Noise suppression method for hydroxyl tagging velocimetry based on generative adversarial networks

Hydroxyl tagging velocimetry (HTV) technology is crucial in the velocimetry diagnosis of combustion flow fields. However, obtaining accurate HTV information in practical engineering applications is difficult because of complex flow fields and background noise interference. Therefore, for noise suppression, we proposed a generative adversarial network method for targeted network training based on the analysis of HTV image noise characteristics in a complex flow field and the construction of a high-confidence noise description model. The proposed method can effectively suppress noise in HTV experimental data, improve the signal-to-noise ratio of HTV images, and improve the accuracy of HTV measurement.