CN117312902A - Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network - Google Patents
Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network Download PDFInfo
- Publication number
- CN117312902A CN117312902A CN202310869007.6A CN202310869007A CN117312902A CN 117312902 A CN117312902 A CN 117312902A CN 202310869007 A CN202310869007 A CN 202310869007A CN 117312902 A CN117312902 A CN 117312902A
- Authority
- CN
- China
- Prior art keywords
- power quality
- layer
- wavelet
- disturbance
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 17
- 230000008859 change Effects 0.000 title claims abstract description 9
- 230000009466 transformation Effects 0.000 claims abstract description 13
- 230000006870 function Effects 0.000 claims description 11
- 210000002569 neuron Anatomy 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 8
- 238000011176 pooling Methods 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000013145 classification model Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims description 2
- 238000013519 translation Methods 0.000 claims description 2
- 230000001351 cycling effect Effects 0.000 claims 1
- 230000011664 signaling Effects 0.000 claims 1
- 238000012360 testing method Methods 0.000 abstract description 11
- 238000012549 training Methods 0.000 abstract description 9
- 238000012544 monitoring process Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 9
- 239000011159 matrix material Substances 0.000 description 8
- 238000012545 processing Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001052 transient effect 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
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Power Engineering (AREA)
- Water Supply & Treatment (AREA)
- Computational Linguistics (AREA)
- Medical Informatics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a power grid power quality on-line state estimation method based on wavelet change and convolutional neural network, which adopts wavelet change to extract power quality monitoring data characteristic values, generates typical power quality disturbance random data samples, and divides training samples and test sample proportions; performing wavelet transformation on the training sample and the test sample respectively, and extracting a plurality of characteristic values of the power quality disturbance according to the power quality characteristic model; and finally, the automatic data classification and the on-line state estimation of the power quality are finished by a convolutional neural network method.
Description
Technical Field
The invention relates to the technical field of power grid power quality estimation methods, in particular to the technical field of power grid power quality on-line state estimation methods.
Background
Under the development background of realizing the carbon neutralization target, pushing energy to transform and constructing a novel electric power system, the new energy technology rapidly advances, the investment cost gradually decreases, and the current new energy power generation capacity is increased year by year. The new energy is connected in a large scale, so that the power quality monitoring points are rapidly increased, the data volume is continuously increased, the data processing speed is slowed down, the application service performance is reduced, and the traditional data acquisition and processing method is difficult to adapt to the requirements of a new generation of power grid; under the background of big data and artificial intelligent power grids, the power quality data of a new generation power system requires cross-platform interaction, support of ultra-large database quantity and high-speed data processing capacity, so that the traditional power quality data acquisition and processing method is difficult to adapt to the development needs of the power grids. And the electric energy quality big data analysis technology is required to be researched, the electric energy quality disturbance characteristics are accurately extracted, and the electric energy quality state monitoring performance is improved.
In the aspect of power quality disturbance feature extraction, the method mainly comprises Fourier transformation, wavelet transformation, S transformation, hilbert yellow transformation and the like; the disturbance signal feature extraction is the basis for grasping the running state of the power grid through data feature classification, and the existing data feature recognition and classification method comprises a neural network, a random forest network, a support vector machine and the like.
Disclosure of Invention
The invention provides a power grid power quality on-line state estimation method based on wavelet change and convolutional neural network. The electric energy quality features are extracted more accurately, and the classification accuracy is higher.
The technical solution adopted by the invention for solving the technical problems is as follows:
the power grid power quality on-line state estimation method based on wavelet change and convolutional neural network comprises the following steps:
first, wavelet transformation feature value extraction
The method adopts the wavelet transformation characteristic extraction processing of multi-resolution analysis, firstly, the power quality signal to be processed is decomposed into a low-frequency signal and a high-frequency signal, the high-frequency signal is reserved, secondly, the low-frequency signal is decomposed again to obtain a new group of high-frequency signal and low-frequency signal, the above processes are circulated until the set decomposition times,
the discrete wavelet function is defined as:
wherein: psi phi type j,k (t) is a wavelet transform function; j, k is an integer; a, a 0 j As a function of the discrete scale parameter,a 0 ＞0；ka 0 j b 0 as a result of the discrete translation parameters,
the discretized wavelet transform coefficients are:
wherein: c (C) j,k Is a discretized wavelet transform function; f (t) is a disturbance signal;as a function of the conjugate discrete wavelet,
when the power quality disturbance characteristics are extracted, 5 layers of multi-analysis wavelet transformation decomposition is carried out on the sampling signals, so that two characteristics of energy and entropy extracted from an approximate component and a detail component are respectively defined as a low-frequency signal cAn and a high-frequency signal cDn, and the characteristic expression is defined as follows:
wherein: j=1, 2,3,4,5,
after wavelet decomposition, the characteristic vector is obtained
The total energy expression of the signal is,
based on the total energy expression (3), the relative energy expression in the transform frequency domain is,
at the same time the relative energy is satisfied,
according to shannon entropy definition, obtaining energy entropy characteristics;
second, classifying the power quality based on the convolutional neural network
Establishing a power quality data classification model based on a convolutional neural network, firstly importing data into an input layer, and performing convolutional operation on different characteristic data through convolutional check; then, obtaining an average characteristic value or a maximum characteristic value through a pooling layer, so that the characteristic dimension of the system is reduced to suppress the fitting; fitting the characteristic data through the full connection layer; finally, the power quality type vector is obtained from the output layer,
the convolution layer convolves the power quality disturbance data with the neuron weight in the convolution kernel to achieve the purpose of feature extraction, and the convolution expression is:
wherein:an ith neuron that is an output of the ith convolutional layer; />Weights between i and j neurons of the layer I; />For the offset of the ith neuron of the ith convolutional layer,
the characteristic data space of the convolution layer is too huge, and the dimension of the power quality disturbance is reduced by a maximum pooling layer in the method, wherein the expression of the maximum pooling layer is as follows:
the full connection layer fits the characteristics obtained in the convolution layer and the pooling layer, and the expression is as follows:
wherein: f () is an activation function;for the parameters that can be learned for the first layer,
the Softmax classification layer carries out corresponding class probability calculation on the extracted power quality disturbance characteristics, namely:
wherein: n represents a disturbance class; p is p i Representing the probability that the disturbance features are of class i (i=1, 2, …, N); b j Neurons to be activated for the j-th class of output layer.
The beneficial effects achieved by adopting the technical proposal of the invention are as follows:
the invention adopts wavelet change to extract the characteristic value of the power quality monitoring data, generates a typical power quality disturbance random data sample, and divides the proportion of a training sample and a test sample; performing wavelet transformation on the training sample and the test sample respectively, and extracting a plurality of characteristic values of the power quality disturbance according to the power quality characteristic model; and finally, completing automatic data classification and power quality on-line state estimation by a convolutional neural network method. Compared with the traditional machine learning method, the electric energy quality characteristic extraction is more accurate, and the classification accuracy is higher.
Drawings
FIG. 1 is a graph of a characteristic profile of the present invention;
FIG. 2 is a training scatter plot in accordance with the present invention;
FIG. 3 is a plot of test scatter in accordance with the present invention;
FIG. 4 is a diagram of a training confusion matrix in accordance with the present invention;
FIG. 5 is a diagram of a test confusion matrix according to the present invention;
FIG. 6 is a diagram of an SVM model confusion matrix in accordance with the present invention;
FIG. 7 is a naive Bayes model confusion matrix diagram in accordance with the present invention;
FIG. 8 is a diagram of a fine tree model confusion matrix in accordance with the present invention;
FIG. 9 is a graph of four conditions of the load according to the present invention;
FIG. 10 is a scatter plot of actual condition test results in the present invention;
FIG. 11 is a graph of a confusion matrix of actual condition test results in the present invention.
Detailed Description
In the invention, ten representative different power quality disturbances are selected, a power quality disturbance mathematical model is established into different disturbance waveforms, as shown in table 1,
meter 1 electric energy quality disturbance signal mathematical model
In Table 1, y (t) is the power quality disturbance model signal, and k is the disturbance signal y (t)The amplitude, alpha and beta are the amplitude and frequency parameters of the voltage flicker disturbance signal, f n For transient oscillation frequency, alpha 1 As fundamental wave parameter in harmonic disturbance, alpha 3 、α 5 、α 7 The parameters of the 3, 5 and 7 harmonics in the harmonic disturbance are respectively.
According to the typical power quality disturbance model of the table 1, the power quality data characteristic distribution can be calculated, wherein the value range of k is [1,1.5 ]]The method comprises the steps of carrying out a first treatment on the surface of the The value range of fn is [300,500 ]]；α 3 、α 5 、α 7 The value range is [0.05,0.15 ]]As shown in fig. 1, the present invention,
the invention extracts 6 power quality characteristic values through wavelet transformation by the established power quality mathematical signal model and convolutional neural model, obtains 2500 groups of data according to the power quality condition, trains 2400 groups of data, tests 100 groups of data, as shown in table 2,
table 2 Power quality sample model
A training scattergram as shown in fig. 2, a test scattergram as shown in fig. 3, a training confusion matrix diagram as shown in fig. 4, and a test confusion matrix diagram as shown in fig. 5 are obtained.
The state recognition effect of the traditional machine learning models such as decision tree, naive Bayes classification, linear discrimination and the like is shown in table 3, the classification precision of a support vector machine, a convolutional neural network and a fine tree is more than 95% except for naive Bayes, and the highest classification precision of the convolutional neural network model is 98.2917%.
Table 3 comparison of different classification methods
As shown in fig. 6, 7 and 8, according to comparison of different model confusion matrixes, the convolutional neural network model is proved to be capable of accurately identifying and classifying the power quality disturbance model.
The invention builds a simulation model of the power system, simulates the typical disturbance working condition of the power system in the table 4, and identifies the disturbance type of the power quality. The output waveform diagram 9 of the 4 working conditions shows that 880 groups of data classification of the 4 working conditions are obtained, wherein 800 groups of data are subjected to convolutional neural network model training, and electric energy quality disturbance characteristics are learned; the 80 sets of data are used to detect model identification classification situations.
Meter 4 Power quality Condition
The accuracy of the 80 groups of data reaches 97.5% through a scatter diagram predicted based on wavelet transformation and a neural network state classification model as shown in fig. 10, which shows that the classification model can accurately classify disturbance types of 4 working conditions of the power system. As can be seen from fig. 2,3,4,5, 6, 7, 8, 9, 10, 11 and table 3, the accuracy and efficiency of the power quality on-line state estimation method of the power grid based on wavelet change and convolutional neural network are higher than those of other cases.
Claims (1)
1. The power grid power quality on-line state estimation method based on wavelet change and convolutional neural network is characterized by comprising the following steps of:
first, wavelet transformation feature value extraction
Firstly decomposing the power quality signal to be processed into a low frequency signal and a high frequency signal, and reserving the high frequency signal, secondly decomposing the low frequency signal again to obtain a new group of high frequency signal and low frequency signal, and cycling the above processes until the set decomposition times,
the discrete wavelet function is defined as:
wherein: psi phi type j,k (t) is a wavelet transform function; j, k is an integer; a, a 0 j Being a discrete scale parameter, a 0 ＞0；ka 0 j b 0 As a result of the discrete translation parameters,
the discretized wavelet transform coefficients are:
wherein: c (C) j,k Is a discretized wavelet transform function; f (t) is a disturbance signal;as a function of the conjugate discrete wavelet,
when the power quality disturbance characteristics are extracted, 5 layers of multi-analysis wavelet transformation decomposition is carried out on the sampling signals, so that two characteristics of energy and entropy extracted from an approximate component and a detail component are respectively defined as a low-frequency signal cAn and a high-frequency signal cDn, and the characteristic expression is defined as follows:
wherein: j=1, 2,3,4,5,
after wavelet decomposition, the characteristic vector is obtained
The total energy expression of the signal is,
based on the total energy expression (3), the relative energy expression in the transform frequency domain is,
at the same time the relative energy is satisfied,
according to shannon entropy definition, obtaining energy entropy characteristics;
second, classifying the power quality based on the convolutional neural network
Establishing a power quality data classification model based on a convolutional neural network, firstly importing data into an input layer, and performing convolutional operation on different characteristic data through convolutional check; then, obtaining an average characteristic value or a maximum characteristic value through a pooling layer, so that the characteristic dimension of the system is reduced to suppress the fitting; fitting the characteristic data through the full connection layer; finally, the power quality type vector is obtained from the output layer,
the convolution layer convolves the power quality disturbance data with the neuron weight in the convolution kernel to achieve the purpose of feature extraction, and the convolution expression is:
wherein:an ith neuron that is an output of the ith convolutional layer; />Weights between i and j neurons of the layer I; />For the offset of the ith neuron of the ith convolutional layer,
the characteristic data space of the convolution layer is too huge, the dimension of the power quality disturbance is reduced through a maximum pooling layer, and the expression of the maximum pooling layer is as follows:
the full connection layer fits the characteristics obtained in the convolution layer and the pooling layer, and the expression is as follows:
wherein: f () is an activation function;for the parameters that can be learned for the first layer,
the Softmax classification layer carries out corresponding class probability calculation on the extracted power quality disturbance characteristics, namely:
wherein: n represents a disturbance class; p is p i Representing the probability that the disturbance features are of class i (i=1, 2, …, N); b j Neurons to be activated for the j-th class of output layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310869007.6A CN117312902A (en) | 2023-07-16 | 2023-07-16 | Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310869007.6A CN117312902A (en) | 2023-07-16 | 2023-07-16 | Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117312902A true CN117312902A (en) | 2023-12-29 |
Family
ID=89259219
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310869007.6A Pending CN117312902A (en) | 2023-07-16 | 2023-07-16 | Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117312902A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117520926A (en) * | 2024-01-04 | 2024-02-06 | 南京灿能电力自动化股份有限公司 | Electric energy quality analysis method and device |
-
2023
- 2023-07-16 CN CN202310869007.6A patent/CN117312902A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117520926A (en) * | 2024-01-04 | 2024-02-06 | 南京灿能电力自动化股份有限公司 | Electric energy quality analysis method and device |
| CN117520926B (en) * | 2024-01-04 | 2024-04-16 | 南京灿能电力自动化股份有限公司 | Electric energy quality analysis method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111562358B (en) | Transformer oil gas content prediction method and system based on combined model | |
| CN107462785B (en) | The more disturbing signal classifying identification methods of power quality based on GA-SVM | |
| CN112710969B (en) | Open-circuit fault diagnosis method for switching tube of single-phase half-bridge five-level inverter | |
| CN117312902A (en) | Power grid power quality on-line state estimation method based on wavelet change and convolutional neural network | |
| CN102279358A (en) | MCSKPCA based neural network fault diagnosis method for analog circuits | |
| CN109884419A (en) | A kind of wisdom grid power quality on-line fault diagnosis method | |
| CN102982347A (en) | Method for electric energy quality disturbance classification based on KL distance | |
| CN111723684A (en) | Method for identifying transient overvoltage type in offshore wind farm | |
| CN113378652A (en) | Disturbance classification method based on EWT-MPE-PSO-BP | |
| CN115271225A (en) | Wind power-wind power modeling method based on wavelet denoising and neural network | |
| CN111506868A (en) | Ultrashort-term wind speed prediction method based on HHT weight optimization | |
| CN114386452A (en) | Method for detecting faults of sun wheel of nuclear power circulating water pump | |
| CN107358323A (en) | A kind of power forecasting method of short-term photovoltaic generation | |
| CN114036977A (en) | Fan power quality disturbance monitoring method based on two-dimensional image coding features | |
| Kong et al. | Vibration fault analysis of hydropower units based on extreme learning machine optimized by improved sparrow search algorithm | |
| CN112636328A (en) | Medium-voltage distribution network equipment load state identification method | |
| CN116340812A (en) | Transformer partial discharge fault mode identification method and system | |
| CN110610203A (en) | Electric energy quality disturbance classification method based on DWT and extreme learning machine | |
| CN110866840A (en) | Database modeling method for power load characteristic quantity training based on knowledge graph | |
| CN111539275B (en) | Electrical load classification method and system based on load characteristic visualization | |
| Mei et al. | Classification of the type of harmonic source based on image-matrix transformation and deep convolutional neural network | |
| Wang et al. | A novel approach for the pattern recognition of hand movements based on EMG and VPMCD | |
| CN112016684A (en) | Electric power terminal fingerprint identification method of deep parallel flexible transmission network | |
| Liu et al. | Wavelet Change and Convolutional Neural Network Based Power Quality Online State Estimation Method | |
| CN116541686B (en) | Electric energy quality disturbance classification method based on multi-domain feature fusion extreme learning machine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |