Detection of six kinds of acid in red wine with infrared spectroscopy based on FastICA and neural network

Abstract

For the rapid detection of the six kinds of acid in red wine, infrared (IR) spectra of 44 wine samples were analyzed. A new method of model construction based on back-propagation artificial neural networks (BP-ANN) regression and fast independent component analysis (FastICA) was proposed. This new chemometric method, named ICA-NNR, has been applied to detect the six kinds of acid in wine samples. Compared with the model built by the common used methods, such as PCR and PLS, ICA-NNR method has advantages in both the correlation coefficient and standard error of calibration. The correlation coefficients (R) between the referenced values and the model predicted values are 0.9833, 0.9759, 0.9585, 0.9989, 0.9643 and 0.9884, respectively. The results show the feasibility of establishing the models with ICA-NNR method for red wine samples? quantitative analysis and provide a foundation for the application and further development of IR on-line red wine analyzer.