煤矿下的通风机、采煤机和掘进机等设备会产生一定分贝的噪声，这些噪声直接或者间接导致煤矿井下不安全行为的发生。传统的噪声对人的心理健康和生理健康研究，是通过对煤矿工作人员采取调查问卷的方式，本课题的研究方向主要是从听觉诱发电位的角度，科学分析煤矿下噪声对人体的影响规律。

（1）针对脑电信号数据量大，且混有大量生理伪迹的问题，本文在传统的奇异值分解重构算法基础上，提出了用特征均值确定奇异值维数的方法，既可以压缩数据，又能有效的去除脑电信号中的伪迹，相比差分谱方法，信噪比提高了16.9%，均方根误差减小了4.5%，相关系数提高了9.2%。

（2）针对现有的脑电信号相关电位提取方法，不适用于长时听觉诱发脑电信号的问题，本文扩展了原有的事件相关电位提取方法，提出了基于小波变换的井下噪声事件点检测方法，提取井下噪声的频率、音强和音色特征，实验证明通过检测到的噪声事件点，对脑电信号叠加平均之后，可以提取出事件相关电位。

（3）针对井下噪声对脑电信号影响个体差异问题，提出了基于能量熵与峰值的最优电极组合方法，即峰值最大、次大，能量熵最大、次大方法，选取最优的电极组合，由脑电地形图证明了电极组合选择的正确性。

实验表明，本文提出的滤波算法能够有效去除脑电信号伪迹，并且通过最优电极组合算法确定井下噪声对大脑额区影响最大。在研究噪声诱发下被试者的听觉脑电信号变化规律时，实验表明，随着刺激噪声频率的增加，额区theta波相对功率平均先减小后增大，alpha波相对平均功率呈上升趋势。随着噪声音强减小，额区theta波相对平均功率，先减小，后增大，alpha波相对平均功率先增大，后减小。

Equipment such as ventilators, coal shearers, and roadheaders underground the coal mine will generate a certain decibel noise. These noises directly or indirectly cause unsafe behavior in the coal mine. The traditional noise research on people's mental health and physical health is through the use of questionnaires on coal mine workers. The research direction of this topic is mainly from the perspective of auditory evoked potentials to scientifically analyze the impact of various noises underground the coal mine on the human body.

(1) Aiming at the current EEG related potential extraction method, which is not suitable for the problem of long-term auditory evoked EEG , this paper extends the original event-related potential extraction method, and proposes a downhole noise event point detection method based on wavelet transform, The frequency, sound Intensity, and timbre characteristics of downhole noise are extracted. Experiments show that the event-related potential can be extracted by superimposing the EEG signals through the detected noise event points.

(2) Aiming at the problem that the EEG are mixed with a large number of physiological artifacts and the amount of data is large, on the basis of the traditional singular value decomposition and reconstruction algorithm, this paper proposes a method of determining the singular value dimension using the feature mean to effectively remove the EEG signals Compared with the differential spectrum method, the signal-to-noise ratio is increased by 16.9%, the rms error is reduced by 4.5%, and the correlation coefficient is increased by 9.2%.

(3) Aiming at the problem of individual differences in the effect of downhole noise on EEG signals, an optimal electrode combination method based on energy entropy and peak value is proposed, that is, the maximum peak value, the next largest, the maximum energy entropy, the second largest method, and the optimal electrode combination is selected. The frontal area of the brain is most sensitive to changes in the sound underground.

Experiments show that the filtering algorithm proposed in this paper can effectively remove the EEG artifacts, and the optimal electrode combination algorithm determines that the downhole noise has the greatest impact on the fr ontal area of the brain. When studying the changes of auditory EEG of the subjects under noise induction, the results show that with the increase of the frequency of the stimulus noise, the relative power of theta wave in the frontal area decreases first and then increases, and the relative power of the alpha wave increases. As the noise intensity decreases, the relative average power of theta wave in the frontal area decreases first and then increases, and the relative average power of alpha wave increases first and then decreases.