Activity recognition in plant: Investigation on actions for preventing human error

Research on recognizing human activities has attracted much interest in pervasive sensing and machine learning. Usually, plants needs to know all information in relation to activities for the sake of security that require all the actions must be observable and recognizable. The existing learning-based approaches are not sufficient to capture the characteristics of human activity efficiently because they don't process the dimensional information enough. This paper describes a novel approach to recognize activity and prevent human error using monitoring data from RFID-based systems. Different from many approaches that analyz the human activity from event sequence first, the proposed approach start it from dimensional information, and the model can be applied to track activities in plants routine and more suitable for discovering the changing of activity. Finally, we prove our approach on data collected in simulation plant environment.