CN101976501B - Principal component analysis and neural network based port road safety prediction method - Google Patents
Principal component analysis and neural network based port road safety prediction method Download PDFInfo
- Publication number
- CN101976501B CN101976501B CN2010105258524A CN201010525852A CN101976501B CN 101976501 B CN101976501 B CN 101976501B CN 2010105258524 A CN2010105258524 A CN 2010105258524A CN 201010525852 A CN201010525852 A CN 201010525852A CN 101976501 B CN101976501 B CN 101976501B
- Authority
- CN
- China
- Prior art keywords
- road
- neural network
- component analysis
- data
- port
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Traffic Control Systems (AREA)
Abstract
The invention relates to a principle component analysis and neural network based port road safety prediction method, belonging to the technical field of road traffic safety. The method includes the following steps: factors influencing port road safety are determined; history traffic data raw data is extracted, the raw data is normalized, then principle component analysis is carried out on the data, thus obtaining principle component influencing port road safety; a BP neural network model is built; the built BP neural network model is trained, and sensitivity analysis is carried out by utilizing the trained model; and port road safety is predicted. Compared with the traditional BP neural network, the invention has simple structure and high prediction accuracy, and the invention can be applied to road safety check, thus providing theory basis for construction and renovation of road.
Description
Technical field
The invention belongs to the technical field of traffic safety, relate to a kind of road traffic accident forecast method.
Background technology
Along with the increase of highway mileage and vehicle population, China's traffic hazard is continuous ascendant trend, and people's safety of life and property has received great threat, has become one of serious social concern that receives much concern now.The research of road safety evaluation has very big effect for the minimizing of traffic hazard.The road safety evaluation is meant angle analysis road traffic environment factor and the relation with traffic accidents from microcosmic, and with the road improvement safety case, formulation technology and policies and measures instruct the road safety design.
The research that has now existed some relevant traffic accident predictions to analyze; But basically all be to analyze from the angle of macroscopic view; The influence factor of choosing is a total population; Macro-datas such as vehicle guaranteeding organic quantity, the analysis result that draws are the predictions of the traffic hazard number of times in a zone, and this does not have too big effect for prevention and minimizing traffic hazard.The present invention is the angle from microcosmic, analyzes the influence of the factor of road to traffic hazard, so just can instruct the design and the trimming of road.
Traffic hazard is the complicated nonlinear systems that is produced by numerous factor actings in conjunction; And the relation between these factors is difficult to describe with the method for resolving; And neural network model is through the study to historical data; Can approach any Nonlinear Mapping with arbitrary accuracy, so the present invention utilizes neural network model that forecast analysis is carried out in traffic hazard.But; The factor that influences traffic hazard is a lot, for prediction accuracy, does not omit important information; Will consider factor as much as possible; And these factors are owing to all be reflection to traffic hazard, and it is overlapping unavoidably to have great deal of information, and this may hide the real characteristic and the inherent law of traffic hazard.Therefore; The present invention has introduced PCA, and raw data is carried out principal component analysis (PCA), between influence factor, extracts some principal ingredients the complicated relation; Carry out quantitative test thereby effectively utilize the large-scale statistical data, disclose the internal relation between the variable.
Summary of the invention
The objective of the invention is to overcome the above-mentioned deficiency of prior art, provide a kind of simple to operate, be applicable to the method for traffic accident prediction in the road safety verification.
A kind of port road safety prediction method based on principal component analysis (PCA) and neural network comprises the following steps:
The first step: the factor of confirming to influence the port road security: confirm driver's quantity, driver's quantity, regional population's sum; Motor vehicle quantity, road total length, cargo throughput of port; The length of road segments, the quantity of small-sized branch road, shoulder width; Number of track-lines, 12 variablees of trackside danger classes and horizontal curve curvedness parameters are for influencing the factor of port road security;
Second step: principal component analysis (PCA): the raw data to relevant above-mentioned 12 variablees in the historical traffic data is extracted, and raw data is carried out normalization, then data is carried out principal component analysis (PCA), obtains influencing the main composition of port road security;
The 3rd the step: set up the BP neural network model: with resulting major component as input variable; With the number of times of traffic hazard as output variable; According to the scope of determined hidden neuron number, set up single hidden layer BP neural network model of a hidden layer neuron number could vary; Raw data is trained; Through the error contrast of each training, confirm best hidden layer neuron number, through the fall off rate of the total error sum of squares of the error sum of squares after each training; Judge that whether selected learning rate is suitable, finally obtains best learning rate.
The 4th step: the BP neural network model to setting up is trained, and utilizes the model that trains to carry out sensitivity analysis, confirms the quantity of small-sized branch road, shoulder width, and number of track-lines, the trackside danger classes, the horizontal curve curvedness parameters is to the influence of traffic hazard number of times;
The 5th step: the security to port road is predicted.
Technique effect of the present invention is following:
1, based on the BP neural network model of principal component analysis (PCA).The classic BP neural network directly utilizes raw data to train as input variable; But there is the problem of two aspects in these raw data; The one, data volume is big, and dimension is many, and the one, there is correlativity between the data; Information has overlapping, and this will cause the reduction of the not high and training speed of result's precision.Raw data is carried out principal component analysis (PCA), between influence factor, extract some principal ingredients the complicated relation, carry out quantitative test, disclose the internal relation between the variable thereby effectively utilize the large-scale statistical data.
2, choosing influence factor from the angle of microcosmic predicts.Traditional traffic accident prediction method all is to utilize some macroscopical influence factors to analyze as input variable, and what obtain is the data of the traffic hazard of an integral body.The present invention is as the input data with the linear factor of road; Analyze road factors and relation with traffic accidents; Thereby obtain the influence of road alignment to traffic hazard, this conclusion can be applied to during traffic safety estimates, for the design and the trimming of road provides theoretical foundation.
Description of drawings
Fig. 1: BP neural network structure.
Fig. 2: based on the BP neural network model framework of principal component analysis (PCA).
Embodiment
In current society, traffic safety gets more and more people's extensive concerning, and the research of road safety evaluation has very big effect for the minimizing of traffic hazard.The present invention uses and based on the BP neural network model of principal component analysis (PCA) road alignment and relation with traffic accidents is analyzed and researched from the angle of microcosmic.
Introduce workflow of the present invention below in detail:
One, data chooses
Road traffic is the dynamic system that is made up of elementals such as people, car, road and environment.Traffic hazard is in the road traffic system, because the cooperation of people, car, road, all key elements of environment is lacked of proper care and the incident of accidental burst.Therefore, when choosing the factor of influence of traffic hazard, analyze from above four aspects.Table 1 has been listed the statistics of China's road traffic accident main cause in 2002.
Table 1 China road traffic accident main cause distribution (2002)
1. human factor
Can find out that by table 1 death toll that causes owing to people's reason accounts for 88.98% of the dead sum of traffic hazard then, wherein automobile driver be main cause account for 78.56%, the bicycle driver accounts for 4.20%, pedestrian and rider account for 6.22%.It is thus clear that in general, the key of traffic hazard is automobile driver, because with respect to pedestrian and cyclist, automobile driver is the traffic powerhouse.Therefore two indexs of driver's quantity and regional population sum have been chosen.
2. the factor of car: causing the second largest key element of traffic hazard is vehicle.At the vehicle of travels down, existing motor vehicle has bicycle and other bicycles again, a kind of vehicles fast during motor vehicle wherein, and energy is maximum, and protective is also best.But this protective is only protected driver and occupant, and therefore, with respect to bicycle and other bicycles, motor vehicle is the traffic powerhouse.But therefore increasing also of bicycle quantity can choose motor vehicle quantity to how much the exerting an influence of traffic hazard, and three indexs of bicycle quantity and average daily traffic volume are as factor of influence.
3. road factors: the statistics from table is that the traffic hazard proportion that causes of main cause is very little with the road defective, but can finds after the serious analysis that true really not so, a lot of accidents are all caused by the poor road condition indirectly.Therefore, among the present invention at first with a road according to its linear plurality of sections that is divided into, choose following data as the factor that influences traffic hazard: the length of road segments; The density parameter of small-sized branch road; Shoulder width; The curvedness parameters of horizontal curve.
4. environmental factor: the environmental factor that influences traffic safety can be divided into physical environment and artificial environment, and physical environment mainly comprises the geographic position, meteorological condition, and the time etc., artificial environment comprises land use situation, trackside interference, road barricade thing etc.The present invention has chosen the trackside danger classes as factor of influence.Table 2 is seen in the classification of trackside danger classes.
The classification of table 2 trackside danger classes
The trackside danger classes | The trackside performance specification |
1 | Trackside headroom district>9m |
2 | Trackside headroom district is between 6~7.5m |
3 | The about 3m of trackside headroom sector width |
4 | Trackside headroom district is between 1.5~3m; Has guardrail (apart from pavement edge 1.5~2m) |
5 | Trackside headroom district is between 1.5~3m; Has guardrail (apart from pavement edge 0~1.5m) |
6 | Trackside headroom district is between smaller or equal to 1.5m; In the outer 2m of pavement edge hard barrier is arranged |
7 | Trackside headroom sector width is smaller or equal to 1.5m; Trackside has cliff or steep steep cliff |
Two, data are carried out principal component analysis (PCA)
Above the factor that influences traffic hazard is analyzed, totally 12 of influence factors are respectively driver's quantity, driver's quantity; Regional population's sum, motor vehicle quantity, road total length, cargo throughput of port; The length of road segments, the quantity of small-sized branch road, shoulder width; Number of track-lines, trackside danger classes, horizontal curve curvedness parameters.Chosen the data of 1999-2007 among the present invention, Tianjin bar road has been divided into 7 sections according to linear, two-way promptly 14 sections, extract above-mentioned road data, formed raw data of the present invention, totally 126 groups.At first raw data is carried out normalization, then these 12 variablees are taken all factors into consideration, obtain 5 major components like table 3.The detail analysis method can referring to " based on the road Traffic Accident Prediction of principal component analysis (PCA). traffic and safety, 2009 (No.204): 86-90 "
Table 3 principal component analysis (PCA) result
Major component 1 | Major component 2 | Major component 3 | Major component 4 | Major component 5 | |
The major component standard deviation | 16.6478 | 6.3349 | 4.9490 | 4.2317 | 3.4334 |
The variance proportion | 38.0760% | 14.4889% | 11.3192% | 9.6786% | 7.8526% |
Contribution rate of accumulative total | 38.0760% | 52.5649% | 63.8841% | 73.5627% | 81.4153% |
The contribution rate of accumulative total of preceding 5 major components surpasses 80%, therefore only needs just can summarize raw data preferably with these 5 major components.
(x
1?x
2?x
3?x
4?x
5?x
6?x
7?x
8?x
9?x
10?x
11?x
12)
T
Three, the BP neural network model confirms
Learn by preceding surface analysis; Traffic hazard is the complicated nonlinear systems that is produced by numerous factor actings in conjunction; And the relation between these factors is difficult to describe with the method for resolving, and the various aspects that can not effectively contain traffic system with traditional forecast model are difficult to, and set up concrete functional form; Even set up model, also be difficult to find an appropriate method for parameter estimation to confirm parameter.And neural network has through learning to approach with arbitrary accuracy the ability of any Nonlinear Mapping, so accuracy of predicting is very high.In addition, neural network is to reflect function through mapping, so can not receive the restriction of the concrete form of function, the mathematic(al) structure that need not suppose just can be predicted through setting up the I/O model.Therefore, predict that with neural network model traffic hazard can overcome the difficulty of modeling problem, and can improve accuracy of predicting.
The present invention has selected the BP network in the forward direction type network that traffic hazard is predicted.According to statistics, 80%~90% neural network model has adopted BP network or its version.The BP network is the core of feedforward network, has embodied in the neural network elite, the most perfect content.
1.BP network structure
The BP network is a kind of multilayer feedforward network of one way propagation, and structure is as shown in Figure 1.The BP network is a kind of neural network that has more than three layers or three layers, comprises input layer, middle layer (latent layer) and output layer.Realize full the connection between the levels, and do not have connection between every layer of neuron.
In the present invention, when the BP network of traffic accident prediction was carried out structural design, the quantity that increases hidden layer can improve the non-linear mapping capability of BP network, but hidden layer outnumber certain value, the performance of network is reduced.And a continuous function of quilt can approach with the BP network of single latent layer in the closed interval for any, thereby one three layers BP network just can be accomplished arbitrarily, and n ties up the mapping that m ties up.So totally three layers on BP network of the present invention.
By the problem decision that will solve, in the present invention, the neuron number of input layer is made up of the major component of influence factor, totally 5 fully for the input layer of network and output layer; The neuron number of output layer is definite by predicting the outcome, and totally 1, i.e. traffic hazard number of times, the framework of model of the present invention is as shown in Figure 2.
2. the design of latent layer
It is a very complicated problems that the neuron number of latent layer is selected, and often need come definitely according to designer's experience and experiment repeatedly, thereby does not exist a desirable analytic expression to represent.The number of the number of hidden neuron and the requirement of problem, I/O unit all has direct relation.The hidden neuron number can cause learning time long, error not necessarily best too much, the sample that also can cause poor fault tolerance, not see before can not discerning, therefore certain hidden unit number that has a best.Below 3 formula be the reference formula when can be used for selecting best hidden unit to count.
(1)
Wherein, k is a sample number, n
1Be the hidden unit number, n is the input block number.If 1>n
1,
(2)
wherein; M is the output neuron number; N is the input block number; A is the constant between [1,10].
(3) n
1=log
2N, wherein, n is the input block number.
According to top empirical design formula, and consider actual conditions of the present invention, the hidden neuron number of the network that addresses this problem should be between 3~12.Therefore, at first designed the BP network of a hidden layer neuron number could vary,, confirmed best hidden layer neuron number through the error contrast.Because the error result of each training all has discrepancy, thus set training 10 times, with the error addition of ten training, obtain hidden layer neuron and be 7 BP network to function to approach effect best.
3. e-learning speed chooses
The weights variable quantity that is produced in the learning rate decision circulation each time.Big learning rate possibly cause the instability of system, but little learning rate will cause learning time longer, and possible speed of convergence is very slow, but the error amount that can guarantee network is not jumped out the low ebb on error surface and finally is tending towards minimum error values.So under one situation, tend to choose less learning rate to guarantee the stability of system.One is chosen at the scope of learning rate between the 0.01-0.7.
In the design of neural network, network will pass through the training of several different learning rates, through observing the error sum of squares ∑ e after the training each time
2Fall off rate judge whether selected learning rate suitable.If ∑ e
2It is very fast to descend, and explains that then learning rate is suitable, if ∑ e
2Reforming phenomena occurs, explain that then learning rate is excessive.BP network of the present invention is trained, and obtaining best learning rate is 0.04.
Four, model is realized and interpretation of result
1. analog result
126 groups of data of collecting are divided into two groups at random, and one group has 100, is used for training, and one group has 26, is used for testing.The present invention has realized the Forecasting Methodology of traffic hazard.The BP network model based on principal component analysis (PCA) that the application front draws carries out training and testing, obtains a result like table 4.Table 5 is raw data directly to be used the classic BP network carry out the result that training and testing obtains.
Table 4 is based on the BP neural network prediction result of principal component analysis (PCA)
● the overall accuracy of training data prediction is 84.2%, and test data is 78%.(N is the traffic hazard number of times)
Table 5BP neural network prediction result
● the overall accuracy of training data prediction is 65.6%, and test data is 75.3%.(N is the traffic hazard number of times)
2. interpretation of result
Draw through contrast, utilize the advantage of the principal component analysis (PCA) more classical neural network of BP neural network afterwards to be:
1) system is reduced to 5 by 12 input variables, has reduced the dimension of input variable, thereby can avoid " dimension disaster " problem of classical neural network effectively, accelerates the neural network prediction of back.When dimension was very high, data volume was big, made the structure of neural network become complicated, speed that will inevitably impact prediction.This model can add more influence factor when practical application, in the present invention because complete inadequately of data have been chosen 12 data, so the effect of principal component analysis (PCA) is so unobvious, when dimension is very big, just can obviously see the raising of predetermined speed.
2) be independent of each other between each major component, and each major component predicts that as the independence input of BP neural network prediction effect is better than the classical neural network of the high relevant input of higher-dimension.Because there be the overlapping of information between the related data, as if directly being trained, the data of not handling will hide true and useful information, be not easy to find accurate mapping relations in the training process, this will cause the reduction of precision of prediction.
But find that easily accuracy for predicting is still waiting to improve.This is because BP neural network self also exists deficiency.The BP algorithm can make weight convergence arrive certain value, but can not guarantee that it is the global minimum of error plane, and this is because adopt the gradient descent method may produce a local minimum.In addition, the number of plies of network hidden layer and the selection of unit number be the guidance on the gear shaper without theoretical still, and one is rule of thumb or through experiment repeatedly to confirm.Therefore, often there is very big redundancy in network, the burden of the e-learning that also increases to a certain extent.
3. sensitivity analysis
The present invention uses sensitivity analysis the influence factor of each input is analyzed the influence of traffic hazard number of times.Thereby obtain the influence of road alignment, building or rectifying and improving theoretical foundation is provided for road to traffic hazard.Sensitivity analysis is to change the influence of certain variable to the traffic hazard frequency distribution under all constant situation of its dependent variable of check.The result is as shown in table 6.The present invention carries out sensitivity analysis to branch road quantity, trackside danger classes and 3 variablees of horizontal curve curvedness parameters, and N is the number of times of traffic hazard, for example; Branch road quantity is 0, and when promptly not having small-sized branch road on the highway section, about 94.5% does not have accident to take place; 5.42% an accident can take place; 0.08% can take place twice, and along with the increase of branch road quantity, the possibility that accident takes place is big more.
Table 6 sensitivity analysis result
Claims (1)
1. the port road safety prediction method based on principal component analysis (PCA) and neural network comprises the following steps:
The first step: the factor of confirming to influence the port road security: confirm driver's quantity, driver's quantity, regional population's sum; Motor vehicle quantity, road total length, cargo throughput of port; The length of road segments, the quantity of small-sized branch road, shoulder width; Number of track-lines, 12 variablees of trackside danger classes and horizontal curve curvedness parameters are for influencing the factor of port road security;
Second step: principal component analysis (PCA): the raw data to relevant above-mentioned 12 variablees in the historical traffic data is extracted, and raw data is carried out normalization, then data is carried out principal component analysis (PCA), obtains influencing the major component of port road security;
The 3rd the step: set up the BP neural network model: with resulting major component as input variable; With the number of times of traffic hazard as output variable; According to the scope of determined hidden layer neuron number, set up single hidden layer BP neural network model of a hidden layer neuron number could vary; Raw data is trained,, confirm best hidden layer neuron number,, judge that whether selected learning rate is suitable, finally obtains best learning rate through the fall off rate of the error sum of squares after each training through the error contrast of each training;
The 4th step: the BP neural network model to setting up is trained, and utilizes the model that trains to carry out sensitivity analysis, confirms the quantity of small-sized branch road, shoulder width, and number of track-lines, the trackside danger classes, the horizontal curve curvedness parameters is to the influence of traffic hazard number of times;
The 5th step: the security to port road is predicted.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2010105258524A CN101976501B (en) | 2010-10-29 | 2010-10-29 | Principal component analysis and neural network based port road safety prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2010105258524A CN101976501B (en) | 2010-10-29 | 2010-10-29 | Principal component analysis and neural network based port road safety prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101976501A CN101976501A (en) | 2011-02-16 |
CN101976501B true CN101976501B (en) | 2012-05-30 |
Family
ID=43576381
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2010105258524A Active CN101976501B (en) | 2010-10-29 | 2010-10-29 | Principal component analysis and neural network based port road safety prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101976501B (en) |
Families Citing this family (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102867182A (en) * | 2012-08-17 | 2013-01-09 | 中国神华能源股份有限公司 | Method and system for detecting water plug loss failure of freight wagon |
CN103093643B (en) * | 2013-01-09 | 2015-06-17 | 东南大学 | Public parking lot berth quantity confirming method |
CN106844663B (en) * | 2017-01-23 | 2020-01-17 | 中国石油大学(华东) | Ship collision risk assessment method and system based on data mining |
CN106780266B (en) * | 2017-02-06 | 2020-10-23 | 重庆大学 | Principal component contribution degree parameter-based accident hotspot internal characteristic analysis and driving guidance method |
DE102017006434A1 (en) * | 2017-07-07 | 2019-01-10 | Wabco Gmbh | Method for predictively assessing a current driving situation and rating module |
CN107644532A (en) * | 2017-09-20 | 2018-01-30 | 中山大学 | A kind of traffic violation menace level Forecasting Methodology based on Bayesian network |
CN108152247B (en) * | 2017-11-22 | 2021-03-09 | 南京航空航天大学 | Multi-angle scattered light signal optimization selection-based collaborative reconstruction method for multiple radiation characteristic parameters of dispersion medium |
CN108198415B (en) * | 2017-12-28 | 2019-07-05 | 同济大学 | A kind of city expressway accident forecast method based on deep learning |
CN108510126A (en) * | 2018-04-11 | 2018-09-07 | 南京邮电大学 | A kind of Predictive Methods of Road Accidents based on PCA and BP neural network |
CN108665093B (en) * | 2018-04-19 | 2021-07-27 | 东南大学 | Deep learning-based expressway traffic accident severity prediction method |
CN109785633A (en) * | 2019-03-14 | 2019-05-21 | 百度在线网络技术(北京)有限公司 | Dangerous road conditions based reminding method, device, car-mounted terminal, server and medium |
CN109993223B (en) * | 2019-03-26 | 2023-05-16 | 南京道润交通科技有限公司 | Pavement use performance prediction method, storage medium and electronic equipment |
CN110458370A (en) * | 2019-08-19 | 2019-11-15 | 南京航空航天大学 | A kind of urban roads safety prediction technique and system based on open data |
CN113256969B (en) * | 2021-04-30 | 2022-08-16 | 山东金宇信息科技集团有限公司 | Traffic accident early warning method, device and medium for expressway |
CN113553391B (en) * | 2021-08-02 | 2024-05-10 | 中国港湾工程有限责任公司 | Port and abdomen land dividing method based on GIS technology |
CN113611114A (en) * | 2021-08-03 | 2021-11-05 | 长安大学 | Urban expressway traffic accident meteorological factor sensitivity analysis method |
CN114512368B (en) * | 2022-03-30 | 2023-05-23 | 广东电网有限责任公司 | Vacuum degree on-line monitoring method and device for high-voltage vacuum circuit breaker |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5798949A (en) * | 1995-01-13 | 1998-08-25 | Kaub; Alan Richard | Traffic safety prediction model |
CN101593424A (en) * | 2009-07-10 | 2009-12-02 | 浙江大学 | A kind of intelligent combination forecasting method for short-term traffic flow |
CN101694745A (en) * | 2009-06-16 | 2010-04-14 | 同济大学 | Safety detection method based on freeway geometry linear comprehensive technical indexes |
CN101465059B (en) * | 2008-12-31 | 2010-06-09 | 公安部交通管理科学研究所 | Early-warning system for inspecting and judging urban road traffic safety situation |
-
2010
- 2010-10-29 CN CN2010105258524A patent/CN101976501B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5798949A (en) * | 1995-01-13 | 1998-08-25 | Kaub; Alan Richard | Traffic safety prediction model |
CN101465059B (en) * | 2008-12-31 | 2010-06-09 | 公安部交通管理科学研究所 | Early-warning system for inspecting and judging urban road traffic safety situation |
CN101694745A (en) * | 2009-06-16 | 2010-04-14 | 同济大学 | Safety detection method based on freeway geometry linear comprehensive technical indexes |
CN101593424A (en) * | 2009-07-10 | 2009-12-02 | 浙江大学 | A kind of intelligent combination forecasting method for short-term traffic flow |
Non-Patent Citations (1)
Title |
---|
闫建华,宇仁德,汪洋.基于主成分分析的道路交通事故预测.《交通与安全》.2009,(第204期),86-89. * |
Also Published As
Publication number | Publication date |
---|---|
CN101976501A (en) | 2011-02-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101976501B (en) | Principal component analysis and neural network based port road safety prediction method | |
Boriboonsomsin et al. | Impacts of freeway high-occupancy vehicle lane configuration on vehicle emissions | |
CN103413263B (en) | Travel time index entropy traffic circulation evaluation method | |
Bauer et al. | Statistical Models of At-Grade Intersection Accidents. Addendum. | |
CN104809554A (en) | Decision-making supporting system and method facing road traffic accident emergency treatment | |
Hosseinlou et al. | Determining optimal speed limits in traffic networks | |
Sun et al. | Comparing the road safety of neighbourhood development patterns: traditional versus sustainable communities | |
CN103646533B (en) | Traffic accident modeling and control method based on sparse multi-output regression | |
CN103632535A (en) | Judgment method for section pedestrian crossing signal lamp arrangement | |
Wang et al. | Conditions for Setting Exclusive Pedestrian Phases at Two‐Phase Signalized Intersections considering Pedestrian‐Vehicle Interaction | |
Persaud et al. | Can crash modification factors be estimated from surrogate measures of safety? | |
Heidarian et al. | Location of Emergency Escape Ramps on Two‐Lane Rural Highways Based on the Fuzzy Multicriteria Decision‐Making Method | |
Cheng et al. | Research and application of urban road traffic safety risk evaluation based on regional disaster system theory | |
Mishra et al. | Safety evaluation of un-signalized intersection using hybrid approach involving empirical and simulation data sources | |
Smit et al. | Simulating fuel consumption and vehicle emissions in an Australian context | |
Abdelrahman | Safety Evaluation of Innovative Intersection Designs: Diverging Diamond Interchanges and Displaced Left-turn Intersections | |
Bassani et al. | Field Investigation on the Effects on Operating Speed Caused by trapezoidal humps | |
Minachi et al. | Development of accident prediction models for safety Evaluation of urban intersections | |
Cohen et al. | Assessing the impact of speed limit changes on urban motorways: A simulation study in Lille, France | |
Donnell et al. | 70 mph Study | |
Islam et al. | Full Bayesian mixed-effect intervention model for before–after speed data analysis | |
Bagheri Khalili et al. | Variable efficiency appraisal in freeway accidents using artificial neural networks—case study | |
Ran | Prediction of Frequencies of Truck-involved and Non-truck-involved Crashes on Roadway Segments in Ontario | |
Russo et al. | Analytical Assessment of Effective Maintenance Operations on At-Grade Unsignalized Intersections | |
Li et al. | A traffic accident predictive model based on neural networks algorithm and rough set theory |
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 | ||
CP01 | Change in the name or title of a patent holder |
Address after: 300051 No. 239, Yingkou Road, Heping District, Tianjin Patentee after: Tianjin municipal engineering design and Research Institute Co.,Ltd. Address before: 300051 No. 239, Yingkou Road, Heping District, Tianjin Patentee before: TIANJIN MUNICIPAL ENGINEERING DESIGN & Research Institute |
|
CP01 | Change in the name or title of a patent holder |