مشخصات پژوهش

خانه /Optimizing predictive models ...
عنوان
Optimizing predictive models for food contaminants: The role of wavelet transforms in ANN and PLSR analysis of heavy metals
نوع پژوهش مقالات در نشریات
کلیدواژه‌ها
Cu2+ and Fe3+ ions Simultaneous determination Schiff base ligand CWT PLS ANNs Vegetable and fruit samples Water samples
چکیده
This research shows multivariate chemometrics models coupled with spectrophotometric data for simultaneous determination of Cu2+ and Fe3+ ions. Spectrophotometric data were acquired by the complexation of metal ions with Schiff base 4,4′-(2,2-dimethylpropane-1,3, diyl)-bis(azan-1-yldene) dipent-2-en-2-ol (DPBDO). The experimental conditions were optimized, and formation constants were calculated by rank annihilation factor analysis. Due to the presence of highly spectral overlap between complexes multivariate chemometrics techniques were selected. Since wavelet transformation able to reveal key features of spectra, it was combined with partial least squares (PLS) and artificial neural networks (ANNs) calibration methods. PLS, CWT-PLS and CWT-ANNs models were optimized, and prediction set was tested. The R.S.E. values were 2.5 and 1.76 and RMSEP% values were 0.023 and 0.344 for Cu2+ and Fe3+, respectively. Hence, this model was applied for determination of ions in some vegetable, legume and Persian Gulf water and algae samples. Also, the results obtained from this method are in good agreement with the results of the flame atomic absorption spectroscopy (FAAS) method. The RSD % values between 0.6 % and 7 % and recovery % values in the range 98–112 % confirm the stability of network and repeatability of results. This approach can be a fast, cost-effective, and reliable alternative to conventional methods for simultaneous target ions determining.
پژوهشگران اعظم آشناگر (نفر اول)، مریم عباسی طریقت (نفر دوم)، غلامرضا عبدی (نفر سوم)
تاریخ انجام 1404-09-24