A Method and System for Detecting Peaks in Transformer Gas Chromatography

Regarding peak detection in transformer gas chromatography, including peak retention time, peak start position, and peak end position detection, it is an important processing step for analyzing and calculating component concentrations. Due to various factors, chromatograms often suffer from noise interference and difficulty in detecting overlapping and weak peaks. However, there is currently no complete solution. In this study, we propose a method that combines discrete wavelet transform (DWT) and continuous wavelet transform (CWT)-gaussian fitting, which is called D-CWTG. DWT is used for denoising, CWT is used for estimating peak parameters, and a combination of wavelet-domain and time-domain feature quantities is introduced to filter spectral peaks. Finally, gaussian fitting is optimized through genetic algorithm (GA) to refine peak parameters. This method improves the detection rate of weak and overlapping peaks while effectively reducing the recognition rate of false peaks. As a complete peak detection system, D-CWTG can accurately detect peak parameters and has been successfully applied in practical transformer gas chromatography.