가스 사용 환경에서의 위험 상황 인지를 위한 딥러닝 예측모델 개발

Development of a Deep Learning Prediction Model to Recognize Dangerous Situations in a Gas-use Environment

Recently, with the development of IoT communication technology, products and services that detect and inform the surrounding environment under the name of smart plugs are being developed. In particular, in order to prepare for fire or gas leakage accidents, products that automatically close and warn when abnormal symptoms occur are used. Most of them use methods of collecting, analyzing, and processing information through networks. However, there is a disadvantage that it cannot be used when the network is temporarily in a failed state. In this paper, sensor information was analyzed using deep learning, and a model that can predict abnormal symptoms was learned in advance and applied to MCU. The performance of each model was evaluated by developing firmware that can judge and process on its own regardless of network and applying a predictive model to the MCU after 3 to 120 seconds.