Eeg-Based Cognitive Overload-Driven Adaptive Training for Enhancing Performance under Simulated High-Stress Space Operation Tasks

Given the potential for highly complex and high-risk space operation tasks to induce cognitive overload in operators, effective training is crucial for maintaining mission performance and safety. Conventional adaptive training often struggles to balance task challenge while preventing overload. To address this limitation, this study proposes an EEG-based cognitive overload-driven adaptive training method that integrates hybrid feature selection, normalized Fast Dynamic Time Warping (NFDTW)-based feature alignment and a shared knowledge layer to improve online cognitive overload detection under nonstationary conditions. A dual-factor task regulation strategy combining task performance and cognitive overload states is further developed to adjust task difficulty based on these two factors. The results showed that the proposed detection method achieved online detection accuracies of 92.07% and 84.67% on the two datasets, respectively, with maximum relative improvements of 7.74% and 8.34% over the comparison methods. The adaptive training strategy was further evaluated in a VR-based robotic arm docking task involving three groups: fixed, control, and adaptive. Mixed-design ANOVA showed that the adaptive group exhibited significant improvements in task performance and SAS as well as more favorable changes in EEG features. Bonferroni-adjusted post hoc comparisons further revealed significant group differences (p [Formula: see text] 0.05). These findings demonstrate the feasibility of EEG-based cognitive overload-driven adaptive training for complex space operation scenarios.