基于深度信念网络的脑电信号疲劳检测系统

EEG fatigue detection system based on deep belief network

摘要:

传统的疲劳驾驶检测系统，一般采用对面部特征进行识别与信息提取的方式，易受到外界因素干扰，检测效率较低。针对这一问题，提出基于深度信念网络（DBM）的脑电信号(EEG)疲劳检测系统。结合深度信念网络工作原理和系统整体框架，设计系统硬件结构和软件功能。采用SAA7115型号信号解码器对数字化信号进行分离，通过采集模块电路图，将解码器连接到低噪声Video接口处，保证分离后的脑电信号为合成信号；通过TMS320DM642的DSP数字信号处理器对端口1信号进行合成、对端口2信号进行复合信号编码，保证信号采集不受外界因素干扰；将受限玻尔兹曼机在硬件采集模块中提取的信号进行疲劳程度检测，根据脑电信号变化强度，区分疲劳和未疲劳状态下脑电信号特征，完成系统设计。实验结果表明，所设计系统具有较高检测效率，可为疲劳驾驶人员生命安全提供保障。

Abstract:

The traditional fatigue driving detection system generally adopts the method of identifying facial features and extracting information, which is easily interfered by external factors and has low detection efficiency. In response to this problem, an EEG fatigue detection system based on Deep Belief Network (DBM) was proposed. Combine the working principle of deep belief network and the overall framework of the system to design the hardware structure and software functions of the system. The digital signal is separated by the SAA7115 model signal decoder. The decoder is connected to the low-noise Video interface through the acquisition module circuit diagram to ensure that the separated EEG signal is a composite signal; the DSP digital signal processor of the TMS320DM642 is used to port 1 The signal is synthesized, and the composite signal of the port 2 signal is encoded to ensure that the signal acquisition is not interfered by external factors; the signal extracted by the limited Boltzmann machine in the hardware acquisition module is tested for fatigue degree, according to the intensity of the change of the EEG signal, Distinguish the characteristics of EEG signals under fatigue and unfatigued conditions, and complete the system design. The experimental results show that the designed system has high detection efficiency and can provide guarantee for the life safety of fatigue drivers.