Analyses of replicated spectrophotometric data by using soft computing methods

Soft computing based chemometric studies are needed to model and to optimize complex processes, such as the extraction stage. In this study, it was aimed to obtain optimal values of ethanol (EtOH) concentration, extraction time, extraction temperature, solvent/solid ratio for the ultrasound-assisted extraction (UAE) of grape seed polyphenols to maximize the total phenolic content (TPC) and total antioxidant activity (TAA) in multi-objective perspective by using soft computing methods. The experimental data set was composed with replicated response measures to see the behavior of the responses. The replicated response measured (RRM) data was recently modeled by using fuzzy linear regression in which the replicated responses were considered as triangular linear fuzzy numbers (TLFNs). Also, polynomial type fuzzy linear regression model parameters were dealt as TLFNs whereas the experiment conditions were crisp. The predicted fuzzy linear models were obtained by using least square (LS) approach. The predicted fuzzy models were optimized through the Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and Pareto set was obtained. The obtained Pareto solution set was experimentally verified. A compromise solution was chosen among many non-dominated solutions of experimental conditions by using a distance function based on root mean square of fuzzy errors. According to the results, it is possible to say that the proposed soft computing based modeling and multi-objective optimization (MOO) approaches can be used as flexible analyses tools in chemometry.