An Evolutionary Algorithm for an Agent-Based Fleet Simulation Focused on Electric Vehicles

摘要: This paper presents a multi-agent simulation for fleet applications with primary focus on electric vehicles. The overall system design is characterized by distributed and decentralized structure. Basic idea to use an evolutionary algorithm in combination Monte Carlo method optimize behavioral patterns of agents. Within iterative process, agent plans are executed, scored modified — if necessary. In doing so, set activity schedules created that compatible given constraints, such as range restrictions cars or space limitations at charging stations. so agents compete each other limited resources within single iteration step. proposed model makes map-based approach takes time variant traffic conditions into account. With help speed profiles longitudinal vehicle dynamic model, travel energy consumption calculations carried out. As the infrastructure configuration input parameters it possible perform case studies several electrification scenarios. total two different scenarios explored. First validated measured data. Second suitability taxi assessed. first study, daily mileage reflected average variation 7.8 % waiting times deviation 0.05 second scenario provides transparent indications choosing both appropriate concept configuration.