| At present,the vast majority of brain-computer interfaces are mainly multi-channel or invasive brain-computer interfaces used in scientific research.They achieve the purpose of human-computer interaction and peripheral control by obtaining cerebral cortex or internal neuron EEG information.The high price of the brain-computer interface and the complicated wearing process make it difficult to get out of the laboratory to be applied in real life;and the invasive brain-computer interface also requires expensive and high-risk brain-computer interface implantation surgery,which is for most physically disabled people.As far as the family is concerned,it is a heavy burden.Relatively speaking,although the single-channel brain-computer interface is inferior to the multi-channel brain-computer interface and invasive brain-computer interface in terms of EEG information transmission and control accuracy,it is convenient to wear and cheap,so there is no need for brain-computer interface implantation.Surgery has good practical significance.This article aims to further improve the accuracy of the EEG feature recognition of the single-channel brain-computer interface,and separately design the brain-computer character and voice interaction system and the brain-computer wheelchair control system to help gradually freeze people to solve the inability to communicate with the outside world,the difficulty of movement,The problem of limited space for activities.The research work mainly includes the following three points:(1)A method for extracting EEG blink features based on CNN and lightweight Mobile Net V1 is proposed to improve the accuracy of blink feature recognition and further subdivide blink features to solve the current single-channel EEG blink based on threshold judgment The feature extraction method has the problem that the accuracy of blink feature recognition is not high and the blink features cannot be further subdivided.(2)Design a brain-computer character and voice interaction system for people with gradual freezing to help aphasia patients with gradual freezing to achieve text and voice communication.First draw the time series EEG signal into a two-dimensional image and color it,use CNN to identify the image with blink features;then,according to the number of consecutive frames of the image with blink features,realize the subdivision of short blink,medium blink,and long blink features;Finally,combining the attention characteristics analyzed in the EEG signal,the designed virtual interactive system outputs characters and voices.The comparative experimental analysis shows that the method effectively improves the accuracy of the interaction.(3)Design a brain-controlled wheelchair control system to help patients with frostbite paralysis solve the problem of difficulty in movement and limited space for movement.The lightweight Mobile Net V1 is deployed in the AI SOC-based brain-controlled wheelchair control system as an EEG blink feature recognition method.Through this method,the braincontrolled wheelchair is separated from the dependence on the deployment of the PC algorithm,and the embedded brain-controlled wheelchair control system is endowed with AI computing ability,and the weight and volume of the wheelchair itself are reduced,and its sports performance is improved. |
