表 3. Comparison of fatigue classification accuracy (in%) based on the combination of denoising algorithm and classifi cation algorithm.
不同去噪方法对疲劳识别算法准确度影响对比
| 去噪方法 | LSTM | CNN | PSO-SVM |
| 硬阈值函数去噪方法 | 80.92% | 77.71% | 73.85% |
| 软阈值函数去噪方法 | 86.47% | 81.83% | 80.66% |
| 改进阈值函数去噪方法 | 95.19% | 92.72% | 90.30% |