Research On Short-Term Speed Prediction And Traffic Safety Evaluation Of Urban Expressway In Rainy Days

As an important part of urban road infrastructure,urban expressway has the characteristics of fast running speed,large service traffic and large capacity,and is responsible for the transit traffic between internal areas of the city.Therefore,sudden traffic accidents on the urban expressway will lead to a decline in capacity and an increase in delays,which will affect the regional road network operating environment and traffic safety.There are many objective causes of road traffic accidents.Unfavorable weather conditions（such as rain,fog,snow,wind,etc.）are the main objective causes of urban expressway traffic accidents.Rainy days are the most common adverse weather,which not only directly affects roads coefficient of friction,but also indirectly affect the driver’s physical and psychological qualities.Therefore,the evaluation of traffic safety on rainy days is of great significance to traffic management and public safety.Firstly,this article analyzes the mechanism of the impact of rainfall on visibility and adhesion coefficient,and studies the relationship between vehicle speed and driver,vehicle,and road alignment.At the same time,there is also a close relationship between vehicle speed and traffic accidents.As vehicle speed increases,the number of road traffic accidents increases.The faster the speed,the more serious the consequences of the accident.Citing the statistic t as the test statistic,I selected the rainy days and no rainy days in Guangzhou from August 1,2016 to September 31,2016 as the research objects,and concluded that rainfall has a significant effect on the speed.However,the Spearman correlation coefficient and Pearson correlation coefficient are not significant,indicating that with the increase of rainfall,the speed drop is not necessarily greater.The speed deviation standard deviation coefficient research results show that different rainfall and speed deviation standard deviation coefficients exist under different traffic conditions.The difference is that from the overall trend,the variation law of vehicle speed dispersion is:heavy rain>storm>heavy storm.Secondly,a short-term vehicle speed prediction model based on a convolutional neural network is constructed.The vehicle speed has spatiotemporal characteristics and can be represented by a matrix as an input to the convolutional neural network.Compared with other models,the advantages of the convolutional neural network model are comprehensive consideration of time and space information of vehicle speed.Take the Guangyuan Expressway in Guangzhou City for a total of 60 days of data from August 1,2016 to September 30,2016 as an example,of which 27 days are rainy days.The data of the 23 days before is selected as training data,September 27 and September 28 is the test data,and the data on the two days of September 29 and September 30 is the prediction data.Convolutional neural network prediction is performed using Python programming.The prediction results are compared with the BP（Back Propagation）neural network optimized by Genetic Algorithm（GA）and Support Vector Regression（SVR）.The Mean Square Error（MSE）,Mean Absolute Error（MAE）and Absolute Percentage Error（MAPE）of the convolutional neural network are all smaller than GA-BP and SVR.It is concluded that the convolutional neural network has a good prediction effect.Thirdly,combining the close relationship between speed and driver,vehicle and road alignment,this paper proposes an expressway traffic safety evaluation based on speed characteristics.With reference to the selection method of speed indicators,the factor contribution degree is determined by the method of gray correlation and the contribution is selected.The large speed index is used as an evaluation index for traffic safety.Finally,the standard deviation of the speed deviation coefficients X1,V₈₅and V₁₅,the speed difference X2,the speed reduction coefficient X3,the difference between the running speed and the speed limit value X4 are selected as speed indicators,and the weather indicators select visibility X5 and rainfall X6 to construct rainy expressway traffic Safety evaluation index system.Finally,a fuzzy comprehensive evaluation method based on a cloud model will be used to evaluate urban expressway traffic safety.The cloud model can combine ambiguity and randomness to realize the conversion between quantitative and qualitative,which improves the authenticity of the evaluation results and reliability.Taking the L8 section of Guangzhou Guangyuan Expressway as an example on September 29,2016 from 7:00 to 9:00,the traffic safety evaluation was carried out by comparing and analyzing the evaluation comprehensive cloud and the evaluation standard cloud,and found the state of the road section.This level of rainfall intensity is relatively small,the visibility is slightly reduced.It has little effect on the line-of-sight of the driving,the driving safety risk is slight,and the vehicle can basically run at normal speed.