面向飞行员疲劳状态监测的脑认知深度模型研究

Brain Cognitive Deep Model for Pilot Fatigue State Monitoring

通过分析飞行员脑电信号,构建了疲劳状态的彩色脑功率图,设计基于高斯牛顿在线变分方法的卷积神经网络参数优化方法,形成一种新型脑功率图深度网络模型,有效实现脑功率图深度网络的模型分类识别能力。相比于其它基于脑电信号的疲劳检测深度模型,疲劳状态认知的准确度提升了3%~5%。

By studying Electroencephalogram(EEG)signals of pilots,the brain power color map of fatigue state is constructed.By designing a parameter optimization method of convolutional neural network using Gauss-Newton online variational method,a new brain power map deep network model is formed,which effectively implementsthe model classification and recognition ability of deep neural network for brain power map.Compared with other deep neural network models based on EEG signals,the accuracy for fatigue detection is improved by 3%~5%.