Abstract:

Brain-computer interface (BCI) establishes a direct information communication channel between the brain and the external environment, which has increasingly highlighted its important scientific significance and application value in many aspects of human motor function rehabilitation, replacement and enhancement. As the most natural brain-computer interaction mode, BCI paradigm based on movement intention has been widely concerned by researchers. The decoding of pre-movement EEG patterns can make BCI more responsive and flexible. However, the EEG features at this stage are weak and difficult to recognize efficiently. Based on the above challenges, research is focused on the enhancement of temporal and spatial lateral characteristics of pre-movement EEG in low frequency. Aleft and right hand self-paced keystroke task is designed, and a new composite spatial filtering algorithm is proposed by combining discriminative canonical pattern matching (DCPM) and task-related component analysis (TRCA). The results show that the average accuracy of EEG signal recognition under this algorithm can reach 78.56%, which is better than previous literature reports, and can provide theoretical basis and technical support for efficient brain-computer interaction based on motion intention.

Key words: movement preparation, electroencephalography, brain-computer interface, spatial filter