基于数据驱动的闭环脑机接口设计

发布时间：2018-06-01 08:58

本文选题：大脑皮层仿真模型 + 无模型控制　；参考：《华东理工大学学报(自然科学版)》2017年05期

【摘要】：基于神经元峰电位的植入式脑机接口开展相关研究,通过搭建大脑皮层仿真模型,并在控制理论分析的基础上进行自发单关节运动任务。使用自适应维纳滤波器完成神经元放电活动的线性解码器设计。通过分析发现在视觉反馈信息缺失时,解码器性能严重下降。针对此问题,使用基于数据驱动算法的紧格式无模型控制算法产生刺激信号来刺激仿真大脑皮层感觉区神经元,使其跟踪存在感官反馈时感觉区神经元的放电活动。由于感觉区神经元放电信息的恢复,解码器在感觉反馈信息缺失时的性能也得到了恢复。最后,通过仿真验证了基于数据驱动算法的人工感官反馈有效性,并与整定的PID控制算法对比,结果验证了本文设计的闭环系统的有效性。
[Abstract]:Based on the neural spike potential embedded brain-computer interface (BCI), a simulation model of cerebral cortex was built, and the task of spontaneous single joint movement was carried out on the basis of control theory analysis. An adaptive Wiener filter is used to design a linear decoder for neuronal activity. It is found that the performance of the decoder is seriously degraded when the visual feedback information is missing. To solve this problem, a compact format modelless control algorithm based on data-driven algorithm is used to generate stimulation signals to stimulate the neurons in the sensory region of the cerebral cortex to track the firing activity of the neurons in the sensory region when sensory feedback exists. The performance of the decoder is also restored when the sensory feedback information is missing due to the recovery of the firing information of the sensory neurons. Finally, the effectiveness of artificial sensory feedback based on data-driven algorithm is verified by simulation, and compared with the tuning PID control algorithm, the results verify the effectiveness of the closed-loop system designed in this paper.
【作者单位】：华东理工大学信息科学与工程学院;西安交通大学电气与控制工程学院;
【基金】：国家自然科学基金青年基金(61603295)
【分类号】：R318;TP273

【相似文献】