Sequential number theoretic optimization (SNTO) method is presented for constrained background bilinearization (CBBL) of two-way multicomponent data and robust multivariate calibration based on least median of squares (LMS) method. compared with variable step generalized simulated annealing (VSGSA) method in both computer simulation on two selected mathematic functions and optimization part for CBBL, SNTO method gives satisfactory results. The LMS method based on SNTO method is apparently superior to the least squares (LS) technique in both real chemical systems and simulated systems. SNTO method is one of practical global optimization methods because of distinctness of its principle and easy realization.

Key words: QUANTITATIVE ANALYSIS