A user-based relocation model for one-way electric carsharing system based on micro demand prediction and multi-objective optimization

Abstract

As a promising sustainable mode of transportation, electric carsharing can relieve the urban traffic pressure and environmental pollution problems, which has been promoted all over the world to reduce or replace fossil-fueled private cars. However, the operation of electric carsharing is faced with difficulties in profitability and management. Vehicle relocation can alleviate the imbalance between supply and demand, and compared with operator-based relocation, user-based relocation is more sustainable and cost-efficient. In this paper, we develop a user-based relocation model with the optimization objectives of profit maximization and using failure rate minimization. Two station pricing schemes of pure preferential and preferential combined with fine are proposed as the constraints. For the input of the relocation model, the back propagation neural network of the quantity demand and the distribution fitting model of the energy demand are constructed to predict user demand. The demand prediction model and relocation model are validated through the operation data of EVCARD in Shanghai. The results demonstrate that the user-based relocation can effectively reduce the using failure rate and has the potential to increase the net profit. Compared with the situation without vehicle relocation, the user-based relocation strategy can reduce using failure rate by 13.6% and increase net profit by 76.9% in the best situation. Through sensitivity analysis of key parameters, it is found that the station pricing of pure preferential can reduce the using failure rate, but the profitability of the electric carsharing system is weakened. The introduction of fine can effectively slow down the erosion of user incentive on the system profit and increase the profit.