2C4-1 モデル内包学習による感染対応病室内の速度・濃度場推定法(2C4 OS:ニューラルネットワーク)

概要

論文の詳細を見る
Air purification in infection-free hospital rooms is becoming more and more important for preventing infection. To design such a room with the air purifier, it is necessary to measure airflow and contamination density distributions over the entire space in the room. It is however very difficult to measure the distributions over the whole flow field for such a real hospital room. This paper proposes a novel method for modelling 3-D flow velocity vector and contamination density distributions from sparsely measured data and boundary conditions. The proposed method is based on the model inclusive learning of a neural network in which the network is trained by using not only teaching data but also a priori knowledge whose model is included in the learning loop. The proposed method makes it possible to estimate simultaneously both 3-D flow velocity vector and contamination density distributions so that they satisfy the continuity and the diffusion equations which are modeled by a priori knowledge on the flow field. In order to evaluate the effectiveness, the proposed method is applied to air flow fields in the full size model of an infection-free hospital room and the results are compared with the measured distribution data obtained by smoke tracer-release experiments
社団法人日本機械学会の論文
2006-09-25