基于FPGA的实时脑机接口应用研究

发布时间：2019-06-20 07:02

【摘要】：脑-机接口(Brain-computer Interface, BCI)是一种不依赖于脑的正常输出通路的通讯系统。通过在人脑与计算机等电子设备之间建立直接通路，脑机接口系统可以把大脑发出的信息直接转换成控制外部设备的命令，进而代替人类肢体或语言器官的功能，以新的途径实现人与外界的交流及对周围环境的控制。脑机接口在康复医学、工业、军事等领域都有巨大的潜在应用价值，已然逐渐成为一个研究热点。然而，脑机接口技术目前仍在发展中，多数研究还处于实验室阶段。面对BCI技术发展的机遇与挑战，本文开展了基于FPGA的实时脑机接口应用研究。所构建的BCI系统以瞬态视觉诱发电位为处理对象，相比于稳态视觉诱发电位，瞬态视觉诱发电位易于检测，而且低刺激频率不易引起视觉疲劳。以FPGA开发板为核心处理平台，，相比于单片机和DSP，FPGA在运算速度和逻辑控制方面具有优势。根据脑机接口应用的要求，采用FPGA设计了新的视觉刺激器。每个刺激模块都是黑白棋盘格交替闪烁的模式，不同点在于模块上的标志信息。在控制短消息发送的脑机接口应用中，设计了两个刺激界面，受试者首先选择短信接收方，然后选择短信内容，用汉字标注选项含义。在控制台灯、风扇运行的脑机接口应用中，刺激界面上的四个选项分别代表台灯的点亮与熄灭、风扇的转动与停止，在刺激模块上，用图形形象地标注各模块所代表的选项。 BCI技术的研究重点是选择合适的算法从强背景噪声中提取视觉诱发电位，识别受试者的选择。研究对比小波分解、主成分分析、K-近邻法、BP神经网络等信号处理算法测试，最终选择用db5小波对累加平均后的脑电信号进行5尺度分解，提取D5、D4两层细节系数作为特征向量，用BP神经网络识别，并用遗传算法对BP网络优化。小波分解和BP网络识别两个处理步骤由Nios II系统实现。本文将BCI系统用于控制TC35通讯模块发送短消息。FPGA将视觉诱发电位识别结果转换成发送短消息的命令，通过串口向TC35模块发送AT指令，TC35向FPGA反馈指令处理信息，以此实现发送短消息的控制。在控制台灯、风扇运行应用中，FPGA将视觉诱发电位识别结果转换成开关控制命令，通过控制继电器的状态实现对台灯、风扇的控制。脑机接口实验表明，所选用算法具有较高的识别率，并且验证了用基于FPGA的实时脑机接口控制发送短消息和台灯、风扇的运行具有可行性。
[Abstract]:Brain-computer interface (Brain-computer Interface, BCI) is a communication system which does not depend on the normal output pathway of the brain. By establishing a direct path between the human brain and the computer and other electronic devices, the brain-computer interface system can directly convert the information sent by the brain into the command to control the external equipment, and then replace the function of the human body or language organ, and realize the communication between man and the outside world and the control of the surrounding environment in a new way. Brain-computer interface has great potential application value in rehabilitation medicine, industry, military and other fields, and has gradually become a research focus. However, brain-computer interface technology is still in development, and most of the research is still in the laboratory stage. In the face of the opportunities and challenges of the development of BCI technology, the application of real-time brain-computer interface based on FPGA is studied in this paper. Compared with steady-state visual potentials, transient visual potentials are easy to detect and low stimulation frequency is not easy to cause visual fatigue in the constructed BCI system. Compared with single chip microcomputer and DSP,FPGA, FPGA development board has advantages in operation speed and logic control. According to the requirements of brain-computer interface application, a new visual stimulator is designed by using FPGA. Each stimulus module is a black and white chessboard alternately flashing mode, the difference lies in the logo information on the module. In the application of brain-computer interface to control the sending of short messages, two stimulation interfaces are designed. The subjects first select the receiver of the short message, then select the content of the short message, and mark the meaning of the option with Chinese characters. In the brain-computer interface application of console lamp and fan running, the four options on the stimulation interface represent the lighting and extinguishing of the lamp, the rotation and stop of the fan, and the options represented by each module are graphically marked on the stimulation module. The research focus of BCI technology is to select the appropriate algorithm to extract the visual evoked potential from the strong background noise and to identify the selection of the subjects. The signal processing algorithms such as wavelet decomposition, principal component analysis, K-nearest neighbor method and BP neural network are studied and compared. Finally, db5 wavelet is used to decompose the accumulated average EEG signals. D5, D4 two-layer detail coefficients are extracted as feature vectors, identified by BP neural network, and optimized by genetic algorithm. Wavelet decomposition and BP network recognition are implemented by Nios II system. In this paper, the BCI system is used to control the TC35 communication module to send the short message. FPGA converts the recognition result of visual evoked potential into the command to send the short message, sends the AT instruction to the TC35 module through the serial port, and the TC35 feedback the instruction processing information to the FPGA, so as to realize the control of sending the short message. In the operation application of console lamp and fan, FPGA converts the recognition result of visual evoked potential into switch control command, and realizes the control of table lamp and fan by controlling the state of relay. The brain-computer interface experiment shows that the selected algorithm has a high recognition rate, and verifies the feasibility of using the real-time brain-computer interface based on FPGA to control the transmission of short messages and lamp, and the operation of the fan.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318

【参考文献】