Research on the fatigue detection method of operators in digital main control rooms of nuclear power plants based on multi-feature fusion

Abstract

Timely waking up the operators in the digital main control rooms of nuclear power plants from fatigue can effectively prevent accidents in nuclear power plants. During the operator's work, real -time evaluation of the fatigue state of operators is very important. Because the operators needs to constantly receive and process information from the computer screen through their eyes, it is easy to make their eyes tired and causes mental fatigue to the operators, so we propose to conduct fatigue detection for the operators. By focusing on the state of eyes, combined with the state of mouth and head movement, the efficiency of fatigue detection for operators can be effectively improved. The fatigue detection method was designed to integrate the classification results of three machine learning classification algorithms, logistic regression (LR), random forest (RF), and support vector machine (SVM), by fusing multi- feature such as slow blinks, PERCLOS, yawns, and head movements. Through experimental comparison, this method has higher accuracy compared with the multi-feature fusion method also based on the NTHU-DDD video dataset.