Research On Grey Prediction Modeling Method Of Traffic Flow Mechanics Based On Data Driven

Traffic flow mechanics is an interdisciplinary subject between the fields of fluid mechanics,applied mathematics,system engineering and traffic engineering.It aims to reveal various traffic flow phenomena by studying the characteristics of traffic flow parameters,establishing appropriate macro and micro models.The short-term traffic flow prediction is the core part of the intelligent transportation system.Through the prediction of traffic flow parameters,real-time,dynamic and accurate traffic flow information can be obtained,and real-time traffic conditions can be provided by travelers.According to the characteristics of traffic flow fluid,the grey prediction modeling method of short-term traffic flow mechanics is studied,focusing on the analysis of dynamic change law of traffic flow,density and speed,revealing the realtime characteristics of traffic system,providing reliable theory for traffic planning,control and optimization on the basis.According to the angle of data-driven mode information,short-term traffic flow can be divided into vector data flow form,matrix data flow form and tensor data flow form.Generally,traffic flow parameters are predicted based on time series vector data flow and section observation matrix data flow.However,The internal characteristics of traffic flow data can be mined by using tensor data flow to predict the change rule of short-term traffic flow parameters.At the same time,the observation scale of shortterm traffic flow is not more than 15 minutes,and the period of traffic guidance is generally 5 minutes,so there are only 12 data in an hour,which belongs to small sample data and has obvious gray system characteristics.Therefore,it is reasonable and feasible to use different models of traffic flow data and establish corresponding shortterm prediction grey model.The main work of this paper is as follows:Aiming at the properties of the traffic flow vector data flow form and the macro hydrodynamics continuous equation,the grey prediction model of macro traffic flow mechanics is established.Through the vehicle conservation equation,the traffic flow state law of the road section is analyzed,and the differential equation of the road section traffic flow is established based on the relationship among the three parameters of traffic flow,speed and density;the relevant grey prediction model is established by using the differential and differential information of the grey system;the parameter identification formula of the model is derived,and the properties of the model are studied.The validity of the new model is illustrated by the vector data flow of two cases,and the model is applied to the short-term traffic flow prediction of a road section in Wuhan.Furthermore,through the new model structure,the matrix least square algorithm is used to effectively calculate the vehicle inflow rate and the blocking traffic volume.Aiming at the traffic flow vector data flow form and the basic idea of micro car following theory,the inertia grey prediction model of micro traffic flow mechanics is established.According to the state laws of free flow and blocking flow of traffic flow,combined with the idea of vehicle following model,the micro vehicle following inertia grey prediction model is established by using the data mechanical characteristics of traffic flow.The important properties of parameter estimation,inertia coefficient,original data and simulation accuracy of the new model are studied.The validity of the model is verified by short-term traffic flow prediction using vector data flow at the same location for three consecutive days.At the same time,the new model is extended to establish three inertia discrete grey prediction models,and the three-phase traffic flow state is determined by six groups of vector data flow data prediction results.Aiming at the linearity of the data flow form of traffic flow tensor and the velocity density model of traffic flow mechanics parameters,the grey prediction model of traffic flow mechanics parameters is established.Based on the linearity of the three-parameter speed density model of traffic flow and the modeling mechanism of the grey prediction model,the parameter grey prediction model of traffic flow mechanics is established,and the parameter estimation and solution of the model are studied.At the same time,the multi-mode characteristics of traffic flow tensor data are studied.The highdimensional tensor multi-mode is used to represent traffic flow data,and the grey prediction model of tensor alternating least square algorithm is established.Two examples are given to illustrate the algorithm of tensor alternating least square,and the features of strong time correlation of approximate tensor data are studied.Finally,the grey prediction model of tensor alternate multiplication algorithm is applied to the short-term traffic flow prediction,which is better than the direct application of traffic flow data modeling.Three phase traffic flow state is determined by six groups of vector data flow data prediction results.Aiming at the strong correlation between the traffic flow tensor data flow form and the data time,the tensor multi-mode coupled grey prediction model is established.Based on the multi-mode characteristics of traffic data and the analysis of traffic flow data volatility,the tensor model of "week day period" is discussed,and the tensor coupled grey prediction model is established.Different traffic flow data are used to predict the traffic flow in the same period from three dimensions of tensor,and the grey NDGM(1,1)model is used to predict the weekly mode.The first mock exam model is built on rolling grey prediction RNDGM model.The BP neural network prediction method is used in the "time period mode".Then,the grey relational analysis method is used to get different weights for the three models,and finally the traffic flow in the same period is forecasted.The example analysis shows that the coupling model is better than the three single models.