Research On Data-driven Cooperative Control Of Hybrid Bus Flow

With the proposal of China's strategic objectives of "carbon peak" and "carbon neutralization",China's transportation system is developing in the direction of "green transportation" and public transportation priority.As the most special part of public transportation,bus plays an essential supporting role in alleviating traffic pressure,reducing environmental pollution and improving public safety.When performing some special tasks,buses tend to run in platoons.With the development of distributed artificial intelligence,the research on vehicle cooperative control through multi-agent system is gradually rising.This paper aims to establish a mixed bus flow model based on data-driven modelfree adaptive control(MFAC)method under the framework of multi-agent systems(MASs),and distinguish the manual / automatic driving buses with mathematical models.A distributed model-free adaptive controller with time delay(IMFAC-TD)for solving the problem of large time delay and a distributed model free adaptive formation controller with data compensation(COM-MFAC)after sensor failure are proposed.The main contents of this paper are as follows:(1)The distributed consensus control problem for a class of discrete nonlinear MASs with time-delay based on MFAC method is considered.For this problem,a distributed data-driven consensus control strategy is proposed by using improved model-free adaptive control with a time delay.First,the unknown nonlinear multiagent time delay system is transformed into a virtual data model with time-varying parameters by using a partial form dynamic linearization method.Then,a controller with MFAC and a tracking differentiator(TD)are designed.The proposed method uses only the input/output data of neighbouring agents during the control process without mathematically modelling multiple agents.For the time delay system,the boundedness of the closed-loop tracking error is theoretically proven for both the tracking problem and the regulation problem.Furthermore,the communication topology is extended from the fixed topology to the switching topology,and the convergence of the controller is proved by the compression mapping.Numerical simulations demonstrate the effectiveness of the controller.(2)The formation control problem of a class of discrete nonlinear multi-agent systems with data drop based on MFAC method is considered.For this problem,a model-free adaptive formation control strategy with data compensation is proposed.First,on the basis of realizing consensus control,Bernoulli distribution model is added to introduce the data packet loss rate,and the communication topology is changed into a time-varying switching topology.Based on the parameter estimation value and the input value to the output estimation value at the previous time,the COM-MFAC formation strategy is obtained and the convergence is proved.Through numerical simulation,multi-agent formation control under different packet loss rates is realized,and finally the trajectory tracking and formation maintenance of agents are realized.(3)The problem of cooperative control of mixed bus flow based on IMFAC-TD and COM-MFAC control strategy is considered.For this problem,the simulation verification of related mixed bus flow is done.First,the mixed bus flow model is established by distinguishing the manual and automatic buses through different time delay and data packet loss rate.Under the condition of time delay,the longitudinal dynamics model of the bus is adopted.Under the condition of straight-line driving,the IMFAC-TD control strategy is used to take the starting torque per unit mass as the control input and the driving speed as the control output,so as to realize the consistency of the speed and position of the bus in the mixed queue.Under the condition of data packet loss,the bus kinematics model is used to describe the transverse process.Under the condition of curve driving,the COM-MFAC control strategy is used to control the front wheel angle and then the body heading angle.Finally,the trajectory tracking of the mixed bus flow under the fixed object is realized and the mixed bus flow keeps driving according to the expected formation.