Safety Assistance Strategy Research Based On Driving Condition Prediction

With the development of social life,the demand for nighttime traffic is increasingly growing.However,the annual number of road accidents at night has also largely increased.In the recent years,artificial intelligence and vehicle and road collaboration technology are rapidly developing.People give more focus on the study of traffic securty control technology combining those leading technologies.To solve the problem of crossroad vehicle collisions at night in the blind area,the thesis presents that under the environment of vehicle and road collaboration technology,the turning intention and the travel speed in the crossing area can be predicted by analyzing the driving conditions,and that auxiliary security strategy has been established so as to achieve auxiliary dynamic lightning and risk early-warning.The thesis mainly focuses on the following several aspects:(1)Driving turning intention prediction.The predicted turning intention is the base of dynamic auxiliary lightning.The turning intention prediction model based on Long Short-Term Memory(LSTM)is established.By analyzing the training results,Bi LSTM and Attention system are established to improve the LSTM model.In this way,the accuracy and discrimination are increased and.The model structure and training process are introduced in detail and the performances of several improved algorithms are compared.(2)Vehicle speed prediction.The predicted driving speed is the base of risk warning of collision.Firstly,the simple grey model,GM(1,1),is established.A makeshift grey prediction model based on the equal-dimension is built because of the low accuracy of GM(1.1)in the long-term prediction.And its accuracy is verified.(3)Auxiliary security strategy.Dynamic turning fuzzy control model is established combining with the driving turning intention and junction illuminance.The speed-replacement model is established combining with speed prediction results.Taking fuzzy control and speed-displacement model as a basis,Auxiliary dynamic lightning and security strategy combining with collision risk evaluation are achieved to evaluate illuminance and collision risk.(4)Smart roadside system based on vehicle and road collaboration system.The system hardware structure and software are designed on the basis of the system demand analysis and crossroad auxiliary security strategy.Simulation and control results have verified auxiliary security strategy.