Transient electromagnetic data noise suppression method

By constructing a CNN-LSTM neural network with improved SPPF_CMA optimization, the data denoising problem of transient electromagnetic method in complex noise environments is solved, and effective denoising of late contaminated data and efficient detection of deep geological bodies is achieved.

CN120234527AActive Publication Date: 2025-07-01INSTITUTE OF GEOLOGY AND GEOPHYSICS CHINESE ACADEMY OF SCIENCES
16 Cites 1 Cited by

Patent Information

Application Number
CN202510303732.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-03-14
Publication Date
2025-07-01
Estimated Expiration
2045-03-14

AI Technical Summary

Technical Problem

Transient electromagnetic method is disturbed by noise in data processing and geological interpretation in complex noise environments, resulting in low resolution of deep geological bodies.

Method used

A CNN-LSTM neural network based on improved SPPF_CMA optimization is adopted to construct a neural network model for transient electromagnetic signal denoising through convolutional layer, pooling layer, SPPF_CMA module, bidirectional long and short-term memory layer and fully connected layer. The model extracts features through CA attention mechanism and multi-scale pooling branches, and designs a mixed loss function for training to generate synthetic training data to improve the denoising effect.

Benefits of technology

Effectively denoising transient electromagnetic data that is contaminated in the late stage improves the efficiency of data usage, enhances the detection ability of deep geological bodies, can adapt to a variety of noise environments, simplifies the denoising process, and improves efficiency.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a transient electromagnetic data noise suppression method based on a neural network, and the method mainly comprises the steps: constructing a CNN-LSTM neural network based on improved SPPFCMA optimization, generating synthetic training data, and designing a mixed loss function. The network comprises a convolutional layer, a pooling layer, a bidirectional long and short time memory layer and a full connection layer. By improving an SPPFCMA module, a CA attention mechanism is added after convolution layer output, and multi-scale pooling branches are adopted. A method of generating synthetic training data includes simulating a noiseless TEM signal using one-dimensional forward modeling and superimposing artificial noise. The training termination condition is that the mixed loss function value of the verification set does not decrease in continuous five iterations, and the method is used for denoising the actually measured data after preprocessing the actually measured data.
Need to check novelty before this filing date? Find Prior Art