基于物理信息神经网络的可控源电磁数据去噪方法及系统
By combining a physical information neural network with a composite loss function that integrates data fidelity and physical constraints, the signal-to-noise ratio (SNR) degradation caused by noise interference in the traditional controllable source electromagnetic method is solved, achieving high-fidelity restoration of electromagnetic signals and improving physical rationality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CENT SOUTH UNIV
- Filing Date
- 2026-06-22
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional controlled-source electromagnetic methods are subject to interference from complex human electromagnetic noise in mining areas during mineral exploration, resulting in a decrease in signal-to-noise ratio. Existing deep learning denoising methods lack physical constraints, leading to physical distortion and poor interpretability of denoising results.
A physical information neural network is used to construct a composite loss function, which combines data fidelity loss and physical constraint loss, including transmit waveform consistency, frequency domain attenuation, time domain smoothness and autocorrelation sidelobe suppression, to ensure that the denoising result conforms to the physical laws of electromagnetic signals.
It achieves high-fidelity restoration of controllable source electromagnetic signals under strong interference environment, improves the physical rationality and interpretability of denoising results, has strong adaptability, and avoids the parameter dependence and black box characteristics of traditional methods.
Smart Images
Figure CN122412969A_ABST