基于EEG的驾驶员疲劳检测：信号处理与分类方法

EEG 疲劳检测方法

背景

EEG（脑电图）是疲劳检测的金标准，直接反映大脑认知状态。

信号处理流程

import numpy as np
from scipy import signal

class EEGFatigueDetector:
    """EEG 疲劳检测器"""
    
    def __init__(self, fs: int = 256):
        self.fs = fs
        self.bands = {
            'delta': (0.5, 4),
            'theta': (4, 8),
            'alpha': (8, 13),
            'beta': (13, 30),
            'gamma': (30, 50)
        }
    
    def extract_features(self, eeg_signal: np.ndarray) -> dict:
        """
        提取频域特征
        
        Args:
            eeg_signal: EEG 信号 (channels, time)
        
        Returns:
            features: 频带功率比
        """
        features = {}
        
        for band_name, (low, high) in self.bands.items():
            # 带通滤波
            sos = signal.butter(4, [low, high], btype='band', fs=self.fs, output='sos')
            filtered = signal.sosfiltfilt(sos, eeg_signal, axis=1)
            
            # 计算功率
            power = np.mean(filtered ** 2, axis=1)
            features[f'{band_name}_power'] = power
        
        # 疲劳指标：theta/alpha 比值
        features['fatigue_index'] = (
            features['theta_power'] / (features['alpha_power'] + 1e-6)
        )
        
        return features
    
    def classify_fatigue(self, features: dict) -> str:
        """分类疲劳等级"""
        fatigue_index = features['fatigue_index'].mean()
        
        if fatigue_index > 2.0:
            return 'severe_fatigue'
        elif fatigue_index > 1.5:
            return 'mild_fatigue'
        else:
            return 'alert'

总结

EEG 提供高精度疲劳检测，但接触式传感器限制车载应用。

---

参考链接：

• 论文：https://www.sciencedirect.com/science/article/abs/pii/S0924424726007168