Optimization of Electricity Delivery Routes for an EV Considering Household Electricity Demand Predicted Using ELM

Abstract

The frequency of disasters has been increasing in recent years. A backup power supply method using an EV could be considered in preparation for power outages caused by disasters. We proposed a route planning algorithm with the estimated electricity demand of each family considered. The method proposed in this paper first estimates each household's future electricity demand using an extreme learning machine based on past household electricity consumption data. Afterward, we use the Dijkstra method to determine the driving route for one EV when delivering electric power to multiple households with the estimation results considered. Optimal routes for the EV are calculated for multiple patterns of electricity delivery situations using actual household electricity consumption data. It was confirmed that the proposed method can deliver electricity more efficiently in most cases compared with the conventional method which determine the driving route of the EV without estimating the electricity demand of each family.