Research On Prediction Method Of Passenger Car Operating Speed Based On Highway Alignment 3D Geometric Features

The road alignment is an important factor that affects driving safety.The main theoretical basis for the ongoing road safety evaluation is based on the continuity of the road alignment.The difference in the speed of the adjacent road sections on the design quality of road alignment and the safety of the road safety analysis and evaluation.As the core of highway safety evaluation,the accuracy of the operating speed prediction model directly determines the reliability of the safety evaluation results,and the existing operating speed prediction model is mainly based on the 2D design indicators and passes through the road section.Therefore,return to the essence of highway 3D space curve,and carry out the operation speed prediction model of highway 3D space curve,to provide a safety analysis evaluation method and evaluation system conforming to highway 3D space characteristics for highway safety assessment.It is great significance to highway safety evaluation.This study is based on the essence of the 3D space curve of the highway alignment,breaking through the limitation of the current 2D angle of"split-and-piece"type to predict the operation speed of the highway.The on-site natural driving test collects the running speed of the vehicle,and finally uses a multi-layer neural network to build a new operation speed prediction model.The main research contents are as follows:Aiming at the calculation of the 3D geometric feature quantity,the differential geometry method is used to divide the alignment into twelve mutually independent alignment combinations according to the combination of the 2D indicators,and the 3D geometric characteristic calculation model of each alignment shape combination is established.The main idea is to first push the wire alignment parametric equation in sections,and then establish a 3D feature calculation model according to the parametric equation and its derivative.This method realizes the fast and accurate calculation from the traditional 2D design index of the highway to the 3D geometric characteristics,and can calculate the 3D geometric characteristics corresponding to any pile,including the 3D curvature-type characteristicsκ,κ_x,κ_y,κ_z;the 3D torsion-type characteristicsτ,τ_x,τ_y,τ_z;unit tangent vector classes:θ_x,θ_y,θ_z;primary normal vector classesβ_x,β_y,β_z;secondary normal vector classesγ_x,γ_y,γ_z.Aiming at the problem that multiple indicators jointly predict the operating speed,a operating speed prediction model based on multi-layer neural network is established.In this paper,the concept of"feeling range"is used to represent the independent variable,that is,the alignment 3D geometric features within a certain feeling range around a certain station are used as the independent variable,and the measured speed of the station is used as the dependent variable,and the multi-layer neural network is used to fit the relationship between the 3D geometric features of the highway alignment and the operation speed in the training set,and the generalization ability of the model in the test set.The prediction results of the model show that the optimal range of alignment combined features is"200m for the front feel+250m for the back feel".When comparing this model with the existing operation speed prediction model,it is found that the test error of the multi-layer neural network model is one-sixtieth of the existing model,and the change trend of the model’s predicted curve is more in line with the measured curve.This study is based on the nature of the 3D space curve of the highway alignment,which provides a more accurate operating speed prediction model for highway safety evaluation,and is of great significance for improving the reliability of highway safety evaluation.