EEG signal analysis for human workload classification

Abstract

This paper provides the results of determining the state of a human pilot operator by using electroencephalograph (EEG) data. The state of a human operator is used to represent the mental (cognitive) workload experienced during task execution. This study used EEG data gathered from a crew-simulation laboratory environment. By using EEG data from twelve subjects encountering six simulated pilot workload levels, we set up a neural network to obtain an overall mean classification accuracy of over 80%. A comparison between the conventional backpropagation method and the resilient backpropagation method also shows that a significant reduction in training time can be achieved.