CN103106793B - Traffic state discriminated method based on real-time driving direction and transit time quantum information - Google Patents
Traffic state discriminated method based on real-time driving direction and transit time quantum information Download PDFInfo
- Publication number
- CN103106793B CN103106793B CN201310010320.0A CN201310010320A CN103106793B CN 103106793 B CN103106793 B CN 103106793B CN 201310010320 A CN201310010320 A CN 201310010320A CN 103106793 B CN103106793 B CN 103106793B
- Authority
- CN
- China
- Prior art keywords
- traffic
- traffic state
- real
- peak period
- traffic flow
- 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.)
- Expired - Fee Related
Links
Landscapes
- Traffic Control Systems (AREA)
Abstract
The invention relates to the technical field of traffic state discrimination, in particular to a traffic state discriminated method based on a real-time driving direction and transit time quantum information. A traffic state, as an overall indicator of evaluating an urban traffic operation state, is influenced by a plurality of factors. The traffic state discriminated method based on the real-time driving direction and the transit time quantum information puts forward a traffic state discriminated module based on the driving direction and the traffic time quantum, develops traffic state discriminated thought from the existing single mode to various comprehensive discrimination, applies a network theory with a self-adaptation characteristic to the traffic state discrimination, digs out objective data (traffic flow parameter data) rules form subjective factors (feeling of a person for the traffic state) and searches data changes which correspond to the state changes. Due to the adoption of dynamic intervals to distinguish changes of the traffic state, a parameter of the traffic state discriminated module can achieve self-adaptation adjustment, truly achieves correspondingly adjusting a traffic state discriminated threshold value according to traffic flow characters of a section of a road and achieves self-adaptation discrimination of the traffic state.
Description
Technical field
The present invention relates to traffic state judging technical field, particularly a kind of traffic state judging method based on real-time direction of traffic and current period information.
Background technology
Traffic state judging is exactly in fact the characteristic according to traffic flow, is a class by similar state demarcation.Traffic state judging can be divided into artificial cognition method and automatic distinguishing method.Although artificial cognition method is simple, directly, convenient, be subject to the restriction of manpower, the personnel amount of needs and working strength are all larger, therefore, artificial cognition method gradually replace by automatic distinguishing method.
Traffic behavior automatic distinguishing method has a lot, such as, and California algorithm and McMaster algorithm etc.California algorithm mainly by comparing the traffic flow parameters such as the occupation rate of adjacent inspection stations detection, differentiates traffic behavior.McMaster algorithm is the algorithm using ordinary traffic jam as traffic state judging object the earliest, this algorithm is based on catastrophe theory, on the basis of a large amount of historical data, flow-occupation rate the data of more crowded and uncongested, as masterplate, by comparing for twice of measured data and flow-occupation rate masterplate, judge whether congested in traffic and congested in traffic type occurs.
For the different sections of highway of urban road, be subject to the restriction of the conditions such as the physical condition of road own and service facility, traffic behavior shows larger difference at different sections of highway; And for same section, traffic behavior neither be unalterable, the variable effect of outside environmental elements driver and is changed the sensation of same place traffic behavior.Such as in the travel directions in section, possible a direction generation traffic congestion, and another direction smooth traffic; In rush-hour, driver may be accustomed to the traffic congestion in section psychologically, and at non-peak period, cannot accept traffic congestion in short-term on the contrary.The change of these outside environmental elements, while affecting driver psychology impression, also have impact on the judgement of driver to traffic behavior.
Prior art does not consider the change of these outside environmental elements, directly according to the traffic behavior of a fixing judgment of standard road, often causes the erroneous judgement of traffic behavior.
In prior art, shorter mention direction of traffic and traffic slot relevant information are on the impact of traffic state judging, but directly according to the traffic behavior of a fixing judgment of standard road, and the influence of single factor is only considered in most of quantitative analysis method.The present invention proposes the traffic state judging model based on direction of traffic and traffic slot, traffic state judging thought develops from the differentiation of original single-mode to many-sided comprehensive distinguishing, self-adaptive features network theory is applied in traffic state judging, from the factor (people is to the impression of traffic behavior) of subjectivity, excavate the rule of objective data (traffic flow parameter data), find the data variation corresponding to state change.Adopt the change of motion interval difference traffic behavior, the parameter of traffic discrimination model realizes adaptive adjustment, really accomplishes the traffic flow feature according to section self, adjusts accordingly traffic state judging threshold values, and the self-adjusting reaching traffic behavior differentiates.The traffic state judging method that the present invention proposes mainly carries out self-adaptation differentiation for different traffic flow modes, no matter the traffic flow modes difference in section is much, model can both adjust according to the traffic behavior threshold values of a large amount of historical datas to this section, traffic state judging is carried out to different sections of highway, makes differentiation result reach optimum.Traffic state judging model of the present invention can realize the self-adaptative adjustment of environmental factor to external world, has very strong practicality, and application prospect is very wide.
Summary of the invention
In view of this, the present invention, according to the traffic stream characteristics of different road, to traffic state judging according to adjusting and revising, reaches the object optimizing traffic state judging.
The present invention adopts following scheme to realize: a kind of traffic state judging method based on real-time direction of traffic and current period information, is characterized in that comprising the following steps:
Step S01: the traffic flow data gathering section, traffic flow data is divided according to direction of traffic and corresponding traffic slot, different direction of traffics and traffic slot are combined into major movement traffic flow peak period, traffic flow peak period of secondary direction of traffic, the period traffic flow of major movement flat peak and traffic flow pattern of secondary direction of traffic flat peak period, set up the traffic flow data collection of associative mode respectively, historical data is as training dataset, and the data of Real-time Collection are as predictive data set;
Step S02: be normalized described training dataset and predictive data set, makes input value be distributed in the interval of [-1,1], eliminates the impact that variable causes because of the difference on the order of magnitude;
Step S03: set up SOM neural network, input layer quantity is determined by the number of adopted traffic flow parameter of classifying;
Step S04: training SOM neural network, for given training dataset, weight data collection is initialized as random value, determines competition layer winning unit, then revise the connection weights of winning unit and neighborhood unit thereof, right value update rule is:
（3）
（4）
In formula,
for the connection weights of winning unit;
for the input of winning unit;
,
represent the weights learning factor of winning unit and neighborhood unit thereof respectively, span is (0,1), and
;
Step S05: after training satisfactory network, substitutes into predictive data set in network, can realize traffic behavior classification; Return step S01 more simultaneously, predictive data set is fed back to training dataset, constantly improves SOM neural network, the automatic adjustment of implementation model.
In an embodiment of the present invention, the unbalancedness directional spreding coefficient of described direction of traffic
represent:
（1）
Direction of traffic distribution coefficient
value is greater than 50%, more close to 50%, shows that the volume of traffic of two direction of traffics gets over convergence balance, more away from 50%, shows that the volume of traffic difference of two direction of traffics is larger.
In an embodiment of the present invention, described traffic slot is divided into peace peak period peak period, with coefficient peak period
represent the volume of traffic difference of peace peak period peak period, expression formula is as follows:
（2）。
In an embodiment of the present invention, adopt the volume of traffic, average speed and traffic density as the index of classification in described step S03, then input neuron is 3; Competition layer neuron number is determined by number of categories, traffic behavior is divided into 3 classes, then competition layer has 3 neurons.
In an embodiment of the present invention, in described step S04 training process, reduce the variable quantity of neighborhood and weights gradually; The weights learning factor of winning unit
and field unit reduces gradually,
with field width
reduce gradually in an iterative process, its adjustable strategies expression formula is:
（5）
（6）
In formula,
,
be respectively
with
initial value,
value between 0.2-0.5,
it is the width of 1/2 or 1/3;
for current iteration number of times;
for the iteration total degree of setting.
Accompanying drawing explanation
Fig. 1 is embodiment of the present invention schematic flow sheet.
Fig. 2 is the traffic state judging model schematic of the embodiment of the present invention based on SOM.
Embodiment
Before explanation specific embodiments of the invention, in order to the effect better allowing those skilled in the art understand the present invention's realization, divide simply introduce lower the technology used in the present invention means below at 4.
1. based on the traffic state analysis of direction of traffic
In the round both direction of same path, traffic flow is not keep balance always.The section coming and going connection due to the difference of different positions land character and urban road is different, and the traffic flow coming and going both direction can exist larger difference.The unbalancedness of this direction of traffic can use directional spreding coefficient
represent:
（1）
Direction of traffic distribution coefficient
value is generally greater than 50%, more close to 50%, shows that the volume of traffic of two direction of traffics gets over convergence balance, more away from 50%, shows that the volume of traffic difference of two direction of traffics is larger.
2. based on the traffic state analysis of traffic slot
Same section, along with time variations, traffic flow is also in continuous change.In commuter time morning and evening, go out line density and increase, the volume of traffic there will be peak, for ordinary municipal road, considers that its peak hour flow stream mode can meet the requirement of traffic control; And Urban Advanced road that is more stable for traffic flow, that continue, peak period is longer than the ordinary municipal road duration, therefore the traffic flow of peak hour can not be considered merely, but consider as a whole with continuous print traffic flow peak period, peace peak period peak period is distinguished, avoids flat peak period traffic behavior all to fall within certain lower interval.With coefficient peak period
represent the volume of traffic difference of peace peak period peak period, expression formula is as follows:
（2）
3. Multifactor Combination analysis
According to the analysis to direction of traffic and traffic slot, separable go out different direction of traffic and traffic slot impact under traffic flow, there is significant difference each other in these traffic flows, and direction of traffic and traffic slot act on the impact of traffic flow sometimes simultaneously, therefore need to combine these factors.As chosen direction of traffic and traffic slot two factors, may be combined with into major movement traffic flow peak period, traffic flow peak period of secondary direction of traffic, the period traffic flow of major movement flat peak and the flat peak period traffic flow of secondary direction of traffic, namely the traffic flow in a certain section is divided into 4 kinds of patterns, then respectively the traffic behavior under these 4 kinds of patterns is classified, the impact that the factor reducing direction of traffic and traffic slot produces traffic behavior classification results.
the Parameter Self of 4.SOM neural network
Traffic state judging process is the characteristic according to traffic flow, is a class similar state demarcation, and for the threshold values of each classification, does not have the clear and definite criteria for classifying.Due to different individualities to same state feel be not quite similar, the threshold values border determined is nonlinear, the threshold values that namely neither one is fixing.And use Self-Organizing Feature Maps (Self-Organizing Maps, the target that SOM) outside can not be used to provide exports the classification found in complex data, namely do not need to predict threshold values, the present invention adopts Self-Organizing Feature Maps to differentiate traffic behavior.
The monolayer neural networks that SOM neural network is made up of input layer and competition layer, input layer is the neuron of one dimension, and neuron number is
; Competition layer is the neuron of two dimension, by
individual neuron forms a two-dimensional planar array.The neuron of input layer has weights to be connected with the neuron of competition layer, also has the lateral inhibition of local to connect between competition layer node; Have two kinds to connect weights in SOM neural network, a kind of is the connection weights of neuron to outside input reaction, and another is the connection weights between neuron, and its size controls interactive power between neuron; Therefore, in the training process, not only to regulate the connection weights of winning unit in competition, also need the weights of the neighborhood unit regulating winning unit.The target that Self-Organizing Feature Maps can not use outside to provide exports the classification found in complex data, does not namely need to predict threshold values, is applicable to processing mass data.
Along with the development of intelligent transport technology, in time, traffic behavior classification accurately and forecast, to alleviation urban traffic blocking with dredge urban traffic pressure and play an important role.Traffic behavior, as the overall target evaluating urban transportation operation conditions, can provide the decision-making foundation of driving path for driver.But traffic behavior is a hazy sensations amount, the vehicle number not only with current on road is relevant, and also have substantial connection with the psychological bearing capability of driver and external environment, each factor is mutually related.Therefore, traffic behavior can not be quantized with traffic flow parameter merely, but the various factors affecting traffic behavior will be considered, comprehensive analysis and quantification is carried out to each factor.
For the different sections of highway of urban road, be subject to the restriction of the conditions such as the physical condition of road own and service facility, traffic behavior shows larger difference at different sections of highway; And for same section, traffic behavior neither be unalterable, the variable effect of outside environmental elements driver and is changed the sensation of same place traffic behavior.Such as in the travel directions in section, possible a direction generation traffic congestion, and another direction smooth traffic; In rush-hour, driver may get used to the traffic congestion in section psychologically, and at non-peak period, cannot accept traffic congestion in short-term on the contrary; The change of these outside environmental elements, while affecting driver psychology impression, also have impact on the judgement of driver to traffic behavior.Therefore, the present invention chooses direction of traffic and traffic slot as the factor affecting traffic behavior classification, proposes, to the traffic state judging method under the combination of each factor, to refer to Fig. 1.
Basic ideas of the present invention are: arrange traffic flow data, set up the SOM neural network of traffic state judging, and training SOM neural network, classifies to real-time traffic flow data according to the network trained.Key step is as follows:
The first step: the traffic flow data gathering section, traffic flow data is divided according to direction of traffic and corresponding traffic slot, different direction of traffics and traffic slot are combined into 4 kinds of patterns, set up the traffic flow data collection of associative mode respectively, historical data is as training dataset, and the data of Real-time Collection are as predictive data set.
Second step: be normalized training dataset and predictive data set, makes input value be distributed in the interval of [-1,1], eliminates the impact that variable causes because of the difference on the order of magnitude.
3rd step: set up SOM neural network, input layer quantity is determined by the number of adopted traffic flow parameter of classifying.The present embodiment adopts the volume of traffic, average speed and traffic density as the index of classification, and therefore input neuron is 3.Competition layer neuron number is determined by number of categories, traffic behavior is divided into 3 classes, and therefore competition layer has 3 neurons.
4th step: training SOM neural network.For given training dataset, weight data collection is initialized as random value, determines competition layer winning unit, then revise the connection weights of winning unit and neighborhood unit thereof, right value update rule is:
（3）
（4）
In formula,
for the connection weights of winning unit;
for the input of winning unit;
,
represent the weights learning factor of winning unit and neighborhood unit thereof respectively, span is (0,1), and
.
In learning process, reduce the variable quantity of neighborhood and weights gradually.The weights learning factor of winning unit
and field unit reduces gradually,
with field width
reduce gradually in an iterative process, its adjustable strategies expression formula is:
（5）
（6）
In formula,
,
be respectively
with
initial value,
value between 0.2-0.5,
be generally the width of 1/2 or 1/3;
for current iteration number of times;
for the iteration total degree of setting.
5th step: after training satisfactory network, can substitute in network by predictive data set, can realize traffic behavior classification; Return the first step more simultaneously, predictive data set is fed back to training dataset, constantly improves SOM neural network, the automatic adjustment of implementation model.
By above-mentioned learning process, can the traffic flow under various combination be classified, thus the traffic behavior under obtaining different affecting factors, model is as shown in Figure 2.
The foregoing is only preferred embodiment of the present invention, all equalizations done according to the present patent application the scope of the claims change and modify, and all should belong to covering scope of the present invention.
Claims (4)
1., based on a traffic state judging method for real-time direction of traffic and current period information, it is characterized in that comprising the following steps:
Step S01: the traffic flow data gathering section, traffic flow data is divided according to direction of traffic and corresponding traffic slot, different direction of traffics and traffic slot are combined into major movement traffic flow peak period, traffic flow peak period of secondary direction of traffic, the period traffic flow of major movement flat peak and traffic flow pattern of secondary direction of traffic flat peak period, set up the traffic flow data collection of associative mode respectively, historical data is as training dataset, and the data of Real-time Collection are as predictive data set;
Step S02: be normalized described training dataset and predictive data set, makes input value be distributed in the interval of [-1,1], eliminates the impact that variable causes because of the difference on the order of magnitude;
Step S03: set up SOM neural network, input layer quantity is determined by the number of adopted traffic flow parameter of classifying;
Step S04: training SOM neural network, for given training dataset, weight data collection is initialized as random value, determines competition layer winning unit, then revise the connection weights of winning unit and neighborhood unit thereof, right value update rule is:
（3）
（4）
In formula,
for the connection weights of winning unit;
for the input of winning unit;
,
represent the weights learning factor of winning unit and neighborhood unit thereof respectively, span is (0,1), and
;
Step S05: after training satisfactory network, substitutes into predictive data set in network, can realize traffic behavior classification; Return step S01 more simultaneously, predictive data set is fed back to training dataset, constantly improves SOM neural network, the automatic adjustment of implementation model; The unbalancedness directional spreding coefficient of direction of traffic
represent:
（1）
Direction of traffic distribution coefficient
value is greater than 50%, more close to 50%, shows that the volume of traffic of two direction of traffics gets over convergence balance, more away from 50%, shows that the volume of traffic difference of two direction of traffics is larger.
2. the traffic state judging method based on real-time direction of traffic and current period information according to claim 1, is characterized in that: described traffic slot is divided into peace peak period peak period, with coefficient peak period
represent the volume of traffic difference of peace peak period peak period, expression formula is as follows:
（2）。
3. the traffic state judging method based on real-time direction of traffic and current period information according to claim 1, it is characterized in that: adopt the volume of traffic, average speed and traffic density as the index of classification in described step S03, then input neuron is 3; Competition layer neuron number is determined by number of categories, traffic behavior is divided into 3 classes, then competition layer has 3 neurons.
4. the traffic state judging method based on real-time direction of traffic and current period information according to claim 1, is characterized in that: in described step S04 training process, reduces the variable quantity of neighborhood and weights gradually; The weights learning factor of winning unit
and field unit reduces gradually,
with field width
reduce gradually in an iterative process, its adjustable strategies expression formula is:
（5）
（6）
In formula,
,
be respectively
with
initial value,
value between 0.2-0.5,
it is the width of 1/2 or 1/3;
for current iteration number of times;
for the iteration total degree of setting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310010320.0A CN103106793B (en) | 2013-01-11 | 2013-01-11 | Traffic state discriminated method based on real-time driving direction and transit time quantum information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310010320.0A CN103106793B (en) | 2013-01-11 | 2013-01-11 | Traffic state discriminated method based on real-time driving direction and transit time quantum information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103106793A CN103106793A (en) | 2013-05-15 |
| CN103106793B true CN103106793B (en) | 2015-02-04 |
Family
ID=48314613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310010320.0A Expired - Fee Related CN103106793B (en) | 2013-01-11 | 2013-01-11 | Traffic state discriminated method based on real-time driving direction and transit time quantum information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103106793B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104573873B (en) * | 2015-01-23 | 2017-08-25 | 哈尔滨工业大学 | A kind of terminal passenger departure's method for predicting based on fuzzy decision-tree |
| CN105679021B (en) * | 2016-02-02 | 2018-11-06 | 招商局重庆交通科研设计院有限公司 | Journey time fusion forecasting and querying method based on traffic big data |
| CN105869417B (en) * | 2016-06-16 | 2019-07-23 | 兰州理工大学 | A kind of traffic signal control method and system based on combination control |
| CN106095963B (en) * | 2016-06-17 | 2019-12-06 | 上海经达信息科技股份有限公司 | Vehicle driving behavior analysis big data public service platform under internet + era |
| CN107545785A (en) * | 2017-07-21 | 2018-01-05 | 华南理工大学 | A kind of river channel running method based on big data |
| CN108399745B (en) * | 2018-03-01 | 2020-10-16 | 北京航空航天大学合肥创新研究院 | Unmanned aerial vehicle-based time-interval urban road network state prediction method |
| CN109686085B (en) * | 2018-12-17 | 2020-05-05 | 北京交通大学 | GPS data based dangerous cargo transport vehicle stop node activity type identification method |
| CN109697849B (en) * | 2018-12-26 | 2020-06-16 | 航天科工广信智能技术有限公司 | Intelligent traffic time interval dividing method based on moving average algorithm |
| CN110276955B (en) * | 2019-07-18 | 2021-04-06 | 中南大学 | Traffic congestion state evaluation method oriented to personal perception of travelers |
| CN112037511B (en) * | 2020-08-28 | 2022-03-08 | 浙江中控信息产业股份有限公司 | Method for identifying single cross port signal timing imbalance state |
| CN112598902B (en) * | 2020-12-09 | 2022-08-16 | 安徽百诚慧通科技股份有限公司 | Intersection turning unbalance degree characterization method and device, electronic equipment and storage medium |
| CN116842018B (en) * | 2023-07-06 | 2024-02-23 | 上海比滋特信息技术有限公司 | Big data screening method and system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101540103B (en) * | 2008-03-17 | 2013-06-19 | 上海宝康电子控制工程有限公司 | Method and system for traffic information acquisition and event processing |
| CN102034350B (en) * | 2009-09-30 | 2012-07-25 | 北京四通智能交通系统集成有限公司 | Short-time prediction method and system of traffic flow data |
| CN102169630B (en) * | 2011-03-31 | 2013-04-24 | 上海电科智能系统股份有限公司 | Quality control method of road continuous traffic flow data |
-
2013
- 2013-01-11 CN CN201310010320.0A patent/CN103106793B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN103106793A (en) | 2013-05-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103106793B (en) | Traffic state discriminated method based on real-time driving direction and transit time quantum information | |
| El Hatri et al. | Fuzzy deep learning based urban traffic incident detection | |
| Araghi et al. | A review on computational intelligence methods for controlling traffic signal timing | |
| CN106205156B (en) | A kind of intersection self-healing combination control method for the mutation of part lane flow | |
| WO2021218235A1 (en) | Road traffic jam early warning method and system | |
| CN105654744B (en) | A kind of improvement traffic signal control method based on Q study | |
| CN103578273B (en) | A kind of road traffic state estimation method based on microwave radar data | |
| CN102542818B (en) | A kind of coordination control method for traffic signal of zone boundary based on organic calculating | |
| CN105701571A (en) | Short-term traffic flow prediction method based on nerve network combination model | |
| CN110889562B (en) | Dynamic city model system based on scenario planning | |
| CN108280998A (en) | Short-time Traffic Flow Forecasting Methods based on historical data dynamic select | |
| CN112950963B (en) | Self-adaptive signal control optimization method for main branch intersection of city | |
| Dahal et al. | GA-based learning for rule identification in fuzzy neural networks | |
| CN111907523B (en) | Vehicle following optimizing control method based on fuzzy reasoning | |
| Quek et al. | A novel self-organizing fuzzy rule-based system for modelling traffic flow behaviour | |
| Chen et al. | Dynamic traffic light optimization and Control System using model-predictive control method | |
| Alam et al. | Intellegent traffic light control system for isolated intersection using fuzzy logic | |
| CN108364490A (en) | Municipal highway transit system vehicle runs regulation and control method | |
| Choy et al. | Real-time coordinated signal control through use of agents with online reinforcement learning | |
| Eze et al. | Fuzzy logic model for traffic congestion | |
| CN108364491A (en) | Driver information based reminding method based on vehicle flowrate prediction | |
| Jiang et al. | Clustering-neural network models for freeway work zone capacity estimation | |
| CN116564088A (en) | Urban traffic state monitoring and regional signal control system oriented to information physical fusion system | |
| CN113688200B (en) | Decision tree-based special population action track collection method and system | |
| CN114444922A (en) | Hybrid traffic efficiency evaluation method under group intelligent control |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150204 Termination date: 20180111 |