This paper investigate the effectiveness of the Empirical Mode Decomposition (EMD) based Power Spectrum analysis (PSA) technique to evaluate the Performance of SSVEP based Brain computer inference (BCI) system in terms of SSEP recognition accuracy and Information transmission rate (ITR). Steady State Visual evoked Potential (SSVEP) is a quasi sinusoidal signal contaminated into recorded EEG signal. The presence of artifacts and spontaneous EEG signal deteriorate the SSVEP Performance. EMD is technique that decomposes the recorded EEG Signal into several oscillating components known as intrinsic mode functions (IMF). The selection of IMF components plays a vital role in recognizing SSVEP signal with high accuracy. Power spectrum density (PSD) as a feature is extracted from the SSVEP Prominent IMF component to recognize the accuracy of SSVEP BCI System. The obtained result compared with the Wavelet-based PSA approach and conventional PSA approach. The result obtained from four subject demonstrate that the improve the SSVEP performance in terms accuracy and ITR about 4.24% and 6.78 bits/minute as compared to DWT-PSA, 6.78% and 10.65 bits/minute as compared to standard PSA respectively.

本文研究了基于经验模态分解（EMD）的功率谱分析（PSA）技术从SSEP识别精度和信息传输率（ITR）方面评估基于SSVEP的脑计算机推理（BCI）系统的性能。稳态视觉诱发电位（SSVEP）是一种准正弦信号，被污染到记录的EEG信号中。伪影和自发EEG信号的存在会降低SSVEP性能。EMD是一种将记录的EEG信号分解成几个称为固有模式函数（IMF）的振荡成分的技术。IMF组件的选择在高精度识别SSVEP信号中起着至关重要的作用。从SSVEP Prominent IMF组件中提取功率谱密度（PSD）作为特征，以识别SSVEP BCI系统的准确性。与基于小波的PSA方法和常规PSA方法进行比较。从四个主题获得的结果表明，与DWT-PSA相比，SSVEP性能在准确性和ITR方面分别提高了约4.24％和6.78位/分钟，与标准PSA相比分别提高了6.78％和10.65位/分钟。