Research On Real-time Electric Taxi Simulation Based On Multi-agent Simulation Technique

Nowadays with the enormous development of new energy technology, the electric vehicles, which holds the advantages of high energy efficiency, low noise and minimum impact to the environment, become a hot topic among different utilities. Compared to other kinds of vehicles, taxis perform the highest travelling density and behavioral complexity together with the relatively stronger ability to participate in social interaction. As to indicate the real-time behavior of electric taxis more accurately, the conventional Up-down Modeling and single-agent Monte-Carlo Modeling inevitably have many limitations.Based on the analysis of existing literatures on agent-based modeling of electric vehicles, this paper firstly illustrate the application of Multi-agent Simulation technique. According to the basis theory and FIPA Specifications of MAS, the distributed electric-taxi simulation platform structure is established with real time and location. Deploying the application of JADE developing toolkit provided from Tilab, this paper built the agents’ model as well as interaction regulation in detail.In addition, in order to reflect the highly complicated behavioral character of electric taxis, Q-Learning Model is used with definition of construction of state and action domains and the setting up of Reward/Penalty function to evaluate the rationality of the action in relative state. In the following simulation test, this papers verifies the convergence and correctness of the QL model.Finally, provided with the confirmation of the physical parameter and initial conditions of the following simulation, this paper establishes different simulation cases on different charging rate and exchanging SOC limitation circumstances in order to analyze the current solution to establish the bottleneck of the low endurance of batteries. This paper also comes up with different scheme of new-constructed quick charging stations when facing operation problems of current settlement and provide suggestions.