Noise Reduction Research on CO Detection in Transformer Oil Based on TDLAS Technology

The noise interference issue in the tunable diode laser absorption spectroscopy detection process for carbon monoxide (CO) in transformer oil was addressed by proposing a denoising algorithm optimized with the grey wolf algorithm for wavelet thresholding. In simulation experiments, this method improved the signal-to-noise ratio of noisy signals from 15 dB to 24.52 dB, significantly outperforming traditional wavelet thresholding, SG filtering, and empirical mode decomposition methods. In practical testing, the fitting R2 between the second harmonic peak-to-peak value and gas concentration reached 0.995; the relative deviation for oil samples with varying concentrations was below 30%; and the relative deviation in 50 stability tests was 11.56%. The results demonstrate that the grey wolf-optimized wavelet thresholding denoising algorithm enhances the measurement stability of CO monitoring devices in transformer oil.