Dyeing Behavior of Enzyme and Chitosan-Modified Polyester and Estimation of Colorimetry Parameters Using Random Forests

Toprak Çavdur T., Aniş P., Bakir M., Sebatli-Saglam A., Çavdur F.

FIBERS AND POLYMERS, vol.24, no.1, pp.221-241, 2023 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 24 Issue: 1
  • Publication Date: 2023
  • Doi Number: 10.1007/s12221-023-00130-x
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED)
  • Page Numbers: pp.221-241
  • Keywords: Machine learning, Random forests, Surface modification, Polyester, Dyeing, SURFACE MODIFICATION, TEXTILE DYES, HYDROLYSIS, CUTINASE, FABRICS, FIBERS, CHITIN
  • Bursa Uludag University Affiliated: Yes


Dyeing of the crystalline structure necessitates a process with a disperse dye either at high temperatures or with a carrier due to its compact and non-ionic structure of polyester. In this study, in order to eliminate these limitations and develop more environmentally friendly dyeing processes, the dyeability of polyester under different conditions with reactive, direct, and acid dyes after surface modifications with enzyme and chitosan was investigated. In addition to the corresponding physical experiments, CIELAB and color strength values were also estimated using random forests. The results of the physical experiments showed that the surface modifications conducted with enzyme and chitosan significantly increased the color depths obtained in dyeing for reactive, direct, and acid dyes, especially at pH 4.5. This was explained by the potentially protonated amine groups in acidic medium of chitosan could have attracted large amounts of anionic dye molecules with physical forces. The highest color depths were obtained from acid dyeing. Washing fastness of the pre-treated and dyed fabrics (except the acid-dyed fabrics) decreased with the shift of the bath pH values to the acidic region. In the next phase of the study, we implemented random forests to estimate CIELAB and color strength values. We considered different random forest designs and trained each design ten times to observe the performance of the corresponding topology. The results of the computational experiments showed that the estimation performance of the random forests is quite satisfactory (with R-values greater than 99%) and random forests could be used to estimate CIELAB and color strength values successfully.