DATA SYSTEM RESEARCH GROUP
In the Great East Japan Earthquake, Tokyo was hit by a tremor with an intensity of 5 or higher on the Japanese seismic scale, causing railroad lines to suspend service, and the long-feared difficulty for people to return home became a reality. For those who have difficulty returning home, a system has been well-maintained to avoid going home en masse and to support people walking home afterward. However, if the railroad service is suspended for a prolonged period, commuting to work and school and returning home may be disrupted. Here, we present a fundamental study on building a route network for alternative transportation by water buses and buses.
For the route network construction problem, we applied a multi-agent system (MAS) consisting of multiple agents acting autonomously with one agent per route. Each route agent ("route Ag") evolves by extending or reducing routes, changing routes and number of buses on its own, and competing with other route Ags for users of the route network. The answer, the route network, consists of a set of routes Ag that have escaped elimination in this evolutionary process. The proposer applied this method to the benchmarking problem, outputting the best route network (*1), and constructed a water bus route network on a river flowing through Tokyo, as shown in Figure 1.
Figure 1 Applications of river and water bus route network
We also developed a system that outputs the distribution of the number of people having difficulty returning home by inputting the time of the disaster and the disrupted sections and routes of the railroad network. Figure 2 shows an example of the results of the analysis. In Figure 2, darker colors indicate more people having difficulty returning home.
Figure 2 Distribution of origins and destinations of those who have difficulty returning home
Furthermore, we have improved the developed MAS to allow the build-up of a route network under mixed conditions between the two types of transportation. As an application of this, we generated a transportation network with a mix of buses and water buses, as shown in Figure 3, using the travel demand input in Figure 2.
Figure 3: Example of analysis of a large-scale route network in the metropolitan area with different modes (bus and water bus) (each route vertically stacked in a different color)
Here, we have outlined the methodology we have developed to address the problem of transporting people who have difficulty returning home by building a route network. In the Tokyo cosmopolitan area, there is concern about a large-scale disaster brought about by an earthquake directly below the Tokyo metropolitan area. We hope that the tools we have developed will provide helpful information for disaster prevention measures.