CN103716180A - Network flow actual forecasting-based network abnormality pre-warning method - Google Patents
Network flow actual forecasting-based network abnormality pre-warning method Download PDFInfo
- Publication number
- CN103716180A CN103716180A CN201310641648.2A CN201310641648A CN103716180A CN 103716180 A CN103716180 A CN 103716180A CN 201310641648 A CN201310641648 A CN 201310641648A CN 103716180 A CN103716180 A CN 103716180A
- Authority
- CN
- China
- Prior art keywords
- network
- layer
- flow
- early warning
- data
- 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
Images
Abstract
The invention relates to a network flow actual forecasting-based network abnormality pre-warning method. By virtue of conducting real-time continuous acquisition of network data to monitor the network flow and performing statistic forecasting analysis of the obtained flow data, a forecasting index value of the network flow can be obtained. Based on the flow forecasting data performance change tendency, a real-time alarm threshold value is set. On the basis of the network flow detection, an administrator monitors by real-time polling of the network to obtain a current value of a defined object. When the network flow is abnormal, an alarm is sent to management personnel through an expanded flow detection and alarm system so that the fault can be discovered timely and treated. In this way, the fault management to a certain extent is realized, and the network flow detection itself also relates to the content in safety management aspect. The application of the network abnormality pre-warning method can detect invasive attacks to a certain degree, effectively help the management personnel with the network performance management, and employ the alarm mechanism to assist the network management personnel with adoption of corresponding safety strategies and protection measures.
Description
Technical field
The field that innovation and creation relate to is mainly communications network security, particularly a kind of Network Abnormal method for early warning of flow actual prediction Network Based.
Background technology
In network security management system, traffic monitoring and statistical analysis are the bases of whole management.The main purpose of flow detection is by network data being carried out to the collection monitoring network flow of real-time continuous, the data on flows obtaining is carried out to statistical computation, thereby obtain the performance index of network main component.
Network manager just can carry out performance evaluation management to network main component according to data on flows, finds performance change trend, and analyzes and affect the factor of network performance and problem place.In addition, the in the situation that of exception of network traffic, by the flow detection warning system of expansion, can also report to the police to administrative staff, find that in time fault is processed.On the basis of detecting in network traffics, keeper monitors the currency of real time polling Network Capture defining objects, if exceed the normal predetermined value of examination value, reports to the police, and the person's of assisting management discovering network bottleneck, so just can realize fault management to a certain extent.And network traffics detection itself also relates to the content of safety management aspect.
As can be seen here, for an effective network security management system, the realization of function all more or less depend on obtaining of flow information.Therefore the collection of network traffic information can be described as the core foundation stone that network security management system is achieved.Its application can detect to a certain extent invasion and attack, can effectively help administrative staff to carry out network performance management, and utilize alarm mechanism to assist webmaster personnel to take corresponding security strategy and safeguard procedures, thereby reduce invasion, attack the loss causing.
Model of network traffic is the basis of carrying out network performance analysis and Network Programe Design, good discharge model and Forecasting Methodology to design new generation network agreement, network management and diagnosis, design the network hardware equipments such as high performance router and load equalizer and raising network service quality all significant.Along with the increase of the network bandwidth and the appearance of various network services, discharge model in the past is also difficult to meet to accurate description and predictions existing and network traffics in the future, therefore, for discharge model in the past and the pros and cons of Forecasting Methodology, for providing model and prediction more accurately, network traffics seem very necessary.Specifically in the network safety filed that paid close attention to by people, for the modeling of network traffics (user network behavior) and prediction, for improving intrusion detection, the abnormal or attack of discovering network in advance provides a direction.
By Mathematical Modeling, describe network traffics, although can carry out certain prediction to network traffics constantly in future, carry out volume forecasting more flexibly, efficiently also needs the certain methods such as artificial intelligence.Simultaneously, network traffics behavior is along with the difference of time and region can present very large variation, therefore wanting to provide a kind of prediction of the generalization for network traffics is that very difficult, desirable prediction mode should be adaptive and must provide enough reliable informations to user.In predicting network flow, also exist predictability problem, the minimum predicted difference in certain prediction step and within the scope of acceptable predicated error maximum prediction step.
Summary of the invention
The traffic conditions of different periods is not considered in conventional flow early warning conventionally, only predicts peak value and to peak value early warning, and lacks learning functionality.The object of the present invention is to provide a kind of Network Abnormal method for early warning of flow actual prediction Network Based, by ANN(artificial neural net) approach is for volume forecasting, and according to volume forecasting, set early warning threshold values at times and carry out exception of network traffic early warning, mainly to solve the different discharge in period of times of enterprise network flow, differ greatly, during exception of network traffic, be difficult to the problem of screening.
In order to achieve the above object, technical scheme of the present invention is to provide a kind of Network Abnormal method for early warning of flow actual prediction Network Based, and it comprises following process:
Step 1, by being carried out to real-time collection and continual collection, network data carrys out monitoring network flow;
Step 2, the network flow data obtaining is carried out to statistical analysis;
Step 4, according to prediction index numerical value and performance change trend thereof, set Realtime Alerts threshold values;
On step 5, the basis detected in network traffics, the currency of keeper's real time polling Network Capture defining objects is monitored;
When if currency exceeds corresponding alarm threshold value, there is exception of network traffic situation in judgement, thereby report to the police to keeper by the flow detection warning system of expansion; If currency does not surpass alarm threshold value, continue monitoring traffic in network;
ANN network in described Network Abnormal method for early warning, is the FIR neural net that adopts three layers of feed forward architecture, comprises: input layer, hidden layer, and output layer; Between each layer, adopt full connection, and before every layer, all added a FIR filter.
Described Network Abnormal method for early warning also comprises the process that the algorithm based on Wan is trained to ANN network using; And after each sample of study, connect the size of weights and synaptic weight in update algorithm.
When different value is got on the rank that described Network Abnormal method for early warning is also investigated FIR filter, on the impact predicting the outcome.The rank of FIR filter between input layer and hidden layer are designated as to Mh, and between hidden layer and output layer, the rank of FIR filter are designated as Mo, preferably, get (Mh, Mo)=(5,5).
The Network Abnormal method for early warning of flow actual prediction Network Based of the present invention, the prediction by time division traffic obtains predicted value at times, according to predicted value, sets alarming threshold value, realizes the flow early warning of higher reliability.The present invention can find that fault is processed in time.So just can realize fault management to a certain extent.And network traffics detection itself also relates to the content of safety management aspect.Its application can detect to a certain extent invasion and attack, and can effectively help administrative staff to carry out network performance management, and utilizes alarm mechanism to assist webmaster personnel to take corresponding security strategy and safeguard procedures.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the Network Abnormal method for early warning of flow actual prediction Network Based of the present invention;
Fig. 2 is neural network structure figure;
Fig. 3 is the schematic diagram that predicts the outcome of different FIR filter order.
Embodiment
The relative theory that the Network Abnormal method for early warning of following paper flow actual prediction Network Based of the present invention relates to.
People extensively utilize artificial intelligence approach to carry out volume forecasting, and such as expert system, fuzzy reasoning, fuzzy neural etc., wherein applying maximum is artificial neural net (ANN).Neural net is suitable for learning the non-linear relation of more complicated, and it is data-driven unceasing study, do not need network traffics to carry out a large amount of mathematical modeling experiment works, thereby the complicated correlation in applicable solution Model of network traffic and the adaptive problem in prediction.
In the ANN of existing volume forecasting model, adopting more is multilayer feedforward structure, especially 3 layers of neural net, such as 1-5-1,5-5-1 etc., according to different prediction, need to utilize the lower flow situation constantly of single output unit prediction, utilize many output units predictions a plurality of moment flow situations or next difference description amount of flow constantly below, and conventional algorithm is backpropagation (BP) algorithm.
Because ANN is relation between carrying out automatic learning input and export by sample learning, thereby when learning, predict, tracking actual flow exists following problem: to what extent follow the tracks of actual flow proper? the appearance on a lot of situation down-offs peak is very of short duration, according to this peak, predicts that next constantly can cause very large error.In actual applications, according to heterogeneous networks service features, actual flow is formulated respectively to different tracking degree and learn, by a plurality of ANN predicted composition systems, network traffics different aspects are predicted to be the developing direction meriting attention in addition.
Fuzzy theory is as another kind of artificial intelligence approach, also can be applicable in seasonal effect in time series prediction, and it has following two advantages conceptive comparing with neural net: the parameter of fuzzy system has clear and definite physical significance, and the control law of fuzzy system has generally comprised the essential information of Unknown Nonlinear Systems.
Based on fuzzy theory, set up an adaptive fuzzy volume forecasting device, adopt nearest-neighbors classification learning algorithm to adjust the parameter of fuzzy logic system, by contrast autoregressive process and time video flowing, verify that it is feasible with more accurate, and be successfully applied in the Dynamic Bandwidth Allocation and queue management of ATM net (asynchronous transfer mode).
In addition, fuzzy theory and ANN combine and form fuzzy neural network (FNN) existing many application in volume forecasting.Utilizing ANN to carry out in the process of volume forecasting, on the one hand, if increase a fuzzy preprocessing process at ANN front end, by classification, can reduce the input node of ANN and ANN system can be processed quantitatively and the data sample of qualitative two aspects; By increasing a fuzzy assembly at ANN output, the result that can predict out to ANN adds the qualitative factor of some network environment to revise on the other hand; In addition, the prognoses system that many ANN form can be connected by fuzzy assembly, fuzzy assembly is classified and to each different ANN, is processed input data, also can synthesize to extract more reasonable or friendly predicting the outcome to the Output rusults of a plurality of ANN.
Wavelet analysis is to process a kind of the most effective method of nonstationary time series, and for the long correlation in network traffics, wavelet transformation has himself distinctive advantage on mathematics.Through wavelet decomposition, in sequential, the network traffics of more complicated can be broken down into approximate incoherent time series, such as the long correlation in real network flow can be broken down into short relevant, and wavelet decomposition carried out smoothly time series, wavelet method can be converted into complicated network traffics by traditional Forecasting Methodology the time series after decomposing is predicted like this.In addition, small echo prediction can reach convergence and less predicated error more fast.
Shown in Figure 1 below, introduce the Network Abnormal method for early warning of flow actual prediction Network Based of the present invention, by network data being carried out to the collection monitoring network flow of real-time continuous, the data on flows obtaining is carried out to statistical forecast analysis, thereby obtain the prediction index numerical value of network traffics.
According to volume forecasting data performance variation tendency, set Realtime Alerts threshold values, on the basis of detecting in network traffics, keeper monitors the currency of real time polling Network Capture defining objects, the in the situation that of exception of network traffic, flow detection warning system by expansion is reported to the police to administrative staff, finds that in time fault is processed.So just can realize fault management to a certain extent.And network traffics detection itself also relates to the content of safety management aspect.Its application can detect to a certain extent invasion and attack, and can effectively help administrative staff to carry out network performance management, and utilizes alarm mechanism to assist webmaster personnel to take corresponding security strategy and safeguard procedures.
As shown in Figure 2, our selected FIR (finite impulse response) neural network structure has three layers: an input layer, a hidden layer and an output layer.Between each layer, adopt full connection, and each connection in three-decker has added a FIR filter.
We are in actual prediction, and selecting input layer is 7 nodes, and wherein, the 1st node is information on working day, then 6 datas on flows that node is continuous 6 moment; Hidden layer is 10 nodes, and output layer is 1 node.Concealed nodes and output node activation primitive all adopt Sigmoid function.
The present invention when neural network training, employing be the algorithm based on Wan:
In formula:
δ
l k(t): reverse propagated error;
W
l ik(t): connect the node of l-1 layer to the connection weights of l layer k node;
C
l ikj(t): the synaptic weight of FIR filter;
X
^l k(t): induction local field;
X
l-1 i(t): the neuronic output of l-1 layer, as the input of FIR filter;
Q
-1: time delay operator;
ε (t): moment error;
N
c: the rank of FIR filter;
D
k(t): t desired output constantly;
Y
k(t): t actual output constantly;
α w, α c: momentum coefficient (connecting weights, synaptic weight);
η w, η c: learning coefficient (connecting weights, synaptic weight).
As Δ ε (t)=ε
t+1-ε
tduring <0, learning coefficient and momentum coefficient increase respectively a η w in a small amount, η c, otherwise, reduce an a small amount of.
While starting to train, get:
Momentum coefficient: α w=0.1, α c=0.1;
Learning coefficient: η w=0.8, η c=0.5;
In a small amount: Δ η=0.02, Δ α=0.001.
The algorithm of Wan is a kind of reverse propagated error algorithm, in training process, in order to improve the randomness of sample training, training adopts serial mode to carry out, after wherein each bout has been trained, the training of next bout is all from one of this sample set random position.Learn according to formula above, to upgrade the size that connects weights and synaptic weight after each sample.Momentum coefficient during adjustment and learning coefficient keep dynamically adjusting.The adjustment that connects weights has strengthened the accurate output valve to training data.
The present invention is when carrying out network traffics actual prediction, and specific practice is: for example, select a whole day 24h on certain Tuesday, are 5min the interval times of traffic statistics, totally 288 data.Because just go to work morning, so the main prediction case of considering after 8 points morning (i.e. the 96th moment).
After initial data normalization, more regular data are trained, the data on flows on Tuesday next week is carried to back and give a forecast.The mode that adopts the prediction of study limit, limit during prediction, sliding time window is 15, before each prediction, all learns the data on flows in front 15 moment.
When different value is got on the rank that we mainly investigate FIR filter, on the impact predicting the outcome.We are designated as Mh the rank of FIR filter between input layer and hidden layer, and between hidden layer and output layer, the rank of FIR filter are designated as Mo.We have investigated following several situation to rank:
(Mh, Mo) gets (3,3) successively, (5,3), (5,5), (5,10), (10,5).
We select (Mh, Mo) to be followed successively by (5,3), and (5,5), during (3,3) three groups of rank, predict the outcome and compare flow, and its Contrast on effect as shown in Figure 3; Abscissa is time series; Ordinate is normal transmission flow.
Experiment situation is analyzed, and we can obtain these points result:
(1) when (Mh, Mo)=(5,5), prediction effect is more satisfactory, and rank are excessive or too small, and estimated performance all can decline.
(2) when the rank of FIR filter progressively reduce from 5, network configuration is tending towards BP neural net, can not reflect the dynamic-change information of data.
(3), when the rank of FIR filter progressively increase from 5, after FIR filter, the input data of each layer are become smoothly, thereby can not reflect the situation of change of real data.
Prediction by time division traffic obtains predicted value at times, according to predicted value, sets alarming threshold value, realizes the flow early warning of higher reliability.
Although content of the present invention has been done detailed introduction by above preferred embodiment, will be appreciated that above-mentioned description should not be considered to limitation of the present invention.Those skilled in the art, read after foregoing, for multiple modification of the present invention with to substitute will be all apparent.Therefore, protection scope of the present invention should be limited to the appended claims.
Claims (6)
1. a Network Abnormal method for early warning for flow actual prediction Network Based, is characterized in that, comprises following process:
Step 1, by being carried out to real-time collection and continual collection, network data carrys out monitoring network flow;
Step 2, the network flow data obtaining is carried out to statistical analysis;
Step 3, based on ANN network, predict at times the prediction index numerical value that obtains network traffics;
Step 4, according to prediction index numerical value and performance change trend thereof, set Realtime Alerts threshold values;
On step 5, the basis detected in network traffics, the currency of keeper's real time polling Network Capture defining objects is monitored;
When if currency exceeds corresponding alarm threshold value, there is exception of network traffic situation in judgement, thereby report to the police to keeper by the flow detection warning system of expansion; If currency does not surpass alarm threshold value, continue monitoring traffic in network;
ANN network in described Network Abnormal method for early warning, is the FIR neural net that adopts three layers of feed forward architecture, comprises: input layer, hidden layer, and output layer; Between each layer, adopt full connection, and before every layer, all added a FIR filter.
2. Network Abnormal method for early warning as claimed in claim 1, is characterized in that,
Described Network Abnormal method for early warning also comprises the process that ANN network is trained, and training adopts the following algorithm based on Wan:
In formula:
δ
l k(t): reverse propagated error;
W
l ik(t): connect the node of l-1 layer to the connection weights of l layer k node;
C
l ikj(t): the synaptic weight of FIR filter;
X
^l k(t): induction local field;
X
l-1 i(t): the neuronic output of l-1 layer, as the input of FIR filter;
Q
-1: time delay operator;
ε (t): moment error;
N
c: the rank of FIR filter;
D
k(t): t desired output constantly;
Y
k(t): t actual output constantly;
α w, α c: the momentum coefficient that connects weights, synaptic weight;
η w, η c: the learning coefficient that connects weights, synaptic weight;
As Δ ε (t)=ε
t+1-ε
tduring <0, corresponding learning coefficient and momentum coefficient increase respectively a η w in a small amount, η c, otherwise, reduce an a small amount of.
3. Network Abnormal method for early warning as claimed in claim 2, is characterized in that,
When ANN network is trained, get:
Momentum coefficient: α w=0.1, α c=0.1;
Learning coefficient: η w=0.8, η c=0.5;
In a small amount: Δ η=0.02, Δ α=0.001.
4. the Network Abnormal method for early warning as described in claim 1 or 3, is characterized in that,
The rank of the FIR filter between described input layer and hidden layer, are designated as Mh;
The rank of the FIR filter between described hidden layer and output layer, are designated as Mo;
, get (Mh, Mo)=(5,5).
5. Network Abnormal method for early warning as claimed in claim 4, is characterized in that,
Input layer in described ANN network has 7 nodes, and wherein the 1st node is information on working day, then 6 network flow datas that node is continuous 6 moment;
Described hidden layer has 10 nodes; Described output layer has 1 node; And the activation primitive of the node of described hidden layer and the node of output layer all adopts Sigmoid function.
6. Network Abnormal method for early warning as claimed in claim 5, is characterized in that,
During prediction, in one is selected workaday 24 hours, with the interval of every 5 minutes, carry out the data that traffic statistics obtain 288 moment, and predict next week same workaday data on flows according to the data of this zero hour work hours on working day, and the data on flows in 15 moment before also learning in each prediction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310641648.2A CN103716180A (en) | 2013-12-04 | 2013-12-04 | Network flow actual forecasting-based network abnormality pre-warning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310641648.2A CN103716180A (en) | 2013-12-04 | 2013-12-04 | Network flow actual forecasting-based network abnormality pre-warning method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103716180A true CN103716180A (en) | 2014-04-09 |
Family
ID=50408786
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310641648.2A Pending CN103716180A (en) | 2013-12-04 | 2013-12-04 | Network flow actual forecasting-based network abnormality pre-warning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103716180A (en) |
Cited By (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105225018A (en) * | 2015-11-04 | 2016-01-06 | 上海斐讯数据通信技术有限公司 | Based on Forecasting Methodology and the prediction unit of the mobile phone flow of BP neural network |
CN105721199A (en) * | 2016-01-18 | 2016-06-29 | 中国石油大学(华东) | Real-time cloud service bottleneck detection method based on kernel density estimation and fuzzy inference system |
CN106126391A (en) * | 2016-06-28 | 2016-11-16 | 北京百度网讯科技有限公司 | System monitoring method and apparatus |
CN106685749A (en) * | 2015-11-09 | 2017-05-17 | 北京国双科技有限公司 | Network traffic checking method and network traffic checking device |
CN106789214A (en) * | 2016-12-12 | 2017-05-31 | 广东工业大学 | It is a kind of based on the just remaining pair network situation awareness method and device of string algorithm |
CN107135125A (en) * | 2017-05-17 | 2017-09-05 | 北京奇艺世纪科技有限公司 | Video IDC predicting bandwidth flow method and devices |
CN108234496A (en) * | 2018-01-05 | 2018-06-29 | 宝牧科技(天津)有限公司 | A kind of method for predicting based on neural network |
CN108574668A (en) * | 2017-03-10 | 2018-09-25 | 北京大学 | A kind of ddos attack peak flow prediction technique based on machine learning |
CN108632863A (en) * | 2018-04-28 | 2018-10-09 | 中国联合网络通信集团有限公司 | Flow method for early warning, device and server |
CN108768750A (en) * | 2018-06-22 | 2018-11-06 | 广东电网有限责任公司 | Communication network failure localization method and device |
CN109542740A (en) * | 2017-09-22 | 2019-03-29 | 阿里巴巴集团控股有限公司 | Method for detecting abnormality and device |
CN109660419A (en) * | 2018-10-08 | 2019-04-19 | 平安科技(深圳)有限公司 | Predict method, apparatus, equipment and the storage medium of network equipment exception |
CN109981358A (en) * | 2019-03-13 | 2019-07-05 | 南京理工大学 | A kind of adaptive network performance method for early warning based on built-up pattern |
CN109976986A (en) * | 2017-12-28 | 2019-07-05 | 阿里巴巴集团控股有限公司 | The detection method and device of warping apparatus |
CN110224852A (en) * | 2019-04-28 | 2019-09-10 | 中电长城网际安全技术研究院(北京)有限公司 | Network security monitoring method and device based on HTM algorithm |
CN110247788A (en) * | 2018-03-09 | 2019-09-17 | 上海诺基亚贝尔股份有限公司 | Method, equipment and computer readable storage medium for network-control |
CN110417577A (en) * | 2019-06-18 | 2019-11-05 | 东华大学 | One specific admixture data flow shunt volume peak value prediction technique |
CN111147899A (en) * | 2019-12-16 | 2020-05-12 | 南京亚信智网科技有限公司 | Fault early warning method and device |
CN113658415A (en) * | 2021-07-30 | 2021-11-16 | 南京凡科信息科技有限公司 | Early warning method and system for intelligent gateway |
CN113691529A (en) * | 2021-08-24 | 2021-11-23 | 珠海市鸿瑞信息技术股份有限公司 | Industrial control system and method based on network security of power industry |
CN114459575A (en) * | 2022-02-10 | 2022-05-10 | 佘以道 | Set information correlation reference platform and method |
CN114760152A (en) * | 2022-06-14 | 2022-07-15 | 湖南警察学院 | Cloud data center virtualization node network security early warning method |
CN117118907A (en) * | 2023-10-25 | 2023-11-24 | 深圳市亲邻科技有限公司 | Entrance guard flow dynamic monitoring system and method thereof |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110154109A1 (en) * | 2009-12-22 | 2011-06-23 | Xerox Corporation | Continuous, automated discovery of bugs in released software |
CN102821002A (en) * | 2011-06-09 | 2012-12-12 | 中国移动通信集团河南有限公司信阳分公司 | Method and system for network flow anomaly detection |
-
2013
- 2013-12-04 CN CN201310641648.2A patent/CN103716180A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110154109A1 (en) * | 2009-12-22 | 2011-06-23 | Xerox Corporation | Continuous, automated discovery of bugs in released software |
CN102821002A (en) * | 2011-06-09 | 2012-12-12 | 中国移动通信集团河南有限公司信阳分公司 | Method and system for network flow anomaly detection |
Non-Patent Citations (1)
Title |
---|
林雪纲,郑成兴等: ""基于FIR神经网络的以太网网络流量预测"", 《计算机工程》 * |
Cited By (37)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105225018A (en) * | 2015-11-04 | 2016-01-06 | 上海斐讯数据通信技术有限公司 | Based on Forecasting Methodology and the prediction unit of the mobile phone flow of BP neural network |
CN106685749A (en) * | 2015-11-09 | 2017-05-17 | 北京国双科技有限公司 | Network traffic checking method and network traffic checking device |
CN106685749B (en) * | 2015-11-09 | 2019-09-20 | 北京国双科技有限公司 | The method of inspection and device of network flow |
CN105721199B (en) * | 2016-01-18 | 2019-06-07 | 中国石油大学(华东) | A kind of real-time cloud service bottleneck detection method based on Density Estimator and fuzzy inference system |
CN105721199A (en) * | 2016-01-18 | 2016-06-29 | 中国石油大学(华东) | Real-time cloud service bottleneck detection method based on kernel density estimation and fuzzy inference system |
CN106126391A (en) * | 2016-06-28 | 2016-11-16 | 北京百度网讯科技有限公司 | System monitoring method and apparatus |
CN106789214A (en) * | 2016-12-12 | 2017-05-31 | 广东工业大学 | It is a kind of based on the just remaining pair network situation awareness method and device of string algorithm |
CN106789214B (en) * | 2016-12-12 | 2019-10-11 | 广东工业大学 | A kind of network situation awareness method and device based on just remaining double string algorithms |
CN108574668A (en) * | 2017-03-10 | 2018-09-25 | 北京大学 | A kind of ddos attack peak flow prediction technique based on machine learning |
CN108574668B (en) * | 2017-03-10 | 2020-10-20 | 北京大学 | DDoS attack flow peak value prediction method based on machine learning |
CN107135125A (en) * | 2017-05-17 | 2017-09-05 | 北京奇艺世纪科技有限公司 | Video IDC predicting bandwidth flow method and devices |
CN107135125B (en) * | 2017-05-17 | 2020-04-21 | 北京奇艺世纪科技有限公司 | Video IDC bandwidth flow prediction method and device |
CN109542740A (en) * | 2017-09-22 | 2019-03-29 | 阿里巴巴集团控股有限公司 | Method for detecting abnormality and device |
CN109542740B (en) * | 2017-09-22 | 2022-05-27 | 阿里巴巴集团控股有限公司 | Abnormality detection method and apparatus |
CN109976986A (en) * | 2017-12-28 | 2019-07-05 | 阿里巴巴集团控股有限公司 | The detection method and device of warping apparatus |
CN109976986B (en) * | 2017-12-28 | 2023-12-19 | 阿里巴巴集团控股有限公司 | Abnormal equipment detection method and device |
CN108234496A (en) * | 2018-01-05 | 2018-06-29 | 宝牧科技(天津)有限公司 | A kind of method for predicting based on neural network |
CN110247788B (en) * | 2018-03-09 | 2023-01-03 | 上海诺基亚贝尔股份有限公司 | Method, apparatus, and computer-readable storage medium for network control |
CN110247788A (en) * | 2018-03-09 | 2019-09-17 | 上海诺基亚贝尔股份有限公司 | Method, equipment and computer readable storage medium for network-control |
CN108632863A (en) * | 2018-04-28 | 2018-10-09 | 中国联合网络通信集团有限公司 | Flow method for early warning, device and server |
CN108632863B (en) * | 2018-04-28 | 2022-01-25 | 中国联合网络通信集团有限公司 | Traffic early warning method and device and server |
CN108768750A (en) * | 2018-06-22 | 2018-11-06 | 广东电网有限责任公司 | Communication network failure localization method and device |
CN109660419A (en) * | 2018-10-08 | 2019-04-19 | 平安科技(深圳)有限公司 | Predict method, apparatus, equipment and the storage medium of network equipment exception |
CN109981358A (en) * | 2019-03-13 | 2019-07-05 | 南京理工大学 | A kind of adaptive network performance method for early warning based on built-up pattern |
CN110224852A (en) * | 2019-04-28 | 2019-09-10 | 中电长城网际安全技术研究院(北京)有限公司 | Network security monitoring method and device based on HTM algorithm |
CN110417577B (en) * | 2019-06-18 | 2021-11-26 | 东华大学 | Method for predicting flow splitting peak value of hybrid data stream |
CN110417577A (en) * | 2019-06-18 | 2019-11-05 | 东华大学 | One specific admixture data flow shunt volume peak value prediction technique |
CN111147899A (en) * | 2019-12-16 | 2020-05-12 | 南京亚信智网科技有限公司 | Fault early warning method and device |
CN113658415A (en) * | 2021-07-30 | 2021-11-16 | 南京凡科信息科技有限公司 | Early warning method and system for intelligent gateway |
CN113658415B (en) * | 2021-07-30 | 2024-03-26 | 江苏湛德医疗用品有限公司 | Early warning method and system of intelligent gateway |
CN113691529A (en) * | 2021-08-24 | 2021-11-23 | 珠海市鸿瑞信息技术股份有限公司 | Industrial control system and method based on network security of power industry |
CN113691529B (en) * | 2021-08-24 | 2022-03-11 | 珠海市鸿瑞信息技术股份有限公司 | Industrial control system and method based on network security of power industry |
CN114459575A (en) * | 2022-02-10 | 2022-05-10 | 佘以道 | Set information correlation reference platform and method |
CN114760152B (en) * | 2022-06-14 | 2022-08-19 | 湖南警察学院 | Cloud data center virtualization node network security early warning method |
CN114760152A (en) * | 2022-06-14 | 2022-07-15 | 湖南警察学院 | Cloud data center virtualization node network security early warning method |
CN117118907A (en) * | 2023-10-25 | 2023-11-24 | 深圳市亲邻科技有限公司 | Entrance guard flow dynamic monitoring system and method thereof |
CN117118907B (en) * | 2023-10-25 | 2024-02-02 | 深圳市亲邻科技有限公司 | Entrance guard flow dynamic monitoring system and method thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103716180A (en) | Network flow actual forecasting-based network abnormality pre-warning method | |
CN108418841B (en) | Next-generation key message infrastructure network Security Situation Awareness Systems based on AI | |
Dudnyk et al. | Development of a method for training artificial neural networks for intelligent decision support systems | |
CN104486141B (en) | A kind of network security situation prediction method that wrong report is adaptive | |
Zhang et al. | Real-time assessment of fault-induced delayed voltage recovery: A probabilistic self-adaptive data-driven method | |
CN105989441A (en) | Model parameter adjustment method and device | |
CN108400895A (en) | One kind being based on the improved BP neural network safety situation evaluation algorithm of genetic algorithm | |
CN201898519U (en) | Equipment maintenance early-warning device with risk control | |
CN108763729A (en) | Process industry Mechatronic Systems couple state appraisal procedure based on network structure entropy | |
CN105825271A (en) | Satellite fault diagnosis and prediction method based on evidential reasoning (ER) | |
CN1996888A (en) | A detection method and detection device for exceptional network traffic | |
Pierini et al. | Prediction of water flows in Colorado River, Argentina | |
Yanchong et al. | A short-term traffic flow prediction method based on wavelet analysis and neural network | |
CN112232604A (en) | Prediction method for extracting network traffic based on Prophet model | |
Ruslan et al. | Flood modelling using artificial neural network | |
Zeng et al. | Predictive model for internet public opinion | |
Lasfer et al. | Neural network design parameters for forecasting financial time series | |
Vafakhah et al. | Application of several data-driven techniques for rainfall-runoff modeling | |
Coulibaly et al. | Neural network‐based long‐term hydropower forecasting system | |
Wang et al. | Network traffic prediction based on improved BP wavelet neural network | |
Ma et al. | Traffic flow forecasting research based on Bayesian normalized Elman neural network | |
Lin et al. | A semi-supervised learning approach for abnormal event prediction on large network operation time-series data | |
Ning et al. | Using chaotic neural network to forecast stock index | |
Ishak et al. | Modelling reservoir water release decision using temporal data mining and neural network | |
Wen et al. | A network security situation awareness method based on GRU in big data environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140409 |
|
RJ01 | Rejection of invention patent application after publication |