Fault Diagnosis Method of Rolling Bearing Based on CBAM_ResNet and ACON Activation Function

In order to cope with the influences of noise interference and variable load on rolling bearing fault diagnosis in real industrial environments, a method based CBAM_ResNet ACON activation function is proposed. Firstly, collected working vibration signals are made into input samples retain original features maximum extent. Secondly, model constructed. By taking advantage convolutional neural network (CNN) classification tasks key feature extraction, block attention module (CBAM) embedded residual blocks, avoid degradation enhance interaction information channel spatial, raise extraction capability model. Finally, Activate or Not (ACON) function, introduced adaptively activate shallow for purpose improving model’s representation generalization capability. The dataset Case Western Reserve University (CWRU) used experiments, average accuracy proposed 97.68% 93.93% under strong load, respectively. Compared other three published methods, results indicate that this has better immunity ability, good application value.