Reliability Prediction Of Intelligent Unmanned Vehicle Lane Changing System Based On DTMC

Intelligent vehicle has gradually become a trend,in which unmanned vehicle is a research hotspot.Reliability is an important index of the intelligent unmanned vehicle.The premise of vigorously developing the intelligent unmanned vehicle is to ensure that the reliability of the intelligent unmanned vehicle reaches a higher level.Among the many driving behaviors of intelligent unmanned vehicle,lane changing behavior,as one of the two basic driving behaviors,is the most common one.At present,the domestic and foreign research on intelligent unmanned vehicle lane changing mainly focuses on the optimization of lane changing technology,and the research on the reliability of lane changing function is less.Lane changing is one of the most common causes of traffic accidents.Reliable lane changing behavior is very important for road traffic safety.It is necessary to predict the reliability of lane changing system of intelligent unmanned vehicle.In this paper,in the early stage of the development of the intelligent unmanned vehicle,the reliability of the lane changing system of the intelligent unmanned vehicle is predicted.Firstly,the lane changing system of intelligent unmanned vehicle based on long short term memory(LSTM)is modeled by unified modeling language(UML).LSTM neural network is the behavior decision module of intelligent unmanned vehicle lane changing system.Then,the fault propagation probability between the sensing module and LSTM module in the lane change system of the intelligent unmanned vehicle is calculated by using the error permeability,so as to obtain more accurate reliability information for reliability prediction.Next,the UML model is transformed into Markov decision process(MDP)model and discrete-time Markov chain(DTMC)model to predict the reliability of the intelligent unmanned vehicle lane changing system from different perspectives.The MDP model is an extension of the DTMC model.Finally,based on the first derivative function between the reliability of each module and the reliability of the system's lane changing function,the influence of the reliability change of each module on the reliability of the intelligent unmanned vehicle lane changing system is obtained.The main research work of this paper is as follows:(1)A reliability prediction method for intelligent unmanned vehicle lane changing system is proposed.DTMC model is usually used to predict the reliability of traditional software system without intelligent module.This paper uses DTMC model to predict the reliability of unmanned vehicle lane changing system with LSTM intelligent module after introducing error permeability.Because of the robustness of LSTM neural network,the output fault of perception module of intelligent unmanned vehicle lane change system does not necessarily penetrate into the output of LSTM module.The fault propagation probability between modules can be calculated through the error permeability,which makes the reliability prediction results more accurate.(2)The model transformation method which is more suitable for the reliability prediction of intelligent unmanned vehicle lane changing system is selected.According to the mapping rules and the idea of component transfer diagram,the UML model of the intelligent unmanned vehicle lane changing system is transformed into the DTMC model,and the reliability of the intelligent unmanned vehicle lane changing system is predicted according to the two DTMC models.The results show that the DTMC model based on mapping rules has a more detailed description of the module granularity,and is more suitable for the reliability prediction of intelligent unmanned vehicle lane changing system.(3)By means of curve fitting and first derivation,the influence degree of each module reliability change on system reliability is obtained.The more influential the modules are improved,the greater the gain of lane changing system of intelligent unmanned vehicle will be,which will provide reference for designers.