24 October 2017 On-line dynamic monitoring automotive exhausts: using BP-ANN for distinguishing multi-components
Author Affiliations +
Proceedings Volume 10461, AOPC 2017: Optical Spectroscopy and Imaging; 104611A (2017) https://doi.org/10.1117/12.2285325
Event: Applied Optics and Photonics China (AOPC2017), 2017, Beijing, China
Remote sensing-Fourier Transform infrared spectroscopy (RS-FTIR) is one of the most important technologies in atmospheric pollutant monitoring. It is very appropriate for on-line dynamic remote sensing monitoring of air pollutants, especially for the automotive exhausts. However, their absorption spectra are often seriously overlapped in the atmospheric infrared window bands, i.e. MWIR (3~5μm). Artificial Neural Network (ANN) is an algorithm based on the theory of the biological neural network, which simplifies the partial differential equation with complex construction. For its preferable performance in nonlinear mapping and fitting, in this paper we utilize Back Propagation-Artificial Neural Network (BP-ANN) to quantitatively analyze the concentrations of four typical industrial automotive exhausts, including CO, NO, NO2 and SO2. We extracted the original data of these automotive exhausts from the HITRAN database, most of which virtually overlapped, and established a mixed multi-component simulation environment. Based on Beer-Lambert Law, concentrations can be retrieved from the absorbance of spectra. Parameters including learning rate, momentum factor, the number of hidden nodes and iterations were obtained when the BP network was trained with 80 groups of input data. By improving these parameters, the network can be optimized to produce necessarily higher precision for the retrieved concentrations. This BP-ANN method proves to be an effective and promising algorithm on dealing with multi-components analysis of automotive exhausts.
© (2017) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yudi Zhao, Yudi Zhao, Ruyi Wei, Ruyi Wei, Xuebin Liu, Xuebin Liu, } "On-line dynamic monitoring automotive exhausts: using BP-ANN for distinguishing multi-components", Proc. SPIE 10461, AOPC 2017: Optical Spectroscopy and Imaging, 104611A (24 October 2017); doi: 10.1117/12.2285325; https://doi.org/10.1117/12.2285325

Back to Top