CN109409598A - Link travel time prediction method and device based on SVM and Kalman filtering - Google Patents
Link travel time prediction method and device based on SVM and Kalman filtering Download PDFInfo
- Publication number
- CN109409598A CN109409598A CN201811239031.7A CN201811239031A CN109409598A CN 109409598 A CN109409598 A CN 109409598A CN 201811239031 A CN201811239031 A CN 201811239031A CN 109409598 A CN109409598 A CN 109409598A
- Authority
- CN
- China
- Prior art keywords
- moment
- travel time
- time
- state
- kalman filtering
- 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
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000001914 filtration Methods 0.000 title claims abstract description 32
- 239000011159 matrix material Substances 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims abstract description 32
- 238000005070 sampling Methods 0.000 claims abstract description 18
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 230000001934 delay Effects 0.000 claims abstract description 11
- 239000000203 mixture Substances 0.000 claims abstract description 6
- 230000008569 process Effects 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 2
- 230000008859 change Effects 0.000 abstract description 4
- 238000012706 support-vector machine Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 230000037452 priming Effects 0.000 description 7
- 238000005457 optimization Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000012731 temporal analysis Methods 0.000 description 3
- 238000000700 time series analysis Methods 0.000 description 3
- 230000009977 dual effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 102000004127 Cytokines Human genes 0.000 description 1
- 108090000695 Cytokines Proteins 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000003595 mist Substances 0.000 description 1
- 210000003739 neck Anatomy 0.000 description 1
- 230000001373 regressive effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to traffic control fields, are based particularly on the link travel time prediction method and device of SVM and Kalman filtering.The training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition obtain support vector regression by the training of Support vector regression algorithm；Initial predicted journey time matrix is obtained according to support vector regression and test set data, state equation is constructed according to intersection time delays and condition of road surface in setting stroke segment information, observational equation is constructed according to continuous sampling moment corresponding predicted travel time, weather conditions and the state equation before any moment；Observational equation, which is solved, according to Kalman filtering algorithm obtains any moment corresponding actual prediction time, pass through the corresponding predicted travel time of continuous setting sampling instant before choosing any moment, change the coefficient matrix of Kalman filtering in real time, so that entire prediction model is more rationally, more accurately.
Description
Technical field
The present invention relates to traffic control fields, are based particularly on the link travel time prediction side of SVM and Kalman filtering
Method and device.
Background technique
Link Travel Time is that urban traffic control controls department to traveler offer effective information, rationally carries out traffic
Induction improves the main foundation of traffic utilization rate, therefore, Forecasting of Travel Time (Travel Time Prediction, TTP)
Therefore become domestic and international each experts and scholars' research hotspot.In recent years, flourishing with intelligent transportation system, this field
Research also obtained major progress, currently, both at home and abroad in terms of Forecasting of Travel Time is based particularly on history Forecasting of Travel Time
There are many researchs.For example, improved K mean cluster method, the improved method of moving average and Bayes classifier and being based on
The classifier predicted travel time of rule, but it is very big to be needed data sample amount.Actual existing formation time prediction
Model can be divided into traffic flow theory model, Time Series Analysis Model and machine learning model by its principle difference；Wherein hand over
Through-flow theory model mainly includes macroscopical AC model and pairing model, and Time Series Analysis Model mainly includes returning mould analysis
Method and Kalman filter model method, machine learning model mainly include neural network model and support vector machines (Support
Vector Machine-SVM) model；In above-mentioned prediction model, traffic flow theory model prediction accuracy is lower, and demarcates more
Difficulty, regression model is poor for applicability in Time Series Analysis Model, and precision of prediction is lower, and Kalman filter model precision of prediction
Height, operability is stronger, and can on-line prediction, but it is poor for abnormal observation anti-interference, meanwhile, be based on machine learning side
The precision of forecasting model of method is high, and wherein neural network model has good learning ability, and it is non-thread can preferably to track fitting
Property, Non-Stationary Time Series, but need a large amount of historical data and there is excessive study and owe study and local minimum
The problems such as point, support vector machines are the new tools that Machine Learning Problems are solved by means of optimal method, have complete statistics
Theories of learning basis and outstanding learning ability, still be able in and the higher situation of dimension less to data sample preferably into
Row prediction, it can be difficult to being used for on-line prediction journey time.
Topic name has been published in the periodical of " Traffic transport system engineering and information " the 4th phase of volume 12 of in August, 2012
The paper referred to as " studied based on SVM and the BRT travel time prediction model of Kalman filter ", proposes combination supporting vector machine
The theoretical Comprehensive Model established with Kalman filtering algorithm forms the time according to the separate of support vector machines and influence public transport
Factor, to study each station on route as division points, support vector machines using v- support vector regression as rudimentary algorithm,
Using gaussian radial basis function core as kernel function, using ε-insensitive loss function as loss function；By a part of composing training matrix of data
For training support vector regression, a part constitutes prediction matrix for prediction and inspection result, by cross validation come really
The most optimized parameter for determining support vector regression is trained support vector regression by the most optimized parameter with training matrix,
And with trained support vector regression predict the priming stroke time, by the priming stroke time input Kalman filter into
The dynamic of row result adjusts, which can either overcome the dependence to amount of training data, and have stronger anti-interference
Ability.But sytem matrix, input variable coefficient and the observing matrix that the Kalman filter in above-mentioned model uses take list
Bit matrix can not convert Kalman filter according to actual road conditions, meanwhile, although it is contemplated that the sleet sky in road
The influence of gas, but the ability without handling emergency, therefore obtained prediction result can not accurately reflect practical feelings
Condition needs further to be improved.
Summary of the invention
The object of the present invention is to provide the link travel time prediction method and devices based on SVM and Kalman filtering, use
Kalman filtering can not be changed in real time according to the leading portion journey time of vehicle to solve existing SVM and Kalman filtering binding model
The coefficient matrix problem that causes current Forecasting of Travel Time result not accurate.
In order to realize the prediction of Link Travel Time, solves existing SVM and Kalman filtering binding model can not be according to vehicle
Leading portion journey time change the coefficient matrix of Kalman filtering in real time and lead to current Forecasting of Travel Time result not accurately
Problem.The present invention provides a kind of link travel time prediction method based on SVM and Kalman filtering, comprising the following steps:
1) training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition, lead to
It crosses Support vector regression algorithm and training set data training obtains support vector regression；
2) initial predicted journey time matrix is obtained according to the support vector regression and the test set data, obtained
The corresponding predicted travel time of continuous setting sampling instant in priming stroke time matrix before any moment；
3) it is shifted using setting intersection time delays in stroke segment information as state-transition matrix according to the state
Matrix and condition of road surface construct state equation, according to the continuous sampling moment corresponding predicted travel time before any moment
Observing matrix is constituted, observational equation is constructed according to observing matrix, weather conditions and the state equation；
4) observational equation is solved according to Kalman filtering algorithm and obtains any moment corresponding actual prediction time.
Beneficial effect is that the requirement based on the prediction technique of support vector regression to sample data volume is low, and to mistake
Study and deficient problem concerning study can also avoid well；It is corresponding by the continuous setting sampling instant before choosing any moment
Predicted travel time, change the coefficient matrix of Kalman filtering in real time, so that entire prediction model is more rationally, more
Accurately.
Further, it is contemplated that the actual conditions of road, in order to accurately obtain actual prediction journey time, the state side
Journey are as follows:
Wherein, x (k) is the state vector at k moment, and k is any moment,It is expressed as state-transition matrix
Intersection time delays, x (k-1) are the state vector at k-1 moment, and u (k-1) is that process noise is expressed as condition of road surface.
Further, it is contemplated that study the concrete condition of traffic, with simply accurately realize prediction, take any moment it
The preceding corresponding predicted travel time of continuous 4 sampling instants, weather conditions and the state equation constructs observational equation are as follows:
Y (k)=A (k) x (k)+w (k)
A (k)=[T (k), T (k-1), T (k-2), T (k-3)]
Wherein, y (k) is the observation vector of k+1, and w (k) is that observation noise is expressed as weather conditions, and T (k) is the pre- of k moment
Journey time is surveyed, T (k-1) is the predicted travel time at k-1 moment, and T (k-2) is the predicted travel time at k-2 moment, T (k-3)
For the predicted travel time at k-3 moment.
For the ease of realizing the above method, a kind of link travel time prediction device based on SVM and Kalman filtering, packet
The computer program that includes memory, processor and storage in memory and can run on a processor, the processor are held
It is performed the steps of when row described program
1) training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition, lead to
It crosses Support vector regression algorithm and training set data training obtains support vector regression；
2) initial predicted journey time matrix is obtained according to the support vector regression and the test set data, obtained
The corresponding predicted travel time of continuous setting sampling instant in priming stroke time matrix before any moment；
3) it is shifted using setting intersection time delays in stroke segment information as state-transition matrix according to the state
Matrix and condition of road surface construct state equation, according to the continuous sampling moment corresponding predicted travel time before any moment
Observing matrix is constituted, observational equation is constructed according to observing matrix, weather conditions and the state equation；
4) observational equation is solved according to Kalman filtering algorithm and obtains any moment corresponding actual prediction time, led to
The corresponding predicted travel time of continuous setting sampling instant before choosing any moment is crossed, changes Kalman filtering in real time
Coefficient matrix, so that entire prediction model is more rationally, more accurately.
Further, it is contemplated that the actual conditions of road, in order to accurately obtain actual prediction journey time, the institute of the device
State state equation are as follows:
Wherein, x (k) is the state vector at k moment, and k is any moment,It is expressed as state-transition matrix
Intersection time delays, x (k-1) are the state vector at k-1 moment, and u (k-1) is that process noise is expressed as condition of road surface.
Further, it is contemplated that study the concrete condition of traffic, simply accurately to realize prediction, take this in the device
The corresponding predicted travel time of continuous 4 sampling instants, weather conditions and state equation building observation before one moment
Equation are as follows:
Y (k)=A (k) x (k)+w (k)
A (k)=[T (k), T (k-1), T (k-2), T (k-3)]
Wherein, y (k) is the observation vector of k+1, and w (k) is that observation noise is expressed as weather conditions, and T (k) is the pre- of k moment
Journey time is surveyed, T (k-1) is the predicted travel time at k-1 moment, and T (k-2) is the predicted travel time at k-2 moment, T (k-3)
For the predicted travel time at k-3 moment.
Detailed description of the invention
Fig. 1 is a kind of flow chart of link travel time prediction method based on SVM and Kalman filtering of the invention；
Fig. 2 is a kind of schematic diagram of link travel time prediction method based on SVM and Kalman filtering of the invention.
Specific embodiment
The present invention will be further described in detail with reference to the accompanying drawing.
Embodiment of the method
The present invention provides a kind of link travel time prediction method based on SVM and Kalman filtering, as shown in Figure 1, packet
Include following steps:
1) training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition, lead to
It crosses Support vector regression algorithm and training set data training obtains support vector regression.
Support vector machines theory includes linear SVM sorting algorithm, Nonlinear Support Vector Machines sorting algorithm and line
Property Support vector regression algorithm, Nonlinear Support Vector Machines regression algorithm, currently, these algorithms have been applied to multiple necks
Domain realizes classification and the regression forecasting of characteristics of objects value, achieves good effect.In Link Travel Time when vehicle travel
Between to change with time not be simple linear relationship, will not unlimitedly increase with the time, will not unlimitedly drop
It is low.For given section, journey time is fluctuated in an interval range.The present invention is with non-linear support vector regression
For machine, particular content is as follows:
The selection of dimension and prediction characteristics of objects value of the high-precision SVM model based on input vector, holiday traffic are early high
The peak period compares morning peak working day, backward delay period, and holiday traffic is mostly for the purpose of amusement and recreation, compared to work
Requirement of the commuting traffic of day to punctuality is less high, therefore, will be right respectively to holiday and Forecasting of Travel Time on ordinary days
To.Weather information includes wide general, the appearance of mist, rain, snowy day gas, or is deteriorated under pavement behavior to driving efficiency and safety
Influence be that cannot be neglected.The vehicle travel time under such atrocious weather situation and good weather conditions, if using same
The model parameter of sample, prediction effect affirmative are bad.Therefore, this method uses svm classifier model with working day, nonworkdays and day
Gas realizes the classification of Forecasting of Travel Time scene as input feature vector value, finally can be obtained that working day weather is good, work
Day bad weather, nonworkdays weather is good and nonworkdays bad weather etc. four in scene.
If learning sample collection (training sample set)It is nonlinear, general
Input sample Space Nonlinear transforms to another high-dimensional feature space, and nonlinear regression letter is constructed in this feature space
Number, and this nonlinear transformation is by defining kernel function K (x appropriatei,yi) Lai Shixian.
Wherein K (xi,yi)=φ (xi)Tφ(xj), φ (x) is a certain nonlinear function.Therefore nonlinear solshing is sought
The problem of be attributed to following optimization problem:
Constraint condition is
||wTφ(xi)+b-yi| |≤ε, i=1,2 ..., l (2)
The Lagrange dual problem of the problem is
Constraint condition is
The dual problem is solved, nonlinear solshing is obtained.When constraint condition can not be realized, introduces two relaxations and become
Amount:
Optimization problem becomes
Constraint condition is
wTφ(xi)+b-yi≤ξi+ ε, i=1,2 ..., l (8)
C > 0 is penalty factor in formula, and C is bigger to indicate that the data point punishment big to error is bigger.It can use Lagrange
Multiplier method solves the constrained optimization problem, constructs following Lagrangian thus:
According to Optimum Theory, by L respectively to w, b, ξi、Seeking partial differential and enabling it is 0, is obtained:
Formula (11) are substituted into formula (10), antithesis optimum solution can be obtained, can further acquire nonlinear solshing.
Using SVM regression forecasting, it is critical that the selection to parameter, being currently being widely used is conventional mesh search
Method, Weigh Direct Determination, linear search method, inverse ratio method, there is deficiencies in some aspects for these methods, and genetic algorithm is in parameter
There is advantage outstanding in optimizing.Therefore, this method establishes the optimization algorithm of regressive prediction model parameter using genetic algorithm.To compile
After the binary system of C, σ and ε after code obtain model by SVM training as an individual, training sample value and training objective value
The model predication value of target value will be obtained in training sample value input model, with the mean square error of simulated target value and predicted target values
Difference is used as fitness function, using individual adaptation degree functional value minimum value corresponding C, σ and ε as final optimization pass result in evolution.
2) initial predicted journey time matrix is obtained according to support vector regression and test set data, obtains priming stroke
The corresponding predicted travel time of continuous setting sampling instant in time matrix before any moment.
Kalman filter based on minimum mean square error criterion solves the problems, such as optimum linear filter.Use recursive algorithm
Optimal estimation is carried out to filter status variable.By calculating state equation and measurement equation, pervious estimation and newest sight
It surveys.
Kalman filter problem can be stated as using observation data vector y (1), y (2) ... y (n), to each point of n >=1
The least-squares estimation of amount；The prediction of city road journey time is carried out used here as Kalman prediction model.Firstly, establishing it
Prediction model is as follows:
With k, k-1 ... certain average travel time for road sections prediction vehicle k+1 moment at k-n+1 moment passes through the road in this section
Section journey time, it is contemplated that study the concrete condition of traffic, take here 4 moment Link Travel Time (i.e. k, k-1, k-2,
K-3) as the influence factor to prediction, prediction model is as follows:
T (k+1)=H0(k)T(k)+H1(k)T(k-1)+
H2(k)T(k-2)+H3(k)T(k-3)+w(k)
In above formula, T (k+1) is the Link Travel Time of prediction；Hi(k) (i=0,1,2,3) is system parameter matrix；w
(k) it is zero-mean white noise, indicates that systematic observation noise, covariance matrix are R (k).
3) state equation is constructed according to intersection time delays and condition of road surface in setting stroke segment information, according to this
Continuous sampling moment corresponding predicted travel time, weather conditions and the state equation before one moment construct observation side
Journey.
Definition status vector:
A (k)=[T (k), T (k-1), T (k-2), T (k-3)]
X (k)=[H0(k),H1(k),H2(k),H3(k),]T
Y (k)=T (k+1)
Then above-mentioned prediction model can be exchanged into state equation and observational equation under Kalman filtering:
Y (k)=A (k) x (k)+w (k)
In formula, x (k) is the state vector at k moment, and k is any moment；It is expressed as state-transition matrix
Intersection time delays；X (k-1) is the state vector at k-1 moment；U (k-1) is process noise, covariance matrix Q (k-
1)；W (k) is observation noise, covariance matrix R (k)；U (k-1) and w (k) is irrelevant zero-mean white noise.
Observation noise has above-mentioned SVM model training to obtain, and process noise includes the station number for originating station, the length in section
Whether degree the signal-controlled intersection number on section, has sleet, otherwise rain and snow weather 1 is 0, visibility, temperature on average, the date and
At least two parameter in frequency, the training set data constituted according to above-mentioned parameter find out optimized parameter by SVM training.
4) observational equation is solved according to Kalman filtering algorithm and obtains any moment corresponding actual prediction time.
According to kalman filtering theory, there is following recurrence formula:
K (k)=P (k | k-1) AT(k)[A(k)P(k|k-1)AT(k)+R(k)]-1
P (k)=[1-K (k) A (k)] P (k | k-1)
Wherein, K (k) represents hybrid cytokine, and P (k) is to walk variance of estimaion error in kth.
Pass through calculatingThis section subsequent time link travel time prediction value can be obtained, be shown below:
Above-mentioned dynamic model is static prediction with dynamically adjusting the model combined.Wherein support vector machines is as prediction
Basis is to predict offline, when trained support vector machines maps out the link travel for needing to predict from a large amount of historical datas
Between.But SVM model effectively cannot timely adjust emergency event, so Kalman filter dynamic algorithm is introduced into
In the model.Using the Link Travel Time of SVM prediction as the priming stroke time, then by the priming stroke time
Input matrix carries out dynamic adjustment to result to Kalman filter.Kalman filter dynamic algorithm will be newest using renewal equation
Observation be added predicted vector in, the precision of prediction of the dynamic model will be effectively improved.
It is illustrated in figure 2 the cardinal principle of section time forecasting methods.This method is based on historical data building SVM training mould
Type.Then SVM training pattern is introduced into Kalman filter, utilizes historical data building state equation, measurement equation and update
Equation completes training process.Determine the minimum absolutely corresponding optimal model parameters of percent error of history predictive result at this
In the process.Finally, introducing real-time road traffic data in the algorithm, the prediction of real-time road traffic data is completed.
Specific embodiment of the present invention is presented above, but the present invention is not limited to described embodiment.
Under the thinking that the present invention provides, to the skill in above-described embodiment by the way of being readily apparent that those skilled in the art
Art means are converted, are replaced, are modified, and play the role of with the present invention in relevant art means it is essentially identical, realize
Goal of the invention it is also essentially identical, the technical solution formed in this way is to be finely adjusted to be formed to above-described embodiment, this technology
Scheme is still fallen in protection scope of the present invention.
Claims (6)
1. a kind of link travel time prediction method based on SVM and Kalman filtering, which comprises the following steps:
1) training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition, pass through branch
It holds vector machine regression algorithm and training set data training obtains support vector regression；
2) initial predicted journey time matrix is obtained according to the support vector regression and the test set data, obtained initial
The corresponding predicted travel time of continuous setting sampling instant in journey time matrix before any moment；
3) to set, intersection time delays is as state-transition matrix in stroke segment information, according to the state-transition matrix
State equation is constructed with condition of road surface, is constituted according to the continuous sampling moment corresponding predicted travel time before any moment
Observing matrix constructs observational equation according to observing matrix, weather conditions and the state equation；
4) observational equation is solved according to Kalman filtering algorithm and obtains any moment corresponding actual prediction time.
2. the link travel time prediction method according to claim 1 based on SVM and Kalman filtering, feature exist
In the state equation are as follows:
Wherein, x (k) is the state vector at k moment,It is expressed as intersection time delays for state-transition matrix,
X (k-1) is the state vector at k-1 moment, and u (k-1) is that process noise is expressed as condition of road surface.
3. the link travel time prediction method according to claim 2 based on SVM and Kalman filtering, feature exist
In taking the corresponding predicted travel time of continuous 4 sampling instants, weather conditions and the state equation before any moment
Construct observational equation are as follows:
Y (k)=A (k) x (k)+w (k)
A (k)=[T (k), T (k-1), T (k-2), T (k-3)]
Wherein, y (k) is the observation vector of k+1, and w (k) is that observation noise is expressed as weather conditions, and T (k) is the prediction row at k moment
Journey time, T (k-1) are the predicted travel time at k-1 moment, and T (k-2) is the predicted travel time at k-2 moment, and T (k-3) is k-
The predicted travel time at 3 moment.
4. a kind of link travel time prediction device based on SVM and Kalman filtering, including memory, processor and storage
In memory and the computer program that can run on a processor, which is characterized in that when the processor executes described program
It performs the steps of
1) training set data and test set data for obtaining setting stroke segment information and its corresponding journey time composition, pass through branch
It holds vector machine regression algorithm and training set data training obtains support vector regression；
2) initial predicted journey time matrix is obtained according to the support vector regression and the test set data, obtained initial
The corresponding predicted travel time of continuous setting sampling instant in journey time matrix before any moment；
3) to set, intersection time delays is as state-transition matrix in stroke segment information, according to the state-transition matrix
State equation is constructed with condition of road surface, is constituted according to the continuous sampling moment corresponding predicted travel time before any moment
Observing matrix constructs observational equation according to observing matrix, weather conditions and the state equation；
4) observational equation is solved according to Kalman filtering algorithm and obtains any moment corresponding actual prediction time.
5. the link travel time prediction device according to claim 4 based on SVM and Kalman filtering, feature exist
In the state equation are as follows:
Wherein, x (k) is the state vector at k moment, and k is any moment,It is expressed as intersecting for state-transition matrix
Crossing time delays, x (k-1) are the state vector at k-1 moment, and u (k-1) is that process noise is expressed as condition of road surface.
6. the link travel time prediction device according to claim 5 based on SVM and Kalman filtering, feature exist
In taking the corresponding predicted travel time of continuous 4 sampling instants, weather conditions and the state equation before any moment
Construct observational equation are as follows:
Y (k)=A (k) x (k)+w (k)
A (k)=[T (k), T (k-1), T (k-2), T (k-3)]
Wherein, y (k) is the observation vector of k+1, and w (k) is that observation noise is expressed as weather conditions, and T (k) is the prediction row at k moment
Journey time, T (k-1) are the predicted travel time at k-1 moment, and T (k-2) is the predicted travel time at k-2 moment, and T (k-3) is k-
The predicted travel time at 3 moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811239031.7A CN109409598A (en) | 2018-10-23 | 2018-10-23 | Link travel time prediction method and device based on SVM and Kalman filtering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811239031.7A CN109409598A (en) | 2018-10-23 | 2018-10-23 | Link travel time prediction method and device based on SVM and Kalman filtering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109409598A true CN109409598A (en) | 2019-03-01 |
Family
ID=65468862
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811239031.7A Pending CN109409598A (en) | 2018-10-23 | 2018-10-23 | Link travel time prediction method and device based on SVM and Kalman filtering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409598A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110031793A (en) * | 2019-04-09 | 2019-07-19 | 中国电子科技集团公司第三十六研究所 | A kind of interferometer direction finding methods, devices and systems |
| CN110992708A (en) * | 2019-12-20 | 2020-04-10 | 斑马网络技术有限公司 | Real-time traffic speed prediction method and device and electronic equipment |
| CN111145535A (en) * | 2019-11-28 | 2020-05-12 | 银江股份有限公司 | Travel time reliability distribution prediction method under complex scene |
| CN112085622A (en) * | 2020-09-11 | 2020-12-15 | 杭州华电下沙热电有限公司 | Photovoltaic power station power generation power prediction algorithm based on CKF-SVR |
| CN112309109A (en) * | 2019-08-01 | 2021-02-02 | 中移(苏州)软件技术有限公司 | Road traffic time prediction method, device and storage medium |
| CN114330148A (en) * | 2022-03-11 | 2022-04-12 | 浙江大学 | Motor insulation life prediction method based on accelerated degradation data |
| CN114333133A (en) * | 2021-12-29 | 2022-04-12 | 杭州海康威视数字技术股份有限公司 | Entrance guard-based commuting authentication method, device, equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102074124A (en) * | 2011-01-27 | 2011-05-25 | 山东大学 | Dynamic bus arrival time prediction method based on support vector machine (SVM) and H-infinity filtering |
| CN102157075A (en) * | 2011-03-15 | 2011-08-17 | 上海交通大学 | Method for predicting bus arrivals |
| WO2014129952A1 (en) * | 2013-02-25 | 2014-08-28 | Telefonaktiebolaget L M Ericsson (Publ) | Grant utilization based other cell interference estimation |
| CN105321345A (en) * | 2015-09-18 | 2016-02-10 | 浙江工业大学 | Road traffic flow prediction method based on ARIMA model and kalman filtering |
| US9799218B1 (en) * | 2016-05-09 | 2017-10-24 | Robert Gordon | Prediction for lane guidance assist |
| CN107480781A (en) * | 2017-03-02 | 2017-12-15 | 南京航空航天大学 | The nuclear accident Source Term Inversion method of neutral net adaptive Kalman filter |
| CN108538050A (en) * | 2017-03-01 | 2018-09-14 | 香港理工大学深圳研究院 | A kind of short-term traffic flow forecast method and device |
-
2018
- 2018-10-23 CN CN201811239031.7A patent/CN109409598A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102074124A (en) * | 2011-01-27 | 2011-05-25 | 山东大学 | Dynamic bus arrival time prediction method based on support vector machine (SVM) and H-infinity filtering |
| CN102157075A (en) * | 2011-03-15 | 2011-08-17 | 上海交通大学 | Method for predicting bus arrivals |
| WO2014129952A1 (en) * | 2013-02-25 | 2014-08-28 | Telefonaktiebolaget L M Ericsson (Publ) | Grant utilization based other cell interference estimation |
| CN105321345A (en) * | 2015-09-18 | 2016-02-10 | 浙江工业大学 | Road traffic flow prediction method based on ARIMA model and kalman filtering |
| US9799218B1 (en) * | 2016-05-09 | 2017-10-24 | Robert Gordon | Prediction for lane guidance assist |
| CN108538050A (en) * | 2017-03-01 | 2018-09-14 | 香港理工大学深圳研究院 | A kind of short-term traffic flow forecast method and device |
| CN107480781A (en) * | 2017-03-02 | 2017-12-15 | 南京航空航天大学 | The nuclear accident Source Term Inversion method of neutral net adaptive Kalman filter |
Non-Patent Citations (1)
| Title |
|---|
| 陈旭梅，龚辉波，王景楠: "基于SVM和Kalman滤波的BRT行程时间预测模型研究", 《交通运输系统工程与信息》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110031793A (en) * | 2019-04-09 | 2019-07-19 | 中国电子科技集团公司第三十六研究所 | A kind of interferometer direction finding methods, devices and systems |
| CN110031793B (en) * | 2019-04-09 | 2023-06-02 | 中国电子科技集团公司第三十六研究所 | Interferometer direction finding method, device and system |
| CN112309109A (en) * | 2019-08-01 | 2021-02-02 | 中移(苏州)软件技术有限公司 | Road traffic time prediction method, device and storage medium |
| CN111145535A (en) * | 2019-11-28 | 2020-05-12 | 银江股份有限公司 | Travel time reliability distribution prediction method under complex scene |
| CN110992708A (en) * | 2019-12-20 | 2020-04-10 | 斑马网络技术有限公司 | Real-time traffic speed prediction method and device and electronic equipment |
| CN110992708B (en) * | 2019-12-20 | 2021-10-01 | 斑马网络技术有限公司 | Real-time traffic speed prediction method and device and electronic equipment |
| CN112085622A (en) * | 2020-09-11 | 2020-12-15 | 杭州华电下沙热电有限公司 | Photovoltaic power station power generation power prediction algorithm based on CKF-SVR |
| CN112085622B (en) * | 2020-09-11 | 2024-05-07 | 杭州华电下沙热电有限公司 | CKF-SVR-based photovoltaic power station generation power prediction algorithm |
| CN114333133A (en) * | 2021-12-29 | 2022-04-12 | 杭州海康威视数字技术股份有限公司 | Entrance guard-based commuting authentication method, device, equipment and storage medium |
| CN114330148A (en) * | 2022-03-11 | 2022-04-12 | 浙江大学 | Motor insulation life prediction method based on accelerated degradation data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109409598A (en) | Link travel time prediction method and device based on SVM and Kalman filtering | |
| Celtek et al. | Real-time traffic signal control with swarm optimization methods | |
| CN103678917B (en) | A kind of real-time arrival time Forecasting Methodology of public transport based on simulated annealing | |
| Zhao et al. | Truck traffic speed prediction under non-recurrent congestion: Based on optimized deep learning algorithms and GPS data | |
| CN105857350B (en) | A kind of high ferro train based on interval section passenger flow starts method | |
| CN105118294B (en) | A kind of Short-time Traffic Flow Forecasting Methods based on state model | |
| CN103996289B (en) | A kind of flow-speeds match model and Travel Time Estimation Method and system | |
| CN106017496A (en) | Real-time navigation method based on road condition | |
| CN104933862A (en) | Urban traffic jam intelligent combination prediction method based on track of floating vehicle | |
| JP2011138487A (en) | Method and system for traffic prediction based on space-time relation | |
| CN106205156A (en) | A kind of crossing self-healing control method for the sudden change of part lane flow | |
| CN109959388A (en) | A kind of intelligent transportation fining paths planning method based on grid extended model | |
| CN113538910A (en) | Self-adaptive full-chain urban area network signal control optimization method | |
| CN110517485A (en) | A kind of Short-time Traffic Flow Forecasting Methods based on Time segments division | |
| CN110570672A (en) | regional traffic signal lamp control method based on graph neural network | |
| CN108648445A (en) | Dynamic traffic Tendency Prediction method based on traffic big data | |
| CN109657878A (en) | A kind of Air-conditioning Load Prediction method and device | |
| CN113392577B (en) | Regional boundary main intersection signal control method based on deep reinforcement learning | |
| CN110674990A (en) | Instant delivery path selection method and system with sliding window updating mechanism | |
| CN109711593A (en) | A kind of method for forecasting wind speed of high speed railway line towards instantaneous calculating decision | |
| CN109063907A (en) | A kind of big step-ahead prediction method of line of high-speed railway extreme wind speed intelligence traversal | |
| Yao et al. | A path planning model based on spatio-temporal state vector from vehicles trajectories | |
| Yu et al. | Short-term road traffic flow prediction based on multi-dimensional data | |
| Yang et al. | A traffic signal control algorithm for an oversaturated isolated intersection based on video detection data | |
| Wang | Optimal method of intelligent traffic signal light timing based on genetic neural network. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190301 |
|
| RJ01 | Rejection of invention patent application after publication |