Integrated Experimental Optimization and Machine-Learning Prediction of Spartium junceum Natural Dyeing with Comparative Wastewater and Functional Assessment
Fibers and Polymers, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s12221-026-01589-0
- Dergi Adı: Fibers and Polymers
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
- Anahtar Kelimeler: Artificial intelligence, Machine learning, Multicriteria decision making, Natural dyes, Spanish broom
- Bursa Uludağ Üniversitesi Adresli: Evet
Özet
Spartium junceum flower extract was investigated as a natural colorant for wool through an integrated framework combining experimental optimization, machine learning (ML), mordant-assisted coloration, functional evaluation, comparative wastewater assessment, and multi-criteria decision analysis. Dyeing parameters, including pH, temperature, dye concentration, and dyeing time, were systematically investigated, and the most favorable dyeing conditions within the experimental range were identified as pH of 4, temperature of 90 °C, dye concentration of 100% (o.w.f.) and dyeing time of 60 min. A Random Forest (RF) regression model was developed to predict wavelength-specific color strength (K/S) using wavelength and the four dyeing parameters as input variables. To prevent information leakage among spectral observations from the same physical experiment, model performance was evaluated using leave-one-experiment-out cross-validation across 19 independent dyeing experiments. The RF model achieved R2 = 0.932, RMSE = 0.685, and MAE = 0.320, demonstrating good predictive performance within the investigated experimental domain. Mordanting significantly influenced apparent color strength and colorimetric properties, with Fe/YM and Fe/TA exhibiting the highest average K/S values of 6.62 ± 0.07 and 5.72 ± 0.07, respectively. K/S was strongly negatively correlated with lightness (L*) (r = − 0.781, p < 0.001). The dyed materials also exhibited enhanced functional properties, with Fe/YM and Sn/PB showing UPF values of 165.52 ± 3.31 and 176.40 ± 3.52, respectively, while several metal–biomordant systems exhibited high DPPH radical-scavenging activity. Comparative wastewater analysis revealed treatment-dependent variations in COD and BOD5 among the mordanting systems, while COPRAS analysis consistently identified Fe/YM as the highest-ranked treatment across seven weighting scenarios within the three-criterion framework based on K/S, COD, and BOD5. Overall, the results demonstrate the potential of S. junceum for multifunctional wool coloration and illustrate the utility of integrating experimental analysis, grouped-validation ML prediction, functional characterization, wastewater quality indicators, and multi-criteria ranking in natural dyeing research.