Disaster Prevention

Interdisciplinary Research for Disaster Prevention

Brief introduction

In 2016, MSIP Lab launched Autonomous River Control Engineering (ARCE), an interdisciplinary fusion research project aimed at developing new technologies to control rivers autonomously. Taking advantage of our research system of interdisciplinary fusion, we aim to conduct cutting-edge research that incorporates the knowledge of experts in various fields.

The following introduces three subprojects in MSIPLab

  • Research on elucidating the mechanisms of and predicting riverbed and ground deformation
  • Research on cyber-physical system (CPS) for controlling river meanders
  • Development of a water level distribution prediction model using river graph structure

This project  is supported by JSPS KAKENHI .

Research on elucidating the mechanisms of and predicting riverbed and ground deformation

Advances in measurement technology have made it possible to acquire a vast array of diverse signals. At the same time, there is a growing demand to understand, predict, and control complex dynamic systems. In order to meet these demands, models that can accurately represent the physical phenomena of interest are needed. In this research, we are studying the derivation of time evolution equations for data-driven riverbed and ground dynamics using filter bank theory.

Publications

International Confs.

  • Yuhei Kaneko, Shogo Muramatsu, Hiroyasu Yasuda, Kiyoshi Hayasaka, Yu Otake, Shunsuke Ono, Masahiro Yukawa: Convolutional-Sparse-Coded Dynamic Mode Decompsition and Its Application to River State Estimation, Proc. of 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.1872-1876, May 2019
  • Eisuke Kobayashi, Hiroyasu Yasuda, Kiyoshi Hayasaka, Yu Otake, Shunsuke Ono, Shogo Muramatsu: Multi-resolution Convolutional Dictionary Learning for Riverbed Dynamics Modeling, Proc. of 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , June 2023
  • C. Zhang, E. Kobayashi, D. Moteki, H. Yasuda, K. Hayasaka, S. Muramatsu: Performance Evaluation of MR-CSC-DMD in River Model Experiment with Groynes, Proc. of ITC-CSCC2023, June 2023

Research on cyber-physical system (CPS) for controlling river meanders

Large-scale river disasters have repeatedly caused the loss of human lives and social assets. We aim to minimize damage to critical infrastructure such as transportation networks in order to secure evacuation routes at the time of disaster and for early recovery after the disaster. Therefore, we propose a river channel active control method for the preservation and restoration of channel health. We will develop a cyber-physical system (CPS) for the prevention of river disasters by integrating different fields of informatics, river engineering, geotechnical engineering, and physics.

Publications

International Confs.

  • Weihang Liao, Gene Cheung, Shogo Muramatsu, Hiroyasu Yasuda, Kiyoshi Hayasaka, Graph Learning & Fast Transform Coding of 3D River Data, Proc. of APSIPA Annual Summit and Conf.，pp.1313-1317, Nov. 2018
  • Dongqi Liu, Yutaka Naito, Chen Zhang, Shogo Muramatsu, Hiroyasu Yasuda, Kiyoshi Hayasaka, and Yu Otake: River Flow Path Control with Reinforcement Learning, Proc. of 2021 IEEE International Conference on Autonomous Systems (ICAS), pp.212-216,  Aug. 2021
  • Yuki TAKAHASHI, Shogo MURAMATSU, Hiroyasu YASUDA, Kiyoshi HAYASAKA, Yu OTAKE: Flow-path Fitting from Images with Fourier Basis for River Health Assessment, Proc. of 2022 IEEE International Conference on Image Processing (ICIP), Bordeaux, Oct. 2022

Development of a water level distribution prediction model using river graph structure

This research establishes a new system of directed graph (DiGraph) signal processing and creates a dynamic modeling method on DiGraph that is necessary to manage the distribution of water levels in rivers. The Ministry of Land, Infrastructure, Transport and Tourism (MLIT) of Japan is urging measures through basin flood control to prepare for water-related disasters caused by climate change, and is calling for the systematization of off-channel water discharge, water stopping, and diversion control. Currently, there is no state-space model that controls the entire water system, and the possibility of controlling water levels in response to rainfall cannot be discussed theoretically. Therefore, we consider the physical quantities on the irregular network as signals on the graph structure, and derive the time evolution equations necessary for their prediction and control.

Publications

International Confs.

  • Yusuke Arai, Shogo Muramatsu, Hiroyasu Yasuda, Kiyoshi Hayasaka, Yu Otake: Sparse-Coded Dynamic Mode Decomposition on Graph for Prediction of River Water Level Distribution, Proc. of 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , DOI: 10.1109/ICASSP39728.2021.9414533, June 2021
  • Hotaka Kitamura, Hiroyasu Yasuda, Yuichi Tanaka, Shogo Muramatsu: Realization of Digraph Filters via Augmented GFT, Proc. of 2023 IEEE International Conference on Image Processing (ICIP), Oct. 2023, to appear

 

Links