CN107153837A - Depth combination K means and PSO clustering method - Google Patents
Depth combination K means and PSO clustering method Download PDFInfo
- Publication number
- CN107153837A CN107153837A CN201710247343.1A CN201710247343A CN107153837A CN 107153837 A CN107153837 A CN 107153837A CN 201710247343 A CN201710247343 A CN 201710247343A CN 107153837 A CN107153837 A CN 107153837A
- Authority
- CN
- China
- Prior art keywords
- msub
- mrow
- particle
- cluster
- mtd
- 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
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Abstract
The invention discloses a kind of depth combination K means and PSO clustering method, including：The particle in particle cluster algorithm is initialized using K means algorithms；Fitness value is calculated according to fitness function, the study part of particle cluster algorithm iteration and equal value part is updated；PSO algorithms on multiple populations are built using cluster, with one cluster of cluster center representative, particle rapidity and position are updated；Judge whether to meet iteration stopping condition, if meeting, export optimal particle；Otherwise iteration is continued.Initialized using K means, PSO algorithms on multiple populations are built using the cluster in cluster process, amplitude limit is carried out to particle rapidity using the PSO on multiple populations of lightweight, and using suitable manner, with more preferable Clustering Effect and convergence rate.
Description
Technical field
The present invention relates to a kind of clustering method, more particularly to a kind of depth combination K-means and PSO clustering method,
With more preferable Clustering Effect and convergence rate.
Background technology
Be exactly to gather numerous objects for cluster one by one for clustering is directly perceived, wherein have identical cluster feature same
One cluster, the obvious object of feature difference occupy different clusters.Clustering is applied to many different types of data acquisition systems.Mould
Numerous application fields such as formula identification, data analysis, image procossing and market survey propose active demand to clustering algorithm.Cluster
The main target of analysis is to obtain a more excellent cluster result of the overall situation.It is typical towards the poly- of prototype as unsupervised learning one
Class algorithm, K-means algorithms are although simple and practical, but unavoidably there is defect caused by some initial methods etc..It is many
For example K-meas++, two points of K-means etc. are suggested innovatory algorithm in succession.Preconditioning technique is also wherein frequently to use
Improve a kind of method of global convergence.
As a kind of typical global optimization approach based on swarm intelligence, particle cluster algorithm (PSO) originate to bird
The behavioral study that group looks for food.It is a kind of global random searching algorithm with stronger global convergence and robustness, and
Data mining, machine learning, Chemical Engineering, economic load dispatching distribution and Non-Linear Programming etc. have been widely applied to it numerous excellent
Change field.
Omran etc. proposed a kind of unsupervised image clustering algorithm based on PSO in 2002, it be it is earliest based on
PSO clustering algorithm.Merwe etc. then proposes a kind of elementary tactics of two kinds of algorithms of knot, and combines mean cluster and grain
Swarm optimization carries out clustering, achieves preferable Clustering Effect.But there is following defect：
1st, population diversity can be lost PSO algorithms in an iterative process；
2nd, structure on multiple populations necessarily causes the time complexity and slower convergence rate of complexity；
3rd, actual scene is not often considered excessively, and practical application effect is poor.
The content of the invention
For above-mentioned technical problem, the present invention seeks to：There is provided a kind of depth combination K-means and PSO
Clustering method, is initialized using K-means, PSO algorithms on multiple populations is built using the cluster in cluster process, using lightweight
PSO on multiple populations, and using suitable manner to particle rapidity carry out amplitude limit, with more preferable Clustering Effect and convergence rate.
The technical scheme is that：
A kind of depth combination K-means and PSO clustering method, comprises the following steps：
S01：The particle in particle cluster algorithm is initialized using K-means algorithms；
S02：Fitness value is calculated according to fitness function, the study part and average portion of particle cluster algorithm iteration is updated
Point；
S03：PSO algorithms on multiple populations are built using cluster, with one cluster of cluster center representative, particle rapidity and position are updated；
S04：Judge whether to meet iteration stopping condition, if meeting, export optimal particle；Otherwise return to step S02, continues
Iteration.
It is preferred that, fitness function is in the step S02：
mj,d(zp,mj) respectively represent cluster center and Euclidean distance, Nc,CjThe total number of cluster is represented respectively and is belonged to each
The number of objects of individual cluster.
It is preferred that, the core iterative equation of particle cluster algorithm is in the step S02：
Vi(t+1)=Vi(t)+c1r1(pbesti(t)-Xi(t))+c2r2(gbest(t)-Xi(t))
Xi(t+1)=X (t)+V (t+1)
Wherein, c1,c2It is Studying factors, r1,r2It is the random number between [0,1], ViAnd X (t)i(t) particle i the is represented respectively
The speed in t generations and position, pbestiAnd gbest (t)i(t) represent that the history optimal location in particle i t generations and population are optimal respectively
Position.
It is preferred that, particle rapidity core iterative equation is changed into following part in the step S03：
Wherein, cbesti(t) the optimal part of history of each subgroup is represented, each iteration is determined more by the fitness of average
It is whether new, Vi old(t) the more new formula of classical particle speed described above is represented, the present invention adds one on its basis
cbesti(t), c3,r3It is the random number between Studying factors and [0,1] respectively.
It is preferred that, amplitude limit is carried out to particle rapidity and position, clipping mode is：Calculate training sample locational space for [-
xmax,xmax], if particle position exceedes this scope in iterative process, it is set to interval endpoint；Particle maximal rate is set to
Vmax=vk × Xmax, vk representation speed limiting figures, then the speed that particle updates is determined by below equation：
It is preferred that, the stop condition is given iterations or global optimum not in change.
Then the present invention is respectively adopted this first by the validity of experimentation preliminary identification PSO and K-means algorithm
Algorithm and other three kinds of associated algorithms in invention is to having label data and carrying out supervised learning and nothing without label data
Supervised learning, and it is good and bad by the index evaluation such as entropy, SSE algorithm.
Compared with prior art, it is an advantage of the invention that：
1st, cluster center is initialized using two points of K-means methods, convergence rate can be greatly improved.
2nd, during in conjunction with PSO algorithm optimizations K-means, a variety of group characters for making full use of K-means to show,
PSO algorithms on multiple populations are built, the forfeiture of population diversity in optimization process can be substantially reduced, the global receipts of particle cluster algorithm are improved
Holding back property.
3rd, on details is calculated, first in iterative process on multiple populations is built, effectively counted using lightweight but row
Calculation method；Next to that carrying out amplitude limit to adapt to actual conditions to flying speed of partcles and position.
Brief description of the drawings
Below in conjunction with the accompanying drawings and embodiment the invention will be further described：
Fig. 1 is the flow chart of clustering method of the present invention；
Fig. 2 is the flow chart of K-means methods；
Fig. 3 is that PSO optimizes case diagram；
Fig. 4 is PSO convergence result；
Fig. 5 is KMEANS, PSO, BPSOKM, DYPSOKM (present invention) cluster result convergence rate design sketch；
Fig. 6 is 4 kinds of typical cluster result design sketch.
Embodiment
Such scheme is described further below in conjunction with specific embodiment.It should be understood that these embodiments are to be used to illustrate
The present invention and be not limited to limit the scope of the present invention.The implementation condition used in embodiment can be done according to the condition of specific producer
Further adjustment, unreceipted implementation condition is usually the condition in normal experiment.
Embodiment：
As shown in figure 1, the present invention a kind of depth combination K-means and PSO clustering method be broadly divided into data processing,
Particle initialization, algorithm iteration, algorithm evaluation four are most of.
First, it is data processing section first, chooses data set, assessment data set, which is divided into, has label data collection and without number of tags
According to collection.The key data collection feature such as following table of selection：
2nd, a particle in particle cluster algorithm is initialized using K-means.The main flow of K-means algorithms
Figure is as shown in Figure 2.The defect of local optimum is absorbed in for displaying K-means algorithms, K=4 is set, K-means algorithms are run.
The distance at object and cluster center is dist (c in K-means algorithmsi, x)=| | ci, x | |, cluster center calculation formula beCriterion SSE：
In above formula, | | ci, x | | Euler's distance at object and cluster center is represented, m represents cluster ciNumber of objects, SSE represents K
Individual cluster kind error sum of squares.
Fitness value is calculated according to fitness function, the study part of particle cluster algorithm iteration and equal value part is updated；
Fitness function is as follows：
mj,d(zp,mj) respectively represent cluster center and Euclidean distance；Nc,CjThe total number of cluster is represented respectively and is belonged to each
The number of objects of individual cluster.Fitness function substantially has no difference with SSE, is equally the function for the degree of polymerization for characterizing cluster.
The core iterative equation of particle cluster algorithm is：
Vi(t+1)=Vi(t)+c1r1(pbesti(t)-Xi(t))+c2r2(gbest(t)-Xi(t))
Xi(t+1)=X (t)+V (t+1)
Wherein, c1,c2It is Studying factors, r1,r2It is the random number between [0,1], ViAnd X (t)i(t) particle i the is represented respectively
The speed in t generations and position, pbestiAnd gbest (t)i(t) represent that the history optimal location in particle i t generations and population are optimal respectively
Position.
Particle renewal speed is determined by 3 parts.Part I is particle previous generation's optimal location, is capable of the overall situation of balanced algorithm
Search and local development ability；Part II is the optimal part of particle history, and Part III is global optimum part, is showed respectively
Particle tends to history learning and social learning.Referred to as know part, history portion, social part.This exactly population
The embodiment of algorithm swarm intelligence.
3rd, PSO algorithms on multiple populations are built using cluster, with one cluster of cluster center representative, updates particle rapidity and position；Utilize
The a variety of characters shown in mean cluster, build particle cluster algorithm on multiple populations and particle are iterated, to improve cluster effect
Really.Each cluster represents a population, and the center of each cluster can represent a cluster to crude density, with the center generation of a cluster
One cluster of table can play fairly good effect in many cases, can play computation complexity and convergence rate and cluster is imitated
The overall balance of fruit.
The core iterative equation of particle rapidity is changed into following part：
Wherein, cbesti(t) the optimal part of history of each subgroup is represented, each iteration is determined more by the fitness of average
It is whether new, Vi old(t) the more new formula of classical particle speed described above is represented, the present invention adds one on its basis
cbesti(t), c3,r3It is the random number between Studying factors and [0,1] respectively.
4th, judge whether to meet iteration stopping condition, if meeting, export optimal particle；Otherwise return to step S02, continues
Iteration.Stop condition is given iterations or global optimum not in change.
5th, algorithm evaluation
Assessed and two class methods of unsupervised assessment using there is supervision.Unsupervised assessment will be using SSE, SSB, TSS, entropy, pure
The methods such as degree；There is supervision assessment to count four kinds of clustering effect patterns.
SSB, TSS expression group quadratic sum and total sum of squares respectively, definition are as follows respectively：
TSS=SSB+SSE
The entropy of the whole result that clusters is determined by following 3 formula：
pij=mij/mi
Wherein, miAnd mijExpression belongs to the number of objects that cluster j is actually belonged in cluster i number of objects and cluster i respectively；Each
The entropy of cluster is by eiDetermine, e represents the entropy of whole sub-clustering.
The purity clustered is determined by following two formula：
Because inventive algorithm main body will use PSO algorithm optimizations, in order to show that PSO algorithms have more preferable global convergence
Property, especially choosing one has 4 locally optimal solutions, the function of a globally optimal solution to be tested, functional schema such as Fig. 3, Fig. 4
Show PSO convergence result, it can be seen that in the case where possessing numerous locally optimal solutions, PSO algorithms are still converged to entirely
Office's optimal solution.
Suitable data set will be respectively adopted to described four kinds of methods KMEANS, PSO, BPSOKM, DYPSOKM (present invention
Method) clustered, finally compare their convergence rate and Clustering Effect.Its Clustering Effect is respectively adopted supervision and nothing
The method of supervision is estimated.Before this, it is necessary to the parameter involved by algorithm is configured, be see the table below：
Wherein N represents population scale, and K represents cluster number, vk representation speed limiting figures.Clipping mode is：Calculate instruction
The locational space for practicing sample is [- xmax,xmax], if particle position exceedes this scope in iterative process, it is set to interval end
Point；Particle maximal rate is set to Vmax=vk × Xmax, then the speed that particle updates is determined by below equation：
Fig. 5 provides 4 kinds of convergence of algorithm speed curve diagrams, it can be seen that the more other algorithms of DYPSOKM algorithms have faster
Convergence rate and smaller convergency value.
Following table gives all results for assessing Cluster Validity：
Likewise, the algorithm of the present invention has less SSE, larger SSB, higher entropy and purity, show cluster result
There are higher condensation degree and separating degree.
Fig. 6 provides 4 kinds of Clustering Effect scatter diagrams in cluster process, and counts the frequency that each algorithm occurs in 10 experiments
Rate, see the table below.
Therefrom it can also be seen that inventive algorithm has more excellent performance.
The foregoing examples are merely illustrative of the technical concept and features of the invention, its object is to allow the person skilled in the art to be
Present disclosure can be understood and implemented according to this, it is not intended to limit the scope of the present invention.It is all smart according to the present invention
Equivalent transformation or modification that refreshing essence is done, should all be included within the scope of the present invention.
Claims (6)
1. a kind of depth combination K-means and PSO clustering method, it is characterised in that comprise the following steps：
S01：The particle in particle cluster algorithm is initialized using K-means algorithms；
S02：Fitness value is calculated according to fitness function, the study part of particle cluster algorithm iteration and equal value part is updated；
S03：PSO algorithms on multiple populations are built using cluster, with one cluster of cluster center representative, particle rapidity and position are updated；
S04：Judge whether to meet iteration stopping condition, if meeting, export optimal particle；Otherwise return to step S02, continues to change
Generation.
2. depth combination K-means and PSO according to claim 1 clustering method, it is characterised in that the step
Fitness function is in S02：
<mrow>
<msub>
<mi>J</mi>
<mi>e</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mi>c</mi>
</msub>
</munderover>
<mo>&lsqb;</mo>
<msub>
<mo>&Sigma;</mo>
<mrow>
<mo>&ForAll;</mo>
<msub>
<mi>z</mi>
<mi>p</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>C</mi>
<mi>j</mi>
</msub>
</mrow>
</msub>
<mi>d</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>z</mi>
<mi>p</mi>
</msub>
<mo>,</mo>
<msub>
<mi>m</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>/</mo>
<mo>|</mo>
<msub>
<mi>C</mi>
<mi>j</mi>
</msub>
<mo>|</mo>
</mrow>
<msub>
<mi>N</mi>
<mi>c</mi>
</msub>
</mfrac>
</mrow>
mj,d(zp,mj) respectively represent cluster center and Euclidean distance, Nc,CjThe total number of cluster is represented respectively and belongs to each cluster
Number of objects.
3. depth combination K-means and PSO according to claim 1 clustering method, it is characterised in that the step
The core iterative equation of particle cluster algorithm is in S02：
Vi(t+1)=Vi(t)+c1r1(pbesti(t)-Xi(t))+c2r2(gbest(t)-Xi(t))
Xi(t+1)=X (t)+V (t+1)
Wherein, c1,c2It is Studying factors, r1,r2It is the random number between [0,1], ViAnd X (t)i(t) particle i t generations are represented respectively
Speed and position, pbestiAnd gbest (t)i(t) history optimal location and the optimal position of population in particle i t generations are represented respectively
Put.
4. depth combination K-means and PSO according to claim 3 clustering method, it is characterised in that the step
Particle rapidity core iterative equation is changed into following part in S03：
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>V</mi>
<mi>i</mi>
<mrow>
<mi>o</mi>
<mi>l</mi>
<mi>d</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>1</mn>
</msub>
<msub>
<mi>r</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>pbest</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>X</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>2</mn>
</msub>
<msub>
<mi>r</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>g</mi>
<mi>b</mi>
<mi>e</mi>
<mi>s</mi>
<mi>t</mi>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>X</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>V</mi>
<mi>i</mi>
<mrow>
<mi>n</mi>
<mi>e</mi>
<mi>w</mi>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msubsup>
<mi>V</mi>
<mi>i</mi>
<mrow>
<mi>o</mi>
<mi>l</mi>
<mi>d</mi>
</mrow>
</msubsup>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>3</mn>
</msub>
<msub>
<mi>r</mi>
<mn>3</mn>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>cbest</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>X</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein, cbesti(t) represent the optimal part of history of each subgroup, each iteration determines to update by the fitness of average and
It is no, Vi old(t) the more new formula of classical particle speed described above is represented, the present invention adds a cbest on its basisi
(t), c3,r3It is the random number between Studying factors and [0,1] respectively.
5. depth combination K-means and PSO according to claim 4 clustering method, it is characterised in that to particle rapidity
Amplitude limit is carried out with position, clipping mode is：The locational space for calculating training sample is [- xmax,xmax], if grain in iterative process
Sub- position exceedes this scope, then is set to interval endpoint；Particle maximal rate is set to Vmax=vk × Xmax, vk representation speed amplitude limits
Coefficient, then the speed that particle updates is determined by below equation：
<mrow>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>min</mi>
<mo>{</mo>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>V</mi>
<mi>max</mi>
</msub>
<mo>}</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>></mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>min</mi>
<mo>{</mo>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>-</mo>
<msub>
<mi>V</mi>
<mi>max</mi>
</msub>
<mo>}</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>V</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo><</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>.</mo>
</mrow>
6. depth combination K-means and PSO according to claim 1 clustering method, it is characterised in that the stopping bar
Part is given iterations or global optimum not in change.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710247343.1A CN107153837A (en) | 2017-04-14 | 2017-04-14 | Depth combination K means and PSO clustering method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710247343.1A CN107153837A (en) | 2017-04-14 | 2017-04-14 | Depth combination K means and PSO clustering method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107153837A true CN107153837A (en) | 2017-09-12 |
Family
ID=59794135
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710247343.1A Pending CN107153837A (en) | 2017-04-14 | 2017-04-14 | Depth combination K means and PSO clustering method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107153837A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108199875A (en) * | 2017-12-29 | 2018-06-22 | 上海上讯信息技术股份有限公司 | A kind of Network Intrusion Detection System and method |
| CN108898636A (en) * | 2018-06-08 | 2018-11-27 | 福州大学 | A kind of camera one-dimension calibration method based on improvement PSO |
| CN110163304A (en) * | 2019-06-14 | 2019-08-23 | 福州大学 | A kind of harmonic source coupling parameter discrimination method clustered using linear relationship |
| CN110696816A (en) * | 2019-10-22 | 2020-01-17 | 河南科技大学 | Dynamic coordination hybrid electric vehicle energy management method based on working condition classification |
| CN111144541A (en) * | 2019-12-12 | 2020-05-12 | 中国地质大学（武汉） | Microwave filter debugging method based on multi-population particle swarm optimization method |
| CN111161879A (en) * | 2020-02-24 | 2020-05-15 | 梅里医疗科技(洋浦)有限责任公司 | Disease prediction system based on big data |
| CN112667876A (en) * | 2020-12-24 | 2021-04-16 | 湖北第二师范学院 | Opinion leader group identification method based on PSOTVCF-Kmeans algorithm |
-
2017
- 2017-04-14 CN CN201710247343.1A patent/CN107153837A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108199875A (en) * | 2017-12-29 | 2018-06-22 | 上海上讯信息技术股份有限公司 | A kind of Network Intrusion Detection System and method |
| CN108898636A (en) * | 2018-06-08 | 2018-11-27 | 福州大学 | A kind of camera one-dimension calibration method based on improvement PSO |
| CN110163304A (en) * | 2019-06-14 | 2019-08-23 | 福州大学 | A kind of harmonic source coupling parameter discrimination method clustered using linear relationship |
| CN110696816A (en) * | 2019-10-22 | 2020-01-17 | 河南科技大学 | Dynamic coordination hybrid electric vehicle energy management method based on working condition classification |
| CN111144541A (en) * | 2019-12-12 | 2020-05-12 | 中国地质大学（武汉） | Microwave filter debugging method based on multi-population particle swarm optimization method |
| CN111161879A (en) * | 2020-02-24 | 2020-05-15 | 梅里医疗科技(洋浦)有限责任公司 | Disease prediction system based on big data |
| CN111161879B (en) * | 2020-02-24 | 2020-10-09 | 吾征智能技术(北京)有限公司 | Disease prediction system based on big data |
| CN112667876A (en) * | 2020-12-24 | 2021-04-16 | 湖北第二师范学院 | Opinion leader group identification method based on PSOTVCF-Kmeans algorithm |
| CN112667876B (en) * | 2020-12-24 | 2024-04-09 | 湖北第二师范学院 | Opinion leader group identification method based on PSOTVCF-Kmeans algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107153837A (en) | Depth combination K means and PSO clustering method | |
| CN111476263B (en) | Bearing defect identification method based on SDAE and improved GWO-SVM | |
| CN112165464B (en) | Industrial control hybrid intrusion detection method based on deep learning | |
| CN102426697B (en) | Image segmentation method based on genetic rough set C-mean clustering | |
| Alfred | Summarizing relational data using semi-supervised genetic algorithm-based clustering techniques | |
| Wang | Fuzzy clustering analysis by using genetic algorithm | |
| Yang et al. | Linearly decreasing weight particle swarm optimization with accelerated strategy for data clustering | |
| CN109242026A (en) | A kind of fuzzy C-means clustering method and system based on hybrid rice algorithm optimization | |
| CN113222035B (en) | Multi-class imbalance fault classification method based on reinforcement learning and knowledge distillation | |
| CN103473599A (en) | Genetic algorithm and Kalman filtering based RBFN (Radial Basis Function Networks) combined training method | |
| Wei et al. | A flight maneuver recognition method based on multi-strategy affine canonical time warping | |
| Shi et al. | PSO-based community detection in complex networks | |
| Li et al. | A feature subset selection method based on the combination of PCA and improved GA | |
| CN113866735A (en) | Radar signal sorting method based on dynamic correction chaotic particle swarm optimization | |
| CN110288606B (en) | Three-dimensional grid model segmentation method of extreme learning machine based on ant lion optimization | |
| CN113341379A (en) | Radar signal sorting method based on adaptive threshold and iterative control | |
| Mahmood et al. | A proposed modification of K-means algorithm | |
| Mousavi | A New Clustering Method Using Evolutionary Algorithms for Determining Initial States, and Diverse Pairwise Distances for Clustering | |
| Hagenbuchner et al. | A supervised self-organizing map for structures | |
| Kim et al. | Proportional voting based semi-unsupervised machine learning intrusion detection system | |
| CN111191717A (en) | Black box confrontation sample generation algorithm based on hidden space clustering | |
| CN113222034B (en) | Knowledge distillation-based fine-grained multi-class unbalanced fault classification method | |
| Wang et al. | Centering SVDD for unsupervised feature representation in object classification | |
| Masuyama et al. | Growing neural gas with correntropy induced metric | |
| Ma et al. | Latent subspace clustering based on deep neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170912 |
|
| RJ01 | Rejection of invention patent application after publication |