为分析综掘工作面光照环境对支护工脑疲劳的影响情况，准确识别综掘工作面支护工的脑疲劳状态，减少煤矿安全事故；设计了脑疲劳诱发实验，采集并分析了主客观数据；采用单因素方差分析法和配对样本T检验法提取了脑疲劳评价指标；使用支持向量机、K近邻、随机森林算法构建了综掘支护工脑疲劳识别模型，并建立混淆矩阵对各模型识别效果进行综合对比，选择出了最优识别模型。结果表明，准确率、反应时间、NASA-TLX值和O₁、O₂、O₆、O_Z、P_O4、P_O8、P_OZ、T₈ 8个电极通道的δ、θ、α、β节律能量值可作为综掘支护工脑疲劳评价指标；3类识别模型的识别准确率均较高，支持向量机的准确率为94.44%，K近邻的准确率为96.30%，随机森林的准确率为90.74%，因此基于K近邻算法的综掘支护工脑疲劳识别模型的识别效果最好。

In order to analyze the influence of light environment on the brain fatigue of support workers, identify and alleviate the brain fatigue of support workers in fully mechanized excavation face, and reduce coal mine safety accidents, a brain fatigue induced experiment was designed, and the subjective and objective data were collected and analyzed. The evaluation indexes of brain fatigue were extracted by single factor analysis of variance and paired sample T-test. Support vector machine, K-nearest neighbor algorithm and random forest algorithm were used to construct the brain fatigue recognition model of the support worker in fully mechanized excavation face, and the confusion matrix was established to comprehensively compare the recognition effect of each model, and the optimal recognition model was selected. The results showed that the accuracy rate, response time, NASA-TLX value and δ,θ, α and β rhythm energy values of O₁, O₂, O₆, O_Z, P_O4, P_O8, P_OZ and T₈ electrode channels could be used as the evaluation indexes of brain fatigue of the support workers in fully mechanized excavation face. The recognition accuracy of the three types of recognition models is high, the accuracy of support vector machine is 94.44%, the accuracy of K-nearest neighbor is 96.30%, and the accuracy of random forest is 90.74%. Therefore, the recognition model of brain fatigue of support workers in fully mechanized excavation face based on K-nearest neighbor algorithm has the best recognition effect.