| Intelligent transport system(ITS)is an important component of smart cities(SCs),and shared electric autonomous vehicles(SEAVs)are envisioned to be the main form of transportation because SEAVs can save energy,protect the environment,improve service efficiency and increase economic benefits.SEAVs are envisaged to evolve into a service rather than a product in the future,and the machine learning(ML)technology that achieved technological breakthroughs continuously improves the big data mining and processing,artificial intelligence(AI),provides technical support for the unified intelligent management of SEAVs in cities.The popularization and deployment of SEAVs can significantly change the transportation system in future SCs.Based on the big data analysis,this thesis excavates the characteristics of urban traffic flow,studies new theories and technologies about the reasonable and dynamic allocation of urban traffic flow resources.For the problems of SEAVs travel,parking and charging scheduling and infrastructure construction,this thesis provides an intelligent management system.In this thesis,the main innovations and work results of the intelligent management of autonomous vehicles traffic flow in smart cities based on big data analysis are as follows:1)In view of the energy storage capacity of SEAVs vehicles and the energy constraints of smart grid electric loads,this thesis proposes an intelligent dispatching scheme based on passenger demands and the deployment of charging piles to jointly dispatch the driving and charging activities of SEAVs in the entire region.The solution uses the proposed constrained vehicle dispatching(CVD)algorithm to intelligently dispatch the traffic flow of SEAVs in the city to achieve reasonable matching of passenger demands and complete energy replenishment behavior under energy constraints.The scheduling process minimizes the total cruise energy consumption of SEAVs while ensuring the quality of service(QoS).This thesis is based on the extended network calculus(NC)to simulate the traffic flow and simulate the real taxi data in Beijing.The results show that the solution can achieve substantial energy saving,and the order completion rate and charging pile utilization rate are increased by 52%and 53.8%respectively.2)Aiming at the difficulty of urban vehicle parking and the shortage of parking spaces,this thesis proposes a dynamic road parking management plan for vehicles based on demand analysis.The plan puts forward the concept of using urban roads as time-sensitive parking lots.According to the difference between road traffic density of different times,the road parking lot implementation strategy is formulated to solve the problem of shortage of urban parking spaces.In addition,the SEAVs parking strategy is modeled through the Markov decision process(MDP).This parking strategy takes the optimization of the profit of each vehicle as the decision-making goal under the condition of energy constraints.Simulation verification shows that the proposed urban road dynamic parking management strategy can greatly alleviate the pressure of urban parking without changing the overall urban planning and occupying new urban land.Intelligent parking scheduling can increase the average daily income by about 17.5%at the same time Energy consumption is reduced by 10.81%.3)This thesis provides a solution for optimizing the deployment of SEAVs in cities and the deployment of centralized placement points-a demands hotspot clustering algorithm based on passenger boarding and disembarking data information mining.First of all,according to the location planning and deployment of the passenger waiting to board the station,an improved K-means clustering algorithm(IKCA)with the clustering center point for conditional judgment is proposed,which combines the passenger and the taxi station.In order to minimize the distance traveled by passengers,the optimal deployment of the waiting station can be obtained.The simulation results show that the coverage rate of the scheme for passengers can reach 91.05%,and the average walking distance between passengers to the boarding station is 276.8m.Compared with the model with stations set at a fixed distance,the performance has been significantly improved.Next,for the deployment of SEAVs centralized placement point,based on the vehicle’s air patrol period and the drop off location point,Density-Based Spatial Clustering of Applications with Noise(DBSCAN)is used to obtain a reasonable deployment. |
