Abstract: In response to the needs of SPO cockpit automation, a self-adaptive architecture for cockpit human-computer interaction was proposed. Firstly, a cognitive status assessment model, a task cognitive demand model and a cognitive prediction model were established based on physiological indicators as well as continuous monitoring of aircraft operating status parameters and environmental parameters. Then, simulation experiments were conducted to verify changes in physiological indicators and cognitive status of pilots with different external conditions and automation levels. Finally, a self-adaptive adjustment model of automation level was developed to adjust automation flexibly and allocate workload and automation dynamically. The results show that the self-adaptive framework for human-computer interaction based on expert system can assess pilots' cognitive status, reasonably allocate cockpit workload through real-time change of automation level, thereby stabilizing pilots' load index at about 0.4 and slowing down fatigue accumulation so as to maintain the best cognitive status.

Key words: single pilot operation (SPO), automation level, self-adaptive adjustment, cockpit, human-computer interaction