A parallel dual output model based on multi-dimensional feature fusion: for spectral separation and concentration...

With the development of industrialization and urbanization, as the main air pollutants, the concentration detection of nitric oxide (NO) and ammonia (NH₃) are very important for environmental management and air quality assessment. Ultraviolet differential optical absorption spectroscopy (UV-DOAS) plays an important role in the field of gas detection with many advantages. However, in the face of mixed gas detection scenarios with overlapping absorption spectra, the detection accuracy still faces challenges. Therefore, a parallel dual output spectral separation model based on multi-dimensional feature fusion is proposed. According to the absorption spectrum features of NO and NH₃ in the model, the corresponding feature extraction branch is designed to realize the synchronous separation of the differential optical density (DOD) of the two gases. Secondly, the polynomial curve concentration inversion model is established, and the relationship between DOD and single-component gas concentration is constructed. Finally, the gas concentration is determined according to the DOD after separation. Based on the data of 200-230 nm band collected by UV-DOAS system, the experiment shows that the uncertainty of NO and NH₃ are 1.71 % and 1.28 %, respectively, and the detection limits are 0.087 ppm·m and 0.05 ppm·m, respectively. This study is expected to be widely applied in the field of multi-component gas detection, contributing to public health and environmental protection.