基于不同调制方式的视觉诱发电位脑机接口研究

发布时间：2018-05-31 20:45

本文选题：脑机接口 + 频率和相位混合调制　；参考：《南昌大学》2016年硕士论文

【摘要】：脑机接口是一座搭建于大脑和外部环境之间的桥梁,通过这座桥梁人可以不经由外围神经肌肉而直接与外界进行通信。随着对脑机接口技术需求人数的增加,脑神经科学和工程技术的日渐成熟,越来越多的机构和科学家致力于脑机接口的研究。在各类脑机接口中,基于稳态视觉诱发电位的脑机接口和基于编码调制的视觉诱发电位脑机接口被研究较多且性能较好,为了进一步提高这两种脑机接口的性能,本文对稳态视觉诱发电位脑机接口和编码调制脑机接口进行深入的研究和探索。传统的基于稳态视觉诱发电位的脑机接口系统采用频率编码,每个刺激块以不同频率闪烁。由于稳态视觉诱发电位响应频带窄且某些受试者存在盲频,因此呈现的目标数有限。为了解决这个问题,本文研究了频率和相位混合调制的稳态视觉诱发电位脑机接口,在这种调制方式下,同一频率可以以不同相位调制多个目标,大大提高频率利用率,增加刺激目标数。目标识别的方法是对脑电信号做各刺激频率处的快速傅立叶变换,得到的各刺激频率处的傅立叶系数投影至各目标的参考相位,最大投影值所对应的目标即为识别目标。时间窗为2s时,受试者的平均目标识别准确率达到89.30%。在编码调制脑机接口中,伪随机码的调制性能对识别准确率有很大的影响,为此本文研究了M序列、近完美序列、Golay互补序列三种伪随机序列的调制性能,目标识别算法均采用模板匹配法。研究结果表明,时间窗为一个刺激周期时三种序列的目标识别率分别为89.71%、94.29%、93.20%。从自相关函数波形来看,与M序列相比,近完美序列、Golay互补序列的峰值更尖锐且整体旁瓣更低,因而这两种序列的调制性能更佳。为了增加刺激目标数同时兼顾识别率和识别速度,本文提出不同编码分组调制方法,并联合多导联脑电信号进行目标识别。在采用近完美序列和Golay互补序列调制两组目标的多导联系统中,平均识别率为92.34%。
[Abstract]:Brain-computer interface is a bridge between the brain and the external environment, through which people can communicate directly with the outside world without passing through the peripheral nerve muscles. With the increasing demand for brain-computer interface technology and the maturation of brain neuroscience and engineering technology, more and more institutions and scientists are devoting themselves to the research of brain-computer interface. Among all kinds of brain-computer interfaces, the brain-computer interface based on steady-state visual evoked potential and the brain-computer interface based on coded modulation have been studied more and better. In order to further improve the performance of these two kinds of BCI, In this paper, the steady-state visual evoked potential brain-computer interface and coded modulation brain-computer interface are deeply studied and explored. The traditional brain-computer interface system based on steady-state visual evoked potentials uses frequency coding, and each stimulus block flashes at different frequencies. Due to the narrow response band of steady-state visual evoked potential (VEP) and blind frequency in some subjects, the number of targets presented is limited. In order to solve this problem, the steady-state VVEP brain-computer interface of frequency and phase mixing modulation is studied in this paper. In this modulation mode, multiple targets can be modulated with different phases at the same frequency, and the frequency utilization ratio can be greatly improved. Increase the number of stimulus targets. The method of target recognition is to do the fast Fourier transform at every stimulation frequency of EEG signal, and the Fourier coefficients of each stimulus frequency are projected to the reference phase of each target. The target corresponding to the maximum projection value is the recognition target. When the time window is 2 s, the average target recognition accuracy is 89.30%. The modulation performance of pseudorandom codes has a great influence on the recognition accuracy in the coded modulation brain-computer interface. In this paper, the modulation performance of three pseudorandom sequences, M sequence, near perfect sequence and Golay complementary sequence, is studied in this paper. Template matching is used in all target recognition algorithms. The results show that when the time window is a stimulus cycle, the target recognition rate of the three sequences is 89.71 and 94.290.93.20, respectively. From the waveform of autocorrelation function, compared with M sequence, the near perfect sequence Golay complementary sequence has more sharp peak value and lower global sidelobe, so the modulation performance of these two sequences is better. In order to increase both the recognition rate and the recognition speed, different coded block modulation methods are proposed and multi-lead EEG signals are combined for target recognition. The average recognition rate is 92.34 in the multi-lead system which uses near-perfect sequence and Golay complementary sequence to modulate two groups of targets.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7;R49

【参考文献】