2020 Volume 76 Issue 2 Pages I_25-I_33
To increase the resilience of facilities over a wide area, it is important to immediately understand the situation in a wide area at the time of a disaster. This study proposes a model to immediately estimate an entire target facility’s behavior from information obtained from limited observation points based on spatio-temporal data obtained by numerical analysis. Firstly, we proposed a method to identify the optimal placements of observation points to effectively make the model work. Here, the QR decomposition with column pivoting was used to optimize the observation points. Secondly, an example is applied to the settlement process of embankment on liquefied, and it demonstrated the effectiveness of the proposed method by comparing the estimation error from random observation points. We confirm that the optimized observation points reflect the physical characteristics of the numerical analysis through the above consideration.