基于空洞卷积和增强型多尺度特征自适应融合的滚动轴承故障诊断

韩康,战洪飞,余军合,王瑞

Rolling bearing fault diagnosis based on dilated convolution and enhanced multi-scale feature adaptive fusion

Kang HAN,Hongfei ZHAN,Junhe YU,Rui WANG

表 8 江南大学数据在不同噪声下8种模型的准确率

Tab.8 Accuracy of 8 models on Jiangnan University data under different noise

模 型 A SNR=4 SNR=6 SNR=8 SNR=10 SNR=12 MSCNN-LSTM[14] 0.6710±0.0402 0.7157±0.0208 0.7939±0.0198 0.8509±0.0216 0.9107±0.0080 WDCNN[15] 0.6396±0.0161 0.6969±0.0149 0.7223±0.0229 0.7703±0.0276 0.8193±0.0197 MSCNN[16] 0.6785±0.0102 0.6985±0.0099 0.7497±0.0074 0.8032±0.0078 0.8164±0.0143 DC 0.6527±0.0212 0.7164±0.0115 0.7654±0.0151 0.8191±0.0182 0.8544±0.0126 Resnet[17] 0.7106±0.0155 0.7552±0.0116 0.8231±0.0091 0.8559±0.0022 0.9072±0.0084 DRSN-CS[18] 0.7437±0.0113 0.7920±0.0138 0.8476±0.0155 0.8760±0.0287 0.9096±0.0250 MA1DCNN[19] 0.6783±0.0474 0.7597±0.0335 0.8238±0.0166 0.8723±0.0270 0.9172±0.0280 DC-MAFFM 0.7556±0.0042 0.8040±0.0063 0.8555±0.0098 0.9077±0.0047 0.9351±0.0048