Study On The Influencing Factors And Prediction Model Of Speed Limit Of Ordinary Highway

Ordinary highway is an important part of the road network in our country.Reasonable speed limit is conducive to improving the efficiency and safety of ordinary highway.Through the investigation of 76 road sections of 22 ordinary highways in Jinan,this paper obtains the road grade,number of lanes,whether there are guardrails,whether there are non motor vehicle lanes,width of single lane,traffic volume,speed and other data of different road sections,aiming at ordinary highway The influencing factors of various highway operation speed are analyzed systematically.Aiming at the operation efficiency and safety,the prediction model of the operation speed of ordinary highway is established,the prediction system of the speed limit value of ordinary highway is established,and the speed management system of ordinary highway is constructed.First of all,the samples of running speed and influencing factors are processed,and the time and space distribution characteristics of running speed are analyzed.It is found that the running speed of different sections of the same road in different time periods and different sections of the same road are quite different,but there are similar change rules,which are the running speed is greatly reduced in the early peak and late peak periods.By constructing the scatter diagram between the influencing factors and the running speed,it is found that the road grade,the number of lanes,the road alignment,the width of single lane,whether there are non motor vehicle lanes,the road condition,the lateral interference source,the nature of the roadside land,whether there are guardrails,the weather,the traffic volume,the saturation and the proportion of large vehicles all have an impact on the running speed.The linear correlation between various influencing factors and operation speed is analyzed.It is found that the linear correlation between various factors and operation speed is very low.Therefore,it is not suitable to use multiple linear regression.It is necessary to find a multiple nonlinear model suitable for classification variable modeling.Taking all kinds of influencing factors of running speed as independent variables,and taking 10 km / h and 5km / h as dependent variables respectively,this paper analyzes the correlation between the variables,and adopts forward stepwise regression method to bring the independent variables into the model,in which the design speed,width of single lane,road alignment,presence or absence of non motorized lane,lateral interference and other factors in the independent variables are successfully brought into the model,The operation speed prediction model I and model II of multivariate ordered logit are established respectively.The significance of each variable in the two models is less than 0.05,and the significance of parallelism test is greater than 0.05.The model passes the test.The regression results are analyzed,and the parameter estimates are introduced into the model to obtain the equation form of the multivariate ordered logit operation speed prediction model.It is found that the increase of design speed and single lane width has a positive impact on the increase of running speed,while the road alignment,non motorized lane and lateral interference factors have a negative impact on the running speed.The multiple ordered logit model only obtains the influence direction and influence magnitude of various factors on the running speed.Without the measured value of running speed,it can not directly get the predicted value of running speed through the model.In order to get the running speed value directly through various influencing factors,the method of deep learning can also be considered to predict the running speed.Firstly,the structure model of BP neural network is built,and the parameters such as number of hidden layers,number of neurons,excitation function,learning rate and iterative steps are calibrated.The neural network is trained with the validation set data,and the training accuracy of 0.85 and the training loss of 0.22 are obtained.The test set data is used to test the existing neural network structure,and the accuracy of the test set is 0.83.In order to facilitate the daily application and overcome the disadvantage of inconvenient use of the model,a speed limit prediction system of ordinary highway is built with machine learning as the key technology.By inputting the processed data of various influencing factors into the speed limit prediction system of ordinary highway,the speed limit value and execution speed can be output.A speed management system with feedback mechanism and real-time correction of speed limit value is constructed based on the speed limit value prediction system of ordinary highway.This paper puts forward the method of setting speed limit signs and speed limit auxiliary strategies for ordinary highways,and puts forward the method of speed limit in different periods for school areas and sections with assembly.This paper uses the data of a class highway in Jinan to apply the speed limit case,and successfully obtains the speed limit value and law enforcement speed of each road section on the class highway.