| 研究実績の概要 |
In this FY, I have achieved three research result with the funding, which are:
A framework to integrate social media and authoritative data for disaster relief detection and distribution optimization. In this paper, we propose an interdisciplinary approach to (natural) disaster relief management. Our framework combines dynamic and static databases, which consist of social media and authoritative data of an afflicted region, respectively, to model rescue demand during a disaster situation. Using Global Particle Swarm Optimization and Mixed-Integer Linear Programming, we then determine the optimal amount and locations of temporal rescue centers.
Global optimization for multi-stage construction of rescue units in disaster response. Considering the real-time location data of trapped people, this paper develops a Mixed Integer Non-linear Programming (MINLP) model to find the highest efficient rescue plan To solve the model accurately and efficiently, a bi-level decomposition (BLD) algorithm is presented to iteratively solve a discretized Mixed Integer Linear Programming (MILP) model and its nonconvex Non-linear Programming (NLP) model until a converged solution is obtained.
Robust Optimization for Emergency Scheduling of Oil Products After Disaster.By taking the shortest emergency scheduling time and the lowest cost as the objective functions, this paper develops a multi-objective mixed integer linear programming (MILP) model accounting for multiple supply sites, multiple disaster sites, multiple transportation modes, uncertain demand and effectiveness of scheduling paths.
|
| 今後の研究の推進方策 |
The previous medical accessibility studies mainly focus on spatial accessibility without considering temporal variation in population distribution which is significant for evaluating access to emergency medical service (EMS). In the next stage, we plan to focus on spatio-temporal accessibility modeling. We plan to apply our method to the greater Tokyo area for a large volume of GPS dataset with millions of users and compare the accessibility difference over space and time. The expected method can illustrate the temporal difference and is suitable for measuring the spatio-temporal accessibility to EMS, thus can guide the hospital location selection and urban planning.
Additionally, we also plan to develop an online Ex-ante online risk assessment model based on multimedia data (eg videos data from surveillance cameras) and deep learning.
|
