phase faults and interphase faults;
2. reclosing characteristics: dividing the reclosing characteristic into reclosing success and reclosing failure (the reclosing success is not overlapped and is also regarded as overlapping failure);
3. time period characteristics: the time period characteristic divides the whole day into 2 time periods, mainly 7-22 hours;
4. weather characteristics: weather characteristics are determined according to the weather conditions of the day when the fault occurs, and are mainly classified into sunny, thunder, wind, rain, fog and ice.
In step 3, for a small sample data set, data type unbalance is one of the common problems in the classifier model training process, in order to solve the data unbalance problem, the embodiment adopts the SMOTE algorithm to perform data expansion, and the basic idea is to analyze and simulate a few types of samples, and add a new sample manually simulated into the data set, so that the types in the original data are not seriously unbalanced any more. The simulation process of the algorithm adopts a K nearest neighbor algorithm (KNN) technology, and comprises the following specific steps:
(1) And constructing a new minority class sample by adopting a K neighbor algorithm.
The method comprises the following specific steps: determining a minority class sample; in this embodiment, a minority class of samples is determined according to the number of different kinds of samples in the dataset.
Calculating K neighbors of each minority class sample by adopting a nearest neighbor algorithm;
randomly selecting N samples from K neighbors to perform random linear interpolation;
a new minority class sample is obtained.
(2) And combining the new minority class samples with the original fault interval characteristic data set to generate a new training set.
In step 4, the invention learns the weight of the discrete features in the data set extracted in the step by utilizing an objective weighting method and a self-adaptive learning mode, trains the continuous features in the data set by utilizing an automatic machine learning mode, and finally realizes the discrimination of the bird damage tripping fault in the power transmission fault wave recording. The discrete features are subjected to weight training by using an objective weighting method, and the specific steps for obtaining the first fault discrimination model are as follows:
s1: and determining initial weights by using an objective weighting method aiming at discrete features in the data set.
Specifically, the structure and the connection weight of the neural network are continuously adjusted, and the model and the parameters are automatically adjusted according to the change of the data and the change of the network structure, so that the learning effect and the adaptability of the model are improved, and the optimal discrete feature weight data are obtained.
In a specific embodiment, the specific step of continuously adjusting the structure and the connection weight of the neural network includes:
1. initializing parameters: at the beginning of training, the connection weights and bias terms of the neural network are initialized.
2. Forward propagation: and forward propagation is carried out through the neural network by using the input data, and a prediction result of the network is calculated.
3. Calculating loss: the predicted result of the network is compared to the actual label and an error in the predicted result is calculated, which error is commonly referred to as a loss (or cost) function.
4. Back propagation: the contribution degree of each connection weight and bias term to the loss is calculated by a back propagation algorithm using the value of the loss function.
5. Parameter updating: the connection weights and bias terms are updated in the opposite direction of the gradient using an optimization algorithm such as gradient descent, thereby reducing the value of the loss function.
6. Repeating the iteration: and repeatedly executing the steps 2-5, and gradually reducing the value of the loss function through multiple iterations, so that the performance of the model on training data is improved.
The present embodiment calculates the gradient by a back-propagation algorithm and then updates the parameters using an optimization algorithm. As the number of iterations increases, the model gradually adjusts its own parameters so that it fits the training data better and performs better on the new data.
S2: and optimizing the initial weight of the discrete feature determined by the objective weighting method by utilizing the self-adaptive neural network to obtain a first fault discrimination model.
In a specific embodiment, for discrete features in the dataset, an objective weighting method is used to determine weights, so as to ensure fairness, credibility and repeatability of initial weights, to reduce influence of subjective bias and uncertainty on decisions, and finally, initial weights of the determined discrete features are set as shown in table 1:
TABLE 1 discrete feature initial weights determined based on objective weighting method
。
The self-adaptive neural network is utilized to optimize the initial weight of the discrete feature determined by the objective weighting method, the model and parameters are automatically adjusted according to the change of data and the change of the network structure by continuously adjusting the structure and the connection weight of the neural network, so that the learning effect and the adaptability of the model are improved, the optimal weight data of the discrete feature are obtained, and finally the judging probability for judging the power transmission tripping fault type according to the discrete feature is obtained.
In step 5, the continuous features are learned using an automatic machine learning framework AutoGluon.
The continuous characteristics in the fault record data of the fault type are learned and trained by utilizing an automatic machine learning framework AutoGluon, proper machine learning models and algorithms are automatically selected by utilizing the automation capability of the automatic machine learning framework AutoGluon, super-parameter tuning is performed, the accuracy of prediction is improved by integrating a plurality of models and algorithms and utilizing a model fusion mode, the performance of each model is compared, the models are not required to be manually selected and adjusted, and finally the fault type discrimination probability of the optimal model is output.
In step 6, the fault type discrimination probability discriminated by the discrete features and the fault type discrimination probability of the continuous features discriminated by the AutoGluon frame are added to obtain the discrimination of the final power transmission line fault tripping bird trouble fault type. In this embodiment, the sum of all probabilities of discrimination using the discrete feature and the continuous feature is added to 1, that is: the sum of weights of a and b is 1, so that the sum of all probabilities finally output is also 1, and the fault corresponding to the highest probability is taken as the final fault discrimination result.
Embodiment two:
the second embodiment of the invention provides a power transmission wave-recording bird damage fault discrimination system based on objective weighting, which comprises the following steps:
the data acquisition module is configured to acquire fault recording data, and preprocess the recording data to obtain real data in the recording data;
the feature extraction module is configured to extract fault interval features according to a set threshold, wherein the continuity features are extracted from the time domain and the frequency domain, and the discreteness features are extracted from the multiple dimensions;
the feature enhancement module is configured to enhance the fault interval features by adopting a synthetic minority class oversampling algorithm;
the first model training module is configured to perform weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model;
the second model training module is configured to learn the continuous features by utilizing an adaptive learning mode to obtain a second fault discrimination model;
the fault judging module is configured to combine judging results of the first fault judging model and the second fault judging model on the data to be tested as final judging results of bird damage faults.
The steps involved in the second embodiment correspond to those of the first embodiment, and reference is made to the relevant description of the first embodiment for the implementation manner.
It will be appreciated by those skilled in the art that the modules or steps of the invention described above may be implemented by general-purpose computer means, alternatively they may be implemented by program code executable by computing means, whereby they may be stored in storage means for execution by computing means, or they may be made into individual integrated circuit modules separately, or a plurality of modules or steps in them may be made into a single integrated circuit module. The present invention is not limited to any specific combination of hardware and software.
While the foregoing description of the embodiments of the present invention has been presented in conjunction with the drawings, it should be understood that it is not intended to limit the scope of the invention, but rather, it is intended to cover all modifications or variations within the scope of the invention as defined by the claims of the present invention.
Claims (12)
1. The method for discriminating the power transmission wave-recording bird damage fault based on objective weighting is characterized by comprising the following steps:
acquiring fault recording data, and preprocessing the recording data to obtain real data in the recording data;
extracting fault interval features according to a set threshold value, wherein continuity features are extracted from two aspects of a time domain and a frequency domain, and discreteness features are extracted from multiple dimensions;
adopting a synthetic minority class oversampling algorithm to enhance the fault interval characteristics;
performing weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model;
learning the continuous features by utilizing a self-adaptive learning mode to obtain a second fault discrimination model;
and combining the discrimination results of the first fault discrimination model and the second fault discrimination model on the data to be detected to serve as the final result of bird damage fault discrimination.
2. The objective weighting-based power transmission recording bird damage discrimination method according to claim 1, wherein the specific steps of preprocessing the recording data are as follows:
and performing irrelevant data rejection operation according to a file format obtained by the recording data to obtain current and voltage data of each phase as real data.
3. The objective weighting-based transmission wave-recording bird fault discrimination method according to claim 1, wherein the continuity features include direct current content of zero sequence current, zero sequence current amplitude variation, zero sequence current 3 rd harmonic content, fault resistance average and zero sequence current wavelet packet transformation energy features.
4. The objective weighting-based power transmission wave-recording bird damage fault discrimination method according to claim 1, wherein the discrete features are extracted through four dimensions of fault phase type, reclosing condition, period and weather.
5. The objective weighting-based power transmission wave-recording bird damage fault judging method according to claim 1, wherein the specific steps of adopting a synthetic minority class oversampling algorithm to enhance the fault interval characteristics are as follows:
constructing a new minority class sample by adopting a K neighbor algorithm;
and combining the new minority class samples with the original fault interval characteristic data set to generate a new training set.
6. The objective weighting-based power transmission recording bird damage discrimination method according to claim 5, wherein the specific steps of constructing a new minority class sample by adopting a K-nearest neighbor algorithm are as follows:
determining a minority class sample;
calculating K neighbors of each minority class sample by adopting a nearest neighbor algorithm;
randomly selecting N samples from K neighbors to perform random linear interpolation;
a new minority class sample is obtained.
7. The method for discriminating power transmission wave-recording bird damage fault based on objective weighting according to claim 1 wherein the specific step of carrying out weight training on discrete features by utilizing the objective weighting method to obtain a first fault discrimination model is as follows:
determining initial weights by using an objective weighting method aiming at discrete features in a data set;
and optimizing the initial weight of the discrete feature determined by the objective weighting method by utilizing the self-adaptive neural network to obtain a first fault discrimination model.
8. The method for discriminating power transmission wave-recording bird damage fault based on objective weighting according to claim 7 wherein the specific step of optimizing the discrete feature initial weight determined by the objective weighting method by utilizing the adaptive neural network is as follows:
the structure and the connection weight of the neural network are continuously adjusted, and the model and the parameters are automatically adjusted according to the change of the data and the change of the network structure, so that the learning effect and the adaptability of the model are improved, and the optimal discrete feature weight data are obtained.
9. The method for discriminating power transmission recording bird damage based on objective weighting according to claim 8 wherein the specific step of continuously adjusting the structure and connection weight of the neural network comprises:
initializing connection weights and bias items of the neural network at the beginning of training;
forward propagation is carried out through a neural network by using input data, and a prediction result of the network is calculated;
comparing the predicted result of the network with the actual label, and calculating the error of the predicted result, namely the loss function;
and gradually reducing the value of the loss function through multiple iterations, so as to obtain the optimal discrete feature weight data.
10. The objective weighting-based power transmission recording bird damage discrimination method according to claim 9, wherein the specific step of gradually decreasing the value of the loss function through a plurality of iterations is:
calculating the contribution degree of each connection weight and the bias term to the loss through a back propagation algorithm by using the value of the loss function;
the connection weights and bias terms are updated in the opposite direction of the gradient using a gradient descent optimization algorithm, thereby reducing the value of the loss function.
11. The method for discriminating power transmission recording bird damage fault based on objective weighting according to claim 1 wherein continuous characteristics are learned by automatic machine learning framework AutoGluon.
12. The utility model provides a transmission of electricity record wave bird trouble discriminating system based on objective empowerment which characterized in that includes:
the data acquisition module is configured to acquire fault recording data, and preprocess the recording data to obtain real data in the recording data;
the feature extraction module is configured to extract fault interval features according to a set threshold, wherein the continuity features are extracted from the time domain and the frequency domain, and the discreteness features are extracted from the multiple dimensions;
the feature enhancement module is configured to enhance the fault interval features by adopting a synthetic minority class oversampling algorithm;
the first model training module is configured to perform weight training on the discrete features by using an objective weighting method to obtain a first fault discrimination model;
the second model training module is configured to learn the continuous features by utilizing an adaptive learning mode to obtain a second fault discrimination model;
the fault judging module is configured to combine judging results of the first fault judging model and the second fault judging model on the data to be tested as final judging results of bird damage faults.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311481469.7A CN117216485B (en) | 2023-11-09 | 2023-11-09 | Objective weighting-based power transmission wave-recording bird damage fault judging method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311481469.7A CN117216485B (en) | 2023-11-09 | 2023-11-09 | Objective weighting-based power transmission wave-recording bird damage fault judging method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117216485A true CN117216485A (en) | 2023-12-12 |
| CN117216485B CN117216485B (en) | 2024-01-30 |
Family
ID=89051490
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311481469.7A Active CN117216485B (en) | 2023-11-09 | 2023-11-09 | Objective weighting-based power transmission wave-recording bird damage fault judging method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117216485B (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108535597A (en) * | 2018-04-13 | 2018-09-14 | 国网福建省电力有限公司 | Singlephase earth fault Section Location based on circuit model |
| CN109270407A (en) * | 2018-11-16 | 2019-01-25 | 国网山东省电力公司电力科学研究院 | Extra high voltage direct current transmission line failure cause discrimination method based on Multi-source Information Fusion |
| CN111898729A (en) * | 2020-06-10 | 2020-11-06 | 国网江苏省电力有限公司电力科学研究院 | Power transmission line fault reason identification method and system |
| CN112147462A (en) * | 2020-09-16 | 2020-12-29 | 国网江西省电力有限公司电力科学研究院 | Power transmission line fault identification method based on deep learning |
| CN112149554A (en) * | 2020-09-21 | 2020-12-29 | 广东电网有限责任公司清远供电局 | Fault classification model training method, fault detection method and related device |
| US20210382103A1 (en) * | 2018-11-01 | 2021-12-09 | University Of Manitoba | Method for Determining Conductors Involved in a Fault on a Power Transmission Line and Fault Location Using Local Current Measurements |
| CN113837477A (en) * | 2021-09-27 | 2021-12-24 | 西安交通大学 | Data dual-drive power grid fault prediction method, device and equipment under typhoon disaster |
| CN113960417A (en) * | 2021-11-19 | 2022-01-21 | 国网湖南省电力有限公司 | Power transmission line fault rapid diagnosis method, device, equipment and medium based on multi-source information fusion |
| CN114113911A (en) * | 2021-12-06 | 2022-03-01 | 国网山东省电力公司电力科学研究院 | Fault waveform-based fault type discrimination method and discrimination system |
| WO2022068074A1 (en) * | 2020-09-29 | 2022-04-07 | 山东大学 | Early warning method and system for slowly developing permanent fault |
| US20220170975A1 (en) * | 2020-11-27 | 2022-06-02 | Mitsubishi Electric Corporation | System and method for sensing fault symptom before occurrence of fault utilizing neural network model |
| WO2022183698A1 (en) * | 2021-03-05 | 2022-09-09 | 国网电力科学研究院武汉南瑞有限责任公司 | Artificial intelligence fault identification system and method based on power transmission line transient waveform |
| CN115422759A (en) * | 2022-09-15 | 2022-12-02 | 西安热工研究院有限公司 | Equipment fault diagnosis model construction and equipment fault diagnosis method and device |
| US20230036218A1 (en) * | 2021-07-27 | 2023-02-02 | Qualitrol Company Llc | Systems and methods for ai continued learning in electrical power grid fault analysis |
| CN116644361A (en) * | 2023-04-20 | 2023-08-25 | 武汉大学 | Overhead transmission line fault diagnosis method and computer readable medium |
-
2023
- 2023-11-09 CN CN202311481469.7A patent/CN117216485B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108535597A (en) * | 2018-04-13 | 2018-09-14 | 国网福建省电力有限公司 | Singlephase earth fault Section Location based on circuit model |
| US20210382103A1 (en) * | 2018-11-01 | 2021-12-09 | University Of Manitoba | Method for Determining Conductors Involved in a Fault on a Power Transmission Line and Fault Location Using Local Current Measurements |
| CN109270407A (en) * | 2018-11-16 | 2019-01-25 | 国网山东省电力公司电力科学研究院 | Extra high voltage direct current transmission line failure cause discrimination method based on Multi-source Information Fusion |
| CN111898729A (en) * | 2020-06-10 | 2020-11-06 | 国网江苏省电力有限公司电力科学研究院 | Power transmission line fault reason identification method and system |
| CN112147462A (en) * | 2020-09-16 | 2020-12-29 | 国网江西省电力有限公司电力科学研究院 | Power transmission line fault identification method based on deep learning |
| CN112149554A (en) * | 2020-09-21 | 2020-12-29 | 广东电网有限责任公司清远供电局 | Fault classification model training method, fault detection method and related device |
| WO2022068074A1 (en) * | 2020-09-29 | 2022-04-07 | 山东大学 | Early warning method and system for slowly developing permanent fault |
| US20220170975A1 (en) * | 2020-11-27 | 2022-06-02 | Mitsubishi Electric Corporation | System and method for sensing fault symptom before occurrence of fault utilizing neural network model |
| WO2022183698A1 (en) * | 2021-03-05 | 2022-09-09 | 国网电力科学研究院武汉南瑞有限责任公司 | Artificial intelligence fault identification system and method based on power transmission line transient waveform |
| US20230036218A1 (en) * | 2021-07-27 | 2023-02-02 | Qualitrol Company Llc | Systems and methods for ai continued learning in electrical power grid fault analysis |
| CN113837477A (en) * | 2021-09-27 | 2021-12-24 | 西安交通大学 | Data dual-drive power grid fault prediction method, device and equipment under typhoon disaster |
| WO2023045278A1 (en) * | 2021-09-27 | 2023-03-30 | 西安交通大学 | Data dual-drive method, apparatus, and device for predicting power grid failure during typhoon |
| CN113960417A (en) * | 2021-11-19 | 2022-01-21 | 国网湖南省电力有限公司 | Power transmission line fault rapid diagnosis method, device, equipment and medium based on multi-source information fusion |
| CN114113911A (en) * | 2021-12-06 | 2022-03-01 | 国网山东省电力公司电力科学研究院 | Fault waveform-based fault type discrimination method and discrimination system |
| CN115422759A (en) * | 2022-09-15 | 2022-12-02 | 西安热工研究院有限公司 | Equipment fault diagnosis model construction and equipment fault diagnosis method and device |
| CN116644361A (en) * | 2023-04-20 | 2023-08-25 | 武汉大学 | Overhead transmission line fault diagnosis method and computer readable medium |
Non-Patent Citations (8)
| Title |
|---|
| LI FAN等: "Fault Identification Technology For Overhead Transmission Lines", 2018 3RD INTERNATIONAL CONFERENCE ON SMART CITY AND SYSTEMS ENGINEERING (ICSCSE) * |
| 刘辉军;李宏军;吴雄;邬邦发;江明奎;吴昌吉;李叶华;翁志海;: "输电线路运行故障的特征与分析识别", 宁德师范学院学报(自然科学版), no. 02 * |
| 廖志伟;岳苓;文福拴;朱炳铨;徐立中;: "基于规则网的高压输电线路故障诊断与保护动作性能评价", 电力建设, no. 02 * |
| 张俊芳;刘鹏;: "基于改进模糊神经网络的配电网故障选线研究", 电力系统保护与控制, no. 22 * |
| 张广斌;束洪春;于继来;: "案例推理式输电线单端电流行波故障测距", 中国电机工程学报, no. 06 * |
| 杨毅;范栋琛;殷浩然;韩佶;苗世洪;: "基于深度-迁移学习的输电线路故障选相模型及其可迁移性研究", 电力自动化设备, no. 10 * |
| 陆帅: "基于深度学习的输电线路故障原因辨识研究", 中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑 * |
| 黄建明;李晓明;瞿合祚;张礼得;: "考虑小波奇异信息与不平衡数据集的输电线路故障识别方法", 中国电机工程学报, no. 11 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117216485B (en) | 2024-01-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112233091B (en) | Wind turbine blade image damage detection and positioning method | |
| CN109873610B (en) | Photovoltaic array fault diagnosis method based on IV characteristic and depth residual error network | |
| CN110082640B (en) | Distribution network single-phase earth fault identification method based on long-time memory network | |
| CN116256602B (en) | Method and system for identifying state abnormality of low-voltage power distribution network | |
| CN110728195B (en) | Power quality disturbance detection method based on YOLO algorithm | |
| CN104459397B (en) | Power quality disturbance recognizing method with self-adaptation multi-resolution generalized S conversion adopted | |
| CN112147462A (en) | Power transmission line fault identification method based on deep learning | |
| CN109936113B (en) | Protection action intelligent diagnosis method and system based on random forest algorithm | |
| CN110969194A (en) | Cable early fault positioning method based on improved convolutional neural network | |
| CN111898729A (en) | Power transmission line fault reason identification method and system | |
| CN111008641A (en) | Power transmission line tower external force damage detection method based on convolutional neural network | |
| CN111999591A (en) | Method for identifying abnormal state of primary equipment of power distribution network | |
| CN117216485B (en) | Objective weighting-based power transmission wave-recording bird damage fault judging method and system | |
| CN112083275A (en) | Distribution network fault type identification method and system | |
| CN116973683A (en) | Method and system for positioning single-phase ground fault section of electric collecting line of wind power plant | |
| CN108594156B (en) | Improved current transformer saturation characteristic identification method | |
| CN110245617B (en) | Artificial intelligence analysis method based on transient recording waveform | |
| CN114415056A (en) | Distributed power supply island fault detection method | |
| CN114325072A (en) | Ferromagnetic resonance overvoltage identification method and device based on gram angular field coding | |
| CN114545147A (en) | Voltage sag source positioning method based on deep learning in consideration of time-varying topology | |
| CN114666236A (en) | Full-automatic signal detection, identification and alarm method | |
| CN114661700A (en) | Artificial influence weather operation effect inspection method based on AI | |
| CN113657149A (en) | Electric energy quality analysis and identification method based on deep learning | |
| CN112163613A (en) | Rapid identification method for power quality disturbance | |
| Murali et al. | Forecasting and Classification Of Power Quality Disturbance In Smart Grid Using Hybrid Networks |
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 |