Evaluation of operation state for operators in NPP Main control room using human behavior recognition

Abstract

The cognitive state of the main control room (MCR) operator is the state of cognitive progress. This paper proposes a video analysis method to assist in the analysis of the operation state. A behavioral coding method is presented for the feature extraction of the MCR operator, and the time-line analysis method is used to continuously sample the operator’s postures and actions. OpenPose algorithm and ST-GCN method are used for the recognition of the operator’s behavior. The level of consciousness and cognition is analyzed based on the operator’s body language and used to evaluate the level of mental stress in performance shaping factors (PSF). A case is presented for the feasibility analysis of the operation state evaluation method. The results of the video analysis help recognize the operator’s bad or error behavior and improve the operator’s operation state.