CN109639524A - Communication network data method for visualizing, device and equipment based on volume forecasting - Google Patents
Communication network data method for visualizing, device and equipment based on volume forecasting Download PDFInfo
- Publication number
- CN109639524A CN109639524A CN201811524648.3A CN201811524648A CN109639524A CN 109639524 A CN109639524 A CN 109639524A CN 201811524648 A CN201811524648 A CN 201811524648A CN 109639524 A CN109639524 A CN 109639524A
- Authority
- CN
- China
- Prior art keywords
- data
- communication network
- visualization
- volume forecasting
- model
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/04—Processing captured monitoring data, e.g. for logfile generation
- H04L43/045—Processing captured monitoring data, e.g. for logfile generation for graphical visualisation of monitoring data
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/0876—Network utilisation, e.g. volume of load or congestion level
- H04L43/0888—Throughput
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Environmental & Geological Engineering (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention provides a kind of communication network data method for visualizing, device and equipment based on volume forecasting, integrated service network flow is predicted using LSTM Network Traffic Forecast Model, and Visualization Model is established based on multi-level node clustering algorithm, to carry out visualization presentation to the acquired original data.While realizing data visualization, also based on LSTM network implementations to the Accurate Prediction of the integrated service network flow of data.The intelligence of increase system, the training speed for promoting prediction model, and improve precision of prediction.
Description
Technical field
The present invention relates to network monitor fields, visualize more particularly to a kind of communication network data based on volume forecasting
Method, device and equipment.
Background technique
In electric integrated data network construction field, the visualization of data network data on flows resource is an important content.Number
The level of electric integrated data network maintenance and management is helped to improve according to net volume forecasting and data visualization, improves operation maintenance personnel
Working efficiency and quality, reduce communication network failure-frequency, reduce equipment manages and maintains expense, promote maintenance and pipe
The innovation of reason mechanism.
Data visualization technique can substantially be divided into three classes at present: the method based on visualization tool, based on repulsion tension
The method of topological layout's model, the method for topological layout's algorithm based on abstract point.
1, based on the method for visualization tool, leading to common visualization tool has Univ California-San Diego USA
MapNet system, the GT-ITM system of Georgia Institute of Technology exploitation of exploitation.Visualization tool can
It is intuitive to clearly indicate network topology structure again, but MapNet system parses data information in planar fashion, cannot show vertical
The hierarchical structure of body.The general scale of network topology that GT-ITM is generated is bigger, it appears more in disorder.
2, the method for topological layout's model based on repulsion tension utilizes the repulsion and tension realization topological structure in network
The balance at midpoint, to realize visualization.The network topology structure of the method autoplacement, which does not become, to cross one another, and is connected
Relationship seems very clear, can realize visualization well, still, the method is laid out for node, can be able to achieve node
The excessively close situation with back gauge influences to be laid out effect.
3, common topological structure ring topology is first abstracted by the method for topological layout's algorithm based on abstract point, the method
For a point, Coordinate generation algorithm when then being shown using directory tree is existed to obtain the later each gateway of abstract
Relative coordinate in whole network topological diagram based on this finally gradually restores abstract point, to obtain entire net
The coordinate-system of network Topology connection.The method can reduce the manual intervention of network management personnel to greatest extent, have good
Practical function.But the method is laid out towards large-scale network topological figure, and small-scale topological diagram not can guarantee clearly.
And above-mentioned data visualization method can not realize data forecast function while realizing data visualization.
Summary of the invention
In view of the foregoing deficiencies of prior art, the purpose of the present invention is to provide a kind of communications based on volume forecasting
Network data method for visualizing, device and equipment, for solve in the prior art cannot be while to realize data visualization pair
The problem of data are accurately predicted.
In order to achieve the above objects and other related objects, the present invention provides a kind of communication network data based on volume forecasting
Method for visualizing, comprising: be acquired according to data of the preset time cycle to integrated service network flow original to obtain
Acquire data；The acquired original data are normalized to obtain normalization data；In conjunction with the normalization data,
LSTM Network Traffic Forecast Model is established to predict the integrated service network flow；It is calculated based on multi-level node clustering
Method establishes Visualization Model, to carry out visualization presentation to the acquired original data.
In a specific embodiment of the invention, the normalized includes: by the standardized transfer function of min-max
Linear transformation is carried out to the acquired original data, to be in the normalization data in [0,1] section, the transfer function
Are as follows:
Wherein, x is the initial data before normalization, x*To carry out normalizing to the initial data x
Normalization data after change, max are the maximum value in acquired original data, and min is the minimum value in acquired original data.
In a specific embodiment of the invention, the activation primitive about the LSTM Network Traffic Forecast Model, damage are determined
Function and the implicit number of plies are lost, and in conjunction with the normalization data, establishes the LSTM Network Traffic Forecast Model to described
Integrated service network flow is predicted.
In a specific embodiment of the invention, the activation primitive is to correct linear unit function.
In an of the invention specific embodiment, determine the input layer parameter about the LSTM Network Traffic Forecast Model,
Export layer parameter and hiding layer parameter, and in conjunction with the normalization data, establish the LSTM Network Traffic Forecast Model with
The integrated service network flow is predicted；Wherein, the input layer parameter includes input layer time step number and input layer
Dimension；The output layer parameter includes output variable dimension；The hiding layer parameter includes hiding number of layers and each hiding
Layer dimension.
In a specific embodiment of the invention, the input layer parameter further includes input layer number, the output layer ginseng
Number further includes output layer number of nodes, and the hiding layer parameter further includes hidden layer number of nodes, and according to the input layer number,
The output layer number of nodes and the regulating constant of setting, are calculated the hidden layer number of nodes.
In a specific embodiment of the invention, the method also includes: continue according to the time cycle to integrated service
The data of network flow are acquired to obtain new acquisition data；It is pre- to the LSTM network flow in conjunction with the new acquisition data
The error for surveying model is analyzed.
In a specific embodiment of the invention, according to the following formula, obtain about the LSTM Network Traffic Forecast Model
Prediction result root-mean-square error:Wherein, p (i) andRespectively network flow
Actual value and predicted value；N is prediction verify data number；I is future position sequence number, CapiOutput sequence maximum value.
In a specific embodiment of the invention, the multi-level node clustering algorithm based on side intensity establishes Visualization Model,
To carry out visualization presentation to the acquired original data.
In order to achieve the above objects and other related objects, the present invention also provides a kind of communication network number based on volume forecasting
According to visualization device characterized by comprising data acquisition module, to according to the preset time cycle to integrated services network
The data of network flow are acquired to obtain acquired original data；Module is normalized, to carry out to the acquired original data
Normalized is to obtain normalization data；Prediction module, it is pre- to establish LSTM network flow in conjunction with the normalization data
Model is surveyed to predict the integrated service network flow；Visualization model, to be based on multi-level node clustering algorithm
Visualization Model is established, to carry out visualization presentation to the acquired original data.
In order to achieve the above objects and other related objects, the present invention also provides a kind of equipment, comprising: memory and processing
Device；The memory stores computer program；The computer program that the processor runs the memory storage is described to enable
Electronic equipment executes as above described in any item communication network data method for visualizing based on volume forecasting.
As described above, communication network data method for visualizing, device and the equipment of the invention based on volume forecasting, utilizes
LSTM Network Traffic Forecast Model predicts integrated service network flow, and being established based on multi-level node clustering algorithm can
Depending on changing model, to carry out visualization presentation to the acquired original data.While realizing data visualization, it is also based on LSTM
Accurate Prediction of the network implementations to the integrated service network flow of data.The intelligence of increase system, the instruction for promoting prediction model
Practice speed, and improves precision of prediction.
Detailed description of the invention
Fig. 1 is shown as the communication network data method for visualizing of the invention based on volume forecasting in one embodiment
Flow diagram.
Fig. 2 is shown as the communication network data method for visualizing of the invention based on volume forecasting in one embodiment
Flow diagram.
Fig. 3 is shown as the training result signal in a specific embodiment of the invention based on LSTM Network Traffic Forecast Model
Figure.
Fig. 4 is shown as the comparison signal of LSTM prediction model predicted value and actual value in a specific embodiment of the invention
Figure.
Fig. 5 is shown as the communication network data visualization device of the invention based on volume forecasting in one embodiment
Module diagram.
Fig. 6 is shown as the composition schematic diagram of equipment of the invention in one embodiment.
Component label instructions
11 data acquisition modules
12 normalization modules
13 prediction modules
14 visualization models
21 memories
22 processors
S11~S14 step
S101~S104 step
Specific embodiment
Illustrate embodiments of the present invention below by way of specific specific example, those skilled in the art can be by this specification
Other advantages and efficacy of the present invention can be easily understood for disclosed content.The present invention can also pass through in addition different specific realities
The mode of applying is embodied or practiced, the various details in this specification can also based on different viewpoints and application, without departing from
Various modifications or alterations are carried out under spirit of the invention.It should be noted that in the absence of conflict, following embodiment and implementation
Feature in example can be combined with each other.
It should be noted that illustrating the basic structure that only the invention is illustrated in a schematic way provided in following embodiment
Think, only shown in diagram then with related component in the present invention rather than component count, shape and size when according to actual implementation
Draw, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its assembly layout kenel
It is likely more complexity.
Referring to Fig. 1, it is specific one to be shown as the communication network data method for visualizing of the invention based on volume forecasting
Flow diagram in embodiment.The communication network data method for visualizing based on volume forecasting, comprising the following steps:
Step S11: it is acquired according to data of the preset time cycle to integrated service network flow original to obtain
Acquire data；The time cycle is, for example, 15 minutes.Number and according to 15 minutes time cycles, in continuous acquisition 30 days
According to.And the data of acquisition can be divided into two parts, front is allocated as the training data of LSTM Network Traffic Forecast Model, latter portion
It is allocated as the data tested for the result to LSTM Network Traffic Forecast Model.
Step S12: the acquired original data are normalized to obtain normalization data；
Step S13: in conjunction with the normalization data, LSTM (Long Short-Term Memory, shot and long term note are established
Recall) Network Traffic Forecast Model to be to predict the integrated service network flow；
Step S14: establishing Visualization Model based on multi-level node clustering algorithm, to carry out to the acquired original data
Visualization is presented.In one embodiment, the multi-level node clustering algorithm based on side intensity establishes Visualization Model, with right
The acquired original data carry out visualization presentation.
In some embodiments, the normalized includes: by the standardized transfer function of min-max to the original
The acquisition data that begin carry out linear transformation, to be in the normalization data in [0,1] section, the transfer function are as follows:
Wherein x is the initial data before normalization, x*For the normalization number after the initial data x is normalized
According to max is the maximum value in acquired original data, and min is the minimum value in acquired original data.
In some embodiments, determine about the activation primitive of the LSTM Network Traffic Forecast Model, loss function with
And the implicit number of plies, and in conjunction with the normalization data, the LSTM Network Traffic Forecast Model is established to the integrated service
Network flow is predicted.Wherein, the activation primitive is, for example, to correct linear unit function, i.e. ReLU (Rectified
Linear unit corrects linear unit).For other activation primitives, ReLU has following advantage: for linear function
For, the ability to express of ReLU is stronger, is especially embodied in depth network；And for nonlinear function, ReLU is due to non-
Gradient between minus zone is constant, therefore gradient disappearance problem (Vanishing Gradient Problem) is not present, so that mould
The convergence rate of type maintains a stable state.Gradient disappearance problem is are as follows: when gradient is less than 1, predicted value and true value
Between error it is every propagate one layer can decay it is primary, it is this existing if using sigmoid as activation primitive in Deep model
As particularly evident, model convergence stagnation will lead to.
In some embodiments, determine that the input layer parameter about the LSTM Network Traffic Forecast Model, output layer are joined
Several and hiding layer parameter, and in conjunction with the normalization data, the LSTM Network Traffic Forecast Model is established to described comprehensive
Business network flow is closed to be predicted；Wherein, the input layer parameter includes input layer time step number and input layer dimension；It is described
Exporting layer parameter includes output variable dimension；The hiding layer parameter includes hiding number of layers and each hidden layer dimension.Institute
Stating input layer parameter further includes input layer number, and the output layer parameter further includes output layer number of nodes, the hidden layer ginseng
Number further includes hidden layer number of nodes, and normal according to the adjusting of the input layer number, the output layer number of nodes and setting
Number, is calculated the hidden layer number of nodes.Such as it is calculated the hidden layer number of nodes according to the following formula.
Wherein: l indicates that node in hidden layer, N indicate that input layer number, M indicate that output layer number of nodes, a indicate to adjust
Constant (takes the number between 1~10).
In some embodiments, the communication network data method for visualizing based on volume forecasting further include: continue root
It is acquired according to data of the time cycle to integrated service network flow to obtain new acquisition data；And in conjunction with described new
Acquisition data analyze the error of the LSTM Network Traffic Forecast Model.For example, according to the following formula, obtaining about institute
State the root-mean-square error of the prediction result of LSTM Network Traffic Forecast Model:
Wherein, p (i) andThe respectively actual value and predicted value of network flow；N is prediction verify data number；I is
Future position sequence number, CapiOutput sequence maximum value.
The communication network data method for visualizing of the invention based on volume forecasting is done below in conjunction with specific embodiment
Further instruction.
In specific embodiment one, as shown in Fig. 2, being shown as that the embodiment of the invention provides a kind of based on volume forecasting
Power communication dispatching and monitoring data visualization method, comprising:
Data collection steps S101 is acquired electric integrated data network data on flows according to Fixed Time Interval；
Collected data are normalized in data processing step S102, conclude the statistical distribution of unified samples
Property；
It is pre- to establish LSTM network flow according to acquisition data cases and application scenarios by predicting network flow step S103
Survey model；
Data visualization model S103 is established according to the various aspects feature of power communication based on multi-level node clustering
Visualization Model.
The embodiment of the present invention realizes that the electric power of volume forecasting is logical using LSTM neural network and multi-level node clustering method
Believe dispatching and monitoring data visualization.By determining input layer time step number in model, input layer dimension, hiding number of layers, each
Hidden layer dimension and output variable dimension establish the LSTM network model for being used for electric integrated data network predicting network flow, right
Network flow is predicted that recent power communication dispatching and monitoring data traffic trend is accurately predicted in realization, detects each communication network
The occupied bandwidth of the data flow of middle different business optimizes network order.It is clustered and is calculated by the multi-layer network based on side intensity
Method remains the determinant attribute of primitive network as much as possible, and carries out visualization presentation to the network on different abstraction hierarchies, real
Now pass through the functional module in the relationship between the model analysis network node, and analysis power communication dispatch network, enliven
The purpose of the special behaviors such as node node or node set.
To obtain enough model trainings and testing the data on flows of station service power ISDN(Integrated Service Digital Network) backbone router,
Preferably, the data collection steps S101, can specifically:
Design corresponding program continuously acquires 30 days every the flow of acquisition in 15 minutes.
It is interfered as caused by data difference for removal, it is preferred that the data processing step S102, it can specifically:
All data are normalized to [0.1,0.9], and data set is divided into training set and test set.Preceding 500h is as network
Training sample, rear 760h is as network test sample.
To realize the prediction to network flow, it is preferred that the predicting network flow step S103, it can specifically:
By selecting suitable activation primitive, Loss function, hiding number of layers, it is pre- to establish high-precision LSTM neural network
Survey model.
Training neural network mainly passes through backpropagation cost, to reduce cost as guiding, adjusting parameter.Major parameter has
The biasing b of connection weight w and each neuron itself between neuron.According to gradient descent algorithm, along gradient direction
Adjusting parameter size.The gradient of w and b derives as follows:
Wherein z indicates the input of neuron, and σ indicates activation primitive.
To avoid gradient from disappearing, ReLuMax (0, x) function is set by activation primitive.
To ensure the good robustness of model, suitable Loss function is designed.Loss function is the core of empirical risk function
The component part that center portion is divided and structure risk function is important.The structure risk function of model includes empiric risk item and canonical
, it is indicated with following formula:
Wherein, what the mean function of front indicated is empirical risk function, and what L was represented is loss function.Loss function is logical
It often writes and is L (y_, y).Y_ represents predicted value, and y represents true value
The rule of thumb method choice hidden layer neuron number that formula and trial and error procedure combine, empirical equation used are as follows:
In formula: l indicates that node in hidden layer, N indicate that input layer number, M indicate that output layer number of nodes, a indicate to adjust
Constant (takes 1~10).
To detect to being built LSTM neural network model, evaluation need to be carried out to it.It is missed according to root mean square
Difference, which passes through prediction, to be evaluated, εRMSEEmbody the performance of model cootrol absolute error, calculation formula are as follows:
In formula: p (i) andThe respectively actual value and predicted value of network flow；N is prediction verify data number；I is
Future position sequence number, CapiOutput sequence maximum value.
To realize data visualization, it is preferred that the data visualization step S104, it can specifically:
According to powerline network visualize in be added network traffic information requirement, using be based on multi-level node clustering
Visual analyzing, power communication dispatching and monitoring data traffic is incorporated into the side intensity of communication network.Further, by being based on
The multi-layer network clustering algorithm of side intensity remains the determinant attribute of primitive network as much as possible, and to different abstraction hierarchies
On network carry out visualization presentation.
In specific embodiment two, according to inventive embodiments, provide a kind of based on the electric integrated of LSTM neural network
Data network network flow prediction method.
According to the concept of LSTM neural network, the parameter of LSTM model need to be selected.It establishes super for photovoltaic power
The LSTM network of short-term forecast it is main it needs to be determined that model 5 hyper parameters, i.e. input layer time step number, input layer dimension, hidden
Hide number, each hidden layer dimension and the output variable dimension of layer.
Input layer time step number is equal to the length for being used to carry out the time series variation of photovoltaic power prediction.Number in embodiment
It is differed between 80-150 according to daily sampling number；After many experiments, using 40 historical datas as a batch for pre-
It surveys；Input layer dimension, that is, variable number, amounts to 39 characteristic values, and input layer dimension is 40*39；Hiding number of layers, that is, LSTM layers of
Number, with increasing for hidden layer, in the case where training sample abundance, the nonlinear fitting ability of model is risen with it, but same
When, the complexity of model and the calculating of training and time cost will also increase.Predict that the present embodiment is only for photovoltaic power
Consider the prediction model of LSTM layers of list, therefore the parameter is set as 1；By repeatedly souning out, the dimension increase of hidden layer can be obtained
To preferable prediction effect；But since this prediction task is to predict the network flow of next step, output according to historical information
Dimension is set as 1.
By selecting suitable activation primitive, Loss function, hiding number of layers, it is pre- to establish high-precision LSTM neural network
Survey model.
Training neural network mainly passes through backpropagation cost, to reduce cost as guiding, adjusting parameter.Major parameter has
The biasing b of connection weight w and each neuron itself between neuron.According to gradient descent algorithm, along gradient direction
Adjusting parameter size.The gradient of w and b derives as follows:
Wherein z indicates the input of neuron, and σ indicates activation primitive.
To avoid gradient from disappearing, ReLuMax (0, x) function is set by activation primitive.
To ensure the good robustness of model, suitable Loss function is designed.Loss function is the core of empirical risk function
The component part that center portion is divided and structure risk function is important.The structure risk function of model includes empiric risk item and canonical
, it is indicated with following formula:
Wherein, what the mean function of front indicated is empirical risk function, and what L was represented is loss function.Loss function is logical
It often writes and is L (y_, y).Y_ represents predicted value, and y represents true value
The rule of thumb method choice hidden layer neuron number that formula and trial and error procedure combine, empirical equation used are as follows:
In formula: l indicates that node in hidden layer, N indicate that input layer number, M indicate that output layer number of nodes, a indicate to adjust
Constant (takes 1~10).
Network Traffic Forecast Model input layer number is 39 in the present embodiment, and output layer number of nodes is 1, then according to above-mentioned
Formula can primarily determine out node in hidden layer between 6~10 ranges.
To detect to being built LSTM neural network model, evaluation need to be carried out to it.It is missed according to root mean square
Difference, which passes through prediction, to be evaluated, εRMSEEmbody the performance of model cootrol absolute error, calculation formula are as follows:
In formula: p (i) andThe respectively actual value and predicted value of network flow；N is prediction verify data number；I is
Future position sequence number, CapiOutput sequence maximum value.
The precision of the electric integrated data network Network Traffic Forecast Model based on LSTM neural network of this experiment is not verified,
It is tested with data, as shown in Figure 3, Figure 4, the embodiment of the present invention effective the result is that:
From figure 3, it can be seen that horizontal axis is the number of iterations of the training of LSTM prediction model, the longitudinal axis is error rate, and is passed through
After the iteration of thousand even higher numbers, the error rate of LSTM data prediction is being gradually decreased, finally close to 0.And in Fig. 4, adopt
It is relatively coincide with the curve of the prediction of LSTM neural network model and actual value, although there is slightly inclined compared with actual value
Difference, but compared with traditional method, faster, precision of prediction also significantly improves training speed, is that a kind of good network flow is pre-
Survey model.
Referring to Fig. 5, Fig. 5 is shown as the communication network data visualization device based on volume forecasting of the application in a tool
Composition schematic diagram in body embodiment.The communication network data visualization device based on volume forecasting includes: data acquisition
Module 11, normalization module 12, prediction module 13 and visualization model 14.
The data acquisition module 11 according to data of the preset time cycle to integrated service network flow to carry out
Acquisition is to obtain acquired original data；
The normalization module 12 is to be normalized to obtain normalization data the acquired original data；
The prediction module 13 is to establish LSTM Network Traffic Forecast Model to described in conjunction with the normalization data
Integrated service network flow is predicted；
The visualization model 14 is to establish Visualization Model based on multi-level node clustering algorithm, to described original
Acquisition data carry out visualization presentation.
The communication network data visualization device based on volume forecasting be and the communication network based on volume forecasting
The corresponding device item of network data visualization method, all about communication network data method for visualizing based on volume forecasting
Description, can be applied in the present embodiment, be not added repeat herein.
Referring to Fig. 6, Fig. 6 is shown as the composition schematic diagram of the equipment of the application in one embodiment.The equipment
Including memory 21 and processor 22；The memory 21 stores computer program；The processor 22 runs the memory
The computer program of 21 storages is to enable the equipment execute the communication network data based on volume forecasting as shown in Figure 1
Method for visualizing.The equipment is the electronic equipment with data intelligence processing function, for example, computer.
In conclusion communication network data method for visualizing, device and the equipment of the invention based on volume forecasting, utilizes
LSTM Network Traffic Forecast Model predicts integrated service network flow, and being established based on multi-level node clustering algorithm can
Depending on changing model, to carry out visualization presentation to the acquired original data.While realizing data visualization, it is also based on LSTM
Accurate Prediction of the network implementations to the integrated service network flow of data.The intelligence of increase system, the instruction for promoting prediction model
Practice speed and improves precision of prediction.So the present invention effectively overcomes various shortcoming in the prior art and has high industrial benefit
With value.
The above-described embodiments merely illustrate the principles and effects of the present invention, and is not intended to limit the present invention.It is any ripe
The personage for knowing this technology all without departing from the spirit and scope of the present invention, carries out modifications and changes to above-described embodiment.Cause
This, institute is complete without departing from the spirit and technical ideas disclosed in the present invention by those of ordinary skill in the art such as
At all equivalent modifications or change, should be covered by the claims of the present invention.
Claims (11)
1. a kind of communication network data method for visualizing based on volume forecasting characterized by comprising
It is acquired according to data of the preset time cycle to integrated service network flow to obtain acquired original data；
The acquired original data are normalized to obtain normalization data；
In conjunction with the normalization data, it is pre- to carry out to the integrated service network flow to establish LSTM Network Traffic Forecast Model
It surveys；
Visualization Model is established based on multi-level node clustering algorithm, to carry out visualization presentation to the acquired original data.
2. the communication network data method for visualizing according to claim 1 based on volume forecasting, which is characterized in that described
Normalized includes: to carry out linear transformation to the acquired original data by the standardized transfer function of min-max, to obtain
Obtain the normalization data in [0,1] section, the transfer function are as follows:
Wherein, x is the initial data before normalization, x*For the normalization data after the initial data x is normalized, max
For the maximum value in acquired original data, min is the minimum value in acquired original data.
3. the communication network data method for visualizing according to claim 1 based on volume forecasting, which is characterized in that determine
About the activation primitive, loss function and the implicit number of plies of the LSTM Network Traffic Forecast Model, and in conjunction with the normalization
Data establish the LSTM Network Traffic Forecast Model to predict the integrated service network flow.
4. the communication network data method for visualizing according to claim 3 based on volume forecasting, which is characterized in that described
Activation primitive is to correct linear unit function.
5. the communication network data method for visualizing according to claim 1 based on volume forecasting, which is characterized in that determine
About the input layer parameter, output layer parameter and hiding layer parameter of the LSTM Network Traffic Forecast Model, and in conjunction with described
Normalization data establishes the LSTM Network Traffic Forecast Model to predict the integrated service network flow；Wherein,
The input layer parameter includes input layer time step number and input layer dimension；The output layer parameter includes output variable dimension；
The hiding layer parameter includes hiding number of layers and each hidden layer dimension.
6. the communication network data method for visualizing according to claim 5 based on volume forecasting, which is characterized in that described
Inputting layer parameter further includes input layer number, and the output layer parameter further includes output layer number of nodes, the hiding layer parameter
It further include hidden layer number of nodes, and according to the regulating constant of the input layer number, the output layer number of nodes and setting,
The hidden layer number of nodes is calculated.
7. the communication network data method for visualizing according to claim 1 based on volume forecasting, which is characterized in that described
Method further include:
Continue to be acquired according to data of the time cycle to integrated service network flow to obtain new acquisition data；
The error of the LSTM Network Traffic Forecast Model is analyzed in conjunction with the new acquisition data.
8. the communication network data method for visualizing according to claim 7 based on volume forecasting, which is characterized in that according to
Following formula obtains the root-mean-square error of the prediction result about the LSTM Network Traffic Forecast Model:
Wherein, p (i) andThe respectively actual value and predicted value of network flow；N is prediction verify data number；I is prediction
Point sequence number, CapiOutput sequence maximum value.
9. the communication network data method for visualizing according to claim 1 based on volume forecasting, which is characterized in that be based on
The multi-level node clustering algorithm of side intensity establishes Visualization Model, to carry out visualization presentation to the acquired original data.
10. a kind of communication network data visualization device based on volume forecasting characterized by comprising
Data acquisition module, to be acquired according to data of the preset time cycle to integrated service network flow to obtain
Acquired original data；
Module is normalized, to be normalized the acquired original data to obtain normalization data；
Prediction module, to establish LSTM Network Traffic Forecast Model to the integrated service in conjunction with the normalization data
Network flow is predicted；
Visualization model, to establish Visualization Model based on multi-level node clustering algorithm, to the acquired original data
Carry out visualization presentation.
11. a kind of equipment characterized by comprising memory and processor；
The memory stores computer program；The computer program that the processor runs the memory storage is described to enable
Electronic equipment executes such as the communication network data method for visualizing according to any one of claims 1 to 9 based on volume forecasting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811524648.3A CN109639524A (en) | 2018-12-13 | 2018-12-13 | Communication network data method for visualizing, device and equipment based on volume forecasting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811524648.3A CN109639524A (en) | 2018-12-13 | 2018-12-13 | Communication network data method for visualizing, device and equipment based on volume forecasting |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109639524A true CN109639524A (en) | 2019-04-16 |
Family
ID=66073514
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811524648.3A Pending CN109639524A (en) | 2018-12-13 | 2018-12-13 | Communication network data method for visualizing, device and equipment based on volume forecasting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109639524A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110223375A (en) * | 2019-05-07 | 2019-09-10 | 深圳供电局有限公司 | Method and system for processing 3D visual data of machine room |
| CN110460605A (en) * | 2019-08-16 | 2019-11-15 | 南京邮电大学 | A kind of Abnormal network traffic detection method based on autocoding |
| CN111200531A (en) * | 2020-01-02 | 2020-05-26 | 国网冀北电力有限公司信息通信分公司 | Communication network traffic prediction method and device and electronic equipment |
| CN112311791A (en) * | 2020-10-29 | 2021-02-02 | 贵州电网有限责任公司 | Statistical method and system suitable for office business flow |
| CN112769972A (en) * | 2020-12-22 | 2021-05-07 | 赛尔网络有限公司 | Flow analysis method and device for IPv6 network, electronic equipment and storage medium |
| CN114424133A (en) * | 2019-09-23 | 2022-04-29 | 西门子股份公司 | Extended trend display of process data and secondary alarms |
| CN114449466A (en) * | 2021-12-16 | 2022-05-06 | 深圳联想懂的通信有限公司 | User charging intelligent monitoring method and device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1764126A (en) * | 2005-11-11 | 2006-04-26 | 上海交通大学 | Method for detecting and monitoring gusty abnormal network flow |
| CN106022614A (en) * | 2016-05-22 | 2016-10-12 | 广州供电局有限公司 | Data mining method of neural network based on nearest neighbor clustering |
| CN106789297A (en) * | 2016-12-29 | 2017-05-31 | 淮海工学院 | Predicting network flow system and its method for predicting based on neutral net |
| CN108900346A (en) * | 2018-07-06 | 2018-11-27 | 西安电子科技大学 | Wireless network traffic prediction technique based on LSTM network |
-
2018
- 2018-12-13 CN CN201811524648.3A patent/CN109639524A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1764126A (en) * | 2005-11-11 | 2006-04-26 | 上海交通大学 | Method for detecting and monitoring gusty abnormal network flow |
| CN106022614A (en) * | 2016-05-22 | 2016-10-12 | 广州供电局有限公司 | Data mining method of neural network based on nearest neighbor clustering |
| CN106789297A (en) * | 2016-12-29 | 2017-05-31 | 淮海工学院 | Predicting network flow system and its method for predicting based on neutral net |
| CN108900346A (en) * | 2018-07-06 | 2018-11-27 | 西安电子科技大学 | Wireless network traffic prediction technique based on LSTM network |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110223375A (en) * | 2019-05-07 | 2019-09-10 | 深圳供电局有限公司 | Method and system for processing 3D visual data of machine room |
| CN110223375B (en) * | 2019-05-07 | 2023-02-03 | 深圳供电局有限公司 | Method and system for processing 3D visual data of machine room |
| CN110460605A (en) * | 2019-08-16 | 2019-11-15 | 南京邮电大学 | A kind of Abnormal network traffic detection method based on autocoding |
| CN114424133A (en) * | 2019-09-23 | 2022-04-29 | 西门子股份公司 | Extended trend display of process data and secondary alarms |
| CN111200531A (en) * | 2020-01-02 | 2020-05-26 | 国网冀北电力有限公司信息通信分公司 | Communication network traffic prediction method and device and electronic equipment |
| CN111200531B (en) * | 2020-01-02 | 2022-01-04 | 国网冀北电力有限公司信息通信分公司 | Communication network traffic prediction method and device and electronic equipment |
| CN112311791A (en) * | 2020-10-29 | 2021-02-02 | 贵州电网有限责任公司 | Statistical method and system suitable for office business flow |
| CN112769972A (en) * | 2020-12-22 | 2021-05-07 | 赛尔网络有限公司 | Flow analysis method and device for IPv6 network, electronic equipment and storage medium |
| CN112769972B (en) * | 2020-12-22 | 2023-02-28 | 赛尔网络有限公司 | Flow analysis method and device for IPv6 network, electronic equipment and storage medium |
| CN114449466A (en) * | 2021-12-16 | 2022-05-06 | 深圳联想懂的通信有限公司 | User charging intelligent monitoring method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109639524A (en) | Communication network data method for visualizing, device and equipment based on volume forecasting | |
| CN108520357B (en) | Method and device for judging line loss abnormality reason and server | |
| CN107077642B (en) | System and method for solving problems usable for quantum computing | |
| CN103257921B (en) | Improved random forest algorithm based system and method for software fault prediction | |
| CN109919684A (en) | For generating method, electronic equipment and the computer readable storage medium of information prediction model | |
| CN109214719A (en) | A kind of system and method for the marketing inspection analysis based on artificial intelligence | |
| CN112686464A (en) | Short-term wind power prediction method and device | |
| CN109145949A (en) | Non-intrusive electrical load monitoring and decomposition method and system based on integrated study | |
| CN107423839A (en) | A kind of method of the intelligent building microgrid load prediction based on deep learning | |
| CN114781254B (en) | Construction method and device of electromagnetic exploration inversion model and storage medium | |
| CN103198139B (en) | The energy analysis method of custom power data | |
| CN109117352B (en) | Server performance prediction method and device | |
| CN107679734A (en) | It is a kind of to be used for the method and system without label data classification prediction | |
| CN109767054A (en) | Efficiency cloud appraisal procedure and edge efficiency gateway based on deep neural network algorithm | |
| CN106651007A (en) | Method and device for GRU-based medium and long-term prediction of irradiance of photovoltaic power station | |
| Feng et al. | [Retracted] Design and Simulation of Human Resource Allocation Model Based on Double‐Cycle Neural Network | |
| CN109449923B (en) | Quantitative analysis method for operation flexibility of active power distribution system and related product | |
| CN114707754A (en) | Intelligent ammeter fault prediction method and system based on BiLSTM-CNN model | |
| CN109633448A (en) | Identify the method, apparatus and terminal device of cell health state | |
| CN114565038A (en) | Intelligent electric meter fault diagnosis method based on improved capsule network | |
| CN110082822A (en) | The method for carrying out earthquake detection using convolutional neural networks | |
| CN109787821A (en) | A kind of Large-scale Mobile customer traffic consumption intelligent Forecasting | |
| CN115545280A (en) | Low-voltage distribution network topology generation method and device | |
| Yi et al. | An Effective Approach for determining Rock Discontinuity sets using a modified Whale optimization Algorithm | |
| Dong | Application of Big Data Mining Technology in Blockchain Computing |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |