CN115311844B - Expressway traffic state estimation method based on self-supervision learning support vector machine - Google Patents

Expressway traffic state estimation method based on self-supervision learning support vector machine Download PDF

Info

Publication number
CN115311844B
CN115311844B CN202210714620.6A CN202210714620A CN115311844B CN 115311844 B CN115311844 B CN 115311844B CN 202210714620 A CN202210714620 A CN 202210714620A CN 115311844 B CN115311844 B CN 115311844B
Authority
CN
China
Prior art keywords
traffic
expressway
traffic state
traffic data
self
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
Application number
CN202210714620.6A
Other languages
Chinese (zh)
Other versions
CN115311844A (en
Inventor
赵妍
芮一康
冉斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Original Assignee
Southeast University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN202210714620.6A priority Critical patent/CN115311844B/en
Publication of CN115311844A publication Critical patent/CN115311844A/en
Application granted granted Critical
Publication of CN115311844B publication Critical patent/CN115311844B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G08G1/0133—Traffic data processing for classifying traffic situation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Landscapes

  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a highway traffic state estimation method based on a self-supervision learning support vector machine, which comprises the following steps: generating a training traffic data set; constructing a traffic state classification model based on a self-supervision learning support vector machine, and training the traffic state classification model by using the expressway sample traffic data feature vector; solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters; and estimating the traffic state of the expressway in the time zone by calculating the interval distance between the traffic data of the expressway in the time zone and the traffic state classifier. The invention realizes the estimation of the traffic state of the expressway, thereby facilitating the expressway manager to conduct targeted control and evaluation on the traffic operation of the expressway and improving the overall traffic performance of the road network.

Description

Expressway traffic state estimation method based on self-supervision learning support vector machine
Technical Field
The invention belongs to the technical field of intelligent traffic systems and traffic state estimation, and particularly relates to a highway traffic state estimation method based on a self-supervision learning support vector machine.
Background
Along with the increasing expansion of the expressway network in China, more information technology support is needed for the real-time supervision and management of the expressway traffic running conditions. At present, aiming at sporadic traffic congestion or accidents caused by sudden events, the main solution is to acquire information in a manual alarm mode, and then take management and rescue measures; and for frequent traffic congestion caused by supply and demand contradiction, the traffic congestion dissipates by itself. The reason for this is mainly that traffic travelers, managers and planners do not have comprehensive knowledge of traffic state information. Therefore, management of highway traffic conditions should require further enhancement of service levels in informatization, intelligence, and humanization.
The highway traffic state estimation is an important component of an intelligent traffic system and is the basis of an intelligent traffic control, induction and coordination system. The highway traffic state estimation is generally based on real-time acquisition of highway traffic data, analysis of various traffic data is realized through various discrimination algorithms, comparison is carried out with prior traffic state standards, what running state the current traffic system is in is obtained, and intelligent control, management and induction of the traffic system are realized according to the discrimination result. The method realizes quick and effective traffic state estimation and is an important guarantee for realizing real-time intelligent and effective control of highway traffic.
In intelligent traffic management systems, various discrimination algorithms are often used to make traffic state estimations. However, no method is available to solve the problem of classifying the high-dimensional traffic data features based on massive unlabeled data.
Disclosure of Invention
The technical problems to be solved are as follows: the invention discloses a highway traffic state estimation method based on a self-supervision learning support vector machine, which is used for mainly calculating and deducing three state variables of traffic flow, interval average speed and space occupancy, carrying out traffic state estimation by considering traffic multidimensional characteristics, and realizing highway traffic state estimation based on the self-supervision learning support vector machine, thereby being convenient for highway managers to carry out targeted control and evaluation on highway traffic operation and improving the overall traffic performance of a road network.
The technical scheme is as follows:
an expressway traffic state estimation method based on a self-supervision learning support vector machine comprises the following steps:
s1: extracting expressway sample traffic data feature vectors, wherein the expressway sample traffic data feature vectors comprise flow, interval average speed and occupancy; constructing a traffic state label pre-judging device based on self-supervision learning, and performing self-supervision pre-training of traffic state labels on input expressway sample traffic data to generate traffic state soft labels; generating a training traffic data set;
s2: constructing a traffic state classification model based on a self-supervision learning support vector machine, and training the traffic state classification model by using the expressway sample traffic data feature vector;
s3: solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters;
s4: and estimating the traffic state of the expressway in the time zone by calculating the interval distance between the traffic data of the expressway in the time zone and the traffic state classifier.
Further, in step S1, the process of generating the training traffic data set includes the following steps:
s11, selecting flow, interval average speed and occupancy as characteristic vectors of traffic data of expressway samples, and carrying out normalization processing, wherein the method specifically comprises the following steps:
Figure BDA0003707362800000021
in the method, in the process of the invention,
Figure BDA0003707362800000022
for highway sample traffic data feature vector, comprising three dimensions, < >>
Figure BDA0003707362800000023
The traffic data is average speed traffic data in expressway intervals, and o is the space occupancy of the expressway;
s12, constructing a traffic state label pre-judging device based on a simple self-training idea in self-supervision learning, and performing self-supervision pre-training of traffic state labels on input expressway sample traffic data to generate traffic state soft labels; the ratio of the average speed of the interval to the free flow rate is selected as a traffic state label pre-judging device identification index:
Figure BDA0003707362800000024
in the method, in the process of the invention,
Figure BDA0003707362800000025
for the j-period traffic state soft label, +.>
Figure BDA0003707362800000026
The average vehicle speed traffic data of the interval of the j time period, v free Is the free flow rate of the expressway;
s13, generating a training traffic data set, specifically:
Figure BDA0003707362800000027
wherein, T is a training traffic data set,
Figure BDA0003707362800000028
for the j-period highway sample traffic data feature vector, j=1, 2,..s, s is the total number of samples.
Further, in step S2, the traffic state classification model
Figure BDA0003707362800000029
The method comprises the following steps:
Figure BDA00037073628000000210
Figure BDA00037073628000000211
wherein sgn represents a sign function, w k Decision hyperplane weights for class k, b k For the bias value of the decision function,
Figure BDA00037073628000000212
for the j period of highway sample traffic data feature vector, < > for>
Figure BDA0003707362800000031
For the j-period traffic state soft tag, j=1, 2.
Further, in step S3, solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters includes the following steps:
s31, deciding weight parameters w of the traffic state classifier k ，b k The solution problem is converted into solution ||w k The problem of minimizing the I is solved by setting a constraint expression of the I as affine function constraint and further refining the constraint expression as a convex optimization problem under affine function constraint according to support vector space constraint, and the specific formula is as follows:
Figure BDA0003707362800000032
Figure BDA0003707362800000033
/>
wherein ε k For the error allowable range of k-type decision hyperplane to a special sample point, C is an error cost coefficient,
Figure BDA0003707362800000034
to relax the variables, g (w k ，b k ) Representing w k And b k Constraint conditions between;
s32, solving a constraint expression by using a Lagrangian multiplier method, wherein the specific formula is as follows:
Figure BDA0003707362800000035
wherein lambda is j Is the corresponding Lagrangian multiplier;
s33, performing bias derivative on each variable to enable the value of each variable to be 0, wherein the specific formula is as follows:
Figure BDA0003707362800000036
Figure BDA0003707362800000037
Figure BDA0003707362800000038
Figure BDA0003707362800000039
pushing out:
Figure BDA00037073628000000310
further, in step S3, the process of solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters further includes the following steps:
s34: the solving problem of the constraint expression is converted into a corresponding dual problem, and the specific formula is as follows:
Figure BDA0003707362800000041
s.t.λ j ≥0
s35: the up-conversion model is performed by using a Gaussian kernel function, and the specific formula is as follows:
Figure BDA0003707362800000042
s.t.λ j ≥0
in the method, in the process of the invention,
Figure BDA0003707362800000043
is a kernel function; delta is the width of a Gaussian kernel function neuron; q (lambda) j ) Lambda is lambda j Is a solution function of (2); lambda (lambda) m Is Lagrangian multiplier +.>
Figure BDA0003707362800000044
For m-period traffic state soft tag, +.>
Figure BDA0003707362800000045
For m-period highway sampleThe traffic data feature vector, m=1, 2,..s, j +.m;
s36: by solving lambda i Obtaining the optimal solution w k * And b k * The specific formula is as follows:
Figure BDA0003707362800000046
further, in step S4, the following formula is adopted to estimate the traffic state of the expressway measured period by calculating the interval distance between the traffic data of the expressway measured period and the traffic state classifier:
Figure BDA0003707362800000047
the beneficial effects are that:
first, the expressway traffic state estimation method based on the Self-supervision learning support vector machine carries out Self-supervision pre-training based on a simple Self-training idea (simple Self-training) in Self-supervision learning (Self-Supervised Learning), automatically constructs supervised soft label data from unsupervised data, and learns a pre-training model. Based on massive unlabeled data, a powerful general representation model is obtained by self-supervision learning. In the upper-layer operation, the aim of realizing a specific task based on supervised learning and the intelligent control based on reinforcement learning are realized. Self-supervised learning can enhance the performance and generalization ability of supervised learning and reinforcement learning models. The method and the device can better utilize the unsupervised data and learn the internal information of the input unsupervised data. And the effect of the follow-up supervision and learning task is improved.
Secondly, the expressway traffic state estimation method based on the self-supervision learning support vector machine is used for designing expressway traffic state estimation based on the support vector machine and effectively solving the classification problem of high-dimensional traffic data features. The obtained decision hyperplane memory occupies less memory, and a kernel function is applied, so that the nonlinear classification problem is flexibly solved. Meanwhile, the method has clear calculation flow and simple calculation, and has higher calculation speed on the premise of ensuring the accuracy of traffic state estimation.
Drawings
FIG. 1 is a flow chart of a highway traffic state estimation method based on a self-supervision learning support vector machine according to an embodiment of the invention;
FIG. 2 is a flow chart of generating a training traffic data set according to an embodiment of the present invention.
Fig. 3 is a schematic flow chart of constructing a traffic state classification model based on a self-supervision learning support vector machine according to an embodiment of the present invention.
Fig. 4 is a schematic flow chart of solving classifier decision weight parameters according to an embodiment of the invention.
Fig. 5 is a schematic diagram of a traffic state estimation result according to an embodiment of the present invention.
Detailed Description
The following examples will provide those skilled in the art with a more complete understanding of the invention, but are not intended to limit the invention in any way.
Referring to fig. 1, the embodiment discloses a highway traffic state estimation method based on a self-supervision learning support vector machine, the highway traffic state estimation method comprises the following steps:
s1: extracting expressway sample traffic data feature vectors, wherein the expressway sample traffic data feature vectors comprise flow, interval average speed and occupancy; constructing a traffic state label pre-judging device based on self-supervision learning, and performing self-supervision pre-training of traffic state labels on input expressway sample traffic data to generate traffic state soft labels; a training traffic data set is generated.
S2: and constructing a traffic state classification model based on a self-supervision learning support vector machine, and training the traffic state classification model by using the expressway sample traffic data feature vector.
S3: and solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters.
S4: and estimating the traffic state of the expressway in the time zone by calculating the interval distance between the traffic data of the expressway in the time zone and the traffic state classifier.
S1: extracting a highway sample traffic data feature vector to generate a training traffic data set;
in an alternative embodiment of the present invention, as shown in FIG. 2, step S1 specifically includes the following substeps S1-1 to S1-3:
in an alternative embodiment of the present invention, the present invention performs normalization processing by selecting the traffic, the section average speed and the occupancy as the characteristic vector of the highway sample traffic data in step S1-1. Specifically, a basic highway section (about 7 km long) of a highway is selected, and a plurality of traffic detectors are arranged in the section to provide required traffic parameters including traffic volume, average speed of a section and space occupancy. The highway traffic state estimation selects 7 days (24 hours in the whole day, the detection time step is 5 minutes) of detection data as a training set and selects 3 days of detection data as a test set.
In an alternative embodiment of the present invention, the ratio of the average vehicle speed to the free flow speed in the interval is selected as the traffic state label pre-identifier in step S1-2, and the specific formula is as follows:
Figure BDA0003707362800000061
in the method, in the process of the invention,
Figure BDA0003707362800000062
for the j-period traffic state soft label, +.>
Figure BDA0003707362800000063
The average vehicle speed traffic data of the interval of the j time period, v free Is the free flow rate of the expressway. The road traffic conditions are classified into four classes in combination with the selected high-speed actual traffic conditions, and the classification is as shown in table 1 below. It should be understood that the identifiers may be defined by themselves, and that the four identifiers of the present embodiment are for illustration only; the division criteria of the fluidity index may be adjusted according to the actual scene, and are not limited to table 1.
TABLE 1
Figure BDA0003707362800000064
In an alternative embodiment of the present invention, the present invention generates a training traffic data set in steps S1-3, specifically:
Figure BDA0003707362800000065
wherein, T is a training traffic data set,
Figure BDA0003707362800000066
and (3) the traffic data feature vector is the sample traffic data feature vector of the expressway in the period j, and s is the total sample number.
S2: training a traffic state classification model by using the expressway sample traffic data feature vector;
in an alternative embodiment of the present invention, as shown in fig. 3, step S2 trains a traffic state classification model by using feature vectors of highway sample traffic data, and solves classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters, and specifically includes the following sub-steps S2-1 to S2-4:
in an alternative embodiment of the present invention, the present invention uses the step S1 to extract the feature vector of the traffic data of the highway sample to train the traffic state classification model in the substep S2-1, and the specific formula is as follows:
Figure BDA0003707362800000067
Figure BDA0003707362800000068
wherein sgn represents a sign function, w k Decision hyperplane weights for class k, b k Is the bias value of the decision function.
In an alternative embodiment of the present invention, the present invention uses the traffic state classifier decision weight parameter w in substep S2-2 k ，b k Solving the problem may be equivalent to solving ||w k The constraint expression of the minimization problem can be set as affine function constraint and further refined to be convex optimization problem under affine function constraint according to support vector space constraint, and the specific formula is as follows:
Figure BDA0003707362800000071
Figure BDA0003707362800000072
wherein ε k For the error allowable range of k-type decision hyperplane to a special sample point, C is an error cost coefficient,
Figure BDA0003707362800000073
is a relaxation variable.
In an alternative embodiment of the present invention, the present invention solves this using the Lagrangian multiplier method in substep S2-3, with the following specific formulas:
Figure BDA0003707362800000074
wherein lambda is j Is the corresponding lagrangian multiplier.
In an alternative embodiment of the present invention, in step S2-4, in order to find the extreme point of the problem, the present invention deflects each variable to have a value of 0, and the specific formula is as follows:
Figure BDA0003707362800000075
from the above, it can be deduced that:
Figure BDA0003707362800000076
s3: solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters;
in an alternative embodiment of the present invention, as shown in FIG. 4, step S3 specifically includes the following substeps S3-1 to S3-3:
in an alternative embodiment of the present invention, the present invention converts the original question into a dual question in step S3-1, and the specific formula is as follows:
Figure BDA0003707362800000077
in an alternative embodiment of the present invention, the present invention performs the up-conversion model by using the gaussian kernel function in the substep S3-2, and the specific formula is as follows:
Figure BDA0003707362800000081
s.t.λ j ≥0，j＝1，2，...，s
in the method, in the process of the invention,
Figure BDA0003707362800000082
is a kernel function.
In an alternative embodiment of the invention, the invention sub-step S3-3 is performed by solving for lambda i Can obtain the optimal solution w k * And b k * The specific formula is as follows:
Figure BDA0003707362800000083
s4: a traffic state classification model based on a self-supervision learning support vector machine is constructed, the traffic state of the expressway in the period is estimated by calculating the interval distance between traffic data of the detected period of the expressway and a traffic state classifier, and the traffic state of the expressway in the period is estimated by the specific formula as follows:
Figure BDA0003707362800000084
in an alternative embodiment of the present invention, step S4 estimates the highway traffic state effect during this period, as shown in fig. 5.
According to the expressway traffic state estimation method based on the self-supervision learning support vector machine, expressway traffic state estimation research is carried out by utilizing the support vector machine and the self-supervision learning thought, and association degrees of different traffic parameters are reflected by fully utilizing a support vector machine classification model and a kernel function thereof. The method has better performance for highway traffic state estimation, and can improve traffic estimation accuracy.
The embodiments of the present invention have been described in detail with reference to the drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present invention.

Claims (1)

1. The expressway traffic state estimation method based on the self-supervision learning support vector machine is characterized by comprising the following steps of:
s1: extracting expressway sample traffic data feature vectors, wherein the expressway sample traffic data feature vectors comprise flow, interval average speed and occupancy; constructing a traffic state label pre-judging device based on self-supervision learning, and performing self-supervision pre-training of traffic state labels on input expressway sample traffic data to generate traffic state soft labels; generating a training traffic data set;
s2: constructing a traffic state classification model based on a self-supervision learning support vector machine, and training the traffic state classification model by using the expressway sample traffic data feature vector;
s3: solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters;
s4: estimating the traffic state of the expressway in the time zone by calculating the interval distance between traffic data of the expressway in the time zone and a traffic state classifier;
in step S1, the process of generating the training traffic data set includes the steps of:
s11, selecting flow, interval average speed and occupancy as characteristic vectors of traffic data of expressway samples, and carrying out normalization processing, wherein the method specifically comprises the following steps:
Figure FDA0004136129480000011
in the method, in the process of the invention,
Figure FDA0004136129480000012
for highway sample traffic data feature vector, comprising three dimensions, < >>
Figure FDA0004136129480000013
The traffic data is average speed traffic data in expressway intervals, o is the space occupancy of the expressway, and q is the traffic data of the expressway;
s12, constructing a traffic state label pre-judging device based on a simple self-training idea in self-supervision learning, and performing self-supervision pre-training of traffic state labels on input expressway sample traffic data to generate traffic state soft labels; the ratio of the average speed of the interval to the free flow rate is selected as a traffic state label pre-judging device identification index:
Figure FDA0004136129480000014
where sgn denotes the sign function,
Figure FDA0004136129480000015
for the j-period traffic state soft label, +.>
Figure FDA0004136129480000016
The average vehicle speed traffic data of the interval of the j time period, v free Is the free flow rate of the expressway;
s13, generating a training traffic data set, specifically:
Figure FDA0004136129480000017
wherein, T is a training traffic data set,
Figure FDA0004136129480000018
for a period j of highway sample traffic data feature vectors, j=1, 2, s, s is the total number of samples;
in step S2, the traffic state classification model
Figure FDA0004136129480000021
The method comprises the following steps:
Figure FDA0004136129480000022
Figure FDA0004136129480000023
wherein sgn represents a sign function, w k Decision hyperplane weights for class k, b k For the bias value of the decision function,
Figure FDA0004136129480000024
for the j period of highway sample traffic data feature vector, < > for>
Figure FDA0004136129480000025
For the j-period traffic state soft tag, j=1, 2,..s, s is the total number of samples:
in step S3, solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters includes the following steps:
s31, deciding weight parameters w of the traffic state classifier k ，b k The solution problem is converted into solution ||w k The problem of minimizing the I is solved by setting a constraint expression of the I as affine function constraint and further refining the constraint expression as a convex optimization problem under affine function constraint according to support vector space constraint, and the specific formula is as follows:
Figure FDA0004136129480000026
Figure FDA0004136129480000027
wherein ε k For the error allowable range of k-type decision hyperplane to a special sample point, C is an error cost coefficient,
Figure FDA0004136129480000028
to relax the variables, g (w k ，b k ) Representing w k And b k Constraint conditions between;
s32, solving a constraint expression by using a Lagrangian multiplier method, wherein the specific formula is as follows:
Figure FDA0004136129480000029
wherein lambda is j Is the corresponding Lagrangian multiplier;
s33, performing bias derivative on each variable to enable the value of each variable to be 0, wherein the specific formula is as follows:
Figure FDA00041361294800000210
Figure FDA00041361294800000211
Figure FDA00041361294800000212
Figure FDA00041361294800000213
pushing out:
Figure FDA0004136129480000031
in step S3, the process of solving the classifier decision weight parameters to obtain different traffic states and corresponding vector classifier decision weight parameters further includes the following steps:
s34: the solving problem of the constraint expression is converted into a corresponding dual problem, and the specific formula is as follows:
Figure FDA0004136129480000032
s.t.λ j ≥0
s35: the up-conversion model is performed by using a Gaussian kernel function, and the specific formula is as follows:
Figure FDA0004136129480000033
/>
Figure FDA0004136129480000034
s.t.λ j ≥0
in the method, in the process of the invention,
Figure FDA0004136129480000035
is a kernel function; delta is the width of a Gaussian kernel function neuron; q (lambda) j ) Lambda is lambda j Is a solution function of (2); lambda (lambda) m Is Lagrangian multiplier +.>
Figure FDA0004136129480000036
For m-period traffic state soft tag, +.>
Figure FDA0004136129480000037
For m-period highway sample traffic data feature vectors, m=1, 2,..s, j+.m;
s36: by solving lambda i Obtaining the optimal solution w k * And b k * The specific formula is as follows:
Figure FDA0004136129480000038
in step S4, the following formula is adopted to estimate the traffic state of the expressway in the time zone by calculating the interval distance between the traffic data of the expressway in the time zone and the traffic state classifier:
Figure FDA0004136129480000039
/>
CN202210714620.6A 2022-06-22 2022-06-22 Expressway traffic state estimation method based on self-supervision learning support vector machine Active CN115311844B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210714620.6A CN115311844B (en) 2022-06-22 2022-06-22 Expressway traffic state estimation method based on self-supervision learning support vector machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210714620.6A CN115311844B (en) 2022-06-22 2022-06-22 Expressway traffic state estimation method based on self-supervision learning support vector machine

Publications (2)

Publication Number Publication Date
CN115311844A CN115311844A (en) 2022-11-08
CN115311844B true CN115311844B (en) 2023-05-16

Family

ID=83855429

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210714620.6A Active CN115311844B (en) 2022-06-22 2022-06-22 Expressway traffic state estimation method based on self-supervision learning support vector machine

Country Status (1)

Country Link
CN (1) CN115311844B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107123265A (en) * 2017-06-12 2017-09-01 东南大学 A kind of traffic status of express way method of estimation based on parallel computation
CN112365359A (en) * 2020-11-26 2021-02-12 腾讯科技（深圳）有限公司 Training method, device, equipment and storage medium for planting decision determination model
CN113139651A (en) * 2020-01-20 2021-07-20 北京三星通信技术研究有限公司 Training method and device of label proportion learning model based on self-supervision learning
CN113313145A (en) * 2021-05-10 2021-08-27 东南大学 Expressway traffic incident detection method based on mixed kernel correlation vector machine
CN113781404A (en) * 2021-08-19 2021-12-10 浙江大学 Road disease detection method and system based on self-supervision pre-training
CN114600130A (en) * 2019-12-10 2022-06-07 赫尔实验室有限公司 Process for learning new image classes without labels
EP4012337A1 (en) * 2020-12-11 2022-06-15 Aptiv Technologies Limited Change detection criteria for updating sensor-based reference maps

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107123265A (en) * 2017-06-12 2017-09-01 东南大学 A kind of traffic status of express way method of estimation based on parallel computation
CN114600130A (en) * 2019-12-10 2022-06-07 赫尔实验室有限公司 Process for learning new image classes without labels
CN113139651A (en) * 2020-01-20 2021-07-20 北京三星通信技术研究有限公司 Training method and device of label proportion learning model based on self-supervision learning
CN112365359A (en) * 2020-11-26 2021-02-12 腾讯科技（深圳）有限公司 Training method, device, equipment and storage medium for planting decision determination model
EP4012337A1 (en) * 2020-12-11 2022-06-15 Aptiv Technologies Limited Change detection criteria for updating sensor-based reference maps
CN113313145A (en) * 2021-05-10 2021-08-27 东南大学 Expressway traffic incident detection method based on mixed kernel correlation vector machine
CN113781404A (en) * 2021-08-19 2021-12-10 浙江大学 Road disease detection method and system based on self-supervision pre-training

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
林嘉豪 ; 章宗长 ; 姜冲 ; 郝建业 ; .基于生成对抗网络的模仿学习综述.计算机学报.2020,(02),全文. *

Also Published As

Publication number Publication date
CN115311844A (en) 2022-11-08

Similar Documents

Publication Publication Date Title
Li et al. Application on traffic flow prediction of machine learning in intelligent transportation
Habtemichael et al. Short-term traffic flow rate forecasting based on identifying similar traffic patterns
CN108985380B (en) Point switch fault identification method based on cluster integration
Dou et al. A hybrid CEEMD-GMM scheme for enhancing the detection of traffic flow on highways
CN108629073A (en) A kind of degenerative process modeling of multi-mode and method for predicting residual useful life
Tang et al. Short-term traffic flow prediction considering spatio-temporal correlation: A hybrid model combing type-2 fuzzy C-means and artificial neural network
CN109376906B (en) Travel time prediction method and system based on multi-dimensional trajectory and electronic equipment
CN114495507B (en) Traffic flow prediction method integrating space-time attention neural network and traffic model
CN113704956A (en) Urban road online microscopic simulation method and system based on digital twin technology
CN114676742A (en) Power grid abnormal electricity utilization detection method based on attention mechanism and residual error network
Zhao et al. Analysis of users’ electricity consumption behavior based on ensemble clustering
CN114034486A (en) Unsupervised transfer learning-based bearing fault diagnosis method for pump mechanical equipment
Pan et al. UAV anomaly detection using active learning and improved S3VM model
Shen et al. An attention-based digraph convolution network enabled framework for congestion recognition in three-dimensional road networks
Siddique et al. State-dependent self-adaptive sampling (SAS) method for vehicle trajectory data
CN114596726B (en) Parking berth prediction method based on interpretable space-time attention mechanism
CN109784562B (en) Smart power grid power load prediction method based on big data space-time clustering
Ren et al. Passenger flow prediction in traffic system based on deep neural networks and transfer learning method
CN115311844B (en) Expressway traffic state estimation method based on self-supervision learning support vector machine
Yin et al. A multivariate time series prediction schema based on multi-attention in recurrent neural network
Wang et al. Contrastive GNN-based Traffic Anomaly Analysis Against Imbalanced Dataset in IoT-based ITS
CN116662899A (en) Noise-containing data anomaly detection method based on self-adaptive strategy
Attanasi et al. A hybrid method for real-time short-term predictions of traffic flows in urban areas
Li et al. Hydropower generation forecasting via deep neural network
CN105930430A (en) Non-cumulative attribute based real-time fraud detection method and apparatus

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
GR01 Patent grant
GR01 Patent grant