CN115348183B - Flow prediction method and device - Google Patents
Flow prediction method and device Download PDFInfo
- Publication number
- CN115348183B CN115348183B CN202210978761.9A CN202210978761A CN115348183B CN 115348183 B CN115348183 B CN 115348183B CN 202210978761 A CN202210978761 A CN 202210978761A CN 115348183 B CN115348183 B CN 115348183B
- Authority
- CN
- China
- Prior art keywords
- prediction
- prediction model
- flow
- data
- time
- 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 43
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 7
- 238000005516 engineering process Methods 0.000 claims abstract description 6
- 230000015654 memory Effects 0.000 claims description 8
- 230000000750 progressive effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000007493 shaping process Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 abstract description 5
- 238000005457 optimization Methods 0.000 description 4
- 238000013480 data collection Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000002790 cross-validation Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012952 Resampling Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Classifications
-
- 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
- H04L41/142—Network analysis or design using statistical or mathematical methods
-
- 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
-
- 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
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Pure & Applied Mathematics (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Mathematical Optimization (AREA)
- Medical Informatics (AREA)
- Mathematical Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Algebra (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a flow prediction method, which comprises the following steps: by introducing an autocorrelation error coefficient of a flow time sequence, in the process of learning a prediction model, a learning sample is converted into difference data of front and rear time sequence band coefficients, the difference data is used as sample data for learning the prediction model, the prediction model is trained, and the flow is predicted by the prediction model obtained by training, so that the relevant influence factors of the front and rear prediction errors are considered in the prediction model training. The task logic of the prediction model provided by the invention is simple, easy to realize and occupies less resources; the improvement of flow prediction accuracy is realized through the introduction of the parameter characteristics of the sequence autocorrelation error; the invention combines the neural network technology in the prediction model modeling, and improves the modeling efficiency. The invention also provides a corresponding flow prediction device.
Description
Technical Field
The invention belongs to the technical field of network management, and particularly relates to a flow prediction method and device.
Background
When network traffic is actually predicted, the traffic may be affected by factors such as network status, user size, device configuration, device operating environment, and the like. Many factors influence the accuracy of network traffic prediction, however, the complete network traffic prediction cannot be fully considered in modeling for various reasons, so that the modeling feature selection cannot achieve full coverage in the network traffic prediction. These uncertainty factors, if converted into correlation error factors, can effectively improve the accuracy of the predictive model when taken into account during modeling.
Disclosure of Invention
Aiming at the defects or improvement demands of the prior art, in order to reduce the influence of other factors such as network environment on flow prediction, the invention provides an improved flow prediction method which is used for optimizing the learning characteristics of a prediction model and improving the flow prediction precision of the prediction model.
To achieve the above object, according to one aspect of the present invention, there is provided a flow prediction method, the method including: by introducing an autocorrelation error coefficient of a flow time sequence, in the process of learning a prediction model, a learning sample is converted into difference data of front and rear time sequence band coefficients, the difference data is used as sample data for learning the prediction model, the prediction model is trained, and the flow is predicted by the prediction model obtained by training, so that the relevant influence factors of the front and rear prediction errors are considered in the prediction model training.
In one embodiment of the present invention, the autocorrelation error coefficient ρ is the correlation coefficient of the error variables e t and e t-1, and the calculation formula is: that is, the autocorrelation error coefficient is calculated by dividing the sum of the products of the front and rear prediction errors by the sum of squares of the prediction model prediction errors at the specified time, where e t represents the prediction model prediction error at time T, e t-1 represents the prediction model prediction error at time T-1, and T represents the length of the traffic time series.
In one embodiment of the present invention, the learning sample is converted into difference data of front and rear time sequence band coefficients, specifically: the input flow sampling data is subjected to data shaping, a new learning sample sequence is obtained by converting a learning sample into an autocorrelation differential form S t-ρSt-1 with coefficients, S t represents an actual observed value at the time t, S t-1 represents an actual observed value at the time t-1, and ρ is an autocorrelation error coefficient.
In one embodiment of the present invention, the prediction model is ：St-ρSt-1＝M(St-w-ρSt-w-1,St-w+1-ρSt-w,……St-1-ρSt-2;λ)+∈t,, where M () represents a function of the prediction model method, including input parameters of the prediction model, including S t-w-1,St-w parameters and prediction model itself parameter λ, where S t-w-1 represents a starting data point of an input prediction sample point, S t-w represents a next sample point of the starting data point of the input prediction sample point, and so on, e t represents random noise in the prediction error of the prediction model at time t.
In one embodiment of the invention, the training of the prediction model is specifically: and carrying the obtained new learning sample sequence S t-ρSt-1 into a prediction model for training, searching for the mean square error of the minimized prediction model by combining a neural network technology, and converging to determine the parameter lambda and the autocorrelation error coefficient rho of the prediction model, thereby finally obtaining the new prediction model.
In one embodiment of the invention, the flow prediction is performed by a prediction model obtained through training, specifically: based on a prediction model obtained by training, the prediction result is an intermediate quantity containing a predicted value at the moment t, and the formula is adoptedAnd obtaining a predicted value at the time t, wherein Fm represents a result of prediction by a prediction model.
In one embodiment of the present invention, the traffic time sequence includes: network traffic data of the current network device is collected, including characteristic data of interest of the network device traffic.
In one embodiment of the invention, for sample range selection of flow data, a preamble progressive mode is adopted to select a training data set and a test data set in batches, and a training set progressive mode is adopted to learn prediction model parameters.
In one embodiment of the present invention, the initial value of the autocorrelation error coefficient ρ is 0, and the prediction model's own parameter λ is a random value.
According to another aspect of the present invention, there is also provided a traffic prediction apparatus, including at least one processor and a memory, the at least one processor and the memory being connected by a data bus, the memory storing instructions to be executed by the at least one processor, the instructions, after being executed by the processor, being configured to perform the traffic prediction method described above.
In general, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:
(1) The task logic of the prediction model provided by the invention is simple, easy to realize and occupies less resources;
(2) The improvement of flow prediction accuracy is realized through the introduction of the parameter characteristics of the sequence autocorrelation error;
(3) The invention combines the neural network technology in the prediction model modeling, and improves the modeling efficiency.
Drawings
FIG. 1 is a block diagram of a primary flow prediction process according to an embodiment of the present invention;
FIG. 2 is a flowchart of learning a predictive model in an embodiment of the invention;
Fig. 3 is a schematic diagram of a progressive cross-validation of a preamble of traffic sample data in an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
As shown in fig. 1, the method introduces an autocorrelation error coefficient ρ to participate in modeling calculation of a prediction model. The method is mainly different from the current method for directly taking the flow time sequence data as a training sample to train the predictive model in the training process of the predictive model. By introducing the autocorrelation error coefficient of the flow time sequence, in the process of learning a prediction model, a learning sample is converted into a difference value of the front time sequence band coefficient and the rear time sequence band coefficient, and the difference value data is used as sample data for learning the prediction model to perform prediction model training, so that the correlation influence factors of the front prediction error and the rear prediction error are considered in the prediction model training. Thus, in the improved method of the present invention, the output of the predictive model is the difference in the flow data at the previous and subsequent moments. Since the flow data at the preamble time is known, prediction of the flow data at the current time by the difference and the preamble flow data can still be ensured.
The prediction error of the prediction model at the time t is designed to have autocorrelation with the prediction error of the prediction model at the time t-1, so the prediction error of the prediction model at the time t is defined as:
et＝ρet-1+∈t
e t denotes the prediction model prediction error at time t, e t-1 denotes the prediction model prediction error at time t-1, and e t denotes random noise in the prediction model prediction error at time t. That is, the prediction error of the prediction model at the specified time is affected by the prediction error of the prediction model at the previous time and the random factor at the specified time.
Therefore, the prediction error of the prediction model at the time t has correlation with the prediction error of the prediction model at the previous time, and error factors caused by random noise at the time t are still reserved.
The autocorrelation error coefficient ρ is the correlation coefficient of the error variables e t and e t-1, and the calculation formula is:
that is, the sequence autocorrelation error coefficient with length T, which represents the length of the flow time sequence, can be calculated by dividing the sum of the products of the front and rear prediction errors by the sum of squares of the prediction errors of the prediction model at the specified time.
Thus, the parameter ρ can be derived from the forward and backward prediction error values: e t and e t-1.
As a result of: Wherein/> The predicted value at time t is shown, and S t is the actual observed value at time t. That is, the error at time t is the difference between the actual time t flow value and the predicted time t flow value. The error at time t-1 is the difference between the actual flow value and the predicted value at time t-1.
According to the above, for the relation between the error at time t and the error at time t-1, the scheme of the invention has the following formula:
Since the flow predictor is obtained by the predictive model method and the historical sample flow values, M () represents a function of the predictive model method, including the input parameters of the predictive model, such as the following S t-w,St-w+1 parameters, and the parameters of the predictive model itself, such as λ. The above formula can thus be transformed into the following form:
St-ρSt-1＝M(St-w,St-w+1,……St-2,St-1；λ)
-ρM(St-w-1,St-w,……St-3,St-2;λ)+∈t
I.e. the flow difference between time t and time t-1 can be obtained by adding a random error to the value predicted by the prediction model. The input of the predicted value of the prediction model is flow sample data before a predicted time point, and the scale of the input sample data is selected by the characteristic engineering. Such as predicting the current time flow data from the first 10 samples. Here S t-w-1 represents the starting data point of the input predicted sample point, S t-w represents the next sample point of the starting data point of the input predicted sample point, and so on. In the case of 10 sample points, w is 10, where λ represents a parameter of the prediction model itself.
Since the prediction model is unchanged, the final prediction model can be simplified by adjusting the input data form of the prediction model, and the following prediction model form can be finally obtained by combining the above forms:
St-ρSt-1＝M(St-w-ρSt-w-1,St-w+1-ρSt-w,……St-1-ρSt-2;λ)+∈t
that is, by inputting the difference value of the front and rear time series flow data with the coefficient as the prediction input value of the prediction model, the prediction of the flow difference value of the corresponding front and rear time series band coefficient can be obtained. Since the predicted flow difference is the front-back flow difference, and the flow data at the previous time is known for the flow at the current time, the flow at the current time can be predicted based on the predicted flow difference and the flow value at the previous time.
In summary, the present invention performs data shaping on the input flow sampling data, and converts the learning sample into the autocorrelation differential form with coefficients: s t-ρSt-1, obtaining a new learning sample sequence, carrying the sample sequence into a prediction model for training, and combining a neural network technology, and determining a parameter lambda and an autocorrelation error coefficient rho of the prediction model by searching a mean square error (Mean Square Error, MSE) of the minimum prediction model and converging. Based on a prediction model obtained by training, the prediction result is an intermediate quantity containing a predicted value at the time t, and the formula is adopted:
the predicted value at the time t can be obtained, and Fm represents the result of prediction by the prediction model.
As shown in fig. 2, the overall scheme needs to combine the collected data and perform sample shaping to learn and predict the prediction model. The specific embodiments are described below. The optimization method is suitable for being realized on the neural network technology.
In step S201, network traffic data of the current network device is collected, where feature data, such as the number of transmission packets, the number of transmission bytes, etc., of interest of the network device traffic is mainly collected. The period of data collection may be minute granularity and the frequency of raw data collected may be different.
Step S202, selecting a proper sample data acquisition period according to the time length to be predicted. Such as predicting a daily flow rate change, the sample data collection period may be set to be hour data. Resampling can ensure that the input granularity for the prediction model meets the prediction time requirement of the prediction model on data preprocessing. And in this step, normalization or unit scaling of the data values may be considered, depending on the characteristics of the particular flow data values. To improve the efficiency of subsequent training.
In step S203, in order to obtain the optimized prediction model parameters, the cross-validation method needs to be considered based on the collected historical data. Particularly for learning of the neural network method, it is necessary to specify a learning processing lot of the sample. As shown in fig. 3, for sample range selection of flow data, the selection of the training data set and the test data set may be performed in batches in a progressive manner, and the prediction model parameter learning may be performed in a progressive manner of the training set.
In step S204, the improvement method needs to consider the autocorrelation error coefficient ρ in addition to the parameter λ of the prediction model itself. To obtain the determined ρ, the initial value of ρ may be set to 0. The prediction model itself parameter lambda is a random value.
In step S205, since a ρ value is determined, the formula can be used: s t-ρSt-1, carrying out differential processing of first-order band coefficients on the training data set and the test data set data, and particularly shown in figure 1.
Step S206, obtaining a predicted value based on the existing rho value and lambda parameter
In step S207, the loss function target of the prediction model targets minimizing the prediction mean square error, and the MSE formula is:
In step S208, each MSE corresponds to a set of λ and ρ. Minimizing the MSE, i.e. finding the predictive model parameter λ by stepwise or GRID SEARCH methods, has a feedback effect on the optimization of λ due to the introduction of ρ by the inventive method optimization. The method is specifically shown in step S209, where the minimum MSE is the minimum value found under the specified maximum convergence parameter and condition by the stepwise or GRID SEARCH method. When the minimum MSE needs to be continuously found, the ρ is triggered to be updated, step S209.
In step S209, since the prediction value is obtained in the iterative learning, and the prediction error is known, the correlation formula can be:
μ represents the mean value, T represents the length of the flow time series;
Since the error e t at time T accords with the standard integral distribution and the average value is 0, the first-order autocorrelation error coefficient of the flow time sequence with the length of T can be obtained:
thus, the autocorrelation error coefficients may be updated based on each time the MSE is obtained
When further iterative learning is needed, the method can be usedUpdating the sample data, and restarting a round of learning process, i.e. returning to step S205;
Step S210, the final ρ parameter and the predictive model parameter λ can be obtained through the above steps.
As shown in the following table, the current network collects flow data of 4 ports, performs flow data collection according to flow collection periods of 15 minutes and 1 hour, and performs prediction model learning effect comparison of a common method and an optimization method through a TCN (time convolutional neural network) and an LSTM (long short term memory neural network) prediction model. The comparison index is the predicted MSE (mean square error) data after the prediction model is learned. The smaller the MSE value, the smaller the prediction error of the learned prediction model, and the higher the prediction precision. The smaller MSE in the contrast data has been marked by bold fonts.
The invention further provides a flow prediction device, which comprises at least one processor and a memory, wherein the at least one processor and the memory are connected through a data bus, the memory stores instructions executed by the at least one processor, and the instructions are used for completing the flow prediction method after being executed by the processor.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (6)
1. A method of traffic prediction, the method comprising: by introducing an autocorrelation error coefficient of a flow time sequence, in the process of learning a prediction model, a learning sample is converted into difference data of front and rear time sequence band coefficients, the difference data is used as sample data for learning the prediction model, the prediction model is trained, and the flow prediction is carried out by the prediction model obtained by training, so that the relevant influence factors of the front and rear prediction errors are considered in the prediction model training, wherein the method comprises the following steps:
The autocorrelation error coefficient ρ is the correlation coefficient of the error variables e t and e t-1, and the calculation formula is:
The autocorrelation error coefficient is calculated by dividing the sum of products of front and rear prediction errors by the square sum of prediction model prediction errors at a specified time, wherein e t represents the prediction model prediction error at the time T, e t-1 represents the prediction model prediction error at the time T-1, and T represents the length of a flow time sequence;
The prediction model is as follows:
St-ρSt-1＝M(St-w-ρSt-w-1,St-w+1-ρSt-w,……St-1-ρSt-2;λ)+∈t
Wherein M () represents a function of the predictive model method, including input parameters of the predictive model, including S t-w-1,St-w parameters and the predictive model itself parameter λ, where S t-w-1 represents a starting data point of the input predictive sample point, S t-w represents a next sample point of the starting data point of the input predictive sample point, and so on, and ε t represents random noise in the predictive model prediction error at time t;
the prediction model training is specifically as follows:
Bringing the obtained new learning sample sequence S t-ρSt-1 into a prediction model for training, searching for the mean square error of the minimized prediction model by combining a neural network technology, and converging to determine the parameter lambda and the autocorrelation error coefficient rho of the prediction model, so as to finally obtain a new prediction model;
the traffic time sequence includes:
network traffic data of the current network device is collected, including characteristic data of interest of the network device traffic.
2. The flow prediction method according to claim 1, wherein the learning samples are converted into difference data of front and rear time-series band coefficients, specifically:
The input flow sampling data is subjected to data shaping, a new learning sample sequence is obtained by converting a learning sample into an autocorrelation differential form S t-ρSt-1 with coefficients, S t represents an actual observed value at the time t, S t-1 represents an actual observed value at the time t-1, and ρ is an autocorrelation error coefficient.
3. The flow prediction method according to claim 1, wherein the flow prediction is performed by a prediction model obtained through training, specifically:
based on a prediction model obtained by training, the prediction result is an intermediate quantity containing a predicted value at the moment t, and the formula is adopted And obtaining a predicted value at the time t, wherein Fm represents a result of prediction by a prediction model.
4. The flow prediction method according to claim 1, wherein for sample range selection of flow data, selection of training data sets and test data sets is performed in batches in a progressive manner, and prediction model parameter learning is performed in a progressive manner of the training sets.
5. The flow prediction method according to claim 1, wherein the initial value of the autocorrelation error coefficient ρ is 0 and the prediction model self parameter λ is a random value.
6. A flow prediction device, characterized in that:
Comprising at least one processor and a memory connected by a data bus, the memory storing instructions for execution by the at least one processor, the instructions, upon execution by the processor, for performing the flow prediction method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210978761.9A CN115348183B (en) | 2022-08-16 | 2022-08-16 | Flow prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210978761.9A CN115348183B (en) | 2022-08-16 | 2022-08-16 | Flow prediction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115348183A CN115348183A (en) | 2022-11-15 |
| CN115348183B true CN115348183B (en) | 2024-05-14 |
Family
ID=83952700
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210978761.9A Active CN115348183B (en) | 2022-08-16 | 2022-08-16 | Flow prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115348183B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06337852A (en) * | 1993-05-31 | 1994-12-06 | Hitachi Ltd | Time series prediction method by neural network |
| CN110460622A (en) * | 2019-09-12 | 2019-11-15 | 贵州电网有限责任公司 | A kind of network anomaly detection method based on Situation Awareness prediction technique |
| WO2021047270A1 (en) * | 2019-09-09 | 2021-03-18 | 中兴通讯股份有限公司 | Network traffic prediction method, communication device and storage medium |
| CN112906982A (en) * | 2021-03-22 | 2021-06-04 | 哈尔滨理工大学 | GNN-LSTM combination-based network flow prediction method |
-
2022
- 2022-08-16 CN CN202210978761.9A patent/CN115348183B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06337852A (en) * | 1993-05-31 | 1994-12-06 | Hitachi Ltd | Time series prediction method by neural network |
| WO2021047270A1 (en) * | 2019-09-09 | 2021-03-18 | 中兴通讯股份有限公司 | Network traffic prediction method, communication device and storage medium |
| CN110460622A (en) * | 2019-09-12 | 2019-11-15 | 贵州电网有限责任公司 | A kind of network anomaly detection method based on Situation Awareness prediction technique |
| CN112906982A (en) * | 2021-03-22 | 2021-06-04 | 哈尔滨理工大学 | GNN-LSTM combination-based network flow prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 基于ARIMA模型的网络流量预测研究;王保民;何月梅;温炎耿;;邯郸学院学报;20080915(03);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115348183A (en) | 2022-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108846517B (en) | Integration method for predicating quantile probabilistic short-term power load | |
| CN111275240B (en) | Load prediction method based on multi-energy coupling scene | |
| CN109376772B (en) | Power load combination prediction method based on neural network model | |
| CN107886160B (en) | BP neural network interval water demand prediction method | |
| Wang et al. | Nonparametric probabilistic forecasting for wind power generation using quadratic spline quantile function and autoregressive recurrent neural network | |
| CN114781723A (en) | Short-term photovoltaic output prediction method based on multi-model fusion | |
| CN114329347B (en) | Method and device for predicting metering error of electric energy meter and storage medium | |
| CN112084461A (en) | KNN real-time correction method based on historical flood learning | |
| CN115348183B (en) | Flow prediction method and device | |
| CN114819178A (en) | Railway construction progress index prediction and online updating method | |
| CN111724004B (en) | Reservoir available water supply amount forecasting method based on improved quantum wolf algorithm | |
| CN111259340B (en) | Saturation load prediction method based on logistic regression | |
| CN114266423B (en) | Multi-model-based power energy demand prediction method and device and storage medium | |
| CN112132334A (en) | Method for predicting yield of urban domestic garbage | |
| CN114970711A (en) | Construction method and application of power load prediction model based on Seq2Seq | |
| CN115860547A (en) | Rural electrification digital formation effect evaluation method based on combined empowerment-TOPSIS | |
| CN115622034A (en) | Internet of things meter power load identification method based on improved genetic neural network | |
| CN109376957B (en) | Method for predicting load of thermal power plant | |
| CN113807019A (en) | MCMC wind power simulation method based on improved scene classification and coarse grain removal | |
| CN110633917A (en) | New energy policy implementation effect comprehensive evaluation method based on electric power big data | |
| CN114742258B (en) | Migratable residential household power load scene prediction method | |
| CN111525553B (en) | New energy output error credible interval estimation method under prediction power optimization segmentation | |
| CN117494888A (en) | Distributed photovoltaic power prediction method and system based on optimized cascade | |
| Kong et al. | Wind Speed Interval Prediction Model Based on Adaptive Decomposition and Parameter Optimization | |
| CN116845879A (en) | Improved AdaRNN-based power load prediction method and system |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230328 Address after: 430000 No. 6, High-tech Fourth Road, Donghu High-tech Development Zone, Wuhan City, Hubei Province Applicant after: FIBERHOME TELECOMMUNICATION TECHNOLOGIES Co.,Ltd. Address before: No.4, Guanshan 2nd Road, Donghu Development Zone, Wuhan City, Hubei Province 430205 Applicant before: WUHAN FIBERHOME TECHNICAL SERVICES Co.,Ltd. Applicant before: FIBERHOME TELECOMMUNICATION TECHNOLOGIES Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |