Simultaneous determination and spectra correction of air toxic organic compounds using PLS and ANN in FTIR spectroscopy

In this paper, the residual spectra library search was used to correct the spectrum containing one interferent. Two groups of which each sample contai ned no or one interferent were supplied to amplify the above method. PLS wa s used to identify the interferents and correct the spectra, the corrected spectra information was sent into the PLS(Partial Least Squares) and ANN(Ar tificial Neural Network) prediction step. The concentration of each compone nt in the unknown mixtures was determined using two different prediction me thod. The RSD% (mean relative standard deviation) for the two groups was ca lculated respectively. The prediction results of PLS and ANN was compared t o illustrate which one is better in solving this problem.