Measurement and Evaluation of Mental Workload Based on Complex Human–Machine Interaction Tasks

This study conducted ergonomic experiments based on three MATB-II tasks with different difficulty levels, collected the subjects’ EEG indicators, task performance, and subjective mental workload indicators during task execution, and proposed a mental workload evaluation method based on the Error Back Propagation (BP) algorithm. The results showed that subjective mental workload increased with the increase in task difficulty and duration. The resource management task scored the lowest among the three tasks, which may reveal the significance of the difficulty of the resource management task. Through EEG data analysis, there were significant differences in the four types of EEG power with different difficulty and duration. The BP algorithm based on difficulty and duration can distinguish the mental workload under different task difficulties in the time dimension, and the accuracy reached 66.61% under fivefold cross-validation. The research results can guide complex human–computer interaction systems’ design, task planning, and mental workload assessment.