Effect of Genetic Parameters in Tour Scheduling and Recommender Services for Electric Vehicles

摘要

This paper assesses the performance of a tour scheduling and recommender service for electric vehicles, aiming at verifying its effectiveness and practicality as a real-life application. The tour service, targeting at electric vehicles suffering from short driving range, generates a time-efficient tour and charging schedule. It combines two computing models, one for user-specified essential tour spots as the traveling salesman problem and the other for service-recommended optional spots as the orienteering problem. As it is designed based on genetic algorithms, this paper intensively measures the effect of the population size and the number of iterations to waiting time, tour length, and the number of visitable spots included in the final schedule. The experiment result, obtained through a prototype implementations, shows that our scheme can stably find an efficient tour schedule having a converged fitness value both on average and overloaded set of user selection.