Research On Dangerous State Identification And Prediction Of Tractor-semitrailer Based On ARIMA-HMM

Along with the national highway traffic volume and heavy cargo rising car ownership,in complex mountainous roads,tractor-semitrailer driving safety is particularly important,its overall vehicle driving state as a multi-factor coupling needs analysis and modeling,the dynamic process of vehicle recognition and prediction of the motion state can determine the happening of the risk in future time series,And then remind the driver to make a correction.Based on this,this paper used HMM(Hidden Markov Model)and ARIMA(Auto Regressive Moving Average Model)theories in machine learning.The identification and prediction of vehicle running state of tractorsemitrailer is studied,and its identification and prediction model is established,so as to achieve the purpose of warning of dangerous state trend and provide basis for reducing the occurrence of road traffic accidents.In this paper,the tractor-semitrailer simulation model was established by Trucksim simulation software,and the vehicle model,driver model and road model were respectively established by human-vehicle-road co-simulation.Based on the idea of hierarchical modeling,TSIP(Tractor-Semitrailer stepwise dangerous state identification and prediction model)was established in Matlab.The whole TSIP model is composed of prediction model ARIMA,low level motion identification model GM-HMM and high level vehicle motion state DHMM.The model is introduced into Simulink / trucksim co simulation system for simulation research.K-means algorithm is used to cluster the threshold of each sub motion data.The preprocessed training database is used to train the TSIP model offline.The trained TSIP model can be verified by a variety of composite conditions.The model can be used to predict the subsequent dangerous state of the tractorsemitrailer train in the mountainous area of the paper 3s in advance and realize the early warning.TSIP model can not only predict the state of the vehicle within a certain time step parameters,but also to the motion of the vehicle can be carried out to identify and forecast,and double prediction results in a model structure,improve the reliability of prediction,the model of state risk early warning system to a certain extent,reduce the occurrence of the accident.The idea of simulation analysis and modeling adopted in this paper can be applied to the design of roll-over warning system of other heavy vehicles,which has certain theoretical and practical significance for improving the driving safety of vehicles on mountain roads.