CN117060984B - Satellite network flow prediction method based on empirical mode decomposition and BP neural network - Google Patents
Satellite network flow prediction method based on empirical mode decomposition and BP neural network Download PDFInfo
- Publication number
- CN117060984B CN117060984B CN202311287098.9A CN202311287098A CN117060984B CN 117060984 B CN117060984 B CN 117060984B CN 202311287098 A CN202311287098 A CN 202311287098A CN 117060984 B CN117060984 B CN 117060984B
- Authority
- CN
- China
- Prior art keywords
- sequence
- satellite network
- model
- prediction
- adopting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 26
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 17
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 12
- 241000282461 Canis lupus Species 0.000 claims abstract description 9
- 238000005457 optimization Methods 0.000 claims abstract description 9
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 claims description 6
- 238000005315 distribution function Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 20
- 238000012549 training Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 13
- 238000004364 calculation method Methods 0.000 description 11
- 238000010606 normalization Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 241000282421 Canidae Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005291 chaos (dynamical) Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/14—Relay systems
- H04B7/15—Active relay systems
- H04B7/185—Space-based or airborne stations; Stations for satellite systems
- H04B7/1851—Systems using a satellite or space-based relay
- H04B7/18519—Operations control, administration or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Environmental & Geological Engineering (AREA)
- Aviation & Aerospace Engineering (AREA)
- Astronomy & Astrophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Physiology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Radio Relay Systems (AREA)
Abstract
The invention relates to the technical field of satellite communication, and particularly discloses a satellite network flow prediction method based on empirical mode decomposition and BP neural network, which comprises the following steps: s10, generating a time sequence according to an ON/OFF source model meeting Pareto distribution; s20, decomposing the time sequence into multi-order inclusion modal components with short-range correlation by adopting empirical mode decomposition; s30, respectively establishing a prediction model for each order of connotation modal components by adopting an autoregressive moving average model; s40, predicting the connotation modal components by adopting each-order prediction model, and summarizing to obtain a linear characteristic sequence; s50, subtracting the linear characteristic sequence from the original flow sequence to obtain a residual sequence of the satellite network flow; s60, optimizing the BP neural network by adopting a wolf algorithm, and constructing an optimization model; s70, predicting the residual sequence according to the optimization model to obtain a residual sequence prediction result; s80, determining a satellite network flow prediction result according to the linear characteristic sequence and the residual sequence prediction result.
Description
Technical Field
The invention relates to the technical field of satellite communication, in particular to a satellite network flow prediction method based on empirical mode decomposition and BP neural network.
Background
The low orbit satellite communication network has the characteristics of wide coverage range and large communication capacity, can break through the limitation of geographic conditions by means of inter-satellite networking, realize uninterrupted signal coverage and provide large-broadband, low-delay and seamless network access service for global users.
In recent years, with the construction of a huge low orbit satellite constellation, the satellite network traffic demand presents a rapid rising trend, and the network bandwidth resources are increasingly tensed. The satellite-borne computing resources and storage resources of the low-orbit satellite are limited by the power consumption and the volume of a satellite platform, and the available bandwidth is limited; meanwhile, with the continuous switching of inter-satellite links (ISLs) between satellites, the low-orbit satellite network topology changes drastically with time, and the traffic of the satellite network presents temporal and spatial non-uniformity. Under the condition that the scale and the flow demand of users are continuously expanded, the factors cause network congestion of the low-orbit satellite network to influence the service quality of the network. By means of flow prediction, the characteristics and the trend of the change of the network flow can be predicted in advance, and the network flow control is changed from a passive response mode to an active sensing mode.
Conventional network traffic models are generally based on poisson processes, including poisson models, markov models, etc., which describe only short correlations of traffic in the time domain. For satellite network traffic that exhibits long correlation processes, it is difficult for conventional models to accurately characterize the network. Since 1994, various traffic prediction models based on self-similarity have been proposed after the self-similarity characteristics of network traffic were discovered. One class is data features observed by constructing a physical model description, including an ON/OFF model of heavy tail distribution,Queuing models, etc.; the other is a statistical model, which model triesThe graph simulates the change trend of network data through a data fitting method. These self-similar traffic prediction models differ from the traditional prediction models in that: the self-similarity forecasting model is established on the basis of network characteristics, can describe the LRD and the burstiness of the flow, is beneficial to forecasting according to the internal law of the network flow, improves the forecasting precision, and has complex and time-consuming calculation process.
With the development of machine learning, neural network models, fuzzy theory, chaos theory and the like have good nonlinear mapping capability, so that the characteristics of network traffic can be better represented, and the performance of network traffic prediction is improved. However, the related algorithm has the problems of a large number of optional parameters and high calculation complexity, and is not suitable for a satellite network platform. Because a single model can only characterize the poisson process or self-similarity of network traffic, satellite network traffic cannot be well described, and many scholars have proposed hybrid models. And the EMD or wavelet model is utilized to decompose the network flow, and then a prediction model is applied to each obtained component, so that the self-similarity characteristic of the network flow is extracted step by step, the calculation complexity is effectively reduced, and the prediction precision is improved. The reliability of the predictions for each component using different models is not fully demonstrated.
Disclosure of Invention
Aiming at the problems, the invention aims to provide a satellite network flow prediction method based on empirical mode decomposition and BP neural network, wherein the prediction precision is superior to that of a traditional satellite network flow prediction model, and the method has higher solving precision and lower calculation complexity.
The invention provides a satellite network flow prediction method based on empirical mode decomposition and BP neural network, which comprises the following steps:
step S10, generating a time sequence of satellite network traffic according to an ON/OFF source model meeting Pareto distribution; the ON/OFF source model meeting Pareto distribution is a data generation model of end-to-end connection in a satellite network;
s20, decomposing the time sequence into multi-order inclusion modal components with short-range correlation by adopting empirical mode decomposition;
step S30, respectively establishing a prediction model for the content modal components of each order by adopting an autoregressive moving average model;
step S40, predicting the connotation modal components by adopting the prediction models of all orders, and summarizing the prediction results to obtain a linear characteristic sequence of the satellite network flow;
step S50, subtracting the linear characteristic sequence from the original flow sequence of the satellite network flow to obtain a residual sequence of the satellite network flow;
step S60, adopting a gray wolf algorithm to optimize the BP neural network, and constructing an optimization model;
step S70, predicting the residual sequence according to the optimization model to obtain a residual sequence prediction result;
and S80, determining a satellite network flow prediction result according to the linear characteristic sequence and the residual sequence prediction result.
In a possible implementation manner, in S10, the probability distribution function of the Pareto distribution is the following formula:
；
wherein,is a positive integer representing a random variableMinimum value that can be taken;determining the mean value of random variablesVariance of random variablesThe method comprises the steps of carrying out a first treatment on the surface of the If it isAverage of Pareto distributionPresence, varianceThere is no upper bound.
In one possible implementation, the S20 includes:
step S21, the data to be analyzed in the ON/OFF source model is processedRespectively fitting all extreme points of the obtained sample by using two cubic spline curves to obtain data to be analyzedIs an extremum envelope of (a);
step S22, making the average value of the extreme value envelope beResidual signalThe method comprises the steps of carrying out a first treatment on the surface of the If it isIf the IMF condition is satisfiedFor the first IMF component, otherwiseAs a means of；
Step S23, throughkAfter the round of iteration, the difference between the obtained signal and the envelope mean value, thekThe difference obtained by 1 iteration isThe method comprises the steps of carrying out a first treatment on the surface of the When the following formula is established, the following will be followedAs a first connotation modality component:
；
wherein,as a result of the threshold value being set,Tnumber of data samples, i.e. passkAfter the round of iterationWith previous iteration resultThe root mean square difference between them is less than a threshold valueWill thenA first connotation modality component that is a satisfying condition;
step S24, willAs a means ofRepeating the above steps; stopping calculation when the residual quantity is a monotonic function and the amplitude is smaller than the threshold value to obtain a plurality of IMF componentsThe data to be analyzed is obtained by the following formula:
；
in the method, in the process of the invention,is the final residual amount.
In one possible implementation, the S30 includes:
determining the predictive model according to the following formula:
；
in the method, in the process of the invention,model parameters for the autoregressive moving average model, i.e. fitting the sum of squares of residuals,the independent error term, the autoregressive order, the differential order and the moving average order are respectively obtained,as an inclusion mode component, the component of the content mode,to fit the standard deviation of the content modality component sequences, AIC is the red pool information criterion.
In one possible implementation, the S50 includes:
the residual sequence is calculated as follows according to the following formula:
；
in the method, in the process of the invention,for the residual sequence to be a sequence of residues,for the original sequence of traffic flows,is a linear characteristic sequence.
In one possible implementation, the S60 includes:
step S61, initializing algorithm parameters of a gray wolf algorithm and parameters of a BP neural network;
step S62, carrying out normalization processing on the parameters to obtain normalized parameters, and dividing the normalized parameters into a training data set and a test data set;
step S63, training the BP neural network according to the training data set to obtain a plurality of training models;
step S64, evaluating the performance of the training model according to the fitness of the individuals in the test data set;
step S65, grading the sirius population according to the fitness, reserving the positions of individuals with the best fitness, and updating the positions of the rest individuals;
and step S66, if the iteration times reach a preset value, taking the individual position with the optimal fitness as the optimal parameter of the BP neural network model, and constructing a prediction model.
In one possible implementation, the S62 includes:
normalization is performed according to the following formula:
；
in the method, in the process of the invention,in order to normalize the parameters,is the firstThe original parameters of the data are used to determine,andrespectively, the maximum and minimum values of the corresponding original parameters.
In one possible implementation, the step S63 includes:
training the BP neural network according to the following formula to obtain a plurality of training models;
；
in the method, in the process of the invention,respectively isAnd (3) withThe distance between the two adjacent plates is equal to the distance between the two plates,is thatThe current location of the current location is indicated,as a random vector of values,for the optimal solution to be a solution that is optimal,and connecting weights for each layer of the BP neural network.
In one possible implementation, the S64 includes:
the fitness of the individual is calculated from the mean square error, which is calculated according to the following formula:
；
in the method, in the process of the invention,Tthe sequence length of the satellite network traffic is the sequence length of the satellite network traffic;is thatIs the first to fit the residual sequencetA value.
In one possible implementation, the step S65 includes:
updating the location of the individual according to the following formula:
；
；
in the method, in the process of the invention,respectively isAnd (3) withThe distance between the two adjacent plates is equal to the distance between the two plates,respectively isThe current location of the current location is indicated,is thatThe current location of the current location is indicated,as a random vector of values,for the number of iterations,the positions of the three operators are respectively.
According to the satellite network flow prediction method based on the empirical mode decomposition and the BP neural network, the satellite network flow with self-similarity is decomposed into multiple orders of IMF components with short-range correlation through the empirical mode decomposition, the IMF components are predicted by adopting ARIMA improved by the self-adaptive order-determining optimizing operator, and the calculation complexity is reduced. Meanwhile, the improved gray wolf algorithm is adopted to optimize BP neural network parameters, the optimized BP neural network is used for forecasting residual errors of an EMD-ARIMA model, a satellite network flow forecasting result is finally obtained, and the forecasting precision is superior to that of a traditional satellite network flow forecasting model, so that the method has higher solving precision and lower calculation complexity.
Drawings
Fig. 1 is a flow chart of a satellite network flow prediction method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a satellite network traffic prediction method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a satellite network architecture according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of an ON/OFF source superposition model subject to Pareto distribution according to an embodiment of the present invention;
FIG. 5 is a time-series data diagram of satellite network traffic provided by an embodiment of the present invention;
FIG. 6 is a schematic diagram of the decomposed IMF components of the various orders provided by embodiments of the present invention;
fig. 7 is a schematic diagram of a satellite network traffic prediction result according to an embodiment of the present invention.
Detailed Description
Embodiments of the present invention are described in further detail below with reference to the accompanying drawings and examples. The following detailed description of the embodiments and the accompanying drawings are provided to illustrate the principles of the invention and are not intended to limit the scope of the invention, i.e. the invention is not limited to the preferred embodiments described, which is defined by the claims.
In the description of the present invention, it is to be noted that, unless otherwise indicated, the meaning of "plurality" means two or more; the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance; the specific meaning of the above terms in the present invention can be understood as appropriate by those of ordinary skill in the art.
The satellite network traffic prediction can provide key information for the routing and resource allocation of the satellite network, and has important significance for the efficient operation of the satellite communication network. However, the satellite network traffic has self-similarity and long-range correlation, and the conventional linear or nonlinear network traffic prediction model cannot achieve enough prediction accuracy. The satellite network flow prediction method based on empirical mode decomposition and BP neural network has prediction accuracy superior to that of the traditional satellite network flow prediction model, and has higher solving accuracy and lower calculation complexity.
Fig. 1 is a flow chart of a satellite network flow prediction method provided by an embodiment of the present invention, and fig. 2 is a schematic diagram of a satellite network flow prediction method provided by an embodiment of the present invention, as shown in fig. 1 and fig. 2, the present invention provides a satellite network flow prediction method based on empirical mode decomposition and BP neural network, including:
step S10, generating a time sequence of satellite network traffic according to an ON/OFF source model meeting Pareto distribution;
fig. 3 is a schematic structural diagram of a satellite network system according to an embodiment of the present invention, where, as shown in fig. 3, the satellite network to which the present invention is applied is a low-orbit satellite network, including a space network and a ground network. Each satellite of the space network is connected with the same-orbit adjacent satellite and the left-right adjacent orbit satellite through four inter-satellite links. The space network is connected with the gateway station of the ground network through a feed link to form a satellite network.
The ON/OFF source model meeting Pareto distribution is a data generation model of end-to-end connection in a satellite network. The ON period is considered as the size of the packet satisfying the Pareto profile and the OFF period is considered as the interval time of packet transmission satisfying the Pareto profile. Pareto distribution is a classical heavy tail distribution. The end-to-end connection in the satellite network is considered as an ON/OFF source satisfying the Pareto distribution. The method comprises the steps that (1) a source node generates a data packet at a constant rate and keeps mutually independent when a message corresponding to ON is sent; and OFF corresponds to the interruption of the message transmission, and is also independent and distributed after a sufficient number of samples are accumulated.
The probability distribution function of the Pareto distribution is as follows:
；
wherein,is a positive integer representing a random variableMinimum value that can be taken;determining the mean value of random variablesVariance of random variablesThe method comprises the steps of carrying out a first treatment on the surface of the If it isAverage of Pareto distributionPresence, varianceThere is no upper bound.
Fig. 4 is a schematic diagram of an ON/OFF source superposition model obeying Pareto distribution provided by an embodiment of the present invention, and fig. 5 is a satellite network traffic time sequence data diagram provided by an embodiment of the present invention.
Step S20, decomposing the time sequence into multiple orders of inclusion modal components with short-range correlation by adopting Empirical Mode Decomposition (EMD);
in one possible implementation, S20 includes:
step S21, the data to be analyzed in the ON/OFF source modelRespectively fitting all extreme points of the obtained sample by using two cubic spline curves to obtain data to be analyzedIs an extremum envelope of (a);
step S22, making the average value of the extreme value envelope beResidual signalThe method comprises the steps of carrying out a first treatment on the surface of the If it isIf the IMF condition is satisfiedFor the first IMF component, otherwiseAs a means of；
Step S23, after k rounds of iteration, the difference between the obtained signal and the envelope mean value, thekThe difference obtained by 1 iteration isThe method comprises the steps of carrying out a first treatment on the surface of the When the following formula is established, the following will be followedAs a first connotation modality component:
；
wherein,as a result of the threshold value being set,Tnumber of data samples, i.e. passkAfter the round of iterationWith previous iteration resultThe root mean square difference between them is less than a threshold valueWill thenA first connotation modality component that is a satisfying condition;
step S24, willAs a means ofRepeating the above steps; stopping calculation when the residual quantity is a monotonic function and the amplitude is smaller than the threshold value to obtain a plurality of IMF componentsThe data to be analyzed is obtained by the following formula:
；
in the method, in the process of the invention,is the final residual amount.
Fig. 6 is a schematic diagram of 10 IMF components obtained by decomposing the time series of fig. 5 through an empirical mode, and fig. 6 is a schematic diagram of IMF components of each order obtained by decomposing provided in the embodiment of the present invention, a, b, c, d, e, f, g, h, i, j are IMF1, IMF2, IMF3, IMF4, IMF5, IMF6, IMF7, IMF8, IMF9, IMF10, respectively.
Step S30, respectively establishing a prediction model for each order of connotation modal components by adopting an autoregressive moving average model;
in one possible implementation, S30 includes:
determining a predictive model according to the following formula:
；
in the method, in the process of the invention,model parameters for the autoregressive moving average model, i.e. fitting the sum of squares of residuals,the independent error term, the autoregressive order, the differential order and the moving average order are respectively obtained,as an inclusion mode component, the component of the content mode,to fit the standard deviation of the content modality component sequences, AIC is the red pool information criterion.
For the decomposed n-order IMF components, an ARIMA model is adopted to respectively establish a prediction model ARIMA (pi, di, qi) for each order component. Where (pi, di, qi) is ARIMA model parameter, i=1, 2, …, n. According to fig. 6, the original flow data is decomposed to obtain 10 IMFs, and the optimal model of each IMF determined by the adaptive fixed-order optimization is shown in table 1:
TABLE 1
Step S40, predicting the connotation modal components by adopting each-order prediction model, and summarizing the prediction results to obtain a linear characteristic sequence of the satellite network flow;
step S50, subtracting the linear characteristic sequence from the original flow sequence of the satellite network flow to obtain a residual sequence of the satellite network flow;
in one possible implementation, S50 includes:
the residual sequence is calculated as follows according to the following formula:
；
in the method, in the process of the invention,for the residual sequence to be a sequence of residues,for the original sequence of traffic flows,is a linear characteristic sequence.
Step S60, adopting a gray wolf algorithm to optimize the BP neural network, and constructing an optimization model IGWO-BPNN;
in one possible implementation, S60 includes: steps S61-S66.
Step S61, initializing algorithm parameters of a gray wolf algorithm IGWO and parameters of a BP neural network;
in one possible implementation, in the wolf algorithm GWO, wolves are divided into four groups from top to bottom according to fitness values:whereinFor the leader level (optimal solution), candidate solutions surroundAnd (5) performing position updating. The invention determines that the population scale of the wolf is set as M; the maximum iteration number isThe method comprises the steps of carrying out a first treatment on the surface of the The upper and lower boundaries of the j-th dimension are set. Each layer of connection weight of the neural network is as followsThe error threshold isThe number of neurons in the hidden layer is。
Step S62, carrying out normalization processing on the parameters to obtain normalized parameters, and dividing the normalized parameters into a training data set and a test data set;
in order to eliminate the influence of the input flow residual sequence dimension on the model result, the data is subjected to normalization processing. Normalizing data toAnd is divided into a training data set and a test data set.
In one possible implementation, the normalization is performed according to the following formula:
；
in the method, in the process of the invention,in order to normalize the parameters,is the firstThe original parameters of the data are used to determine,andrespectively, the maximum and minimum values of the corresponding original parameters.
Before training, the termination condition (the current solution is the minimum) is determined as: if one value is satisfied that remains unchanged for successive iterations, it is deemed to be the minimum value. The selected sample performs the following steps S63-S66 until the termination condition is satisfied, and then exits.
Step S63, training the BP neural network according to the training data set to obtain a plurality of training models;
in one possible implementation, each individual location contains BPNN parameters, and the BP neural network is trained according to the following formula, resulting in a plurality of training models;
；
in the method, in the process of the invention,respectively isAnd (3) withThe distance between the two adjacent plates is equal to the distance between the two plates,is thatThe current location of the current location is indicated,as a random vector of values,for the optimal solution to be a solution that is optimal,and connecting weights for each layer of the BP neural network.
Step S64, evaluating the performance of the training model according to the fitness of individuals in the test data set;
in one possible implementation, the fitness of the individual is calculated from the mean square error, which is calculated according to the following formula:
；
in the method, in the process of the invention,Tthe sequence length of the satellite network traffic is the sequence length of the satellite network traffic;is thatIs the first to fit the residual sequencetA value.
Step S65, grading the sirius population according to the fitness, reserving the positions of individuals with the best fitness, and updating the positions of the rest individuals;
in one possible implementation, the location of the individual is updated according to the following formula:
；
；
in the method, in the process of the invention,respectively isAnd (3) withThe distance between the two adjacent plates is equal to the distance between the two plates,respectively isThe current location of the current location is indicated,is thatThe current location of the current location is indicated,as a random vector of values,for the number of iterations,respectively are provided withIs the position of three types of operators.
Step S66, if the iteration number reaches the preset value, namelyAnd (5) ending the parameter optimization process and jumping out of the loop. Individual location with optimal fitnessAnd constructing a prediction model as an optimal parameter of the BP neural network model.
Step S70, predicting the residual sequence according to the optimization model to obtain a residual sequence prediction result;
and S80, determining a satellite network flow prediction result according to the linear characteristic sequence and the residual sequence prediction result.
In one possible implementation, the satellite network traffic prediction result is determined according to the following formula:
；
in the method, in the process of the invention,as a result of the prediction of the satellite network traffic,as a sequence of linear features,and predicting a result for the residual sequence.
The accuracy index of the prediction result of the EMD-ARIMA-BPNN model is shown in Table 2:
TABLE 2
The calculation formulas of the indexes in the table are as follows:
；
；
wherein,Tis the length of the flow sequence;as a residual of the traffic sequence,is the predicted value of the traffic sequence residual,is the average value of the predicted value of the residual error of the flow sequence.
Fig. 7 is a schematic diagram of a satellite network traffic prediction result according to an embodiment of the present invention. The method provided by the invention has good prediction performance on satellite network flow data with long-range correlation characteristics, and the predicted network flow is close to the actual flow data.
According to the satellite network flow prediction method based on empirical mode decomposition and BP neural network, self-similarity characteristics of satellite network flow are quantitatively analyzed, and the short-range correlation of multi-order IMF components obtained by EMD decomposition of the satellite network flow is respectively proved through theoretical analysis and experiments. The satellite network flow with self-similarity is decomposed into multiple orders of IMF components with short-range correlation through EMD, the IMF components are predicted by ARIMA improved by the self-adaptive fixed order optimizing operator, and the calculation complexity is reduced. And meanwhile, optimizing the BPNN network weight by adopting an IGWO algorithm, and forecasting the residual error of the EMD-ARIMA model by using the optimized BPNN to finally obtain a satellite network flow forecasting result. Compared with the traditional flow forecasting model and the mixed model, the EMD-ARIMA-BPNN model has higher forecasting precision on satellite network flow, can describe the stepwise change trend of the satellite network flow, and can meet the high-efficiency forecasting of the satellite network flow.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (2)
1. The satellite network flow prediction method based on empirical mode decomposition and BP neural network is characterized by comprising the following steps of:
step S10, generating a time sequence of satellite network traffic according to an ON/OFF source model meeting Pareto distribution; the ON/OFF source model meeting Pareto distribution is a data generation model of end-to-end connection in a satellite network;
s20, decomposing the time sequence into multi-order inclusion modal components with short-range correlation by adopting empirical mode decomposition;
step S30, respectively establishing a prediction model for the content modal components of each order by adopting ARIMA improved by the self-adaptive order-determining optimizing operator;
step S40, predicting the connotation modal components by adopting the prediction models of all orders, and summarizing the prediction results to obtain a linear characteristic sequence of the satellite network flow;
step S50, subtracting the linear characteristic sequence from the original flow sequence of the satellite network flow to obtain a residual sequence of the satellite network flow;
step S60, adopting an improved gray wolf algorithm IGWO to optimize the BP neural network, and constructing an optimized model IGWO-BPNN;
step S70, predicting the residual sequence according to the optimization model IGWO-BPNN to obtain a residual sequence prediction result;
and S80, determining a satellite network flow prediction result according to the linear characteristic sequence and the residual sequence prediction result.
2. The method according to claim 1, wherein in S10, the probability distribution function of the Pareto distribution is the following formula:
where k is a positive integer, representing the minimum value that the random variable x can take.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311287098.9A CN117060984B (en) | 2023-10-08 | 2023-10-08 | Satellite network flow prediction method based on empirical mode decomposition and BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311287098.9A CN117060984B (en) | 2023-10-08 | 2023-10-08 | Satellite network flow prediction method based on empirical mode decomposition and BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117060984A CN117060984A (en) | 2023-11-14 |
| CN117060984B true CN117060984B (en) | 2024-01-09 |
Family
ID=88666595
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311287098.9A Active CN117060984B (en) | 2023-10-08 | 2023-10-08 | Satellite network flow prediction method based on empirical mode decomposition and BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117060984B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109802862A (en) * | 2019-03-26 | 2019-05-24 | 重庆邮电大学 | A kind of combined network flow prediction method based on set empirical mode decomposition |
| CN111030889A (en) * | 2019-12-24 | 2020-04-17 | 国网河北省电力有限公司信息通信分公司 | Network traffic prediction method based on GRU model |
| CN113206756A (en) * | 2021-04-22 | 2021-08-03 | 大连大学 | Network flow prediction method based on combined model |
| CN113205698A (en) * | 2021-03-24 | 2021-08-03 | 上海吞山智能科技有限公司 | Navigation reminding method based on IGWO-LSTM short-time traffic flow prediction |
| CN113905391A (en) * | 2021-09-27 | 2022-01-07 | 湖北工业大学 | Ensemble learning network traffic prediction method, system, device, terminal, and medium |
| DE202022100907U1 (en) * | 2022-02-17 | 2022-03-14 | Himansu Sekhar Behera | A novel hybridization system for time series prediction |
| CN114676645A (en) * | 2022-05-30 | 2022-06-28 | 湖南大学 | Non-stationary time sequence prediction method and system |
| CN116703466A (en) * | 2023-06-15 | 2023-09-05 | 中国平安财产保险股份有限公司 | System access quantity prediction method based on improved wolf algorithm and related equipment thereof |
-
2023
- 2023-10-08 CN CN202311287098.9A patent/CN117060984B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109802862A (en) * | 2019-03-26 | 2019-05-24 | 重庆邮电大学 | A kind of combined network flow prediction method based on set empirical mode decomposition |
| CN111030889A (en) * | 2019-12-24 | 2020-04-17 | 国网河北省电力有限公司信息通信分公司 | Network traffic prediction method based on GRU model |
| CN113205698A (en) * | 2021-03-24 | 2021-08-03 | 上海吞山智能科技有限公司 | Navigation reminding method based on IGWO-LSTM short-time traffic flow prediction |
| CN113206756A (en) * | 2021-04-22 | 2021-08-03 | 大连大学 | Network flow prediction method based on combined model |
| CN113905391A (en) * | 2021-09-27 | 2022-01-07 | 湖北工业大学 | Ensemble learning network traffic prediction method, system, device, terminal, and medium |
| DE202022100907U1 (en) * | 2022-02-17 | 2022-03-14 | Himansu Sekhar Behera | A novel hybridization system for time series prediction |
| CN114676645A (en) * | 2022-05-30 | 2022-06-28 | 湖南大学 | Non-stationary time sequence prediction method and system |
| CN116703466A (en) * | 2023-06-15 | 2023-09-05 | 中国平安财产保险股份有限公司 | System access quantity prediction method based on improved wolf algorithm and related equipment thereof |
Non-Patent Citations (3)
| Title |
|---|
| A Combined Method of Improved Grey BP Neural Network and MEEMD-ARIMA for Day-Ahead Wave Energy Forecast;Feng Wu 等;《IEEE TRANSACTIONS ON SUSTAINABLE ENERGY》;全文 * |
| 基于时间相关的网络流量建模与预测研究;高波;《中国博士学位论文全文数据库 信息科技辑》;全文 * |
| 挠性陀螺EMD-ARIMA漂移模型设计与应用;蔡曜 等;《北京航空航天大学学报》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117060984A (en) | 2023-11-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112351503B (en) | Task prediction-based multi-unmanned aerial vehicle auxiliary edge computing resource allocation method | |
| WO2021036414A1 (en) | Co-channel interference prediction method for satellite-to-ground downlink under low earth orbit satellite constellation | |
| EP3977559A1 (en) | Neural network circuit remote electrical tilt antenna infrastructure management based on probability of actions | |
| López et al. | Primary user characterization for cognitive radio wireless networks using a neural system based on deep learning | |
| CN110213784B (en) | Flow prediction method and device | |
| CN109495898B (en) | Quantitative prediction method and device for indexes of wireless network coverage | |
| CN109246495A (en) | A kind of optical network service method for evaluating quality of oriented multilayer, multi objective | |
| CN118157797A (en) | Low orbit satellite parallel simulation optimization method and system based on LSTM and Bayesian optimization | |
| CN114254734B (en) | Flow matrix modeling method supporting deterministic application | |
| Sova et al. | Development of methodological principles of routing in networks of special communication in conditions of fire storm and radio-electronic suppression | |
| Abdallah et al. | Combining fuzzy logic and neural networks in modeling landfill gas production | |
| CN111414927A (en) | Method for evaluating seawater quality | |
| CN117060984B (en) | Satellite network flow prediction method based on empirical mode decomposition and BP neural network | |
| CN117420443A (en) | LSTM hydrogen fuel cell residual service life prediction method based on genetic algorithm | |
| Geng et al. | A LSTM based campus network traffic prediction system | |
| CN116843016A (en) | Federal learning method, system and medium based on reinforcement learning under mobile edge computing network | |
| Ehiagwina et al. | Development of Neural Network-Based Spectrum Prediction Schemes for Cognitive Wireless Communication: A Case Study of Ilorin, North Central, Nigeria | |
| CN116506863A (en) | Decision optimization method, decision optimization device, electronic equipment and readable storage medium | |
| Park et al. | FedQOGD: Federated Quantized Online Gradient Descent with Distributed Time-Series Data | |
| Aguiar et al. | Learning Flow Functions from Data with Applications to Nonlinear Oscillators | |
| CN118381683B (en) | Distributed monitoring method and device for industrial control network attack event | |
| CN118612754B (en) | Three-in-one terminal control system and method capable of intelligent networking | |
| Viveros et al. | Estimation of future occupation of spectral channels by licensed users in cognitive radio networks applying Neuro-Fuzzy models | |
| Askari et al. | A multi-objective subtractive FCM based TSK fuzzy system with input selection, and its application to dynamic inverse modelling of MR dampers | |
| Mohammed et al. | Multiparametric Assesment of the Conditional Channel of Multitarent Radio Communication Systems using Fuzzy Sets |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |