A spectral power-based real-time BCI neuro-feedback (decision machine) system has been proposed in this paper. It is used to reflect the intended task of the movement imagination of hand movements. As a consequence of the cost-effective and greater temporal resolution, the proposed decision machine considered the electroencephalogram (EEG) signal. EEG data modeling is therefore an important step in seeking a rational solution for a specific application. The most potential temporal relative spectral power (TRSP) based feature extraction algorithm, and probabilistic Bayesian classifier has been used to model the EEG data for the reproduction of the thoughts. In this paper, the author has considered two channels of information for command generation. This paper also focuses on the maximization of information retrieval by analyzing incoming two channels' brain signals. This initiates to devise an appropriate reactive frequency band for each subject before applying the feature extraction algorithm. In this direction, this paper uses relative power spectral intensity (RPSI) for reactive frequency band estimation. The spectral power-based feature extraction method is responsible for the improvement of the accuracy, reducing the computational complexity, lowers the computational time, and improves the information transfer rate. The performance of the proposed BCI system shows better results compared to the different well established conventional EEG based BCI systems. The proposed real-time neuro-feedback system is a generalized BCI system that will be readily applicable for controlling any external devices. The proposed system shows 90.37% accuracy over the BCI Competition II dataset and 81% average accuracy with 64 ms computational time in the real-time application.

本文提出了一种基于频谱功率的实时BCI神经反馈（决策机）系统。它用于反映手部动作的动作想象力的预期任务。由于具有成本效益且具有更高的时间分辨率，因此建议的决策机考虑了脑电图（EEG）信号。因此，EEG数据建模是为特定应用寻求合理解决方案的重要步骤。基于最潜在的时间相对频谱功率（TRSP）的特征提取算法和概率贝叶斯分类器已被用于对脑电数据进行建模以再现思想。在本文中，作者考虑了用于命令生成的两个信息通道。本文还着重于通过分析输入的两个通道的大脑信号来最大化信息检索。这开始在应用特征提取算法之前为每个对象设计一个合适的无功频段。在这个方向上，本文将相对功率谱强度（RPSI）用于无功频带估计。基于频谱功率的特征提取方法有助于提高精度，降低计算复杂度，减少计算时间并提高信息传输率。与不同的完善的基于常规EEG的BCI系统相比，所提出的BCI系统的性能显示出更好的结果。所提出的实时神经反馈系统是一种通用的BCI系统，将很容易应用于控制任何外部设备。所提出的系统在BCI Competition II数据集上显示90.37％的精度，在实时应用中具有64 ms的计算时间，平均精度为81％。