Friction signal denoising based on deep learning

Addressing the issue of noise present in friction signals collected by scanning probe microscopy instruments, a denoising method for friction signals based on deep learning is proposed. This article proposes a denoising model that combines convolutional autoencoder, deep residual network, and attention mechanism. By using convolutional blocks and attention mechanism, different weights can be assigned to signal features. The residual network can effectively prevent gradient explosion and network degradation caused by increasing network depth. Generate simulated friction signals through numerical calculations and add Gaussian white noise to construct a training set for training and testing the model. This method successfully increased the Signal-to-Noise Ratio (SNR) of the signal by about 26dB, demonstrating excellent noise reduction performance.