Details

Autor(en) / Beteiligte

• Yan, Fuxue
• Pi, Shuai
• Yu, Qi
• Lin, Jun

Titel

Transient Electromagnetic Data Noise Suppression Method Based on RSA-VMD-DNN

Ist Teil von

• IEEE geoscience and remote sensing letters, 2024, Vol.21, p.1-5

Ort / Verlag

Piscataway: IEEE

Erscheinungsjahr

2024

Quelle

IEEE Xplore

Beschreibungen/Notizen

• Random noise greatly affects the quality of transient electromagnetic (TEM) signals in urban environments, leading to reduced detection accuracy. The effectiveness of denoising methods such as filtering or modal decomposition is limited by manually selecting parameters, and some noises are bound to be retained. In this letter, a novel approach combining reptile search algorithm (RSA) optimized variational mode decomposition (VMD) with deep neural network (DNN) is proposed to identify and eliminate noise. First, RSA is used to select key parameters in VMD. Then, based on the optimized parameters, the noisy signal is decomposed into different intrinsic mode functions (IMFs) via VMD, and the cross-correlation (CC) coefficient is used to select and reconstruct the effective signal. To solve the problem of residual noise, the convolutional neural network (CNN) and long short-term memory are combined to extract the time-related features of the data and further improve the signal-to-noise ratio (SNR). RSA-VMD-DNN is a denoising method that is better suited for random noise in nonlinear and nonstationary TEM signals. It can effectively select and reconstruct signals while eliminating residual noise. The results show its great potential for improving the accuracy and reliability of TEM detection.