Skip to main content
Figure 1 | Chinese Journal of Mechanical Engineering

Figure 1

From: Denoising Fault-Aware Wavelet Network: A Signal Processing Informed Neural Network for Fault Diagnosis

Figure 1

SPINN (Filterbank and feature stages are designed from SP-based methods but realized by DL-based methods. They can extract the information we need (green circle). Then the extracted information is input into a normal neural network (NN) as the final decision stage. The overall SPINN is trained by backpropagation with the target loss until desired training epoch (E))

Back to article page