Research Of Rollover Prediction Algorithm For Heavy Goods Vehicles Based On AR-CHMM Model

The heavy goods vehicle has its specific characteristics of higher gravity-center,large loads and large volume, which can affect the dynamic roll stability of the vehicle.Especially when the vehicle needs to change lanes or emergency steering, can easilycause rollover. In this paper, based on analysis, summary and learning the relevantliterature for heavy goods vehicles states identification and prediction of the domesticand foreign, and combining with the Project of National Natural Science Fund“hierarchical hidden Markov model for heavy truck vehicle rollover state identificationand prediction algorithm”, the goal build the accuracy, real-time,off-line stateidentification of heavy goods vehicles， since the establishment ofAutoregressive-Continuous Hidden Markov model, and the model has been off-linetraining. On the basis of these, the method for vehicle state identification, proposed inthis paper, is verified on off-line with the Matlab/Simulink, HMM toolbox andTrucksim software. The results show that the proposed vehicle state identificationmethod has higher precision, and the real-time characteristic.The main research work is summarized as follows：(1)Heavy goods vehicle test data acquisition and pre-processingUsing Trucksim software for the heavy goods vehicle was selected multipleloading configuration (no load,500KG,1000KG), thus collected to the selected workingcase(fishhook steering, double lane change) corresponding to the experimental data, andsegment processing and the critical value of the vehicle motion state.(2) Autoregressive-Continuous Hidden Markov model for the state identification ofheavy goods vehicleBased on Hidden Markov theory, by combining the HMM toolbox to establishedthe AR-CHMM model and off-line training. Then, the off-line identification for vehiclestate is studied by using the Matlab/Simulink, HMM toolbox and Trucksim softwareseamlessly integrates development.The results show that the proposed vehicle state identification method has higherprecision, and the real-time characteristic(3)Heavy goods vehicle motion state parameters prediction algorithmThrough the application of knowledge from the auto-regressive prediction to rollrate of the vehicle, steering angle, speed, lateral acceleration was predicted. Autoregressive prediction.