基于双分支网络的表面肌电信号识别方法

王万良,潘杰,王铮,潘家宇

Recognition method of surface electromyographic signal based on two-branch network

Wanliang WANG,Jie PAN,Zheng WANG,Jiayu PAN

表 3 不同受试者情况下各方法的识别效果

Tab.3 Recognition accuracy of each method under different subject

方法 Pc/% Po/% HC HO RF WE WF KNN+FE[30] 97.18±0.00 72.99±0.00 47.95±0.00 93.50±0.00 42.14±0.00 70.75±0.00 SVM+FE[31] 64.27±0.00 86.24±0.00 51.71±0.00 100.00±0.00 52.99±0.00 71.04±0.00 CNN[8] 87.35±3.45 66.63±0.58 58.44±2.09 98.94±0.65 40.85±1.05 70.49±0.86 CNN+FE 71.99±4.16 90.51±1.85 80.87±7.59 99.82±0.49 31.39±7.51 75.06±0.18 CNN+FE+ETD 82.63±2.37 80.13±3.88 82.22±6.97 99.25±0.78 33.68±8.55 77.57±0.78 RESNET+FE[32] 73.79±3.96 85.22±2.78 81.17±6.19 99.22±0.65 32.11±6.95 74.96±0.26 Bi-GRU[33] 82.06±6.71 66.26±7.42 37.35±3.84 94.65±2.52 63.98±5.02 68.93±0.38 LSTM[34] 78.79±6.81 60.85±5.56 40.32±3.96 99.27±0.72 54.03±6.77 67.08±1.11 TRANSFORMER[35] 83.09±2.81 59.05±6.56 33.22±3.88 97.57±0.92 55.13±6.44 65.33±1.08 LCNN[36] 95.95±2.11 80.78±3.46 62.25±5.33 99.69±0.15 46.95±3.16 77.26±0.53 TDACAPS[14] 98.25±3.89 84.98±7.12 82.48±5.96 98.24±0.62 44.75±5.50 81.90±1.09 ETDTBN 99.75±0.15 75.51±1.85 95.27±2.54 99.93±0.05 50.21±0.45 84.15±0.41