基于稳态视觉诱发电位（Steady-state visual evoked potentials，SSVEP）的脑-机接口系统以其信息传递速率高、稳定性强、适用性广等特点而受到广泛的关注。目前，该方向的研究工作主要聚焦于范式研究、算法设计及系统开发三个方面。其中范式研究是最为重要的环节之一，好的刺激范式不仅可以产生更强的脑电诱发响应，还有助于实现更加丰富的控制指令。以往的SSVEP刺激范式往往局限于对刺激目标块进行时间、频率、相位上的编码，未充分利用视野空间信息。本文在视野刺激中引入空间信息，探索SSVEP的时空频联合编码范式，有效提高了枕区及耳后的SSVEP响应的识别性能，并最终实现了一套便捷的全新脑-机接口系统。在双频空间相位编码方面，本文设计了一种双频空间相位编码范式。该范式可有效提升备选目标数量。在该范式基础上，本文进一步提出的基于刺激信号的编码优化方法，可有效缩短范式刺激编码的设计时间。实验表明，本研究设计的双频空间相位刺激范实验被试平均识别正确率达到96.1±4.0%，信息传递速率达到196.1±15.3bits/min，性能优于现有的其他双频SSVEP刺激范式。在此基础上，开发了一套160目标脑-机接口应用系统，实现了在线平均86.6±10.1%的识别正确率及175.2±30.2bits/min的信息传递速率，验证了范式在编码特征空间上的优势。在多相位空间频率编码方面，本文着重研究了空间多相位刺激的脑电响应叠合模式，提出了一套基于SSVEP相位锁定特性的多相位刺激响应延迟估计方法。在此基础上，开发了基于左、右视野刺激的多相位空间频率刺激范式。该范式可提高大部分被试的耳后SSVEP响应识别性能，可将耳后系统适用被试比例从58.3%提升至75.0%，并可将现有最佳标准刺激的74.6±20.0%识别正确率提高至84.2±14.7%, 信息传递速率从14.4±6.4bits/min提高至17.8±5.7bits/min。在系统应用方面，本文基于多相位空间频率编码开发了一套便捷化使用的耳后SSVEP脑-机接口系统。该系统采用凝胶干电极采集耳后区域的脑电信号，使用者无需洗头即可快速开展BCI系统应用，有效降低了脑-机接口系统的应用准备时间。实验结果表明，虽然耳后SSVEP响应通常较弱，但该系统有效提高了大部分被试的耳后SSVEP识别性能，最终绝大多数被试实现了在线平均84.2±11.1%的识别正确率及24.9±6.7bits/min的信息传递速率。最后本文总结了SSVEP视觉刺激编码和优化的一些基本思路和关键技术。

The brain-computer interface system based on steady-state visual evoked potential (SSVEP) has been widely researched due to its characteristics of less training time, ease-of-use, and high information transfer rate. At present, the research work in this field mainly focuses on three aspects: paradigm research, algorithm design, and system development. Among them, paradigm design is one of the most important parts. A good stimulation paradigm can not only generate stronger EEG-evoked responses but also help to achieve more abundant control instructions. Previous studies on the SSVEP paradigm were mainly limited to encoding the stimulus target block in time, frequency, and phase, and did not make full use of visual field information. In this paper, spatial information is introduced into visual field stimuli, and the spatiotemporal-frequency coding paradigm of SSVEP is explored, which effectively improves the recognition performance of SSVEP responses in the occipital and retroauricular regions. Finally, this paper implements a new brain-computer interface system that is convenient and easy to use.In aspect of the visual field multi-frequency stimulation paradigm, this paper designs a dual-frequency phase-encoded stimulation paradigm. This paradigm can effectively increase the number of candidate targets through dual-frequency coding. Based on this paradigm, the coding optimization method based on stimulus signal further proposed in this paper can effectively shorten the designing time of paradigm stimulus coding. Experiments show that the average recognition accuracy of the dual-frequency phase stimulation paradigm designed in this study reaches 96.1±4.0%, and the information transfer rate reaches 196.1±15.3 bits/min, which is superior to other existing dual-frequency SSVEP stimulation paradigms. Besides, a 160-target SSVEP-BCI application system is developed based on dual-frequency phase-encoded stimulation paradigm, which achieves an average online accuracy of 86.6±10.1% and an ITR of 175.2±30.2bits/min. This work verifies the advantages of the dual-frequency phase encoding paradigm in encoding feature space.In the aspect of visual field multi-phase stimulation paradigm, this paper studies the EEG response overlapping pattern of spatial multi-field stimulation, and designs a multi-field stimulation response delay estimation method based on the phase-locking property of SSVEP. On this basis, a multiphase spatial frequency stimulation paradigm based on left and right visual field stimulation was developed. This paradigm can improve the recognition performance of most of the subjects behind the ear SSVEP response, and can increase the proportion of subjects suitable for the behind-the-ear system from 58.3% to 75.0%, and can improve the accuracy rate of 74.6±20.0% of the existing best standard stimulus to 84.2±14.7%, ITR increased from 14.4±6.4bits/min to 17.8±5.7bits/min.In terms of system application, this paper develops a convenient SSVEP-BCI system based on the left and right visual field biphasic stimulation paradigm. The system uses gel dry electrodes to collect SSVEP signals behind the ear, and users can quickly apply the BCI system without washing their hair, thus effectively reducing the application preparation time of the brain-computer interface system. Although the SSVEP response behind the ear is usually weak, the system effectively improves the recognition performance of the SSVEP behind the ear for most of the subjects. In the end, most subjects achieved average online recognition accuracy of 84.2±11.1% and an information transfer rate of 24.9±6.7bits/min.Finally, this paper summarizes some basic ideas and key technologies of SSVEP visual stimulus encoding and optimization.