CN106971548B - The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines - Google Patents
The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines Download PDFInfo
- Publication number
- CN106971548B CN106971548B CN201710352326.4A CN201710352326A CN106971548B CN 106971548 B CN106971548 B CN 106971548B CN 201710352326 A CN201710352326 A CN 201710352326A CN 106971548 B CN106971548 B CN 106971548B
- Authority
- CN
- China
- Prior art keywords
- traffic flow
- short
- support vector
- vector machines
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/065—Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to a kind of Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines.Gaussian kernel function and Polynomial kernel function are combined to construct multi-core adaptive support vector machines (AMSVM)；Parameter optimization is carried out to AMSVM using APSO algorithm (APSO)；Historical data and real time data are considered simultaneously, propose the short-term traffic flow forecasting model based on AMSVM；Traffic flow data collection is inputted, the prediction result of short-term traffic flow is generated using prediction model；According to the prediction result of traffic flow and actual traffic data, evaluation analysis is carried out to prediction error.The method of the present invention can improve the deficiency that existing support vector machines (SVM) method is predicted only with single kernel function, the non-linear variation characteristic with randomness of traffic flow can sufficiently be adapted to, realize real-time, the adaptive prediction to short-term traffic flow, the speed and precision of prediction result is improved simultaneously, and there is certain theoretical reference and realistic meaning.
Description
Technical field
The present invention relates to machine learning and intelligent transportation field, especially a kind of optimizable multi-core adaptive supporting vector
The Short-time Traffic Flow Forecasting Methods of machine.
Background technique
Traffic flow guidance and control are the basic functions of intelligent transportation system (ITS), by issuing effective traffic in real time
Trip information, induction traveler select best trip route, avoid further gathering to congestion regions, guidance and equilibrium traffic
The distribution in the time and space is flowed, realizes the active control to traffic congestion, road network traffic efficiency can be effectively improved, alleviated
Urban traffic blocking, while mitigating resulting environmental pollution and problem of resource waste.And Traffic flow guidance and control system
Normal operation, key foundation be that short-term traffic flow it is real-time, dynamic and precisely prediction.
So far, the traffic flow forecasting method proposed already both at home and abroad main having time serial method, mixes at Kalman filtering
Ignorant theoretical, neural network and support vector machines (SVM) etc..The factor for influencing traffic flow variation is varied, and each other
Has the characteristics that non-linear and randomness.SVM has good self study and nonlinear prediction ability, and can be in small trained sample
Preferable precision of prediction is obtained in the case where this, therefore possesses very important status in short-time traffic flow forecast field.SVM is
A kind of statistical learning method for classification and regression problem, establish VC dimension is theoretical and structural risk minimization it
On, for solving the problems, such as small sample and overlearning, it can guarantee that required extreme value is globally optimal solution.However, existing be based on SVM
Method in actual prediction substantially only with a kind of kernel function, be not sufficient enough to adapt to the variation characteristic of traffic flow.For
Improve the accuracy rate of prediction result, it is necessary to further be improved algorithm and propose better prediction model.
Summary of the invention
The purpose of the present invention is to provide a kind of short-time traffic flow forecasts of optimizable multi-core adaptive support vector machines
Method, this method can improve the deficiency that existing support vector machines (SVM) method is predicted only with single kernel function, can be sufficiently
The non-linear variation characteristic with randomness of traffic flow is adapted to, realizes to real-time, the adaptive prediction of short-term traffic flow, improves simultaneously
The speed and precision of prediction result has certain theoretical reference and realistic meaning.
To achieve the above object, the technical scheme is that a kind of optimizable multi-core adaptive support vector machines
Short-time Traffic Flow Forecasting Methods include the following steps,
Step S1: combination gaussian kernel function and Polynomial kernel function are to construct multi-core adaptive support vector machines；
Step S2: parameter optimization is carried out to multi-core adaptive support vector machines using APSO algorithm；
Step S3: while considering historical data and real time data, it proposes based on multi-core adaptive support vector machines in short-term
Forecasting traffic flow model；
Step S4: input traffic flow data collection generates the prediction result of short-term traffic flow using prediction model；
Step S5: according to the prediction result of traffic flow and actual traffic data, evaluation analysis is carried out to prediction error.
In an embodiment of the present invention, in the step S1, the multi-core adaptive support vector machines passes through such as lower section
Formula building:
Step S11: the mixed kernel function that building gaussian kernel function and Polynomial kernel function are composed:
K(x,xi)=β exp (- γ | | x-xi||2)+(1-β)·[γ(x·xi)+1]q
Wherein, x and xiIndicate that any two reality input vector in sample set, β ∈ [0,1] are the weight system of mixed kernel function
Number, γ are the intrinsic parameter of kernel function, and q is power；
Step S12: weight is adaptively adjusted according to the real-time change of traffic flow trend, that is, numerical value slope size:
Wherein,The first two period traffic flow magnitude of expression corresponding point on two-dimensional surface
(xi-1,yi-1) and (xi-2,yi-2) slope.
In an embodiment of the present invention, in the step S2, the APSO algorithm is used to using adaptive
Property weight, Studying factors and the flight time factor with inertia weight dynamic adjustment learning strategy；In APSO algorithm
The current location of each particle is the current value of undetermined parameter: χ=(C, ε, γ), by optimization process search for obtain it is required from
Adapt to the optimized parameter of multi-kernel support vector machine.
In an embodiment of the present invention, in the step S3, the short-term traffic flow forecasting model detailed process is such as
Under:
Step S31: adaptive population environmental parameter is initialized；
Step S32: by parameter to be asked as the position vector of particle, and particle rapidity and position are initialized；
Step S33: input history average traffic flow data is carried out to training module by multi-core adaptive support vector machines
Mapping and recurrence；
Step S34: the fitness value of particle is calculated, and according to the speed of more new particle and position；
Step S35: being directed to each particle, repeats step S32~S34；
Step S36: being iterated update according to step S33~S35, limits until the training error of sample meets precision,
Optimized parameter C, ε and γ are exported at this time to prediction module；
Step S37: the real-time traffic flow data of input top n period to prediction module: X=[Xt-N,…,Xt-1,Xt], root
β value is calculated according to real time data, then training multi-core adaptive support vector machines under the conditions of optimized parameter, and predicts to export next
The magnitude of traffic flow of period.
In an embodiment of the present invention, in the step S4, traffic flow data collection is pre-processed, step is input to
Prediction model described in S3, short-time traffic flow forecast of the output target road section in subsequent period after model is optimized and trained
Value.
In an embodiment of the present invention, in the step S5, prediction error is evaluated using evaluation criterion below
Analysis:
1) average absolute percentage error (MAPE):
2) root-mean-square error (RMSE):
3) related coefficient (R):
Wherein, YiFor actual traffic amount,For the magnitude of traffic flow of prediction, n is number of samples.
Compared to the prior art, the invention has the following advantages: one kind proposed by the invention is optimizable adaptive
The Short-time Traffic Flow Forecasting Methods of multi-kernel support vector machine are answered, existing SVM method can be improved and carried out in advance only with single kernel function
The deficiency of survey can sufficiently adapt to the non-linear variation characteristic with randomness of traffic flow, realize to the real-time, adaptive of short-term traffic flow
It should predict, while improve the speed and precision of prediction result.
Detailed description of the invention
Fig. 1 is the short-time traffic flow forecast mould based on optimizable multi-core adaptive support vector machines in the embodiment of the present invention
Type.
Fig. 2 (a) be utilized in the embodiment of the present invention invented method predict certain section generated peak period in short-term
The magnitude of traffic flow.
Fig. 2 (b) is short to peak period using ADAPTIVE MIXED kernel function and fixed mixed kernel function in the embodiment of the present invention
When forecasting traffic flow Performance Evaluating Indexes.
Fig. 3 is the comparison of the method invented in the embodiment of the present invention and the prior art to Forecasting Short-term Traffic.
Specific embodiment
With reference to the accompanying drawing, technical solution of the present invention is specifically described.
The present invention proposes a kind of Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines, specifically presses
It is realized according to following steps:
Step S1 combines gaussian kernel function and Polynomial kernel function to construct multi-core adaptive support vector machines.
In the present embodiment, since the distribution of data in different characteristic space is different, the performance of support vector machines very great Cheng
The selection of kernel function is depended on degree.Kernel function can be divided into local kernel function and global kernel function, local kernel function by type
Habit ability is strong, but generalization ability is weaker；Global kernel function generalization ability is strong, and learning ability is then relatively weak.Gaussian radial basis function
Kernel function (RBF) and Polynomial kernel function (Poly) are respectively typical local kernel function, global kernel function, therefore they are often
It is used for forecasting traffic flow.The formula of the two is as follows:
K(x,xi)=exp (- γ | | x-xi||2) (1)
K(x,xi)=[γ (xxi+1)]q (2)
Wherein, (1) formula is RBF, and (2) formula is Poly, x and xiIndicate that any two reality input vector in sample set, γ are
The intrinsic parameter of kernel function simultaneously determines the distribution that data are mapped to after new feature space, and q is the power of Poly.To adapt to hand over
Through-flow non-linear and randomness, while the reliability and accuracy of prediction result are improved, multi-core adaptive of the invention is supported
The mixed kernel function that vector machine is composed using gaussian kernel function and Polynomial kernel function, expression formula are as follows:
K(x,xi)=β exp (- γ | | x-xi||2)+(1-β)·[γ(x·xi)+1]q (3)
Wherein, β ∈ [0,1] is the weight coefficient of mixed kernel function.
It further, is the advantage both made full use of, the present invention is according to real-time change trend, that is, numerical value of traffic flow
Slope size adaptively adjusts weight, and formula is as follows:
Wherein,The first two period traffic flow magnitude of expression corresponding point on two-dimensional surface
(xi-1,yi-1) and (xi-2,yi-2) slope.When slope | when k reduces, curve tends to be flat, should enhance the overall situation of kernel function at this time
Generalization ability increases the weight of Polynomial kernel function, the corresponding value for reducing β；When slope | k | when increase, curve tends to sharply,
The local learning ability of kernel function should be enhanced at this time, that is, increase the weight of gaussian kernel function, the corresponding value for increasing β.
Step S2 carries out parameter optimization to AMSVM using APSO algorithm.
In the present embodiment, it is known that undetermined parameter includes: penalty coefficient C, insensitive loss coefficient ε and the choosing of parameter γ, q value
It is taken as 2.The main thought of APSO is using adaptive inertia weight, with season Studying factors and the flight time factor with inertia
Weight dynamic adjusts.When Optimized model, enable the current location of each particle in APSO for the current value of undetermined parameter: χ=(C, ε,
γ), then by algorithm optimization process searches to global optimum position be element in vector optimal solution, that is, it is required
AMSVM optimized parameter.
Step S3, while considering historical data and real time data, propose the short-term traffic flow forecasting model based on AMSVM.
In the present embodiment, target road section subsequent period the magnitude of traffic flow mainly with the history average of corresponding period,
The real value of preceding several periods is related.In order to accurately reflect influence of the traffic flow changing rule to predicted value, mould is predicted in building
When type, using history average traffic flow data, former period traffic flow datas as input variable.In addition, definition backtracking coefficient N table
Show and is predicted in real time using the traffic data of top n period.First the history average traffic flow data of input as training
Sample carries out parameter optimization to AMSVM using APSO algorithm proposed in this paper；Then the real time data of input is utilized
AMSVM after training optimization, and predict the magnitude of traffic flow of subsequent period.Model detailed process is as follows:
1) adaptive population environmental parameter is initialized；
2) by parameter to be asked as the position vector of particle, and particle rapidity and position are initialized；
3) input history average traffic flow data is mapped and is returned by AMSVM to training module；
4) fitness value of particle is calculated, and according to the speed of more new particle and position；
5) it is directed to each particle, repeats step 2)~4)；
6) according to step 3)~5) it is iterated update, until the training error (i.e. fitness value) of sample meets precision limit
System exports optimized parameter C, ε and γ to prediction module at this time；
7) the real-time traffic flow data of top n period is inputted to prediction module: X=[Xt-N,…,Xt-1,Xt], according to formula
(4) β value is calculated, then training AMSVM under the conditions of optimized parameter, and predicts the magnitude of traffic flow of output subsequent period.
Step S4 inputs traffic flow data collection, and the prediction result of short-term traffic flow is generated using prediction model.
In the present embodiment, traffic flow data collection is pre-processed, the prediction model being input in step S3, model warp
Short-time traffic flow forecast value of the output target road section in subsequent period after optimization and training.
Step S5 carries out evaluation analysis to prediction error according to the prediction result of traffic flow and actual traffic data.
In the present embodiment, evaluation analysis is carried out to prediction error using evaluation criterion below:
1) average absolute percentage error (MAPE):
2) root-mean-square error (RMSE):
3) related coefficient (R):
Wherein, YiFor actual traffic amount,For the magnitude of traffic flow of prediction, n is number of samples.Average absolute percentage error and
Root-mean-square error is smaller, indicates that the precision of model prediction result is higher；Related coefficient is bigger, indicates predicting traffic flow amount and reality
The degree of fitting of the magnitude of traffic flow is better.
In order to allow those skilled in the art to further appreciate that a kind of optimizable multi-core adaptive branch proposed by the invention
The Short-time Traffic Flow Forecasting Methods for holding vector machine, elaborate combined with specific embodiments below.The present embodiment is with skill of the present invention
Implemented premised on art scheme, the detailed implementation method and specific operation process are given.
As shown in Figure 1, being the short-term traffic flow forecasting model based on optimizable multi-core adaptive support vector machines.
The present embodiment comprises the following specific steps that:
Step 1: by wagon detector collect target road section traffic flow data, after uploading to traffic information center, can under
Carry corresponding vehicle flowrate data.Assuming that predict target road section in the Short-Term Traffic Flow of certain Monday, need in advance to carry out data
Pretreatment.With 5 minutes for a cycle statistical vehicle flowrate, 288 periods were divided by 24 hours one day.The last week of selection
Working day, that is, the week magnitude of traffic flow calculates their sample data of the average value as training module.Enable backtracking system
Number N=3 predicted the traffic flow of subsequent period (i.e. 5 minutes following) using 15 minutes before current time real time datas
Amount.
Step 2: history average traffic flow data in input step one to training module.Normalizing is carried out to data first
Change processing, to reduce the training time, improve model efficiency.The environmental variance of APSO algorithm is provided that the number of iterations is 200,
Population quantity is 30, and inertia weight value interval is [0.4,0.9], and Studying factors value interval is [0.8,2], ginseng to be optimized
The constant interval of number C, ε, γ are disposed as [0.001,0.1], are limited to 10 with season precision-4.APSO iterative process each time
In, AMSVM carries out cross validation to training sample, until training error meets precision limitation.After optimization, model can be obtained
Optimized parameter C, ε and γ be respectively 0.086,0.021 and 0.045.
Step 3: real-time traffic flow data and optimized parameter are inputted to prediction module.When according to the first two at current time
The flow value of section calculates slope, and the weight of mixed kernel function at this time is calculated according to formula (4), thereby determines that prediction subsequent period
The composition of used kernel function when flow value.Based on optimized parameter and self-adaptive kernel function, worked as using AMSVM mapping and recurrence
The data on flows of (i.e. 3 periods) 15 minutes before the preceding moment, and predict the magnitude of traffic flow of output subsequent period.
As shown in Fig. 2, being that invented method is utilized to predict the target road section generated in the Short-Term Traffic Flow of peak period
And Performance Evaluating Indexes.
The present embodiment comprises the following specific steps that:
Step 1: input target road section is in peak period (15:00-18:00) corresponding history average traffic flow data to instruction
Practice module, after optimized, optimized parameter C, ε and the γ for obtaining model are respectively 5,0.01 and 0.002.Fig. 2 (a) show peak
The prediction result of the phase magnitude of traffic flow, wherein the line of bottommost indicates the slope size of actual traffic fluxion value, and variation tendency can
To reflect the change procedure of mixed kernel function indirectly.At the same time, preset parameter C, ε and γ value is constant, make weight beta [0.0,
1.0] section value obtains estimated performance shown in Fig. 2 (b) and compares.
Step 2:, can be with flexible adaptation by Fig. 2 (a) it is found that optimizable multi-core adaptive support vector machines of the invention
The non-linear and random nature of traffic flow.By Fig. 2 (b) it is found that compared to fixed mixed kernel function weight, possess adaptive
The AMSVM of weight has lower average absolute percentage error and root-mean-square error and higher related coefficient.Thus illustrate
Feasibility and superiority of the AMSVM in autoregressive prediction.
As shown in figure 3, the comparison for institute's inventive technique and the prior art to Forecasting Short-term Traffic.
The present embodiment comprises the following specific steps that:
Step 1: advantage in order to further illustrate the present invention chooses PSO-SVM in the prior art, APSO-SVM-R
It is compared with APSO-SVM-P with APSO-AMSVM.Without loss of generality, Fig. 3 only shows these methods to peak period traffic flow
Prediction result.Table 1 is evaluation index of these methods to 24 hours forecasting traffic flow performances.
The performance of 1 distinct methods forecasting traffic flow result of table compares
Step 2: as shown in Table 1, MAPE, RMSE of APSO-AMSVM method are respectively 10.2608% and 12.3632%,
Both it is lower than other three kinds of methods.Meanwhile R value is 0.9654, is higher than other three kinds of methods.Thus illustrate, compared to existing
There is technology, the present invention possesses higher precision when predicting short-term traffic flow.
Above-mentioned analytic explanation, the traffic proposed by the invention in short-term based on optimizable multi-core adaptive support vector machines
Prediction model is flowed, can be good at the prediction for realizing short-term traffic flow, there is certain theoretical reference and reality with certain
Meaning.
The above are preferred embodiments of the present invention, all any changes made according to the technical solution of the present invention, and generated function is made
When with range without departing from technical solution of the present invention, all belong to the scope of protection of the present invention.
Claims (6)
1. a kind of Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines, it is characterised in that: including such as
Lower step,
Step S1: combination gaussian kernel function and Polynomial kernel function are to construct multi-core adaptive support vector machines；
Step S2: parameter optimization is carried out to multi-core adaptive support vector machines using APSO algorithm；
Step S3: while considering historical data and real time data, propose the traffic in short-term based on multi-core adaptive support vector machines
Flow prediction model；
Step S4: input traffic flow data collection generates the prediction result of short-term traffic flow using prediction model；
Step S5: according to the prediction result of traffic flow and actual traffic data, evaluation analysis is carried out to prediction error.
2. the Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines according to claim 1,
Be characterized in that: in the step S1, the multi-core adaptive support vector machines constructs in the following way:
Step S11: the mixed kernel function that building gaussian kernel function and Polynomial kernel function are composed:
K(x,xi)=β exp (- γ | | x-xi||2)+(1-β)·[γ(x·xi)+1]q
Wherein, x and xiIndicate that any two reality input vector in sample set, β ∈ [0,1] are the weight coefficient of mixed kernel function, γ
For the intrinsic parameter of kernel function, q is power；
Step S12: weight is adaptively adjusted according to the real-time change of traffic flow trend, that is, numerical value slope size:
Wherein,The first two period traffic flow magnitude of expression corresponding point (x on two-dimensional surfacei-1,
yi-1) and (xi-2,yi-2) slope.
3. the Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines according to claim 1,
Be characterized in that: in the step S2, the APSO algorithm is using adaptive inertia weight, Studying factors and flies
The learning strategy that row time factor is adjusted with inertia weight dynamic；The current location of each particle is in APSO algorithm
The current value of undetermined parameter: χ=(C, ε, γ) searches for obtain required multi-core adaptive support vector machines by optimization process
Optimized parameter.
4. the Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines according to claim 1,
Be characterized in that: in the step S3, the short-term traffic flow forecasting model detailed process is as follows:
Step S31: adaptive population environmental parameter is initialized；
Step S32: by parameter to be asked as the position vector of particle, and particle rapidity and position are initialized；
Step S33: input history average traffic flow data is mapped to training module by multi-core adaptive support vector machines
And recurrence；
Step S34: the fitness value of particle, and the speed of more new particle and position are calculated；
Step S35: being directed to each particle, repeats step S32~S34；
Step S36: being iterated update according to step S33~S35, limits until the training error of sample meets precision, at this time
Optimized parameter C, ε and γ are exported to prediction module；
Step S37: the real-time traffic flow data of input top n period to prediction module: X=[Xt-N,…,Xt-1,Xt], according to reality
When data calculate β value, then under the conditions of optimized parameter training multi-core adaptive support vector machines, and predict output subsequent period
The magnitude of traffic flow.
5. the Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines according to claim 1,
It is characterized in that: in the step S4, traffic flow data collection being pre-processed, be input to prediction mould described in step S3
Type, short-time traffic flow forecast value of the output target road section in subsequent period after model is optimized and trained.
6. the Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines according to claim 1,
It is characterized in that: in the step S5, evaluation analysis being carried out to prediction error using evaluation criterion below:
1) average absolute percentage error (MAPE):
2) root-mean-square error (RMSE):
3) related coefficient (R):
Wherein, YiFor actual traffic amount,For the magnitude of traffic flow of prediction, n is number of samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710352326.4A CN106971548B (en) | 2017-05-18 | 2017-05-18 | The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710352326.4A CN106971548B (en) | 2017-05-18 | 2017-05-18 | The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106971548A CN106971548A (en) | 2017-07-21 |
| CN106971548B true CN106971548B (en) | 2019-06-07 |
Family
ID=59326183
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710352326.4A Active CN106971548B (en) | 2017-05-18 | 2017-05-18 | The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106971548B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108427915A (en) * | 2018-02-07 | 2018-08-21 | 浙江理工大学 | A kind of near-infrared method for detecting human face based on improvement Multiple Kernel Learning frame |
| CN109598939B (en) * | 2018-12-24 | 2019-08-27 | 中国科学院地理科学与资源研究所 | A kind of prediction of short-term traffic volume method based on multitask multiple view learning model |
| CN110390419A (en) * | 2019-05-20 | 2019-10-29 | 重庆大学 | Freeway toll station method for predicting based on PSO-LSSVM model |
| CN111833596B (en) * | 2019-11-19 | 2022-04-29 | 东南大学 | Day-by-day road section flow prediction method considering decision inertia of travelers |
| CN113313145B (en) * | 2021-05-10 | 2024-03-15 | 东南大学 | Expressway traffic event detection method based on mixed kernel correlation vector machine |
| CN113763710B (en) * | 2021-09-16 | 2023-01-24 | 汕头大学 | Short-term traffic flow prediction method based on nonlinear adaptive system |
| CN114186168A (en) * | 2021-11-24 | 2022-03-15 | 北京邮电大学 | Correlation analysis method and device for intelligent city network resources |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101964061A (en) * | 2010-09-02 | 2011-02-02 | 北京航空航天大学 | Binary kernel function support vector machine-based vehicle type recognition method |
| CN103150609A (en) * | 2013-02-18 | 2013-06-12 | 健雄职业技术学院 | Modeling method for short time traffic flow predicting model |
| CN103730006A (en) * | 2014-01-26 | 2014-04-16 | 吉林大学 | Short-time traffic flow combined forecasting method |
| CN104091216A (en) * | 2014-07-29 | 2014-10-08 | 吉林大学 | Traffic information predication method based on fruit fly optimization least-squares support vector machine |
| CN105160866A (en) * | 2015-08-07 | 2015-12-16 | 浙江高速信息工程技术有限公司 | Traffic flow prediction method based on deep learning nerve network structure |
-
2017
- 2017-05-18 CN CN201710352326.4A patent/CN106971548B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101964061A (en) * | 2010-09-02 | 2011-02-02 | 北京航空航天大学 | Binary kernel function support vector machine-based vehicle type recognition method |
| CN103150609A (en) * | 2013-02-18 | 2013-06-12 | 健雄职业技术学院 | Modeling method for short time traffic flow predicting model |
| CN103730006A (en) * | 2014-01-26 | 2014-04-16 | 吉林大学 | Short-time traffic flow combined forecasting method |
| CN104091216A (en) * | 2014-07-29 | 2014-10-08 | 吉林大学 | Traffic information predication method based on fruit fly optimization least-squares support vector machine |
| CN105160866A (en) * | 2015-08-07 | 2015-12-16 | 浙江高速信息工程技术有限公司 | Traffic flow prediction method based on deep learning nerve network structure |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106971548A (en) | 2017-07-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106971548B (en) | The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines | |
| Amjady et al. | Day-ahead electricity price forecasting by modified relief algorithm and hybrid neural network | |
| Adamowski | Peak daily water demand forecast modeling using artificial neural networks | |
| CN103390200A (en) | Photovoltaic power station electricity generation output power forecasting method based on similar days | |
| CN105046953A (en) | Short-time traffic-flow combination prediction method | |
| CN104636822A (en) | Residential load prediction method of elman-based neural network | |
| CN108205889A (en) | Freeway traffic flow Forecasting Methodology based on convolutional neural networks | |
| CN108694673A (en) | A kind of processing method, device and the processing equipment of insurance business risk profile | |
| Azad et al. | Prediction of river flow using hybrid neuro-fuzzy models | |
| CN103679263A (en) | Thunder and lightning approach forecasting method based on particle swarm support vector machine | |
| Li et al. | Identifying explicit formulation of operating rules for multi-reservoir systems using genetic programming | |
| CN112801342A (en) | Adaptive runoff forecasting method based on rainfall runoff similarity | |
| Shiau | Optimization of reservoir hedging rules using multiobjective genetic algorithm | |
| CN103093288A (en) | Partition power grid bus load prediction system based on weather information | |
| CN108074004A (en) | A kind of GIS-Geographic Information System short-term load forecasting method based on gridding method | |
| Karami et al. | Optimization of energy management and conversion in the water systems based on evolutionary algorithms | |
| CN108280998A (en) | Short-time Traffic Flow Forecasting Methods based on historical data dynamic select | |
| CN105429134A (en) | Grid voltage stability prediction method based on power big data | |
| Tolabi et al. | A review on classification and comparison of different models in solar radiation estimation | |
| CN109671272A (en) | A kind of freeway traffic flow prediction technique | |
| CN106251242A (en) | A kind of wind power output interval combinations Forecasting Methodology | |
| Nguyen et al. | Flood forecasting in large rivers with data-driven models | |
| CN106100922A (en) | The Forecasting Methodology of the network traffics of TCN and device | |
| Ahmad et al. | Forecast-informed hydropower optimization at long and short-time scales for a multiple dam network | |
| Huang et al. | A hybrid method for one-day ahead hourly forecasting of PV power output |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |