Research on Rotating Parts Fault of Industrial Robots Based on Improved Multi-scale Residual Networks
The traditional residual neural network does not consider the difference of fault characteristics at different scales.An Improved Multiscale Residual Neural Network(IMRNN)based fault diagnosis model for ro-tating parts of industrial robots is proposed.First,wide convolution layers of different sizes are used to extract the macro features of the fault signals and combine them into initial feature vectors.Secondly,multi-scale adaptive selection convolution blocks are constructed to extract features of different scales.Then,the residual architecture is applied to multiple scale layers to improve the generalization of the model.At the same time,the channel attention mechanism is used to perform weighted fusion of features to complete fault diagnosis.The results show that the model has higher accuracy and better anti-jamming ability.
万方数据
维普
