Publication: Antiperspirant formüllerinin yapay sinir ağları ve genetik algoritma kullanılarak analizi ve optimizasyonu
Abstract
Bu çalışmada, koltuk altı ter kokusunun önlenmesi için en etkili ürünler olan antiperspiran çubuklar; stearil alkol, hidrojene hint yağı, talk ve siklometikon temelinde hazırlandılar. Etken madde olarak alüminyum zirkonyum tetraklorhidreks gliserin kompleksi FDA'nın (Food and Drug Administration) belirlediği maksimum oranda (% 20) kullanıldı Stearil alkol 3 farklı konsantrasyonda (% 10, % 20, % 30), hidrojene hint yağı 3 farklı seviyede (% 1, % 3, % 5) ve talk da yine 3 farklı seviyede (% 1, % 3, % 5) kullanılarak toplam 31 adet formül hazırlandı. Bu çalışmada, 3 farklı seviyede çalışılarak yapılan tasarım ile formülasyondaki maddelerin antiperspiranın uygunluğunu gösteren kırılma kuvveti ve kalıntı özellikleri üzerinde olan etkileri belirlenmiştir. Siklometikon formüllere yeter miktar olarak ilave edildi. Bu formüllerde kırılma kuvveti ve kalıntı ölçümleri yapıldı. Veri analizi ve optimizasyon için Yapay Sinir Ağı (YSA) ve Genetik Algoritma (GA) sistemleri kullanılmıştır. YSA ve GA modelleriyle ölçülen cevaplar için 0.6474'den 0.9984'e kadar R2 (determinasyon katsayısı) değerleri elde edilmiştir. Yapılan validasyon sonucunda da, genetik algoritma modeli antiperspiran özellikleri tahmin etmede daha başarılı olmuştur. Genetik algoritma kullanılarak yapılan optimizasyonda, kabul edilebilir, en az kalıntı bırakan ve kolay sürülebilen optimum antiperspiran formülasyonun % 18-22 konsantrasyonunda stearil alkol, % 2-4 konsantrasyonunda talk ve % 4-5 konsantrasyonunda hidrojene hint yağı içerdiği belirlenmiştir. ANALYSIS AND OPTIMIZATION OF ANTIPERSPIRANT FORMULATIONS BY USING THE ARTIFICIAL NEURAL NETWORKS AND GENERAL ALGORITHMS
In this study, antiperspirant stick formulations were prepared based on stearyl alcohol, hydrogenated castor oil, talc and cyclomethicone to prevent underarm odor. As an active ingredient aluminum zirconium tetrachlorohydrex glycine complex was used at the permisible level (20 %) by the FDA (Food and Drug Administration). A total of 31 formulations were prepared in which stearyl alcohol was used at 3 levels (10 %, 20 %, 30 %), hydrogenated castor oil and talc were used at 3 levels (1 %, 3 %, 5%). In this study, 3 level design was used to evaluate the effect of concentration and formulation variables on the performation of antiperspirants. Cyclomethicon was added to formulations as q.s.. Breaking force and residue (spreadability) were measured as the responses. The Artificial Neural Networks methodology (ANN) along with Genetic Algorithms (GA) were used for data analysis and optimization. ANN and genetic models provided R2 values between 0.6474 and 0.9984 for measured responses. When a set of validation experiments was analyzed, genetic algorithm (GA) predictions of antiperspirant characteristics were much better than the ANN. Optimization based on genetic algorithm showed that using stearyl alcohol at 18-22 % with talc at 2-4 % and hydrogenated castor oil at 4-5 % would produce acceptable and the best antiperspirant stick that leave the least amount of the residue and are easy to spread.
In this study, antiperspirant stick formulations were prepared based on stearyl alcohol, hydrogenated castor oil, talc and cyclomethicone to prevent underarm odor. As an active ingredient aluminum zirconium tetrachlorohydrex glycine complex was used at the permisible level (20 %) by the FDA (Food and Drug Administration). A total of 31 formulations were prepared in which stearyl alcohol was used at 3 levels (10 %, 20 %, 30 %), hydrogenated castor oil and talc were used at 3 levels (1 %, 3 %, 5%). In this study, 3 level design was used to evaluate the effect of concentration and formulation variables on the performation of antiperspirants. Cyclomethicon was added to formulations as q.s.. Breaking force and residue (spreadability) were measured as the responses. The Artificial Neural Networks methodology (ANN) along with Genetic Algorithms (GA) were used for data analysis and optimization. ANN and genetic models provided R2 values between 0.6474 and 0.9984 for measured responses. When a set of validation experiments was analyzed, genetic algorithm (GA) predictions of antiperspirant characteristics were much better than the ANN. Optimization based on genetic algorithm showed that using stearyl alcohol at 18-22 % with talc at 2-4 % and hydrogenated castor oil at 4-5 % would produce acceptable and the best antiperspirant stick that leave the least amount of the residue and are easy to spread.