CN102567781A - Network behavior prediction method based on neural network - Google Patents
Network behavior prediction method based on neural network Download PDFInfo
- Publication number
- CN102567781A CN102567781A CN2011104273396A CN201110427339A CN102567781A CN 102567781 A CN102567781 A CN 102567781A CN 2011104273396 A CN2011104273396 A CN 2011104273396A CN 201110427339 A CN201110427339 A CN 201110427339A CN 102567781 A CN102567781 A CN 102567781A
- Authority
- CN
- China
- Prior art keywords
- network
- algorithm
- neural network
- chaotic
- behavior prediction
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a network behavior prediction method based on a neural network, which relates to a network behavior prediction method. The network behavior prediction method is characterized in that the method comprises the steps of firstly, performing data acquisition and processing, and secondly performing model learning and testing. The method introduces a chaotic mechanism into an SMF Elman network, adopts the intrinsic global motility of the chaotic mechanism to escape from a local minimum point existing during the weight optimization process, researches the chaotic features mapped by Tent, and establishes a corresponding chaotic training algorithm, and the algorithm has better ergodic uniformity, can search a global optimum solution, has high search speed, and also avoids problems of the standard Elman neural network that the convergence speed is low, the standard Elman neural network is easy to involve in a local minimum value, and the like.
Description
Technical field:
The present invention relates to a kind of network behavior Forecasting Methodology, be specifically related to a kind of network behavior Forecasting Methodology based on neural network.
Background technology:
Along with the continuous development of complexity science, the achievement in research of chaos continues to bring out, and The empirical results shows that network system is one and has the extremely complicated nonlinear kinetics system of chaos characteristic, and the variation of network traffics has non-linear, time-varying Characteristics.The tradition neural network model is only studied the time series forecasting with taxis; In addition; The tradition neural network adopts the BP algorithm or basically based on the hybrid algorithm of neural network; These algorithms all have weak point, need find the solution the derivative of objective function based on the gradient descent algorithm of BP, and the scope of application has limitation; The simple problem complicacy is talked about by hybrid algorithm regular meeting based on neural network, also is easier to occur the local convergence problem.
Summary of the invention:
The purpose of this invention is to provide a kind of network behavior Forecasting Methodology based on neural network; It is introduced mechanism of chaos in the SIMF Elman network, utilizes the intrinsic overall situation of mechanism of chaos to move about, and escapes from the local minimum point that exists in the weights optimizing process; Studied the chaotic characteristic of Tent mapping; And constructed corresponding chaos training algorithm, and algorithm has better traversal homogeneity, and this algorithm can be searched for globally optimal solution; And search speed is faster arranged, also avoid the speed of convergence of standard Elman neural network slow, be prone to be absorbed in problem such as local minimum.
In order to solve the existing problem of background technology, the present invention takes following technical scheme: it adopts following steps: data acquisition and processing (DAP) is carried out in (1); (2) carry out model learning and test.
The concrete grammar of described step (2) is: confirming of hidden layer carried out in (21); (22) through the search of Chaos Search algorithm optimization, carry out weights optimization, can make algorithm jump out local optimum, can keep colony's diversity, improve the global search performance of algorithm; (23) training of improvement Elman network.
The present invention introduces mechanism of chaos in the SIMF Elman network, utilizes the intrinsic overall situation of mechanism of chaos to move about, and escapes from the local minimum point that exists in the weights optimizing process; Studied the chaotic characteristic of Tent mapping; And constructed corresponding chaos training algorithm, and algorithm has better traversal homogeneity, and this algorithm can be searched for globally optimal solution; And search speed is faster arranged, also avoid the speed of convergence of standard Elman neural network slow, be prone to be absorbed in problem such as local minimum.
Embodiment:
This embodiment is taked following technical scheme: it adopts following steps: data acquisition and processing (DAP) is carried out in (1); (2) carry out model learning and test.
The concrete grammar of described step (2) is: confirming of hidden layer carried out in (21): the selection of hidden layer neuron number is a problem of difficulty: neuron number very little, network can not well be learnt, and needs the number of times of training also many, precision is not high yet; Neuron number is too many, and the training time is longer, even possibly cause not restraining.Can select best hidden layer neuron number according to
; Wherein n ' is the hidden layer neuron number; M is the output neuron number; N is the input neuron number, and α is the constant between (1,10).So the hidden layer node number can be got 10,11,12,13,14,15,16,17 respectively, through repeatedly debugging experiment, through repeatedly debugging experiment; Neural network hidden layer number is 16 o'clock; Neural network is effective to approaching of function, so improved Elman neural network adopts 48-15-24 formula structure, promptly imports nodes for 48; 16 hidden layer nodes, 24 output nodes;
(22) through the search of Chaos Search algorithm optimization, carry out weights optimization, can make algorithm jump out local optimum; Can keep colony's diversity, improve the global search performance of algorithm: the essence of BP algorithm is to find the solution optimum solution by the direction that gradient descends, when initial value is selected immediately; The route that gradient descends just confirmed, if when the BP algorithm is absorbed in local minimum, can select initial value again at random; And then calculate with gradient method; At this moment the value of being calculated might be the zone of having searched for, and therefore, algorithm can be done skimble-skamble work.Chaos is a kind of comparatively general phenomenon that is present in the NLS; The variation of Chaos Variable has randomness, ergodicity and regularity in certain scope; Therefore can utilize these characteristic optimization search of Chaos Variable, carry out weights optimization, can make algorithm jump out local optimum; Can keep colony's diversity, improve the global search performance of algorithm.
Yet; Different chaotic maps operators have very big influence to the chaos searching process; Quoting more at present is Logistic mapping operator; Through pointing out that relatively Tent mapping has better traversal homogeneity and iteration speed faster than the Logistic mapping, the strict mathematical reasoning of going forward side by side has been proved Tent and has been shone upon and have the precondition as the optimized Algorithm chaos sequence.
Tent mapping is called the tent mapping again, its expression formula as shown in the formula:
Theoretical research shows: can be expressed as after Tent mapping process Bei Nuli displacement changes: X
N+1=(2X
n) mod1.
Adopt the basic step that respectively connects weights of chaos sequence optimization SIMF Elman network following:
Step1 initialization weight vector produces random number on (1,1) interval, compose and give the network initial weight;
Step2 adopts the BP algorithm that neural network weight is learnt, the computational grid error, if satisfy accuracy requirement after the study, this moment, the weights of network were designated as W, and initialization does, and then algorithm finishes, otherwise, change Step3 over to;
Step3 makes W '=W/2+1/2, initial weight is mapped in the field of definition of Chaos Variable, with chaos optimization algorithm correction weights, by W "=2W ' mod1 produces chaos value W "; " 1 shines upon back in the original scope of weights, again the computational grid error then with W '=2W;
Step4 is if network error satisfies accuracy requirement, and then algorithm finishes; Otherwise, change Step3 over to;
If Step5 is constant through several times, finishes, otherwise return Step2;
(23) training of improvement Elman network: according to the Chaos Search algorithm, use train function and adapt function that network is trained, iterative process all comprises several steps each time:
Step1 as whole input vector fan-in network, calculates its output after the True Data that collects is handled;
Step2 is before adjustment connects power; Must connect power to each earlier and give initial value at random; Hope that generally initial weight can make each neuronic state value approach zero when input adds up, but can not equal same number, otherwise system can not train;
Step3 with the result of calculation object vector relatively produces error vector;
Step4 asks weights to change through the gradient descent method and the backpropagation of error, obtains the gradient of error function for each weights and deviation;
Step5 revises weights and deviation, the computational grid error according to Chaos Search algorithm and the Grad that calculates again;
Step6 is if network error satisfies accuracy requirement, and then algorithm finishes; Otherwise, change Step4 over to;
If Step7 is constant through several times, finish.
This embodiment is introduced mechanism of chaos in the SIMF Elman network, utilizes the intrinsic overall situation of mechanism of chaos to move about, and escapes from the local minimum point that exists in the weights optimizing process; Studied the chaotic characteristic of Tent mapping; And constructed corresponding chaos training algorithm, and algorithm has better traversal homogeneity, and this algorithm can be searched for globally optimal solution; And search speed is faster arranged, also avoid the speed of convergence of standard Elman neural network slow, be prone to be absorbed in problem such as local minimum.
Claims (2)
1. network behavior Forecasting Methodology based on neural network, it is characterized in that its adopts following steps: data acquisition and processing (DAP) is carried out in (1); (2) carry out model learning and test.
2. a kind of network behavior Forecasting Methodology based on neural network according to claim 1 is characterized in that the concrete grammar of described step (2) is: confirming of hidden layer carried out in (21); (22) through the search of Chaos Search algorithm optimization, carry out weights optimization, can make algorithm jump out local optimum, can keep colony's diversity, improve the global search performance of algorithm; (23) training of improvement Elman network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104273396A CN102567781A (en) | 2011-12-20 | 2011-12-20 | Network behavior prediction method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104273396A CN102567781A (en) | 2011-12-20 | 2011-12-20 | Network behavior prediction method based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102567781A true CN102567781A (en) | 2012-07-11 |
Family
ID=46413153
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011104273396A Pending CN102567781A (en) | 2011-12-20 | 2011-12-20 | Network behavior prediction method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102567781A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102946613A (en) * | 2012-10-10 | 2013-02-27 | 北京邮电大学 | Method for measuring QoE |
| CN103914985A (en) * | 2014-04-25 | 2014-07-09 | 大连理工大学 | Method for predicting future speed trajectory of hybrid power bus |
| CN103967963A (en) * | 2014-05-21 | 2014-08-06 | 合肥工业大学 | Method for measuring temperature of DCT wet clutches on basis of neural network prediction |
| CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大***研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
| CN104636801A (en) * | 2013-11-08 | 2015-05-20 | 国家电网公司 | Transmission line audible noise prediction method based on BP neural network optimization |
| CN108491923A (en) * | 2018-04-10 | 2018-09-04 | 吉林大学 | Based on the pest image-recognizing method for improving wolf pack algorithm optimization Bayesian network |
| TWI646808B (en) * | 2016-01-29 | 2019-01-01 | 中華電信股份有限公司 | Request traffic prediction method |
| CN110035559A (en) * | 2019-04-25 | 2019-07-19 | 重庆邮电大学 | A kind of contention window size intelligent selecting method based on chaos Q- learning algorithm |
| CN110261735A (en) * | 2019-06-18 | 2019-09-20 | 西华大学 | Based on the electrical power distribution network fault location method for improving quantum cuckoo algorithm |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102056183A (en) * | 2010-12-10 | 2011-05-11 | 北京交通大学 | Network flow prediction method and device based on cognitive network |
-
2011
- 2011-12-20 CN CN2011104273396A patent/CN102567781A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102056183A (en) * | 2010-12-10 | 2011-05-11 | 北京交通大学 | Network flow prediction method and device based on cognitive network |
Non-Patent Citations (2)
| Title |
|---|
| 党小超等: "基于改进Elman神经网络的网络流量预测", 《计算机应用》, vol. 30, no. 10, 31 October 2010 (2010-10-31), pages 2648 - 2652 * |
| 党小超等: "新型 Elman 混沌神经网络的流量预测", 《计算机工程》, vol. 37, no. 3, 28 February 2011 (2011-02-28), pages 173 - 174 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102946613A (en) * | 2012-10-10 | 2013-02-27 | 北京邮电大学 | Method for measuring QoE |
| CN102946613B (en) * | 2012-10-10 | 2015-01-21 | 北京邮电大学 | Method for measuring QoE |
| CN104636801B (en) * | 2013-11-08 | 2018-03-16 | 国家电网公司 | A kind of prediction transmission line of electricity audible noise method based on Optimized BP Neural Network |
| CN104636801A (en) * | 2013-11-08 | 2015-05-20 | 国家电网公司 | Transmission line audible noise prediction method based on BP neural network optimization |
| CN103914985A (en) * | 2014-04-25 | 2014-07-09 | 大连理工大学 | Method for predicting future speed trajectory of hybrid power bus |
| CN103914985B (en) * | 2014-04-25 | 2015-10-28 | 大连理工大学 | A kind of hybrid power passenger car following speed of a motor vehicle trajectory predictions method |
| CN103967963A (en) * | 2014-05-21 | 2014-08-06 | 合肥工业大学 | Method for measuring temperature of DCT wet clutches on basis of neural network prediction |
| CN103967963B (en) * | 2014-05-21 | 2016-08-17 | 合肥工业大学 | The measuring method of DCT wet clutch temperature based on neural network prediction |
| CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大***研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
| TWI646808B (en) * | 2016-01-29 | 2019-01-01 | 中華電信股份有限公司 | Request traffic prediction method |
| CN108491923A (en) * | 2018-04-10 | 2018-09-04 | 吉林大学 | Based on the pest image-recognizing method for improving wolf pack algorithm optimization Bayesian network |
| CN110035559A (en) * | 2019-04-25 | 2019-07-19 | 重庆邮电大学 | A kind of contention window size intelligent selecting method based on chaos Q- learning algorithm |
| CN110035559B (en) * | 2019-04-25 | 2023-03-10 | 重庆邮电大学 | Intelligent competition window size selection method based on chaotic Q-learning algorithm |
| CN110261735A (en) * | 2019-06-18 | 2019-09-20 | 西华大学 | Based on the electrical power distribution network fault location method for improving quantum cuckoo algorithm |
| CN110261735B (en) * | 2019-06-18 | 2021-07-20 | 西华大学 | Power distribution network fault positioning method based on improved quantum cuckoo algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102567781A (en) | Network behavior prediction method based on neural network | |
| CN105389980B (en) | Short-time Traffic Flow Forecasting Methods based on long short-term memory recurrent neural network | |
| Liu et al. | A novel hybrid methodology for short-term wind power forecasting based on adaptive neuro-fuzzy inference system | |
| Klissarov et al. | Learnings options end-to-end for continuous action tasks | |
| CN104914835A (en) | Flexible job-shop scheduling multi-objective method | |
| Duan | Short-time prediction of traffic flow based on PSO optimized SVM | |
| CN103968841B (en) | Improved fireflyalgorithm based AUV (autonomous underwater vehicle) three-dimensional track planning method | |
| CN109978283B (en) | Photovoltaic power generation power prediction method based on branch evolution neural network | |
| CN108091135A (en) | Parking position multistep forecasting method based on Optimization of Wavelet neutral net | |
| CN104217258B (en) | A kind of electric load sigma-t Forecasting Methodology | |
| CN103823504B (en) | A kind of maximum power tracking and controlling method based on least square method supporting vector machine | |
| CN103886395A (en) | Reservoir optimal operation method based on neural network model | |
| CN110007235A (en) | A kind of accumulator of electric car SOC on-line prediction method | |
| CN104240520A (en) | GSO (glowworm swarm optimization) based highway traffic prediction method | |
| CN110530373B (en) | Robot path planning method, controller and system | |
| CN105760213A (en) | Early warning system and method of resource utilization rate of virtual machine in cloud environment | |
| CN104375478A (en) | Method and device for online predicting and optimizing product quality in steel rolling production process | |
| CN113641722A (en) | Long-term time series data prediction method based on variant LSTM | |
| CN104517035A (en) | Planar array antenna active scattering directional diagram predication method | |
| CN103559541A (en) | Back propagation method for out-of-order data stream in big data | |
| CN106896724A (en) | For the tracking system and tracking of sun tracker | |
| Çarkıt et al. | Comparison of the performances of heuristic optimization algorithms PSO, ABC and GA for parameter estimation in the discharge processes of Li-NMC battery | |
| Linlin et al. | Research on the back propagation neural network haze prediction model based on particle swarm optimization | |
| Jamil et al. | Dynamic weight-based multi-objective reward architecture for adaptive traffic signal control system | |
| US20230131614A1 (en) | Systems and methods for coordinated vehicle lane assignment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120711 |