TY  - JOUR
PY  - 2020//
TI  - A knowledge extraction method for the prediction of incidents with low frequency of occurrence
JO  - Journal of the Japan Society for Safety Engineering
A1  - Kurahashi, Setsuya
SP  - 368
EP  - 372
VL  - 59
IS  - 6
N2  - In this paper, we introduce methods and experimental examples for acquiring knowledge related to accident signs that occur infrequently, which is difficult to find by statistical processing. This method was designed using a process simulation model and a gaming & simulation method created from process data using a time series analysis method. Then, the operator virtually experiences the accident that actually occurred, and analyzes the log that the operator himself operated using the simulator and the record that discussed the countermeasures against the crisis through gaming. As a result of the experiment, we were able to discover five new accident-predicting hiyari hats by operating the simulator.[Google Translate]    本論では，統計処理では見出しにくい，発生頻度の低い事故予兆に関わる知識を獲得するための手法と実験例を紹介する．本手法は，プロセスデータから時系列解析手法を用いて作成したプロセスシミュレーションモデルおよびゲーミング＆シミュレーション手法を用いて設計した．そして，実際に発生した事故を運転員が仮想体験することで，運転員自身がシミュレータを用いて操作したログや，ゲーミングを通して危機への対応策を議論した記録を分析する．実験の結果，シミュレータの操作から5 件の事故予兆となる新たなヒヤリハットを発見することができた<p />  <p>Language: ja</p>
LA  - ja
SN  - 0570-4480
UR  - http://dx.doi.org/10.18943/safety.59.6_368
ID  - ref1
ER  -