CN105868435B - It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence - Google Patents
It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence Download PDFInfo
- Publication number
- CN105868435B CN105868435B CN201510568792.7A CN201510568792A CN105868435B CN 105868435 B CN105868435 B CN 105868435B CN 201510568792 A CN201510568792 A CN 201510568792A CN 105868435 B CN105868435 B CN 105868435B
- Authority
- CN
- China
- Prior art keywords
- hot spot
- model
- construction
- feature vector
- linear dependence
- 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
Abstract
It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence, belong to communication technique field, various dimensions Optimal Decomposition is carried out to optical access network hot spot using the linear dependence of feature vector, matching construction model the most suitable meets the construction demand of hot spot.It increases optical access network hot spot in the present invention newly and construction Optimized model carries out linear matched, find out one group of approximate set, building Bayesian network makes decisions judgement, chooses the efficient construction of optimal construction model realization hot spot.
Description
Technical field
The invention belongs to communication technique fields, more particularly to a kind of control method of optical-fiber network construction.
Background technique
With the continuous development of the information society, people are growing day by day to the demand of information service, recent P2P, network video
The rapid growth of the data services such as frequency and mobile Internet has had resulted in " bandwidth is hungered and thirst ".And smart city, Internet of Things planning
Construction proposes bigger challenge to communication bandwidth, still continues to use Traditional photovoltaic heterogeneous network and has been unable to satisfy multimedia, big bandwidth
Business demand, all-optical network development is imperative, and optical-fiber network transformation has been not limited in traditional construction mode, i.e., according to single
Fixed Combination factor determine convergence or access level hot spot construction mode or type, dynamic, comprehensive considering various effects,
How feature and mass data analysis, processing, including the network architecture and bearer service type, flow etc., build development of optical network
If mode and the more preferable combination of mass data analysis, will become the important subject of all-optical network development.
The accuracy that the key of mass data analysis is not required included data reaches to a certain degree, but is logarithm
Range according to collection and tolerance and fault-tolerance to abnormal data.And in " light entering and copper back " implementation process, for various types of
The hot spot scene and magnanimity hot spot feature and element of type, without effective planning, design means.It is a set of reasonable to need at present
Optimization method according to the different demands in region, different hot spot features, user group, class of service realize the conjunction of network construction
Reason planning, the efficient deployment of network element device, network service are distributed rationally, accelerate engineering construction progress, in traditional design mode
On the basis of must seek more powerful technical support in due course, such as with mass data analysis and cloud computing technology, pass through building
Data warehouse or Data Supermarket, visual analyzing and powerful predictive ability, realize the quality management function of mass data, mention
The quality and efficiency, network construction quality of high planning, reduce engineering construction cost, accelerate network deployment, for the following light net
Solid technical know-how basis and effective analysis, appraisal procedure are established in network development.
Summary of the invention
The invention proposes a kind of efficient control methods analyzed based on linear dependence and realize that optical-fiber network is built.Using spy
The linear dependence for levying vector carries out various dimensions Optimal Decomposition to optical access network hot spot, and matching construction model the most suitable meets
The construction demand of hot spot.In the present invention increase newly optical access network hot spot and construction Optimized model carry out linear matched, find out one group it is close
Like collection, constructs Bayesian network and make decisions judgement, choose the efficient construction of optimal construction model realization hot spot.
This method includes the following steps:
Step 1: the descriptions such as optical access network hot spot construction scene are arranged, by characteristic element by grade, classification point
Class, and different weighted values is assigned according to significance level, form corresponding feature vector；
Practical construction demand is depended on for hot spot scene characteristic individual hierarchical, classifying and dividing, is such as connect with large-scale consumer
Based on entering, it is contemplated that building type, number of users of scene etc. will be accessed and be used as I grades of elements (important element in vector), it is such as excellent
First consider that existing cable resource situation realizes access, then by existing cable resource situation (such as existing optical cable item number, fibre core number and fibre
Core occupancy etc.) I grades of elements are used as, and then complete the characteristic element accessed to new hot spot in the planning stage and be classified predefined, structure
Vector (i.e. feature vector) is built convenient for decomposing, calculating.
The feature vector, X of certain hot spot scene is set, vector interior element is xij, wherein serial number i is classification (point of hot spot scene
Grade), j is therefore the x in the characteristic element feature of serial number j under i grades (i.e. description) under i gradeijFor j characteristic element under i grades；
Classification and corresponding element quantity can be adjusted according to practical construction demand, but classification used by a kind of research object, classification element
Strict conformance must be kept, is analyzed convenient for linear dependence, and assigns different weighted value T according to the significance level that feature describes
(tij: tijxijMiddle xijThe weighted value of element).That is:
X={ TX }={ tijxij}；
Step 2: the feature vector of magnanimity hot spot scene is determined into similitude, phase by the cosine function of linear dependence
It is classified as one kind like hot spot, realizes the feature vector group model of magnanimity hot spot, and is that information fingerprint collection is deposited by Hash mapping
Storage, to save memory space；
Set of eigenvectors (quantity N) linear dependence that one group of hot spot is calculated by cosine function, according to feature quantity and
Construction model granularity sets least model collection quantity S, threshold valueWherein Q is Models Sets capacity, is vector
Space granularity (can be consistent with least model number S, according to existing optical access network hot spot build scene be about bordering on 60,
So Q value can fix tentatively the cosine value for guaranteeing two feature vectors for 60), α is regulation coefficient in reasonable threshold value range
Interior, hot spot and the scene characteristic of construction model are similar enough, and usual value range is [3~10].
It is compared two-by-two between hot spot, cosine function value >=1- α/Q that the linear dependence of feature vector calculates can determine that two
A hot spot construction model is similar；Cosine function absolute value < 1- α/Q that the linear dependence of its feature vector calculates, can determine that two
A hot spot construction model is uncorrelated.Maximum model quantity L is set, characteristic quantity set is more than predetermined value L after the above classification, then considers
Increase granularity, using regulation coefficient α, increases vector capacity.
The linear dependence of the above magnanimity hot spot is quickly sorted out, and new set of eigenvectors (quantity M) is formed, convenient for data
Compression storage, by hash function map information fingerprint, is mapped as information fingerprint collection for existing set of eigenvectors, it may be assumed that
{Ha(X1),Ha(X2),…,Ha(XM)}。
Wherein Ha (X) is completed using Hash mapping
Step 3: according to step 2 model, decompose to new hot spot and signature analysis, find out one group of close feature to
Amount；Shown in specific step is as follows:
Step A: using cosine function to new hot spot XOWith feature vector, X existing in Models SetsACarry out similarity analysis；
Cos θ is the cosine value of hot spot scene characteristic vector.
Step B: the following (threshold value of the determination method of calculated valueWherein Q is Models Sets capacity, and α is adjustment
Coefficient)；
1) cos θ >=1- α/Q can determine that new hot spot XOWith construction model XAIt is similar；
2) cos θ < 1- α/Q can determine that new hot spot XOWith construction model XAIt is unrelated.
Step C: model relevant to new hot spot is unique, then directly determines the reasonable construction that new hot spot is concentrated in hot spot scene
Model；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model is unique, then directly determines new
The reasonable construction model that hot spot is concentrated in hot spot scene；
Step D: model relevant to new hot spot is one group, enters step four；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model quantity is greater than 1, then enters step
Rapid four；
Step E: with new hot spot non-correlation, then hot spot model collection, and appropriate adjustment model dependency is added in the hot spot
Granularity；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model quantity is 0, and existing hot spot is special
Vector model collection quantity N=N+1 is levied, as N > L, adjusts α, which is added hot spot feature vector model collection, and be mapped as
Information fingerprint is stored in fingerprint base；
Step 4: the feature vector group that step 3 is obtained and new hot spot construct Bayesian network model, feature vector group
Maximum middle probability value is new hot spot most reasonable construction scene；
Step A: the feature vector of feature vector group and new hot spot is set as one group of event { R }；
That is event RXWith feature vector group event RA,RB,…,RN；
Step B: feature vector group is set as precondition, new hot spot is set as calculating as a result, by Bayesian network model
Maximum value out；
That is max { P (RX|RA),P(RX|RB),…,P(RX|RN)}
Step C: according to calculated result, reasonable construction model is chosen.
Beneficial effects of the present invention:
The present invention is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence, it is intended to pass through extraction
Hot spot builds the feature description of scene, composition characteristic vector, using such hot spot scene of the linearly related sex determination of feature vector
Construction similitude, this calculation method be particularly suitable for magnanimity hot spot scene construction, improve hotspot's classification planning, construction efficiency,
The deviation of manual analysis is avoided, provides an efficient new method and thinking for optical-fiber network construction.
Detailed description of the invention
Attached drawing is flow chart of the invention.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and detailed description.
The present invention chooses the construction mode of hot spot scene in certain districts and cities' optical-fiber network transformation process, including hot spot scene
17242, the description item of each access hot spot is 73, and the present invention is directed to realize light based on linear dependence analysis using a kind of
The efficient control method of network construction divides hot spot construction model collection, and realizes the Rapid matching of new hot spot Yu construction model collection,
The following steps are included:
Step 1: the descriptions such as optical access network hot spot construction scene are arranged, by characteristic element by grade, classification point
Class, and different weighted values is assigned according to significance level, form corresponding feature vector；
Optical access network construction demand is deeply understood, to the device configuration in access network planning, implementation, maintenance link, construction
Step etc. is analyzed, and then is completed the characteristic element accessed to new hot spot classification in project period and predefined, and it is (i.e. special to construct vector
Levy vector) convenient for decomposition, calculating.
Hot spot scene is set as vector X, feature vector element is xij, wherein i is classification, and j is the characteristic element under i grades
Plain serial number, therefore xijFor j characteristic element under i grades；Classification and corresponding element quantity can be adjusted according to practical construction demand, but one
Classification used by class research object, classification element must keep strict conformance, analyze convenient for linear dependence, and according to feature
The significance level of description assigns different weighted value T.That is:
X={ TX }={ tijxij}；
According to hot spot construction information table, 73 contents that existing hot spot is described are divided into 5 grades, I, II, III, IV, V,
Middle I grades is important Element-Level, comprising: service area, building type, construction area, resident/trade company's number etc.；II grades are time important member
Plain grade, comprising: existing cable resource situation (whether optical cable access, optical cable item number, fibre core quantity and occupancy) etc.；III level is
General element grade, IV are insignificant Element-Level, and V is negligible Element-Level, such as serial number, regional code accessory ID element.
Every level element is using number distribution；Such hot spot feature element decomposes, can be according to practical business demand and construction center of gravity to member
Element classification, classification carry out appropriate adjustment, when hot spot description point is more, do not consider V grades of elements when forming hot spot feature vector.
Weighted value T setting range is as follows:
I grades: T value, which takes, determines section [4,9]；
II grades: T value, which takes, determines section [1,4]；
III level: T value, which takes, determines section [0.6,1.0]；
IV grades: T value, which takes, determines section [0.2,0.6]；
V grades: T value, which takes, determines section [0,0.1]；
Step 2: the feature vector of magnanimity hot spot is measured into similitude, similar heat by the cosine function of linear dependence
Point is classified as one kind, realizes the feature vector group model of magnanimity hot spot, and is that information fingerprint collection is stored by Hash mapping, with
Save memory space；
Set of eigenvectors (quantity N) linear dependence that one group of hot spot is calculated by cosine function, according to feature quantity and
Construction model granularity is set least model collection quantity 60 (S=60), is compared two-by-two between Q=60, α=3 (tentative) hot spot,
Cosine function value >=0.95 that the linear dependence of feature vector calculates, can determine that two hot spot construction models are similar；Its feature
The cosine function value < 0.95 that the linear dependence of vector calculates, can determine that two hot spot construction models are uncorrelated.Setting is maximum
Model quantity L, characteristic quantity set is more than predetermined value 100 (L=100) after the above classification, then considers to increase similitude particle angle value,
Suitably increase α value.
The linear dependence of the above magnanimity hot spot is quickly sorted out, and forming new set of eigenvectors, (quantity M, construction model are held
Amount≤60 increases after linear analysis or according to the actual situation, but≤100), convenient for the compression storage of data and rapid computations, lead to
Hash function map information fingerprint is crossed, existing set of eigenvectors is mapped as information fingerprint library, it may be assumed that
{Ha(X1),Ha(X2),…,Ha(XM)}。
The formation in information fingerprint library uses for reference google about web crawlers to net depending on the conversion method of hash function
Page search and storage method can be by construction model features using PRNG (pseudorandom number generator algorithm) or Mason's Rotation Algorithm
Vector set is mapped as information fingerprint library, when building contextual data amount at million grades, can pass through the line of information fingerprint feature vector
The similitude of property correlation analysis hot spot, in this way can be by original O (N2) operand be reduced to O (N), while reduce O (N) storage
The consuming in space.
An Intel I5 processor 8G memory is configured, operating system win7 acquires Excel table by C programmer card
The feature vector content converted in lattice is completed the feature vector classification of existing 17242 hot spot scenes for time-consuming about 13 minutes, is formed
The set of eigenvectors of construction model is 47.
Step 3: according to step 2 model, decompose to new hot spot and signature analysis, find out one group of close feature to
Amount；Shown in specific step is as follows:
Using cosine function to new hot spot XOWith feature vector, X existing in Models SetsACarry out similarity analysis；
Cos θ is the cosine value of hot spot scene characteristic vector.
Step B: given threshold, calculated value, which is less than threshold value, can determine that the two is related；It can determine that the two is uncorrelated greater than threshold value；
3) θ >=0.95 cos can determine that new hot spot Xo and construction model XAIt is similar；
4) cos θ < 0.95 can determine that new hot spot Xo and construction model XAIt is unrelated.
Step C: model relevant to new hot spot is unique, then directly determines the reasonable construction that new hot spot is concentrated in hot spot scene
Model；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model is unique, then directly determines new
The reasonable construction model that hot spot is concentrated in hot spot scene；
Step D: model relevant to new hot spot is one group, enters step four；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model quantity is greater than 1, then enters step
Rapid four；
Step E: with new hot spot non-correlation, then hot spot model collection, and appropriate adjustment model dependency is added in the hot spot
Granularity；
It according to step B mode, is compared one by one with existing Models Sets feature vector, scale model quantity is 0, and existing hot spot is special
Vector model collection quantity N=N+1 is levied, as N > L, which is added hot spot feature vector model collection, gone forward side by side by regulation coefficient α
Row mapping deposit information fingerprint library；
Step 4: the feature vector group that step 3 is obtained and new hot spot construct Bayesian network model, feature vector group
Maximum middle probability value is new hot spot most reasonable construction scene；
Step A: the feature vector of feature vector group and new hot spot is set as one group of event set (R)；
That is event RXWith feature vector group event RA,RB,…,RN；
Step B: feature vector group is set as precondition, new hot spot is set as calculating as a result, by Bayesian network model
Maximum value out；
That is max { P (RX|RA),P(RX|RB),…,P(RX|RN)}
Step C: according to calculated result, reasonable construction model is chosen.
In conjunction with Step 3: introducing one group of new hot spot described in four, hot spot quantity is 10, passes through existing 47 construction models
Set of eigenvectors is brought into Step 3: four contents, wherein cosine function value is equal to greatly after 10 hot spot feature vectors traversals
It is unique that 0.95 and 9 new hot spot corresponds to construction model；
Other 1 hot spot is similar to three, existing feature vector library unit, hence into step 4, introduces Bayesian network
Model, new hot spot are set as feature vector, X o, and existing feature vector model is respectively XA,XB,XC, introduced as four events public
Formula:
max{P(RX|RA),P(RX|RB),P(RX|RC)}
Big number statistic law usually can be used in Probit Analysis, and (I, II grade element of A, B, C inspire the cumulative system of mass data of X
The method of meter), normal distribution method (meet normal state statistical number point distribution) and expert prediction method (with reference to A, B, C situation, use for reference
Plan expert and engineering construction personnel's experience, provide with reference to probability), this is taken expert prediction method to calculate separately and obtains:
P(RX|RA)=0.37, P (RX|RB)=0.52, P (RX|RC)=0.61
I.e. new hot spot Xo refers to XCClass construction model completes planning construction.
Claims (4)
1. a kind of analyze the efficient control method for realizing optical-fiber network construction based on linear dependence, it is characterised in that including following step
It is rapid:
Step (1) arranges optical access network hot spot construction scene description, and characteristic element is pressed grade, category classification, and root
Different weighted values is assigned according to significance level, forms corresponding feature vector；
The feature vector of magnanimity hot spot is measured similitude by the cosine function of linear dependence by step (2), and similar hot spot is returned
For one kind, the feature vector group model of magnanimity hot spot is realized, and be that information fingerprint collection is stored by Hash mapping；
Step (3) carries out decomposition and signature analysis according to step (2) model, to new hot spot, finds out one group of close feature vector；
Step (4): feature vector group that step (3) is obtained and new hot spot construct Bayesian network model, in feature vector group
Maximum probability value is new hot spot most reasonable construction scene.
2. a kind of efficient control method that realization optical-fiber network construction is analyzed based on linear dependence according to claim 1,
It is characterized by: the feature vector achievement formed described in step (2) according to step (1), passes through the cosine letter of linear dependence
Number is sorted out, and constructs the set of eigenvectors of hot spot model and is mapped as information fingerprint library.
3. a kind of efficient control method that realization optical-fiber network construction is analyzed based on linear dependence according to claim 2,
It is characterized by: being decomposed to new hot spot by the set of eigenvectors of construction hot spot model, finding out one by linear dependence
Group similar features vector, includes the following steps；
Step A: it is analyzed by feature vector of the linear dependence to the set of eigenvectors of new hot spot and hot spot model；
Step B: given threshold, calculated value, which is less than threshold value, can determine that the two is related；It can determine that the two is uncorrelated greater than threshold value；
Step C: model relevant to new hot spot is unique, then directly determines the reasonable construction mould that new hot spot is concentrated in hot spot scene
Type；
Step D: model relevant to new hot spot is one group, enters step (4)；
Step E: with new hot spot non-correlation, then the set of eigenvectors of hot spot model, and appropriate adjustment model phase is added in the hot spot
Closing property granularity.
4. a kind of efficient control method that realization optical-fiber network construction is analyzed based on linear dependence according to claim 1,
It is characterized by: the obtained one group of feature vector of step (3), constructs Bayesian network model, feature vector group with new hot spot
Maximum middle probability value is new hot spot most reasonable construction scene, comprising the following steps:
Step A: the feature vector of feature vector group and new hot spot is set as one group of event；
Step B: feature vector group is set as precondition, new hot spot is set as calculating most as a result, by Bayesian network model
Big value；
Step C: according to calculated result, reasonable construction model is chosen.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510568792.7A CN105868435B (en) | 2015-09-09 | 2015-09-09 | It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510568792.7A CN105868435B (en) | 2015-09-09 | 2015-09-09 | It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105868435A CN105868435A (en) | 2016-08-17 |
| CN105868435B true CN105868435B (en) | 2019-02-15 |
Family
ID=56624242
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510568792.7A Active CN105868435B (en) | 2015-09-09 | 2015-09-09 | It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105868435B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110247822B (en) * | 2019-06-11 | 2021-08-03 | 北京全路通信信号研究设计院集团有限公司 | Network element for monitoring network service and service processing method thereof |
| CN110417612B (en) * | 2019-06-11 | 2021-05-28 | 北京全路通信信号研究设计院集团有限公司 | Network flow monitoring system and method based on network elements |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1908945A (en) * | 2006-08-24 | 2007-02-07 | 上海交通大学 | Method for implementing mesh based optical mesh emulation platform |
| CN103020359A (en) * | 2012-12-17 | 2013-04-03 | 江苏省电力公司扬州供电公司 | Power fiber to the home (PFTTH) access method based on EPON (Ethernet passive optical network) |
| CN103647598A (en) * | 2013-11-05 | 2014-03-19 | 国家电网公司 | Communication multi-level protection method based on intelligent optical network |
| CN104115165A (en) * | 2012-02-15 | 2014-10-22 | 阿尔卡特朗讯 | Method for mapping media components employing machine learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9014022B2 (en) * | 2010-11-15 | 2015-04-21 | Level 3 Communications, Llc | Wavelength regeneration in a network |
-
2015
- 2015-09-09 CN CN201510568792.7A patent/CN105868435B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1908945A (en) * | 2006-08-24 | 2007-02-07 | 上海交通大学 | Method for implementing mesh based optical mesh emulation platform |
| CN104115165A (en) * | 2012-02-15 | 2014-10-22 | 阿尔卡特朗讯 | Method for mapping media components employing machine learning |
| CN103020359A (en) * | 2012-12-17 | 2013-04-03 | 江苏省电力公司扬州供电公司 | Power fiber to the home (PFTTH) access method based on EPON (Ethernet passive optical network) |
| CN103647598A (en) * | 2013-11-05 | 2014-03-19 | 国家电网公司 | Communication multi-level protection method based on intelligent optical network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105868435A (en) | 2016-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | A hybrid machine learning model for demand prediction of edge-computing-based bike-sharing system using Internet of Things | |
| Huang et al. | Forecasting hourly solar irradiance using hybrid wavelet transformation and Elman model in smart grid | |
| Lin et al. | Particle swarm optimisation aided least‐square support vector machine for load forecast with spikes | |
| CN103235825A (en) | Method used for designing large-quantity face recognition search engine and based on Hadoop cloud computing frame | |
| Zhang et al. | Towards green metaverse networking: Technologies, advancements and future directions | |
| WO2020098728A1 (en) | Cluster load prediction method and apparatus, and storage medium | |
| CN108549907B (en) | Data verification method based on multi-source transfer learning | |
| Li et al. | Research on QoS service composition based on coevolutionary genetic algorithm | |
| Hüsch et al. | CorClustST—Correlation-based clustering of big spatio-temporal datasets | |
| Kim et al. | Role and challenge of technology toward a smart sustainable city: Topic modeling, classification, and time series analysis using information and communication technology patent data | |
| Cong | Personalized recommendation of film and television culture based on an intelligent classification algorithm | |
| Xu et al. | Hybrid holiday traffic predictions in cellular networks | |
| CN105868435B (en) | It is a kind of to analyze the efficient control method for realizing optical-fiber network construction based on linear dependence | |
| Seo et al. | Resource-efficient federated learning with non-iid data: An auction theoretic approach | |
| Hong et al. | Development of EV charging templates: an improved K‐prototypes method | |
| Liu et al. | Data mining for energy systems: Review and prospect | |
| Li et al. | Research on the inheritance and protection of data mining technology in national sports | |
| Wang et al. | Applications of Machine Learning in Public Security Information and Resource Management | |
| Shafi et al. | Feature selection and extraction along with electricity price forecasting using big data analytics | |
| Gorricha et al. | A framework for exploratory analysis of extreme weather events using geostatistical procedures and 3D self-organizing maps | |
| Zhang et al. | Research on a supply–demand matching method for cloud 3D printing services based on complex networks | |
| Wang et al. | The combination forecasting of electricity price based on price spikes processing: a case study in south Australia | |
| Chen et al. | Research and application of cluster analysis algorithm | |
| Yin et al. | Development and design of intelligent gymnasium system based on K-Means clustering algorithm under the internet of things | |
| Yuan et al. | Resource-Aware Federated Neural Architecture Search over Heterogeneous Mobile Devices |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |