脑机接口是一种新型的人机交互技术。该技术可直接由脑信号重建运动控制，可以战略性地用于军事目的，也可为严重运动残疾人和正常人提供辅助控制，从而改善生活质量。传统基于运动想象的BCI仅识别想象运动的肢体类型，不提供运动的参数，只提供少量的方向控制命令，难于满足灵活的运动控制。虽然新的基于运动参数想象的BCI在一定程度上可弥补此局限，但该类BCI的被试难于产生差异显著、可分性好的脑电特征。本课题拟聚焦探索神经反馈对运动参数想象相关脑电活动的调节机理，以寻找被试能产生差异显著、可分性好的脑电特征的策略和方法。采用多因素方差分析和多重比较方法，揭示由时-频分析方法、平均技术和相干算法提取的与运动速度和握力变化想象相关的神经振荡、运动相关皮层电位和运动功能脑区之间的同步活动规律。同时，利用该规律来优化神经反馈技术，采用基于脑电运动参数想象的单次识别算法，构建该类BCI灵活控制仿人机器人运动的原型系统。

Brain-computer interface（BCI）is a new type human-computer interaction technology by which motor control can be directly restructed by brain signal bypass peripheral nerves and muscles. BCI can be strategically used for military purposes and also provide auxiliary control for severely movement disabled persons and healthy persons so as to improve the quality of their lives. The traditional BCI based on motor imagery only identifies limb types involved in imagined movement,and thus it does not provide kinematic parameters and only a small amount of direction control commands which are difficult to meet the flexible motion control.Although a new BCI based on motor parameters imagery can compensate for this limitation to some extent, subjects operating the BCI is difficult to produce EEG features which have significant difference and good separability among motor parameters imagery.This project intends to focus on exploring the mechanism of neurofeedback regulating EEG activity related to motor parameters imagery in order to look for the strategies and methods which subjects can produce EEG features having significant difference and good separability.Using multivariate analysis of variance and multiple comparisons method, we will reveal the laws for neural oscillations, movement-related cortices potentials (MRCPs) and synchronization activity between brain areas related to motor functions which are associated with imagined motion speed and gripping force variation and extracted by time-frequency analysis method, average technique and coherent algorithm.At the same time, the neurofeedback technology is optimized by these laws and the prototype system is builded for the BCI to flexibly control of a humanoid robot ' movement by the single-trial recognition algorithm of imagined motor parameters based on EEG.