Vanillin (VA), vanillic acid (VAA) and syringaldehyde (SYR) are additive foods and they have well defined UV spectra. However, these overlapped seriously, it is difficult to determine them individually from their binary and ternary mixtures without a pre-separation. In this study, multivariate calibration methods such as partial least square and artificial neural networks (ANNs) were applied to resolve the overlapping spectra and to determine these compounds simultaneously in the binary and ternary mixtures. The chemical parameters affecting for each of system were studied and optimized. Under the optimum experimental conditions, for each compound a linear calibration was obtained in the concentration range of 0.61-20.99, 0.67-23.19 and 0.73-25.12 μg mL-1 for VA, VAA and SYR, respectively. Four calibration sets of standard samples were designed by combination of a full and fractional factorial designs with the use of the seven (49 samples) and three (27 samples) levels for each factor for binary and ternary mixtures, respectively. The analytical performance of these chemometric methods were characterized by percentage relative error of prediction (%REP), root mean squares error prediction (RMSEP) and squared regression coefficient (R2), their predictive ability was tested with the use of synthetic samples and in general satisfactory results were obtained. The results of this study reveal that ANN and PLS methods can be used for determination VA, VAA and SYR in binary and ternary binary mixtures, respectively. Finally, in order to test the applicability and accuracy of the proposed artificial neural network method was applied for determination simultaneous VA, VAA and SYR in tap samples.
