EEG-Based Crew Workload Identification by Maritime Virtual Simulator

With the rapid development of ship automation technology, crew workload during navigation has been increasing, significantly impacting both navigational safety and work efficiency. In this study, we designed and conducted an experiment to collect electroencephalogram (EEG) signals in a maritime virtual simulator to investigate the workload of ship operators performing different tasks. Based on Multiple Resource Theory (Visual, Auditory, Cognitive, Psychomotor; VACP), operational tasks-including normal navigation, ship encounter, emergencies, and berthing-were categorized into high and low workload levels. After preprocessing the EEG signals, the Greedy Adjacent Algorithm (GAA) was applied to extract clustering coefficient features from the brain functional networks. A Support Vector Machine (SVM) was then employed for workload classification, achieving an average accuracy of 81.87%. In terms of computational efficiency, including average processing time and network density, GAA outperformed the fully connected network. Visualization of brain region activation revealed significantly higher activity in the occipital and temporal lobes across different navigation tasks, with particularly prominent activation observed in the beta frequency band. This study provides valuable insights for developing crew workload identification and monitoring technologies, which are essential for accident prevention and the optimization of crew operations.