Calibration-free physics-informed multi-task residual U-Net for simultaneous denoising and gas pressure retrieval from noisy voigt spectra
We present a machine learning framework based on a one-dimensional U-Net (1D U-Net) that simultaneously performs spectral denoising and pressure estimation within a unified architecture. The model is trained on simulated Voigt profiles of the P(21) CO absorption line over pressures ranging from of 1 mbar to 2 bar. To ensure realistic conditions, simulated spectra are superimposed with experimentally captured noise, making them nearly indistinguishable from real experimental s
A novel machine learning framework using a 1D U-Net architecture has been developed for simultaneously denoising spectral data and estimating gas pressure. The model was trained on simulated Voigt profiles of a CO absorption line, incorporating realistic experimental noise to mimic real-world conditions. Quantitative evaluations demonstrated high accuracy in reconstructing clean spectra from noisy inputs, achieving near-unity correlation coefficients and low error metrics. The U-Net successfully predicted pressures for unseen spectra without relying on traditional linewidth analysis, showing minimal error at specific pressure points. Experimental validation on a difference-frequency generation spectrometer confirmed the model's robust performance and high signal-to-noise ratios. The research highlights this U-Net as an efficient and reliable alternative to conventional spectroscopy techniques, simplifying workflows while maintaining high fidelity.
This research presents a significant advancement in spectroscopic analysis, offering a more efficient and accurate method for data processing and measurement in scientific applications.
📌 Kaynak
Bu özet naturecom kaynağından otomatik derlenmiştir. Tamamı için orijinal habere gidin.
Orijinal haberi oku →