Stable Carpooling Matching Model Based On Evolutionary Algorithm For Commuting Private Cars

With rapid economic development in recent years,the number of private cars has increased year by year.Private car commuting has become one of the most important way of traveling.A large volume of commuting private cars and the low occupancy rate causes serious traffic congestion and a huge waste of resources,especially during morning and evening rush hours.Traffic congestion during morning and evening rush hours,waste of resources,and a large amount of exhaust pollution can be alleviated by carpooling.Electronic Registration Identification of the motor vehicle（ERI）is a vehicle identification and tracking technology that applies radio frequency identification（RFID）,which can effectively collect travel data of all motor vehicles.The research on large-scale private vehicles’ERI data provides a driving force for research on carpooling of commuting private vehicles.Carpooling is divided into real-time dynamic carpooling and long-term stable carpooling.Existing vehicle matching research mainly focuses on real-time dynamic carpooling which is based on passenger request mode.However,this thesis aims at the demand of commuters and based on the ERI data of commuting private cars in Chongqing for the long-term stable carpooling.This thesis studies the carpooling matching model and its optimization algorithm.The main work is as follows:（1）A carpooling matching approach based on improved ant colony algorithm is proposed,which is aiming at the carpooling matching model of commuter private vehicles with similar origins（ACMMCSC,Ant Colony Based Matching Approach for Commuter Stable Carpooling）.The objective of this model is minimizing carpool miles and traffic flow.The ACMMCSC is divided into two steps:commuting private car grouping and carpool matching.The experimental verification is carried out based on the ERI data of two groups of commuter private cars in Chongqing.Firstly,the best parameter combination is found,weighted values of pheromone visibilityα=1.5,β=2,and evaporation rateρ=0.3;Secondly,the ACMMCSC algorithm is compared with GA and SA,and the performance is the best;Finally,the visualization of carpool route planning is shown by an example and all the benefits of carpooling and the overall beneficial effect of carpooling is analyzed.The results show that carpooling reduces the traffic flow by 61.78%and the mileage by 55.51%.（2）A carpooling matching approach based on improved particle swarm algorithm is proposed,which is aiming at the carpooling matching model of commuter private vehicles with similar trajectories（PSMMCSC,Particle Swarm Based Matching Approach for Commuter Stable Carpooling）.The objective of this model is minimizing carpool miles and traffic flow.In the proposed PSMMCSC,the finess calculation rules are redesigned to make it more suitable for the model.The experimental verification is carried out based on the ERI data of two groups of commuter private cars in Chongqing.Firstly,the best parameter combination is found,acceleration constant C₁=0.5,C₂=2,inertia factorω=1;Secondly,the PSMMCSC algorithm is compared with GA,SA and HC,and the performance is the best;Finally,the visualization of carpool route planning is shown by an example and all the benefits of carpooling and the overall beneficial effect of carpooling is analyzed.The results show that carpooling reduces the traffic flow by 70.55%and the mileage by 64.61%.