CN110677297A - Combined network flow prediction method based on autoregressive moving average model and extreme learning machine - Google Patents

Combined network flow prediction method based on autoregressive moving average model and extreme learning machine Download PDF

Info

Publication number
CN110677297A
CN110677297A CN201910934741.XA CN201910934741A CN110677297A CN 110677297 A CN110677297 A CN 110677297A CN 201910934741 A CN201910934741 A CN 201910934741A CN 110677297 A CN110677297 A CN 110677297A
Authority
CN
China
Prior art keywords
model
moving average
prediction
autoregressive
network
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
Application number
CN201910934741.XA
Other languages
Chinese (zh)
Inventor
张延华
杨思成
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Technology
Original Assignee
Beijing University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing University of Technology filed Critical Beijing University of Technology
Priority to CN201910934741.XA priority Critical patent/CN110677297A/en
Publication of CN110677297A publication Critical patent/CN110677297A/en
Pending legal-status Critical Current

Links

Images

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
    • H04L43/08—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
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a network flow prediction method based on an autoregressive moving average model and an extreme learning machine, which comprises four main steps of data processing, machine learning, flow prediction and result comparison. The method is used for solving the problems of the conventional model and improving the accuracy of the prediction result.

Description

Combined network flow prediction method based on autoregressive moving average model and extreme learning machine
Technical Field
The invention belongs to the technical field of computer networks, and particularly relates to a network flow prediction method.
Background
With the rapid development of the internet, the network becomes an important platform for people to communicate and communicate. The service types and the flow of the network are increased sharply, and the prediction of the network flow becomes the focus of attention of people. The network flow is influenced by various factors, the change of the network flow is very complex, the network flow data in the future is reasonably predicted, and the method has very important significance for knowing the impending network behavior, analyzing the network safety condition and guiding the network safety monitoring and control.
In order to realize more accurate network traffic data prediction, analysis is performed on the characteristics of network traffic, so that establishment of a more accurate and effective network traffic prediction model has become one of many research hotspots. The method mainly comprises the steps of designing and establishing a proper model by analyzing the time period characteristic, the self-similarity, the length correlation and the chaos characteristic of the network flow time sequence, and realizing the modeling and prediction of the network flow. Many learning methods of machine learning are used for modeling and prediction, and are also widely used in network traffic prediction nowadays. Numerous researchers have proposed a plurality of prediction methods and prediction models in turn to simulate the characteristics of network traffic, and the prediction effect and accuracy are continuously improved. According to the predicted network traffic data and network traffic characteristics, network resources can be reasonably configured, a better network structure is designed, network load is balanced, network congestion is avoided, the network resources are optimized, network services are safer and more stable, and the method has huge economic, technical and social research values. The modeling and prediction of the network flow are developed and researched for a long time, and become more flexible and accurate. And selecting a proper model according to the advantages and disadvantages of the model and applying the model to the prediction of the network traffic, which is the most important step for realizing the network traffic prediction.
The invention designs a network flow prediction method based on an autoregressive moving average model and an extreme learning machine. The method comprises four main steps of data processing, machine learning, flow prediction and result comparison, wherein wavelet transformation is used for preprocessing the original data of the network flow time sequence, then an autoregressive moving average model and an improved extreme learning machine are used for modeling and predicting the decomposed network flow time sequence, and finally the effect of the evaluation model is detected.
Disclosure of Invention
The invention provides a network traffic prediction method based on an autoregressive moving average model and an extreme learning machine, namely a network traffic prediction method combining active online learning, the extreme learning machine and the autoregressive moving average model.
The method is used for solving the problems of the conventional model and improving the accuracy of the prediction result.
The technical scheme is as follows:
a network flow prediction method based on an autoregressive moving average model and an extreme learning machine is characterized in that: the method provides a combined prediction method applying wavelet transformation, phase space reconstruction, autoregressive moving average model (ARMA) and Extreme Learning Machine (ELM) technologies through chaotic characteristic and self-similarity analysis of network flow time sequences.
The invention discovers that the network flow time sequence has self-similarity, long (short) correlation and chaotic characteristics by analyzing the network flow time sequence, and the characteristics have important functions for realizing the establishment and the prediction of a network flow model. Aiming at different characteristics of high and low frequency components of network flow, the invention designs a wavelet transform-based network flow time sequence original data preprocessing method, which can divide the network flow time sequence into a high frequency part and a low frequency part and has the characteristic of quick operation; aiming at self-similarity, the invention adopts a phase space reconstruction method to carry out chaotic analysis and processing on the network flow time sequence; aiming at low-frequency components, the invention adopts the modeling analysis of an autoregressive moving average model; aiming at high-frequency components, the extreme learning machine is adopted for modeling analysis; aiming at the characteristic that the network flow time sequence is continuously generated, the invention combines an Extreme Learning Machine (ELM) with an online learning technology to realize the function of real-time modeling and prediction. By using the method to carry out network flow modeling prediction, the accuracy of the prediction result is higher, and the calculation speed is faster.
The prediction method comprises the following steps:
1. a combined network traffic prediction method based on an autoregressive moving average model and an extreme learning machine is characterized by comprising the following steps:
(1) network original data preprocessing method based on wavelet transform design
Wherein x (t) is original data of network traffic with length N, N is any positive integer greater than 0, c (t) is high-frequency component of network traffic, r (t) is low-frequency component of network traffic, t e (1, N), H is high-frequency filter coefficient matrix, H [ -0.482960.83652-0.22414-0.12941 ], G is low-frequency filter coefficient matrix, G [ -0.129410.224140.836520.48296 ], l is decomposition scale, and l is number less than positive infinite maximum; lambda is a translation coefficient, and lambda is an arbitrary value in an interval [0,1 ];
(2) establishing a high-frequency component prediction model based on an extreme learning machine
Figure BDA0002221313130000032
Wherein, L is the number of hidden layer nodes of the extreme learning machine, c (t) is the input high-frequency component, f (·) is the excitation function, viFor the connection weights, v, of hidden layer nodes to input layer nodesiRandomly initializing to an arbitrary value; deltaiFor the connection weights, δ, of hidden layer nodes to output layer nodesiRandomly initializing to an arbitrary value; biBias value for hidden layer node, biRandomly initializing to an arbitrary value; y isc(t) is an output predicted value of the sample through the extreme learning machine model, t is a time sequence, and i is a sequence number of a hidden layer;
(3) establishing a low-frequency component prediction model based on an autoregressive moving average model
Step 1, determining an autoregressive order p and a moving average order q of an ARMA model by using a minimum information criterion:
wherein min () is a function of taking a minimum value,
Figure BDA0002221313130000042
the calculation method of (2) is shown as formula (4);
in the formula, r (t) is a low-frequency component output after wavelet transformation of network original data, and N is the length of original network flow data;
step 2, estimating unknown parameters of the ARMA model by using a least square estimation method, wherein the unknown parameters comprise autoregressive coefficients
Figure BDA0002221313130000044
Coefficient of autocorrelation
Figure BDA0002221313130000045
Sum partial correlation coefficient
Figure BDA0002221313130000046
Figure BDA0002221313130000048
Wherein p is the autoregressive order determined in step 1, q is the moving average order determined in step 1, R,
Figure BDA0002221313130000049
As shown in formula (7) and formula (8);
Figure BDA00022213131300000410
wherein p is the autoregressive order determined in the step 1, and q is the moving average order determined in the step 1; epsilon (t) is a time sequence which is independently distributed with r (t) and is equal to the expectation and the variance of the r (t) sequence, and values in the epsilon (t) sequence are initialized randomly;
and 3, establishing an ARMA model according to the obtained parameters, wherein the mathematical model of the ARMA is represented as:
Figure BDA00022213131300000412
in the formula, yr(t) is the output prediction value of the sample through the ARMA model,
Figure BDA0002221313130000051
for the autoregressive coefficients determined in step 2,
Figure BDA0002221313130000052
The autocorrelation coefficient determined in the step 2, p is the autoregressive order determined in the step 1, q is the moving average order determined in the step 1, and epsilon (t) is a time sequence which is independently distributed with r (t) and determined in the step 2;
(4) component reconstruction process
Performing wavelet reconstruction on output components of each model to realize single-step or multi-step prediction of network flow;
Y＝{yr,t+yc,t,yr,t+λ+yc,t+λ,yr,t+2λ+yc,t+2λ,…,yr,t+(m-1)λ+yc,t+(m-1)λ} (10)
in the formula, Y is a multidimensional network flow predicted value obtained after wavelet reconstruction; y isr(t) is the predicted value of the high frequency component of the ARMA model outputted in step (3), yc(t) is the low-frequency component predicted value of the extreme learning machine model output in the step (2), m is the dimension of the original network flow data, lambda is a translation coefficient, and lambda is in the interval [0,1]]Any value of (c).
The invention is mainly characterized in that: the invention discovers that the network flow time sequence has self-similarity, long (short) correlation and chaotic characteristics by analyzing the network flow time sequence, and the characteristics have important functions for realizing the establishment and the prediction of a network flow model. Aiming at different characteristics of high and low frequency components of network flow, the invention designs a wavelet transform-based network flow time sequence original data preprocessing method, which can divide the network flow time sequence into a high frequency part and a low frequency part and has the characteristic of quick operation; aiming at self-similarity, the invention adopts a phase space reconstruction method to carry out chaotic analysis and processing on the network flow time sequence; aiming at low-frequency components, the invention adopts the modeling analysis of an autoregressive moving average model; aiming at high-frequency components, the extreme learning machine is adopted for modeling analysis; aiming at the characteristic that the network flow time sequence is continuously generated, the invention combines an Extreme Learning Machine (ELM) with an online learning technology to realize the function of real-time modeling and prediction.
The online extreme learning machine is one that deals with the arrival of samples in groups or one after another. In the basic extreme learning machine algorithm, after a new sample is added to a training sample, the original sample is often repeatedly trained along with the new sample, and the model updating time is increased. The online extreme learning machine overcomes the problems, and is based on a basic extreme learning machine as a basic model and mainly comprises two parts: the first part is to initialize an extreme learning machine, and set the number, weight and bias of hidden nodes of the extreme learning machine to obtain the output weight of the hidden nodes; the second part is an online sequential learning part, and when one or a batch of new samples come, the output weight of the single hidden layer feedforward neural network is updated.
Drawings
Fig. 1 is a schematic flow chart of a network traffic prediction method of a combined model according to the present invention.
FIG. 2 is a flow chart of an improved ELM part of the network traffic prediction method of the combined model of the invention.
Fig. 3 is a comparison diagram of the network traffic prediction results of the network traffic prediction method of the combination model of the present invention.
FIG. 4 is a test set prediction error graph of a network traffic prediction method of a combined model of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a network traffic prediction method of a combined model according to the present invention. In the overall design process, the method comprises the following steps:
1. a combined network traffic prediction method based on an autoregressive moving average model and an extreme learning machine is characterized by comprising the following steps:
(1) network original data preprocessing method based on wavelet transform design
Figure BDA0002221313130000061
Wherein x (t) is original data of network traffic with length N, N is any positive integer greater than 0, c (t) is high-frequency component of network traffic, r (t) is low-frequency component of network traffic, t e (1, N), H is high-frequency filter coefficient matrix, H [ -0.482960.83652-0.22414-0.12941 ], G is low-frequency filter coefficient matrix, G [ -0.129410.224140.836520.48296 ], l is decomposition scale, and l is number less than positive infinite maximum; lambda is a translation coefficient, and lambda is an arbitrary value in an interval [0,1 ];
(2) establishing a high-frequency component prediction model based on an extreme learning machine
Figure BDA0002221313130000071
In the formula, L is the number of hidden layer nodes of the extreme learning machine, c (t) is the input high-frequency component, f (.) is the excitation function, viFor the connection weights, v, of hidden layer nodes to input layer nodesiRandomly initializing to an arbitrary value; deltaiFor the connection weights, δ, of hidden layer nodes to output layer nodesiRandomly initializing to an arbitrary value; biBias value for hidden layer node, biRandomly initializing to an arbitrary value; y isc(t) is an output predicted value of the sample through the extreme learning machine model, t is a time sequence, and i is a sequence number of a hidden layer;
(3) establishing a low-frequency component prediction model based on an autoregressive moving average model
Step 1, determining an autoregressive order p and a moving average order q of an ARMA model by using a minimum information criterion:
Figure BDA0002221313130000072
wherein min () is a function of taking a minimum value,
Figure BDA0002221313130000073
the calculation method of (2) is shown as formula (4);
Figure BDA0002221313130000074
in the formula, r (t) is a low-frequency component output after wavelet transformation of network original data, and N is the length of original network flow data;
step 2, estimating unknown parameters of the ARMA model by using a least square estimation method, wherein the unknown parameters comprise autoregressive coefficients
Figure BDA0002221313130000075
Coefficient of autocorrelation
Figure BDA0002221313130000076
Sum partial correlation coefficient
Figure BDA0002221313130000077
Figure BDA0002221313130000078
Figure BDA0002221313130000081
Wherein p is the autoregressive order determined in step 1, q is the moving average order determined in step 1, R,
Figure BDA0002221313130000082
As shown in formula (7) and formula (8);
wherein p is the autoregressive order determined in the step 1, and q is the moving average order determined in the step 1; epsilon (t) is a time sequence which is independently distributed with r (t) and is equal to the expectation and the variance of the r (t) sequence, and values in the epsilon (t) sequence are initialized randomly;
and 3, establishing an ARMA model according to the obtained parameters, wherein the mathematical model of the ARMA is represented as:
in the formula, yr(t) is the output prediction value of the sample through the ARMA model,
Figure BDA0002221313130000086
for the autoregressive coefficients determined in step 2,
Figure BDA0002221313130000087
The autocorrelation coefficient determined in the step 2, p is the autoregressive order determined in the step 1, q is the moving average order determined in the step 1, and epsilon (t) is a time sequence which is independently distributed with r (t) and determined in the step 2;
(4) component reconstruction process
Performing wavelet reconstruction on output components of each model to realize single-step or multi-step prediction of network flow;
Y＝{yr,t+yc,t,yr,t+λ+yc,t+λ,yr,t+2λ+yc,t+2λ,…,yr,t+(m-1)λ+yc,t+(m-1)λ} (10)
in the formula, Y is a multidimensional network flow predicted value obtained after wavelet reconstruction; y isr(t) is the predicted value of the high frequency component of the ARMA model outputted in step (3), yc(t) is the low-frequency component predicted value of the extreme learning machine model output in the step (2), m is the dimension of the original network flow data, lambda is a translation coefficient, and lambda is in the interval [0,1]]Any value of (c).
And predicting the network traffic time series or predicting in multiple steps in real time by using the established model. When multi-step real-time prediction is carried out, the length of a training sample is increased, and a new data bureau is continuously merged into an original data set. And finally, after all the training samples are learned, terminating learning and establishing a corresponding regression model. And finally, evaluating the effect of the method, and mainly comparing the training speed, the prediction speed and the prediction error of the method.
The output of the network flow prediction method of the combined model is a network flow prediction value; fig. 3 shows a comparison of network traffic prediction results, X-axis: prediction sample number, unit is one, Y-axis: network traffic value, unit: kilobytes (kb) are realized as true values of the network traffic data, and dotted lines are predicted values of the network traffic data; fig. 4 shows the test set prediction error, X-axis: test set sample number, in units of one, Y-axis: predicted absolute error, unit: kilobytes (kb).

Claims (1)

1. A combined network traffic prediction method based on an autoregressive moving average model and an extreme learning machine is characterized by comprising the following steps:
(1) network original data preprocessing method based on wavelet transform design
Figure FDA0002221313120000011
Wherein x (t) is original data of network traffic with length N, N is any positive integer greater than 0, c (t) is high-frequency component of network traffic, r (t) is low-frequency component of network traffic, t e (1, N), H is high-frequency filter coefficient matrix, H [ -0.482960.83652-0.22414-0.12941 ], G is low-frequency filter coefficient matrix, G [ -0.129410.224140.836520.48296 ], l is decomposition scale, and l is number less than positive infinite maximum; lambda is a translation coefficient, and lambda is an arbitrary value in an interval [0,1 ];
(2) establishing a high-frequency component prediction model based on an extreme learning machine
Figure FDA0002221313120000012
In the formula, L is the number of hidden layer nodes of the extreme learning machine, c (t) is the input high-frequency component, f (.) is the excitation function, viFor the connection weights, v, of hidden layer nodes to input layer nodesiRandomly initializing to an arbitrary value; deltaiFor the connection weights, δ, of hidden layer nodes to output layer nodesiRandomly initializing to an arbitrary value; biBias value for hidden layer node, biRandomly initializing to an arbitrary value; y isc(t) is an output predicted value of the sample through the extreme learning machine model, t is a time sequence, and i is a sequence number of a hidden layer;
(3) establishing a low-frequency component prediction model based on an autoregressive moving average model
Step 1, determining an autoregressive order p and a moving average order q of an ARMA model by using a minimum information criterion:
Figure FDA0002221313120000013
wherein min () is a function of taking a minimum value,the calculation method of (2) is shown as formula (4);
Figure FDA0002221313120000021
in the formula, r (t) is a low-frequency component output after wavelet transformation of network original data, and N is the length of original network flow data;
step 2, estimating unknown parameters of the ARMA model by using a least square estimation method, wherein the unknown parameters comprise autoregressive coefficients
Figure FDA0002221313120000022
Coefficient of autocorrelation
Figure FDA0002221313120000023
Sum partial correlation coefficient
Figure FDA0002221313120000025
Wherein p is the autoregressive order determined in step 1, q is the moving average order determined in step 1, R,
Figure FDA0002221313120000027
As shown in formula (7) and formula (8);
Figure FDA0002221313120000028
Figure FDA0002221313120000029
wherein p is the autoregressive order determined in the step 1, and q is the moving average order determined in the step 1; epsilon (t) is a time sequence which is independently distributed with r (t) and is equal to the expectation and the variance of the r (t) sequence, and values in the epsilon (t) sequence are initialized randomly;
and 3, establishing an ARMA model according to the obtained parameters, wherein the mathematical model of the ARMA is represented as:
Figure FDA00022213131200000210
in the formula, yr(t) is the output prediction value of the sample through the ARMA model,
Figure FDA00022213131200000211
for the autoregressive coefficients determined in step 2,
Figure FDA00022213131200000212
For the autocorrelation coefficients determined in step 2, p is the autoregressive order determined in step 1Q is the moving average order determined in step 1, and epsilon (t) is the time sequence determined in step 2 and independently distributed with r (t);
(4) component reconstruction process
Performing wavelet reconstruction on output components of each model to realize single-step or multi-step prediction of network flow;
Y＝{yr,t+yc,t,yr,t+λ+yc,t+λ,yr,t+2λ+yc,t+2λ,…,yr,t+(m-1)λ+yc,t+(m-1)λ} (10)
in the formula, Y is a multidimensional network flow predicted value obtained after wavelet reconstruction; y isr(t) is the predicted value of the high frequency component of the ARMA model outputted in step (3), yc(t) is the low-frequency component predicted value of the extreme learning machine model output in the step (2), m is the dimension of the original network flow data, lambda is a translation coefficient, and lambda is in the interval [0,1]]Any value of (c).
CN201910934741.XA 2019-09-29 2019-09-29 Combined network flow prediction method based on autoregressive moving average model and extreme learning machine Pending CN110677297A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910934741.XA CN110677297A (en) 2019-09-29 2019-09-29 Combined network flow prediction method based on autoregressive moving average model and extreme learning machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910934741.XA CN110677297A (en) 2019-09-29 2019-09-29 Combined network flow prediction method based on autoregressive moving average model and extreme learning machine

Publications (1)

Publication Number Publication Date
CN110677297A true CN110677297A (en) 2020-01-10

Family

ID=69080119

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910934741.XA Pending CN110677297A (en) 2019-09-29 2019-09-29 Combined network flow prediction method based on autoregressive moving average model and extreme learning machine

Country Status (1)

Country Link
CN (1) CN110677297A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111355633A (en) * 2020-02-20 2020-06-30 安徽理工大学 Mobile phone internet traffic prediction method in competition venue based on PSO-DELM algorithm
CN112308169A (en) * 2020-11-10 2021-02-02 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN113411216A (en) * 2021-06-21 2021-09-17 国网宁夏电力有限公司信息通信公司 Network flow prediction method based on discrete wavelet transform and FA-ELM
CN115022210A (en) * 2022-07-26 2022-09-06 中国银行股份有限公司 Construction method, prediction method and device of network traffic prediction model
CN115022210B (en) * 2022-07-26 2024-06-21 中国银行股份有限公司 Construction method, prediction method and device of network traffic prediction model

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105243259A (en) * 2015-09-02 2016-01-13 上海大学 Extreme learning machine based rapid prediction method for fluctuating wind speed
CN105976051A (en) * 2016-04-29 2016-09-28 武汉大学 Wavelet transformation and improved firefly-optimized extreme learning machine-based short-term load prediction method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105243259A (en) * 2015-09-02 2016-01-13 上海大学 Extreme learning machine based rapid prediction method for fluctuating wind speed
CN105976051A (en) * 2016-04-29 2016-09-28 武汉大学 Wavelet transformation and improved firefly-optimized extreme learning machine-based short-term load prediction method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
张洋等: ""基于组合模型的网络流量预测"", 《华中科技大学学报（自然科学版）》 *
李巧侠: ""基于组合模型的网络流量预测"", 《微型电脑应用》 *
田中大等: ""基于ARIMA补偿ELM的网络流量预测方法"", 《信息与控制》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111355633A (en) * 2020-02-20 2020-06-30 安徽理工大学 Mobile phone internet traffic prediction method in competition venue based on PSO-DELM algorithm
CN112308169A (en) * 2020-11-10 2021-02-02 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN112308169B (en) * 2020-11-10 2022-05-03 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN113411216A (en) * 2021-06-21 2021-09-17 国网宁夏电力有限公司信息通信公司 Network flow prediction method based on discrete wavelet transform and FA-ELM
CN113411216B (en) * 2021-06-21 2022-11-04 国网宁夏电力有限公司信息通信公司 Network flow prediction method based on discrete wavelet transform and FA-ELM
CN115022210A (en) * 2022-07-26 2022-09-06 中国银行股份有限公司 Construction method, prediction method and device of network traffic prediction model
CN115022210B (en) * 2022-07-26 2024-06-21 中国银行股份有限公司 Construction method, prediction method and device of network traffic prediction model

Similar Documents

Publication Publication Date Title
CN109120462B (en) Method and device for predicting opportunistic network link and readable storage medium
Szorenyi et al. Gossip-based distributed stochastic bandit algorithms
CN112418482B (en) Cloud computing energy consumption prediction method based on time series clustering
CN110677297A (en) Combined network flow prediction method based on autoregressive moving average model and extreme learning machine
Peng et al. A hybrid forward algorithm for RBF neural network construction
CN108304679A (en) A kind of adaptive reliability analysis method
CN106529701B (en) Optical fiber state prediction method for optimizing neural network based on improved firefly algorithm
CN111783209B (en) Self-adaptive structure reliability analysis method combining learning function and kriging model
CN114239840A (en) Quantum channel noise coefficient estimation method and device, electronic device and medium
CN113935489A (en) Variational quantum model TFQ-VQA based on quantum neural network and two-stage optimization method thereof
CN113377964A (en) Knowledge graph link prediction method, device, equipment and storage medium
CN112861066A (en) Machine learning and FFT (fast Fourier transform) -based blind source separation information source number parallel estimation method
CN116663419A (en) Sensorless equipment fault prediction method based on optimized Elman neural network
CN111310996A (en) User trust relationship prediction method and system based on graph self-coding network
CN111539558A (en) Power load prediction method adopting optimized extreme learning machine
Feng et al. A FOM/ROM hybrid approach for accelerating numerical simulations
CN113177078B (en) Approximate query processing algorithm based on condition generation model
CN107967395A (en) A kind of nonlinear time_varying system Fast Identification Method based on the expansion of beta wavelet basis functions
CN115183969A (en) Method and system for estimating BWBN model parameters
Wu et al. Parallel optimal statistical design method with response surface modelling using genetic algorithms
Abbas et al. Volterra system identification using adaptive genetic algorithms
Ozoh et al. An In-Depth Study of Typical Machine Learning Methods via Computational Techniques
CN114492007A (en) Factor effect online identification method and device based on hierarchical error control
CN113743485A (en) Data dimension reduction method based on Fourier domain principal component analysis
CN111462479A (en) Traffic flow prediction method based on Fourier-recurrent neural network

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: 20200110