Search

The increasing concern over global warming has led to the rapid development of the electric vehicle industry. Electric vehicles (EVs) have the potential to reduce the greenhouse effect and facilitate more efficient use of energy resources. In this paper, we study several EV route planning problems that take into consideration possible battery charging or swapping operations. Given a road network, the objective is to determine the shortest (travel time) route that a vehicle with a given battery capacity can take to travel between a pair of vertices or to visit a set of vertices with several stops, if necessary, at battery switch stations. We present polynomial time algorithms for the EV shortest travel time path problem and the fixed tour EV touring problem , where the fixed tour problem requires visiting a set of vertices in a given order. Based on the result, we also propose constant factor approximation algorithms for the EV touring problem , which is a generalization of the traveling salesman problem. [ABSTRACT FROM AUTHOR]

• A novel framework for large-scale EV charging energy predictions is introduced. • The MAPE retains at 2.5–3.8% with a testing/training ratio varying from 0.1 to 1000. • MICs and PCCs are combined for feature analyses of charging energy predictions. • Multiple data sources are coupled by linking the timestamps and location data. Large-scale and high-precision predictions of the charging energy required for electric vehicles (EVs) are essential to ensure the safety of EVs and provide reliable inputs for grid-load calculations. However, the complex and dynamic operating conditions of EVs make it challenging to accurately predict the charging energy under real-world conditions, especially for large-scale EV utilization. In this study, a novel data-driven framework for large-scale charging energy predictions is developed by individually controlling the strongly linear and weakly nonlinear contributions. The proposed framework concurrently addresses the overfitting of nonlinear networks using a low proportion of training data as well as the poorly descriptive ability of linear networks under complex environments. For each charging session, the charging energy predictions appropriately account for important factors such as the variations in the state of charge (SOC) of the battery, ambient temperatures, charging rates, and total driving distances. The results suggest that, compared with existing prediction models (such as the random forest, xgboost, and neural network), the proposed framework persists with evidently higher accuracy and stability over a wide range of the ratio between the number of EVs used for testing and training; its mean absolute percentage error (MAPE) is maintained at 2.5–3.8% when the ratio ranges from 0.1 to 1000. The proposed models can be further utilized for cloud-based battery diagnoses and large-scale forecasting of the energy demands of EVs. [ABSTRACT FROM AUTHOR]

Highlights • Joint location and capacity design problem for charging stations. • A novel expanded network structure. • Queuing time approximation formula. • Customized neighborhood search heuristics. Abstract This study is devoted to designing locations and capacities of charging stations for supporting long-distance travel by electric vehicles (EVs). We first establish an expanded network structure to model the set of valid charging strategies for EV drivers, and then a variational inequality (VI) is formulated to capture the equilibrated route-choice and charging behaviors of EVs by incorporating an approximated queuing time function for a capacitated charging facility. Next, we formulate the problem of designing the locations and capacities of charging facilities under a fixed budget constraint and solve the optimization problem with a customized neighborhood search strategy. A lower bound for the system cost is also developed to evaluate the qualities of solutions acquired using our proposed heuristic. Numerical examples with a toy network and a highway network extracted from the Yangtze River Delta are used to show the effectiveness of the proposed methodology, and we observe that our strategy can solve a large-scale problem within an optimality gap of less than 5%. [ABSTRACT FROM AUTHOR]