CN114545510B - Underwater magnetic target identification method based on social culture civilization evolution strategy - Google Patents
Underwater magnetic target identification method based on social culture civilization evolution strategy Download PDFInfo
- Publication number
- CN114545510B CN114545510B CN202011334224.8A CN202011334224A CN114545510B CN 114545510 B CN114545510 B CN 114545510B CN 202011334224 A CN202011334224 A CN 202011334224A CN 114545510 B CN114545510 B CN 114545510B
- Authority
- CN
- China
- Prior art keywords
- space
- neural network
- individual
- social
- underwater
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000013528 artificial neural network Methods 0.000 claims abstract description 38
- 238000012360 testing method Methods 0.000 claims abstract description 6
- 238000004364 calculation method Methods 0.000 claims abstract description 4
- 238000012545 processing Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 16
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 2
- 230000002457 bidirectional effect Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 230000002349 favourable effect Effects 0.000 claims 1
- 238000005065 mining Methods 0.000 claims 1
- 238000005070 sampling Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 10
- 238000013473 artificial intelligence Methods 0.000 abstract description 5
- 238000013135 deep learning Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 3
- 239000011159 matrix material Substances 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000005415 magnetization Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/08—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices
- G01V3/081—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices the magnetic field is produced by the objects or geological structures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/38—Processing data, e.g. for analysis, for interpretation, for correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Geology (AREA)
- Geophysics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Environmental & Geological Engineering (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Electromagnetism (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Feedback Control In General (AREA)
Abstract
The invention belongs to the field of artificial intelligence and underwater signal processing, and relates to an intelligent recognition method of an underwater target based on artificial intelligence. According to the technical scheme, the underwater magnetic target identification method based on the social culture civilization evolution strategy is designed, the LSTM neural network is organically combined with the social culture civilization evolution strategy as a classifier, key parameters in the classifier are globally optimized, the advantages of a deep learning technology are fully exerted, the internal characteristics and the dependence of data are mined, a target identification system can obtain higher accuracy, the identification accuracy and the operation amount are optimized, the target identification system can adaptively adjust the parameters in different underwater environments to achieve the optimal identification effect, the system calculation load is reduced while the identification accuracy is improved, and the underwater magnetic target can be quickly and accurately identified in a classified mode under the condition of less first-test knowledge.
Description
Technical Field
The invention relates to an intelligent identification method of an underwater target based on artificial intelligence, and belongs to the field of artificial intelligence and underwater signal processing.
Background
The target identification is a key technology widely applied to various underwater systems, and has important significance for situation awareness, comprehensive navigation, intrusion detection and strategy development of the underwater systems. In practical engineering application, the transmission channel of the target underwater signal is complex and is more interfered, and the traditional target identification method is insufficient in utilization information and is difficult to obtain a reliable identification result, so that the method for classifying and identifying the target by utilizing the magnetic field is more and more important.
The magnetic field of the underwater target comprises a fixed magnetic field and an induced magnetic field, such as a fixed magnetic field generated by magnetization of a target shell; eddy current magnetic fields generated by target motion or sway; the equipment on the object works as a corrosion-resistant system or a magnetic field generated by an electric propulsion system. The frequency of the underwater magnetic field of the target is generally lower, the traditional magnetic target identification method needs more priori information and performs more complex artificial data analysis, and along with the development of artificial intelligence technology, the design of the intelligent underwater magnetic target identification method has important significance. The cyclic neural network in the deep learning technology can analyze the time sequence, wherein the long-term dependence between the time steps of data can be better analyzed and learned by the long-time memory (LSTM) neural network, but aiming at the data of different scenes, the LSTM neural network is sensitive to parameter setting, the difficulty of manually selecting the parameters of the LSTM neural network is high, and unreasonable parameter setting can cause over fitting or slow convergence speed, and even greatly reduce the classification accuracy.
The intelligent optimization technology evolution deep learning is an emerging technical means, so that the recognition system can adaptively adjust parameters according to actual conditions, and intelligently trade-off between recognition accuracy and calculation cost, so that the system obtains better instantaneity, recognition accuracy and anti-interference capability, the application range of the system is expanded, and the engineering application problem is solved.
Through the search discovery of the prior art document, she Pingxian et al published "mine magnetic fuze target identification technology initial detection" in the naval engineering college journal (1993 (05): 19-26) uses a binary tree classifier to identify the tonnage of the ship, but the required training data is more and the accuracy is not high; xu Jie et al in "technology of ship electricity" (2011, 31 (09): 51-54), "a neural network identification model based on a ship magnetic field" published on proposes a model for applying a neural network data fusion technology to ship target identification, and the method needs to use an inefficient trial and error method to determine the optimal number of hidden nodes of the neural network, and is difficult to obtain a globally optimal network structure and parameters.
The retrieval results of the existing documents show that in the existing underwater magnetic target identification method, the mode of extracting and utilizing information in the underwater magnetic signals is deficient, and the mode of selecting parameters is complex, so that the underwater magnetic target identification method with higher intelligence, higher precision and wider application range is required to be designed.
Disclosure of Invention
Aiming at the defects and shortcomings of the existing method, the invention designs an intelligent underwater magnetic target identification method, and the key parameters in the deep learning classifier are globally evolved through a social culture civilization evolution strategy, so that a target identification system can adaptively adjust the parameters in different underwater environments to achieve the optimal identification effect, and the calculation load of the system is reduced while the identification precision is improved.
The invention has the following effects and benefits: compared with the prior art, the invention designs an intelligent underwater magnetic target identification method based on a social culture civilization evolution strategy aiming at the problems that the existing underwater magnetic target identification method is low in identification precision, is greatly influenced by parameter selection and is difficult to manually select parameters, takes an LSTM neural network as a classifier to be organically combined with the social culture civilization evolution strategy, carries out global optimization on key parameters in the classifier, fully plays the advantages of a deep learning technology, and mines the internal characteristics and the dependence of data, so that a target identification system can obtain higher accuracy, and optimizes the identification accuracy and the operation amount to adapt to a complex actual engineering environment. The designed method has wide application range and universality, and can effectively classify the size, tonnage, navigational speed and the like of the underwater magnetic target according to actual requirements.
Drawings
FIG. 1 is a schematic diagram of an underwater magnetic target recognition method based on a social culture civilization evolution strategy.
Fig. 2 is a graph of large target magnetic field passing characteristics.
In fig. 3, the target magnetic field passes through the characteristic curve.
Fig. 4 small target magnetic field passing characteristic curve.
Fig. 5 uses the raw data for classification recognition to obtain a fuzzy matrix.
FIG. 6 uses raw data and gradient features for classification recognition to obtain a fuzzy matrix.
Detailed Description
The invention is described in further detail below with reference to the drawings and the detailed description.
Step one, receiving magnetic field signals and carrying out data preprocessing.
And arranging a magnetic sensor under water, simultaneously measuring magnetic field components of an underwater target in three directions by adopting a three-component magnetometer, arranging D magnetometers with different orientations according to a straight line, and recording a space passing characteristic signal obtained after the target passes through a measuring area. The space passing characteristic signal is sampled N times at equal intervals to obtain D discrete space passing characteristic signals x d＝[xd(1),xd(2),…,xd (N) ], d=1, 2, …, D with a length of N.
Feature extraction of the spatial pass characteristic signal is advantageous for improving classification performance, thus calculating a gradient sequence Δ d＝[Δd(1),Δd(2),…,Δd (N) of the discrete spatial pass characteristic signal, wherein
In order to make the classifier converge more quickly, the normalization processing with zero as the center is carried out on the original signals and gradient characteristic sequences with different mean and variance. Firstly, calculating the average value and standard deviation of all sequences, subtracting the average value of the sequences from each value in the sequences, and dividing the average value by the standard deviation of the sequences to obtain adjusted data.
Initializing individuals and forming a civilized space.
The group scale of the civilization space is set as P, the maximum iteration number is G, the iteration number label is G, and G is [1, G ]. In the g-th iteration, the civilized space is represented asConsists of P M-dimensional individuals, the g-th individual being specifically expressed asP=1, 2, …, P, each dimension of each individual has an initial value of a uniform random number within [0,1 ].
And step three, defining the architecture of the LSTM neural network.
The LSTM neural network can learn the dependency and the correlation between time steps of time series data, and the LSTM neural network is used as a classifier in an identification system. The flow of data through the neural network is as follows: firstly inputting data into a bidirectional LSTM layer with n 1 neurons, wherein an activation function uses a hyperbolic tangent function, and outputting the last element of a sequence; and finally, through a full-connection layer with the size of n 2 and a softmax layer and a classification layer, the neural network can automatically mine and learn the internal dependency relationship of the data.
And step four, designing a social culture civilization evolution strategy evolution LSTM neural network classifier and defining a fitness function.
In the invention, a socio-cultural relic evolution strategy is used for embedded optimization of key parameters in an LSTM neural network, including the number of neurons in a hidden layer of the neural network, the number of times of training the neural network and the batch size of each training. Thereby defining the fitness function value of each individual in the socio-cultural civilization evolution strategy asWherein/>Corresponding to the number of neurons in the hidden layer,/>Corresponding to training times/>Corresponding batch size,/>Representing the accuracy of the neural network to classify the test set after training the LSTM neural network using the parameters represented by the current individual, c 1、c2 and c 3 are constants. In the off-line stage, the social culture civilization evolution strategy can adaptively adjust and optimize the parameters so as to achieve the aim of obtaining higher recognition rate by using less operation amount.
And fifthly, dividing the social space according to the civilized space.
Randomly classifying all individuals in the current civilization space into K clustersIs defined as K social spaces, and the social spaces meet the conditionsAnd S g(i)∩Sg (j) =Φ, i=1, 2, …, K, j=1, 2, …, K, i+.j. The individuals in the kth social space are denoted/>
Step six, selecting a leader and establishing a belief space.
Firstly, determining an individual with the largest fitness function value in the current civilized space as a leaderThen, independent belief spaces are respectively built in each social space, and the belief space of the kth society is expressed asWherein/>Is the m-th dimension canonical knowledge interval in the kth society, wherein/>The initial value is 0 for the lower bound of the variable; /(I)The variable upper bound is the initial value of 1; /(I)Is thatCorresponding fitness function values; /(I)For/>Corresponding fitness function value,/>And/>The initial value of (2) is minus infinity. The leader and each belief space are used to guide the direction of evolution of the individual.
And step seven, all individuals in the civilized space evolve.
All individuals in each society evolve according to probability according to belief space or evolve to a leader, so that each dimension of the society is updated. The probabilities of implementing the two evolution modes are p 1 and 1-p 1, respectively. For each individual, probability judgment is firstly carried out before updating, and if the probability judgment result determines that the individual is updated according to belief space, the specific evolution mode of the mth dimension of the qth individual in the kth social space is as follows Wherein N (0, 1) is a random number with mean value of 0 and variance of 1 subject to normal distribution, and ρ and η are constants; if the probability decision result determines that the individual evolves towards the leader, firstly calculating the distance between the individual and the leader asSpecifying a specific evolution of the (th) individual(s) dimension(s) in the (th) social space as/>
And step eight, updating belief space of the social space according to the evolved individuals.
The beta individuals with the maximum fitness function values are selected to be used for updating the upper bound or the lower bound of the standard knowledge in each social space and the corresponding fitness function values according to equal probability, and the specific mode is as follows: if the condition is satisfiedOr (b)Updating the lower bound of the canonical knowledge and its fitness function value asIf the condition is satisfiedOr/>Updating the standard knowledge upper bound and fitness function value of each society asRepeating the above operation for each social space to obtain updated belief space as/>
And step nine, generating a new generation of civilization space.
And merging all the individuals in the social space after evolution, and replacing the individual with the smallest fitness function value in the current civilization space by a leader to obtain a new generation civilization space C g+1.
Step ten, judging whether the maximum iteration times are reached, if the current iteration number G is smaller than G, enabling G to be equal to g+1, and returning to the step five; otherwise, outputting the leader in the last generation and completing training of the neural network according to the parameters in the leader.
And step eleven, in the actual measurement stage, preprocessing actual measurement data in the same way as in the step one, and identifying the underwater magnetic target by using an optimal neural network obtained after training and optimization.
Step twelve, simulation experiment verification
In a simulation experiment, the performance of the method designed by the invention is tested by taking the classification and identification of three kinds of underwater magnetic targets, namely large, medium and small. Simulating magnetic field passing characteristic curves of three kinds of underwater targets, namely a large, a medium and a small, by using the magnetic ellipsoids, and determining the length of the major axis and the short axis of the magnetic ellipsoids according to the length and the width of the targets, wherein the classification standards of the three kinds of targets refer to industry standards used by China ship group Limited company; specific modeling methods refer to Xu Jie et al in "Marine electric technology" (2011, 31 (09): 51-54), "a neural network identification model based on a ship magnetic field". After the total 432 groups of passing characteristic simulation values of each type of targets at different positions and different angles are obtained, 70% of data are randomly selected from the passing characteristic simulation values to serve as a training set, the rest 30% of data are served as a test set, and the passing characteristic curves of the three types of targets are respectively shown in figures 2, 3 and 4. Parameters obtained by optimizing the social culture civilization evolution strategy are as follows: the number of neurons in the hidden layer is n 1 =45, the training times are 25, and the batch size is 300; other parameters were set as follows: n 2 =3, gradient threshold of 1, initial learning rate of 0.01, p=30, g=20, k=3, m=3, c 1＝0.1,c2＝0.3,c3 = -0.1, p=0.5, η=0.06, β=2. The fuzzy matrix obtained by training and testing the neural network by using the original data is shown in fig. 5, the recognition accuracy is 89.9%, and the fuzzy matrix obtained by training and testing the neural network by using the original data and the gradient characteristics is shown in fig. 6, and the recognition accuracy is 93%. The experimental result shows that the method can quickly and accurately classify and identify the underwater magnetic targets under the condition of less prior knowledge.
Claims (1)
1. An underwater magnetic target identification method based on a social culture civilization evolution strategy comprises the following steps:
Firstly, receiving magnetic field signals and carrying out data preprocessing, arranging magnetic sensors under water, adopting three-component magnetometers to measure magnetic field components of an underwater target in three directions simultaneously, arranging D magnetometers with different orientations according to a straight line, recording space passing characteristic signals obtained after the target passes through a measuring area, carrying out N times of equidistant sampling on the space passing characteristic signals to obtain D discrete space passing characteristic signals x d＝[xd(1),xd(2),…,xd (N) ], wherein d=1, 2, …, D, carrying out feature extraction on the space passing characteristic signals is favorable for improving classification performance, and therefore calculating gradient sequences delta d＝[Δd(1),Δd(2),…,Δd (N) ]ofthe discrete space passing characteristic signals, wherein the D discrete space passing characteristic signals are obtained In order to enable the classifier to converge more quickly, carrying out normalization processing with zero as a center on original signals and gradient feature sequences with different mean values and variances, firstly calculating the mean values and standard deviations of all sequences, subtracting the mean values of the sequences from each value in the sequences, and dividing the mean values by the standard deviations of the sequences to obtain adjusted data;
Initializing individuals and forming a civilized space, setting the population scale of the civilized space as P, setting the maximum iteration number as G, marking the iteration number as G, and expressing the civilized space as in the G-th iteration Consists of P M-dimensional individuals, the g-th individual being specifically denoted/>P=1, 2, …, P, each dimension of each individual has an initial value of a uniform random number within [0,1 ];
Defining an LSTM neural network architecture, wherein the LSTM neural network can learn the dependence and the correlation between time steps of time sequence data, and the LSTM neural network is used as a classifier in an identification system, and the flow of data passing through the neural network is as follows: firstly inputting data into a bidirectional, LSTM, layer with n 1 neurons, wherein an activation function uses a hyperbolic tangent function, and outputting the last element of the sequence; through a full-connection layer with the size of n 2, and finally through a softmax layer and a classification layer, the automatic mining and learning of the data internal dependency relationship by the neural network are realized;
Designing a socio-cultural relic evolution strategy evolution LSTM neural network classifier and defining a fitness function, wherein the socio-cultural relic evolution strategy is used for embedded optimization of key parameters in the LSTM neural network, including the number of neurons of a hidden layer of the neural network, the training times of the neural network and the batch size of each training, so that the fitness function value of each individual in the socio-cultural relic evolution strategy is defined as Wherein/>Corresponding to the number of neurons in the hidden layer,/>Corresponding to training times/>Corresponding batch size,/>The accuracy of the neural network to the classification of the test set after the LSTM neural network is trained by using the parameters represented by the current individuals is represented, c 1、c2 and c 3 are constants, and the parameters can be adaptively adjusted and optimized by the socio-cultural civilization evolution strategy in an off-line stage so as to achieve the purpose of obtaining higher recognition rate by using less calculation amount;
dividing the social space according to the civilization space, and randomly classifying all individuals in the current civilization space into K clusters K=1, 2, …, K, defined as K social spaces, which satisfy the conditionsK=1, 2, …, K, and S g(i)∩Sg (j) =Φ, i=1, 2, …, K, j=1, 2, …, K, i+.j, then individuals in the kth social space are denoted/>q＝1,2,…,P/K，k＝1,2,…,K；
Step six, selecting a leader and establishing a belief space, and firstly determining an individual with the largest fitness function value in the current civilization space as the leaderThen, independent belief spaces are respectively built in each social space, and the belief space of the kth society is expressed as/>K=1, 2, …, K, m=1, 2, …, M, whereinIs the m-th dimension canonical knowledge interval in the kth society, wherein/>The initial value is 0 for the lower bound of the variable; /(I)The variable upper bound is the initial value of 1; /(I)For/>Corresponding fitness function values; /(I)For/>Corresponding fitness function value,/>And/>The initial value of (1) is minus infinity, and the leader and each belief space are used to guide the evolution direction of the individual;
Step seven, all individuals in the civilized space evolve, all individuals in each society evolve according to probability according to belief space or evolve to a leader, update of each dimension of the society is realized, the probability of implementing two evolution modes is p 1 and 1-p 1 respectively, probability judgment is firstly carried out before update for each individual, and if the probability judgment result determines that the individual is updated according to belief space, the specific evolution mode of the m dimension of the q-th individual in the k-th social space is that K=1, 2, …, K, q=1, 2, …, P/K, m=1, 2, …, M, where N (0, 1) is a random number with mean 0 and variance 1 subject to normal distribution, ρ, η are constants; if the probability decision result determines that the individual evolves towards the leader, firstly calculating the distance between the individual and the leader asK=1, 2, …, K, q=1, 2, …, P/K, m=1, 2, …, M, prescribing the specific evolution of the mth dimension of the qth individual in kth social space as/>k＝1,2,…,K，q＝1,2,…,P/K，m＝1,2,…,M；
Step eight, according to the belief space of the updated social space of the evolved individuals, selecting the beta individuals with the largest fitness function value to be used for updating the upper bound or the lower bound of the standard knowledge in each social space and the corresponding fitness function value according to equal probability, wherein the specific mode is as follows: if the condition is satisfiedOr/>Updating the lower bound of canonical knowledge and its fitness function value as/> K=1, 2, …, K, q e [1,2, …, β ], m=1, 2, …, M; if the condition/>Or/>Updating the standard knowledge upper bound and fitness function value of each society as/>K=1, 2, …, K, q e [1,2, …, β ], m=1, 2, …, M, repeating the above operations for each social space to obtain updated belief space ask＝1,2,…,K，m＝1,2,…,M；
Step nine, generating a new generation civilized space, combining all individuals in the social space after evolution, and replacing the individual with the smallest fitness function value in the current civilized space by a leader to obtain a new generation civilized space C g+1;
Step ten, judging whether the maximum iteration times are reached, if the current iteration number G is smaller than G, enabling G to be equal to g+1, and returning to the step five; otherwise, outputting the leader in the last generation, and completing training of the neural network according to the parameters in the leader;
Step eleven, in the actual measurement stage, preprocessing actual measurement data in the same way as in the step one, and identifying the underwater magnetic target by using an optimal neural network obtained after training and optimization;
Step twelve, simulation experiments prove that in the simulation experiments, the performance of the designed underwater magnetic target identification method is tested, and the underwater magnetic target can be quickly and accurately identified in a classified mode under the condition of less prior experimental knowledge.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011334224.8A CN114545510B (en) | 2020-11-25 | 2020-11-25 | Underwater magnetic target identification method based on social culture civilization evolution strategy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011334224.8A CN114545510B (en) | 2020-11-25 | 2020-11-25 | Underwater magnetic target identification method based on social culture civilization evolution strategy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114545510A CN114545510A (en) | 2022-05-27 |
| CN114545510B true CN114545510B (en) | 2024-06-11 |
Family
ID=81659525
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011334224.8A Active CN114545510B (en) | 2020-11-25 | 2020-11-25 | Underwater magnetic target identification method based on social culture civilization evolution strategy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114545510B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109188536A (en) * | 2018-09-20 | 2019-01-11 | 成都理工大学 | Time-frequency electromagnetism and magnetotelluric joint inversion method based on deep learning |
| CN109579827A (en) * | 2018-12-24 | 2019-04-05 | 中国船舶重工集团公司第七〇九研究所 | A kind of magnetic target detection and localization method based on arcuate array |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20190017454A (en) * | 2017-08-11 | 2019-02-20 | 고려대학교 산학협력단 | Device and method for generating location estimation model |
-
2020
- 2020-11-25 CN CN202011334224.8A patent/CN114545510B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109188536A (en) * | 2018-09-20 | 2019-01-11 | 成都理工大学 | Time-frequency electromagnetism and magnetotelluric joint inversion method based on deep learning |
| CN109579827A (en) * | 2018-12-24 | 2019-04-05 | 中国船舶重工集团公司第七〇九研究所 | A kind of magnetic target detection and localization method based on arcuate array |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114545510A (en) | 2022-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070060B (en) | Fault diagnosis method for bearing equipment | |
| CN110232319B (en) | Ship behavior identification method based on deep learning | |
| CN109581339B (en) | Sonar identification method based on automatic adjustment self-coding network of brainstorming storm | |
| CN111814871A (en) | Image classification method based on reliable weight optimal transmission | |
| CN110751108B (en) | Subway distributed vibration signal similarity determination method | |
| CN112989708A (en) | Well logging lithology identification method and system based on LSTM neural network | |
| CN111368920A (en) | Quantum twin neural network-based binary classification method and face recognition method thereof | |
| CN109002848B (en) | Weak and small target detection method based on feature mapping neural network | |
| CN109239669B (en) | Self-evolution radar target detection algorithm based on deep learning | |
| CN114676742A (en) | Power grid abnormal electricity utilization detection method based on attention mechanism and residual error network | |
| CN108549866B (en) | Remote sensing airplane identification method based on dense convolutional neural network | |
| CN114019370B (en) | Motor fault detection method based on gray level image and lightweight CNN-SVM model | |
| CN112633328A (en) | Dense oil reservoir transformation effect evaluation method based on deep learning | |
| CN113269274B (en) | Zero sample identification method and system based on cycle consistency | |
| CN112686372A (en) | Product performance prediction method based on depth residual GRU neural network | |
| CN116805051A (en) | Double convolution dynamic domain adaptive equipment fault diagnosis method based on attention mechanism | |
| CN115705393A (en) | Radar radiation source grading identification method based on continuous learning | |
| CN112881987A (en) | Airborne phased array radar behavior prediction method based on LSTM model | |
| CN116383747A (en) | Anomaly detection method for generating countermeasure network based on multi-time scale depth convolution | |
| CN112149502A (en) | Unfavorable geology positioning forecasting method based on convolutional neural network | |
| CN117171702A (en) | Multi-mode power grid fault detection method and system based on deep learning | |
| CN114545510B (en) | Underwater magnetic target identification method based on social culture civilization evolution strategy | |
| CN117312885A (en) | Dynamic threshold division and multi-factor early warning algorithm model for gas emission of working face | |
| CN112819087A (en) | Effluent BOD sensor abnormity detection method based on modular neural network | |
| CN117093922A (en) | Improved SVM-based complex fluid identification method for unbalanced sample oil reservoir |
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 |