用于控制短消息发送的实时脑机接口系统

目的：设计一种基于视觉诱发电位的实时脑机接口,用于控制短消息发送。方法：实时脑机接口系统由视觉刺激器、脑电采集电路、FPGA开发板、通讯模块组成。脑机接口界面包括短消息发送的目标选项和内容选项界面,受试者每次实验注视刺激界面中的一个模块,通过检测视觉诱发脑电来确定受试者做出的选择。利用基于FPGA的VGA视觉刺激器为受试者提供视觉刺激,采集脑电信号并在FPGA平台上对其进行在线的实时分析处理。选用小波分解提取视觉诱发电位特征向量,输入BP神经网络进行模式识别,产生脑机接口控制信号,其中,小波分解和BP神经网络两部分由NIOS II实现。脑机接口控制命令用于控制TC35无线模块发送短消息。结果：通过对五名受试者做实验,识别准确率最高可以达到100%,脑机接口系统能有效控制短消息的发送。采用小波滤波、BP神经网络识别的算法优于时域波形匹配识别法。结论：实验表明本文提出的实现脑机接口短消息发送系统的方法具有可行性。

A Real-time Brain-computer Interface System for Controlling Short Message Sending

Objective: To design a real-time brain-computer interface system for controlling short message sending.Methods: The system is made up of visual stimulator,EEG acquisition circuit,FPGA development board and communication module.The brain-computer interface consists of an object choosing interface and a content choosing interface.The subject can choose one option by staring at one module in the interface for several seconds.The system can determine which option the subject has chosen by analyzing the VEP signal of the subject.The VGA monitor based on FPGA was used to produce visual stimulation.Collected EEG signal is processed on FPGA development board in real-time.Wavelet filtering was used to extract the feature vectors of VEP(visual evoked potential) signal,which was input to the BP neural network to recognize VEP and produce brain-computer interface control signal.Wavelet transformation and BP neural network were realized by NIOS II.The BCI control signal was used to control TC35 wireless module to send short message.Results: The recognition can reach highly to 100% when five testers are tested separately short message can be sent out using the designed BCI system.The algorithm of wavelet filtering and BP neural network is better than the waveform matching method in time domain for VEP recognition.Conclusions: The experiments show the proposed method of sending short message based on BCI is workable.