The service management based on battery heterogeneity has become an increasingly important research problem in battery swapping technology. In this paper, with the method of bipartite matching, we first theoretically analyse the offline optimization problem of battery swapping service under battery heterogeneity. Nevertheless, the information of global view used in offline optimization solution cannot be known in advance during real-time operation. To address the disadvantage, an online framework comprising several sub-procedures is proposed for heterogeneous battery implementation. Firstly, by incorporating battery swapping station (BSS) local status such as charging and waiting queue of heterogeneous batteries, a charging slot allocation mechanism is designed. Utilizing the proposed allocation method, the charging priority is determined by the proportion of heterogeneous batteries demand, so as to guarantee charging fairness. Secondly, with the help of reservation information, the proposed allocation method can further be improved by predicting the future arrival distribution of heterogeneous types of electric vehicles. Thirdly, according to the service demand prediction based on long short-term memory neural network, joint optimization of BSS-selection and charging cost can be achieved by charging power adjustment. Simulation results indicate the desirable performance of proposed scheme in balancing the demands of multi-party participators.

Benefited from the development of intelligent transportation system and autonomous vehicle, the autonomous valet parking has been attracting attentions worldwide. Compared with short-range autonomous valet parking, long-range autonomous valet parking further alleviates users from parking during long-range traveling. Although autonomous vehicles can finish parking autonomously under long-range autonomous valet parking, they still need to estimate parking cost of each parking lot and decide which one to park. However, existing works on parking cost estimation do not consider the competition with uncertain parking demand, leading to inaccurate cost estimation. Besides, fixed parking prices intensify parking competition and cause local congestion at parking lot. Therefore, we propose a Long-Range Autonomous Valet Parking Considering Uncertain Demand and Parking Pricing. Based on the uncertain demand analysis, we predict the future occupancy of parking lot and adjust its price dynamically, so as to balance parking demand. Moreover, the accurate estimation of parking cost is achieved through mathematical analysis on uncertain demand, with vision that autonomous vehicles can select parking lot with lower parking cost. Simulation results show that compared with other schemes, our scheme can outperform on reducing parking cost and balancing parking demand.