CN116933114A - CNN-LSTM-based direct-current micro-grid detection method and device - Google Patents
CNN-LSTM-based direct-current micro-grid detection method and device Download PDFInfo
- Publication number
- CN116933114A CN116933114A CN202310692457.2A CN202310692457A CN116933114A CN 116933114 A CN116933114 A CN 116933114A CN 202310692457 A CN202310692457 A CN 202310692457A CN 116933114 A CN116933114 A CN 116933114A
- Authority
- CN
- China
- Prior art keywords
- matrix
- prey
- lstm
- grid
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 29
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 31
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 26
- 230000004927 fusion Effects 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000002372 labelling Methods 0.000 claims abstract description 7
- 239000011159 matrix material Substances 0.000 claims description 40
- 238000004590 computer program Methods 0.000 claims description 13
- 230000010355 oscillation Effects 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 238000010248 power generation Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 5
- 230000000368 destabilizing effect Effects 0.000 claims description 4
- 244000062645 predators Species 0.000 claims description 4
- 230000005653 Brownian motion process Effects 0.000 claims description 3
- 238000005537 brownian motion Methods 0.000 claims description 3
- 230000001687 destabilization Effects 0.000 abstract description 2
- 238000012423 maintenance Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 7
- 230000008054 signal transmission Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000003064 k means clustering Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 241001334134 Rugopharynx epsilon Species 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application discloses a direct-current micro-grid detection method based on CNN-LSTM, which comprises the following steps: step 1, acquiring a historical voltage signal of a DC power grid bus, and labeling the historical voltage signal according to a destabilization fault type and a working condition; step 2, arranging and reconstructing one-dimensional historical voltage signals in time sequence to obtain two-dimensional image data, and forming a data set with the corresponding labels; step 3, constructing a convolutional neural network, wherein the convolutional neural network comprises a feature extraction module, a feature fusion module and a classification module; step 4, training a convolutional neural network and a K-means algorithm by adopting a data set to obtain a detection model; and step 5, collecting voltage signals and inputting the voltage signals into a detection model to judge the type of the unsteady fault of the direct current power grid and the corresponding working condition. The application also provides a direct-current micro-grid detection device. The method can accurately judge the stability and the working condition identification of the direct current micro-grid, thereby providing accurate guidance for power grid rush repair and daily maintenance.
Description
Technical Field
The application belongs to the technical field of direct-current micro-grids, and particularly relates to a method and a device for detecting a direct-current micro-grid based on CNN-LSTM.
Background
The DC micro-grid comprises various forms of energy storage devices, micro sources, energy conversion devices and the like, so that the problem of system instability can be caused, but due to the fact that the working conditions are complex, such as power fluctuation can be caused by the connection of photovoltaic/wind power generation, the stability of the system can be negatively influenced by the connection of nonlinear loads, the voltage characteristics of the represented DC buses are similar, and difficulty is brought to instability detection.
The current instability judging method mainly comprises a model analysis method and a data driving method, wherein the model analysis method is used for analyzing the stability of the power grid according to an impedance criterion, and is complex and difficult to operate; data-driven methods, such as Decision Trees (DTs), support Vector Machines (SVMs), convolutional Neural Networks (CNNs), etc., are proposed, however, the dependence on the number of data samples is high, and the space-time characteristics of the ignored sample data are not wide in application range.
Patent document CN 115828165A discloses a new energy intelligent micro-grid data processing method and system, the method collects time series data of a digital intelligent energy-saving cabinet; generating derivative class features A, B and C based on the preprocessed time series data, inputting the preprocessed data in the step S2 into an LSTM to mine out time features related to the fault of the intelligent energy-saving cabinet, inputting the derivative class features A, B and C generated in the step S3 into a convolutional neural network CNN to mine out space features related to the fault of the intelligent energy-saving cabinet, and fusing the time features and the space features; and constructing a fault classification model SVM of the intelligent energy-saving cabinet to realize fault classification. However, the method requires a big data sample to train the model to realize accurate fault classification.
Patent document CN 113536607A discloses a substation signal transmission system evaluation method and system, the method comprising: step S1, establishing a simulation model of a substation signal transmission system, and performing fault signal transmission simulation by using the simulation model to obtain fault evaluation sample data; s2, training the CNN-LSTM time sequence prediction model based on the fault evaluation sample data to construct a fault path prediction model for predicting a signal transmission fault path; and S3, applying the fault path prediction model to a substation signal transmission system for online evaluation. The model cannot automatically optimize parameters and has certain errors.
Disclosure of Invention
The application aims to provide a CNN-LSTM-based direct-current micro-grid detection method and device, which can accurately judge the stability of a direct-current micro-grid and identify the working condition, thereby providing accurate guidance for power grid rush repair and daily maintenance.
In order to achieve the first object, the present application provides a technical solution, including the following steps:
step 1, acquiring a historical voltage signal of a DC power grid bus, labeling the historical voltage signal according to a destabilizing fault type and working conditions, performing similarity matching on the historical voltage signal which cannot be labeled and the historical voltage signal of the existing label by adopting a K-means algorithm, and classifying and labeling the historical voltage signal which cannot be labeled based on a matching result.
And 2, arranging and reconstructing the one-dimensional historical voltage signals in time sequence to obtain two-dimensional image data, and forming a data set by the reconstructed two-dimensional image data and the corresponding labels.
Step 3, constructing a convolutional neural network based on CNN-LSTM, wherein the convolutional neural network comprises a feature extraction module, a feature fusion module and a classification module, the feature extraction module is used for extracting image features and time sequence features of two-dimensional image data, the feature fusion module is used for fusing the image features and the time sequence features to generate corresponding fusion features, the classification module classifies according to the input fusion features to obtain classification results, and the classification results comprise fault types and corresponding working conditions.
And step 4, training the convolutional neural network and the K-means algorithm by adopting the data set to obtain a detection model for judging whether the direct current micro-grid is faulty or not.
And step 5, collecting voltage signals of the DC power grid bus and inputting the voltage signals into the detection model to judge the instability fault type and the corresponding working condition of the DC power grid.
Specifically, the instability fault types include no fault, instability caused by impedance mismatch between systems, power oscillation caused by photovoltaic/wind power generation, and bus harmonic oscillation caused by nonlinear loads.
Specifically, the specific process of the K-means algorithm is as follows:
step 1-1, setting K as 4 types, and calculating an expression of a clustering center by using the real tagged data as follows:
wherein, c k Represents a K-class cluster center, x i Representing all tagged data belonging to class k, a i Indicating the category to which the current sample belongs.
Step 1-2, setting super parameters to determine the offset degree of the clustering center, wherein the expression of the clustering center point at the moment is as follows:
c‘ k ＝c k +δ
wherein delta represents a superparameter, c' k Representing the cluster center after the offset.
And step 1-3, performing similarity matching on the history voltage signals which cannot be marked and the history voltage signals of the existing labels, and classifying the history voltage signals which cannot be marked based on a matching result.
The expression of similarity matching is as follows:
specifically, during training, a marine predator optimization algorithm is adopted to optimize a convolutional neural network and a K-means algorithm.
Specifically, the training process is as follows:
step 4-1, randomly initializing elite matrix and prey matrix based on parameters of a convolutional neural network and a K-means algorithm, wherein the expression of the prey matrix is as follows:
wherein n represents population size, d represents dimension of generation solution, and Prey represents Prey matrix;
taking the classification accuracy of the data set as an fitness function value, and copying n parts of fitness values corresponding to each individual in the prey matrix to form an elite matrix by using the individual with the optimal fitness value;
step 4-2, updating the hunting matrix at different stages based on iteration step length, and performing overall optimization based on vortex problem, wherein the expression of the optimization process is as follows:
wherein CF represents an iteration step, R represents a vector of random numbers, R e (0, 1), R1 and R2 represent random subscripts, U represents a binary vector comprising 0 and 1, d represents a dimension representing the solution, prey i A prey matrix representing an ith iteration step;
and step 4-3, repeating the step 4-2 until the iteration termination condition is met, so as to obtain the optimal parameters of the convolutional neural network and the K-means algorithm.
Specifically, the parameters include the number of channels of a convolutional layer of the convolutional neural network, the size of a convolutional kernel, the change interval of a pooling layer, the number of memory units of LSTM and the offset value of a K-means algorithm.
Specifically, the specific process of updating the hunting matrix of different stages based on the iteration step is as follows.
When the iteration step is less than 1/3 of the maximum iteration step, the game of Brownian is adopted to update the prey matrix.
When the iteration step length is greater than 1/3 of the maximum iteration step length and less than 2/3 of the maximum iteration step length, the front half population of the prey matrix is updated by adopting Levy motion, and the rear half population of the prey matrix is updated by adopting Brownian motion.
When the iteration step is greater than 2/3 of the maximum iteration step, then the Levy motion is used to update the prey matrix.
In order to achieve the second object, the present application further provides an apparatus, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to perform the detection model, wherein the specific steps are as follows:
and inputting the voltage signal of the direct current micro power grid into a detection model to judge the instability fault type and the corresponding working condition of the direct current power grid.
The instability fault types include:
type 1, system stability.
Type 2, instability caused by impedance mismatch between systems.
Type 3, power oscillation due to photovoltaic/wind power generation.
Type 4, bus harmonic oscillation due to nonlinear loading.
Compared with the prior art, the application has the beneficial effects that:
(1) Based on LSTM-CNN, a K-means clustering algorithm is introduced to solve the problem of lack of fault characteristics in the process of destabilization of the direct-current micro-grid.
(2) And optimizing the super parameters of the convolutional neural network and the K-means clustering algorithm by adopting a predator algorithm, so that an optimal training model can be obtained.
Drawings
Fig. 1 is a flowchart of a method for detecting a dc micro-grid according to the present embodiment;
FIG. 2 is a schematic diagram of the K-means clustering method provided in the present embodiment to process unlabeled data;
FIG. 3 is a structural framework of a convolutional neural network+long-short-term memory network provided in this embodiment;
FIG. 4 is a flow chart for a marine predator optimization algorithm to find the optimal hyper-parameters.
Detailed Description
The exemplary embodiments of the present application will now be described with reference to the accompanying drawings, however, the present application may be embodied in many different forms and is not limited to the examples described herein, which are provided to fully and completely disclose the present application and fully convey the scope of the application to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the accompanying drawings is not intended to be limiting of the application. In the drawings, like elements/components are referred to by like reference numerals.
Unless otherwise indicated, terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. In addition, it will be understood that terms defined in commonly used dictionaries should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense.
As shown in fig. 1, a CNN-LSTM based direct current micro grid detection method.
Step 1, acquiring a historical voltage signal of a DC power grid bus, labeling the historical voltage signal according to a destabilizing fault type and working conditions, performing similarity matching on the historical voltage signal which cannot be labeled and the historical voltage signal of the existing label by adopting a K-means algorithm, and classifying and labeling the historical voltage signal which cannot be labeled based on a matching result.
More specifically, the original bus historical voltage signal of the direct current power grid is sampled, wherein the original bus historical voltage signal comprises waveform characteristic values and system stability (type 1), instability is caused by impedance mismatch among systems (type 2), power oscillation is caused by photovoltaic/wind power generation (type 3), and bus harmonic oscillation is caused by nonlinear load (type 4).
As shown in fig. 2, the specific procedure of the K-means algorithm is as follows:
step 1-1, setting K as 4 types, and calculating an expression of a clustering center by using the real tagged data as follows:
wherein, c k Represents a K-class cluster center, x i Representing all tagged data belonging to class k, a i Indicating the category to which the current sample belongs.
Step 1-2, setting super parameters to determine the offset degree of the clustering center, wherein the expression of the clustering center point at the moment is as follows:
c‘ k ＝c k +δ
wherein delta represents a super parameter, c ‘ k Representing the cluster center after the offset.
And step 1-3, performing similarity matching on the history voltage signals which cannot be marked and the history voltage signals of the existing labels, and classifying the history voltage signals which cannot be marked based on a matching result.
And 2, arranging and reconstructing the one-dimensional historical voltage signals in time sequence to obtain two-dimensional image data, and forming a data set by the reconstructed two-dimensional image data and the corresponding labels.
Step 3, constructing a convolutional neural network based on CNN-LSTM, wherein the convolutional neural network comprises a feature extraction module, a feature fusion module and a classification module, the feature extraction module is used for extracting image features and time sequence features of two-dimensional image data, the feature fusion module is used for fusing the image features and the time sequence features to generate corresponding fusion features, the classification module classifies according to the input fusion features to obtain classification results, and the classification results comprise fault types and corresponding working conditions.
As shown in fig. 3, the convolutional neural network includes: the 2-dimensional convolutional neural network (2D-CNN) comprises 3 convolutional layers, 2 max pooling layers, a global pooling layer, and is generated from CNN modelThe feature vector sequence is put forward in the feature map, the feature sequence is input into LSTM for learning, and the LSTM layer adopts a network structure with 2 layers and k memory units, wherein h is i Are implicit state variables for different time steps of LSTM. Finally, to establish the relationship between the feature and fault class, the LSTM layer is connected to a global averaging pool, and then a softmax classifier is used to implement the destabilizing source classification.
And step 4, training the convolutional neural network and the K-means algorithm by adopting the data set to obtain a detection model for judging whether the direct current micro-grid is faulty or not.
More specifically, as shown in fig. 4, the training process is as follows:
step 4-1, first determining an initial range of the super parameter, including: offset value of K-means algorithm, number of convolution layer channels of convolution neural network, convolution kernel size, change interval of pooling layer, number of memory units of LSTM, elite matrix (Elite) and Prey matrix (Prey) which are randomly initialized, wherein the expression of Prey matrix is as follows:
where n represents population size, d represents the dimension of the generation solution, prey represents the Prey matrix, and Elite moment Elite is constructed from top-level Prey and has the same dimension as the Prey matrix.
And taking the classification accuracy of the data set as an fitness function value, and copying n parts of fitness values corresponding to each individual in the hunting matrix by using the individual with the optimal fitness value to form an elite matrix.
Step 4-2, updating the hunting matrix at different stages based on iteration step length, and performing overall optimization based on vortex problem, wherein the expression of the optimization process is as follows:
where CF represents the iteration step, R represents a vector of random numbers, R.epsilon. (0, 1),r1 and r2 represent random subscripts, U represents a binary vector containing 0 and 1, d represents a dimension of the solution, prey i Representing the prey matrix for the ith iteration step.
The specific process of updating the prey matrix at different stages based on the iteration step length is as follows:
when the iteration step is less than 1/3 of the maximum iteration step, the game of Brownian is adopted to update the prey matrix.
When the iteration step length is greater than 1/3 of the maximum iteration step length and less than 2/3 of the maximum iteration step length, the front half population of the prey matrix is updated by adopting Levy motion, and the rear half population of the prey matrix is updated by adopting Brownian motion.
When the iteration step is greater than 2/3 of the maximum iteration step, then the Levy motion is used to update the prey matrix.
And step 4-3, repeating the step 4-2 until the iteration termination condition is met, so as to obtain the optimal parameters of the convolutional neural network and the K-means algorithm.
The present embodiment also provides a dc micro-grid detection device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to perform the detection model as described in the foregoing embodiments.
And inputting the voltage signal of the direct current micro power grid into a detection model to judge the instability fault type and the corresponding working condition of the direct current power grid.
The instability fault types include:
type 1, system stability.
Type 2, instability caused by impedance mismatch between systems.
Type 3, power oscillation due to photovoltaic/wind power generation.
Type 4, bus harmonic oscillation due to nonlinear loading.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. The scheme in the embodiment of the application can be realized by adopting various computer languages, such as object-oriented programming language Java, an transliteration script language JavaScript and the like.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The application does not depend on expert experience, greatly saves manpower, is beneficial to improving the running stability of the direct current power distribution network, and has great engineering application value and popularization prospect.
Claims (8)
1. The direct-current micro-grid detection method based on the CNN-LSTM is characterized by comprising the following steps of:
step 1, acquiring a historical voltage signal of a DC power grid bus, labeling the historical voltage signal according to a destabilizing fault type and working conditions, performing similarity matching on the historical voltage signal which cannot be labeled and the historical voltage signal of the existing label by adopting a K-means algorithm, and classifying and labeling the historical voltage signal which cannot be labeled based on a matching result;
step 2, arranging and reconstructing one-dimensional historical voltage signals in time sequence to obtain two-dimensional image data, and forming a data set by the reconstructed two-dimensional image data and corresponding labels;
step 3, constructing a convolutional neural network based on CNN-LSTM, wherein the convolutional neural network comprises a feature extraction module, a feature fusion module and a classification module, the feature extraction module is used for extracting image features and time sequence features of two-dimensional image data, the feature fusion module is used for fusing the image features and the time sequence features to generate corresponding fusion features, and the classification module classifies according to the input fusion features to obtain classification results, and the classification results comprise fault types and corresponding working conditions;
step 4, training a convolutional neural network and a K-means algorithm by adopting the data set to obtain a detection model for judging whether the direct current micro-grid is faulty or not;
and step 5, collecting voltage signals of the DC power grid bus and inputting the voltage signals into the detection model to judge the instability fault type and the corresponding working condition of the DC power grid.
2. The CNN-LSTM based direct current micro grid detection method according to claim 1, wherein the type of instability fault includes instability due to impedance mismatch between systems, power oscillation due to photovoltaic/wind power generation, and bus harmonic oscillation due to nonlinear load.
3. The CNN-LSTM based direct current micro grid detection method according to claim 1, wherein the specific process of the K-means algorithm is as follows:
step 1-1, setting K as 4 types, and calculating an expression of a clustering center by using the real tagged data as follows:
wherein, c k Represents a K-class cluster center, x i Representing all tagged data belonging to class k, a i Representing the category to which the current sample belongs;
step 1-2, setting super parameters to determine the offset degree of the clustering center, wherein the expression of the clustering center point at the moment is as follows:
c‘ k ＝c k +δ
wherein delta represents a superparameter, c' k Representing the cluster center after the offset;
and step 1-3, performing similarity matching on the history voltage signals which cannot be marked and the history voltage signals of the existing labels, and classifying the history voltage signals which cannot be marked based on a matching result.
4. The CNN-LSTM based direct current micro-grid detection method according to claim 1, wherein the convolutional neural network and the K-means algorithm are optimized by using a marine predator optimization algorithm during training.
5. The CNN-LSTM based direct current microgrid detection method according to claim 1 or 4, wherein the training process is as follows:
step 4-1, randomly initializing elite matrix and prey matrix based on parameters of a convolutional neural network and a K-means algorithm, wherein the expression of the prey matrix is as follows:
wherein n represents population size, d represents dimension of generation solution, and Prey represents Prey matrix;
taking the classification accuracy of the data set as an fitness function value, and copying n parts of fitness values corresponding to each individual in the prey matrix to form an elite matrix by using the individual with the optimal fitness value;
step 4-2, updating the hunting matrix at different stages based on iteration step length, and performing overall optimization based on vortex problem, wherein the expression of the optimization process is as follows:
wherein CF represents an iteration step, R represents a vector of random numbers, R e (0, 1), R1 and R2 represent random subscripts, U represents a binary vector comprising 0 and 1, d represents a dimension representing the solution, prey i A prey matrix representing an ith iteration step;
and step 4-3, repeating the step 4-2 until the iteration termination condition is met, so as to obtain the optimal parameters of the convolutional neural network and the K-means algorithm.
6. The CNN-LSTM based direct current micro grid detection method according to claim 5, wherein the parameters include a number of convolutional layer channels of a convolutional neural network, a convolutional kernel size, a change interval of a pooling layer and a number of memory cells of the LSTM, and an offset value of a K-means algorithm.
7. The CNN-LSTM based direct current micro grid detection method according to claim 5, wherein the specific process of updating the hunting matrix of different phases based on iteration step length is as follows:
when the iteration step is smaller than 1/3 of the maximum iteration step, the game of Brownian is adopted to update the prey matrix;
when the iteration step length is greater than 1/3 of the maximum iteration step length and less than 2/3 of the maximum iteration step length, updating the front half population of the prey matrix by adopting Levy motion, and updating the rear half population of the prey matrix by adopting Brownian motion;
when the iteration step is greater than 2/3 of the maximum iteration step, then the Levy motion is used to update the prey matrix.
8. A direct current micro grid detection device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the detection model according to claim 1 when executing the computer program; the method comprises the following specific steps:
and inputting the voltage signal of the direct current micro power grid into a detection model to judge the instability fault type and the corresponding working condition of the direct current power grid.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310692457.2A CN116933114B (en) | 2023-06-12 | 2023-06-12 | CNN-LSTM-based direct-current micro-grid detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310692457.2A CN116933114B (en) | 2023-06-12 | 2023-06-12 | CNN-LSTM-based direct-current micro-grid detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116933114A true CN116933114A (en) | 2023-10-24 |
| CN116933114B CN116933114B (en) | 2024-06-21 |
Family
ID=88383439
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310692457.2A Active CN116933114B (en) | 2023-06-12 | 2023-06-12 | CNN-LSTM-based direct-current micro-grid detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116933114B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117767250A (en) * | 2023-12-22 | 2024-03-26 | 浙江稳山电气科技有限公司 | Direct-current micro-grid coordinated control method and system based on fault monitoring |
| CN117872038A (en) * | 2024-03-11 | 2024-04-12 | 浙江大学 | DC micro-grid instability fault source positioning method and device based on graph theory |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019214133A1 (en) * | 2018-05-08 | 2019-11-14 | 华南理工大学 | Method for automatically categorizing large-scale customer complaint data |
| CN112288034A (en) * | 2020-11-19 | 2021-01-29 | 江南大学 | Semi-supervised online anomaly detection method for wireless sensor network |
| US20210133569A1 (en) * | 2019-11-04 | 2021-05-06 | Tsinghua University | Methods, computing devices, and storage media for predicting traffic matrix |
| CN113222072A (en) * | 2021-06-10 | 2021-08-06 | 上海海事大学 | Lung X-ray image classification method based on K-means clustering and GAN |
| WO2022041394A1 (en) * | 2020-08-28 | 2022-03-03 | 南京邮电大学 | Method and apparatus for identifying network encrypted traffic |
| CN114295377A (en) * | 2021-12-13 | 2022-04-08 | 南京工业大学 | CNN-LSTM bearing fault diagnosis method based on genetic algorithm |
| WO2022147965A1 (en) * | 2021-01-09 | 2022-07-14 | 江苏拓邮信息智能技术研究院有限公司 | Arithmetic question marking system based on mixnet-yolov3 and convolutional recurrent neural network (crnn) |
| CN114897129A (en) * | 2022-03-24 | 2022-08-12 | 华北电力大学 | Photovoltaic power station short-term power prediction method based on similar daily clustering and Kmeans-GRA-LSTM |
| CN115034126A (en) * | 2022-04-18 | 2022-09-09 | 宁波杭州湾新材料研究院 | Method and system for optimizing LSTM neural network model through wolf algorithm |
| CN115293400A (en) * | 2022-06-23 | 2022-11-04 | 国网浙江省电力有限公司营销服务中心 | Power system load prediction method and system |
-
2023
- 2023-06-12 CN CN202310692457.2A patent/CN116933114B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019214133A1 (en) * | 2018-05-08 | 2019-11-14 | 华南理工大学 | Method for automatically categorizing large-scale customer complaint data |
| US20210133569A1 (en) * | 2019-11-04 | 2021-05-06 | Tsinghua University | Methods, computing devices, and storage media for predicting traffic matrix |
| WO2022041394A1 (en) * | 2020-08-28 | 2022-03-03 | 南京邮电大学 | Method and apparatus for identifying network encrypted traffic |
| CN112288034A (en) * | 2020-11-19 | 2021-01-29 | 江南大学 | Semi-supervised online anomaly detection method for wireless sensor network |
| WO2022147965A1 (en) * | 2021-01-09 | 2022-07-14 | 江苏拓邮信息智能技术研究院有限公司 | Arithmetic question marking system based on mixnet-yolov3 and convolutional recurrent neural network (crnn) |
| CN113222072A (en) * | 2021-06-10 | 2021-08-06 | 上海海事大学 | Lung X-ray image classification method based on K-means clustering and GAN |
| CN114295377A (en) * | 2021-12-13 | 2022-04-08 | 南京工业大学 | CNN-LSTM bearing fault diagnosis method based on genetic algorithm |
| CN114897129A (en) * | 2022-03-24 | 2022-08-12 | 华北电力大学 | Photovoltaic power station short-term power prediction method based on similar daily clustering and Kmeans-GRA-LSTM |
| CN115034126A (en) * | 2022-04-18 | 2022-09-09 | 宁波杭州湾新材料研究院 | Method and system for optimizing LSTM neural network model through wolf algorithm |
| CN115293400A (en) * | 2022-06-23 | 2022-11-04 | 国网浙江省电力有限公司营销服务中心 | Power system load prediction method and system |
Non-Patent Citations (8)
| Title |
|---|
| DONGJING WANG等: "Sequential Recommendation Based on Multivariate Hawkes Process Embedding With Attention", IEEE TRANSACTIONS ON CYBERNETICS, vol. 52, no. 11, 7 June 2021 (2021-06-07), pages 11893, XP011923978, DOI: 10.1109/TCYB.2021.3077361 * |
| YUE YUN等: "Interpretable Educational Recommendation: An Open Framework based on Bayesian Principal Component Analysis", 2022 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC), 18 November 2022 (2022-11-18), pages 3409 - 3414 * |
| 刘秋丽: "小功率IGBT的失效建模及失效预测", 万方数据, 12 January 2023 (2023-01-12), pages 1 - 91 * |
| 吕志星等: "基于K-Means和CNN的用户短期电力负荷预测", 计算机***应用, vol. 29, no. 03, 15 March 2020 (2020-03-15), pages 161 - 166 * |
| 池永为等: "基于LSTM-RNN的滚动轴承故障多标签分类方法", 振动.测试与诊断, vol. 40, no. 03, 15 June 2020 (2020-06-15), pages 563 - 571 * |
| 闾曾怡: "基于CNN-LSTM深度学习的燃烧不稳定性检测方法研究", 燃烧科学与技术, vol. 28, no. 2, 15 April 2022 (2022-04-15), pages 119 - 125 * |
| 陈泽慧: "基于最优加权组合模型的时间序列预测研究", 中国优秀硕士学位论文全文数据库 (基础科学辑), no. 1, 15 January 2023 (2023-01-15), pages 002 - 969 * |
| 陈臣鹏等: "基于并联CNN-LSTM的弱受端直流输电***故障诊断", 电机与控制应用, vol. 49, no. 06, 10 June 2022 (2022-06-10), pages 83 - 91 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117767250A (en) * | 2023-12-22 | 2024-03-26 | 浙江稳山电气科技有限公司 | Direct-current micro-grid coordinated control method and system based on fault monitoring |
| CN117767250B (en) * | 2023-12-22 | 2024-05-24 | 浙江稳山电气科技有限公司 | Direct-current micro-grid coordinated control method and system based on fault monitoring |
| CN117872038A (en) * | 2024-03-11 | 2024-04-12 | 浙江大学 | DC micro-grid instability fault source positioning method and device based on graph theory |
| CN117872038B (en) * | 2024-03-11 | 2024-05-17 | 浙江大学 | DC micro-grid instability fault source positioning method and device based on graph theory |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116933114B (en) | 2024-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116933114B (en) | CNN-LSTM-based direct-current micro-grid detection method and device | |
| CN111352971A (en) | Bank system monitoring data anomaly detection method and system | |
| CN116245033B (en) | Artificial intelligent driven power system analysis method and intelligent software platform | |
| CN111680820A (en) | Distributed photovoltaic power station fault diagnosis method and device | |
| KR20230171966A (en) | Image processing method and device and computer-readable storage medium | |
| CN113434685A (en) | Information classification processing method and system | |
| CN113901623A (en) | SVM power distribution network topology identification method based on cuckoo search algorithm | |
| Thomas et al. | Neural architecture search algorithm to optimize deep transformer model for fault detection in electrical power distribution systems | |
| CN116402352A (en) | Enterprise risk prediction method and device, electronic equipment and medium | |
| CN113673622B (en) | Laser point cloud data labeling method, device, equipment and product | |
| CN115965135A (en) | New energy prediction error modeling method and system based on naive Bayes classification | |
| CN110795934A (en) | Sentence analysis model training method and device and sentence analysis method and device | |
| CN117093260B (en) | Fusion model website structure analysis method based on decision tree classification algorithm | |
| CN112579777B (en) | Semi-supervised classification method for unlabeled text | |
| CN114266676A (en) | Abnormal electricity utilization detection method of genetic optimization Bagging heterogeneous integration model | |
| Xu et al. | An improved ELM-WOA–based fault diagnosis for electric power | |
| CN113780402A (en) | User electricity stealing detection method based on improved generation countermeasure network | |
| CN117669656A (en) | TCN-Semi PN-based direct-current micro-grid stability real-time monitoring method and device | |
| CN114974398A (en) | Information processing method and device and computer readable storage medium | |
| CN116720095A (en) | Electrical characteristic signal clustering method for optimizing fuzzy C-means based on genetic algorithm | |
| CN116566061A (en) | Grid-connected inverter system stability on-line monitoring method and system | |
| CN115840920A (en) | Photovoltaic group string single-day separable anomaly classification method, device, equipment and storage medium | |
| CN116415147A (en) | Pre-training method for photovoltaic system equipment operation data based on self-learning | |
| CN116204771A (en) | Power system transient stability key feature selection method, device and product | |
| CN113128130B (en) | Real-time monitoring method and device for judging stability of direct-current power distribution system |
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 |