Optimization of Synergistic Fermentation Parameters for Enzyme–Microbial Systems in Machine-Harvested Cotton Residue Using TOPSIS and Response Surface Methodology

Machine-harvested cotton residue is a major byproduct of cotton processing whose utilization is limited by a lignocellulosic matrix and antinutritional factors. This study combined an L49(76) orthogonal design with entropy-weighted TOPSIS for multi-indicator selection and then used Plackett–Burman screening and Box–Behnken response surface methodology (RSM) to optimize NDF degradability as a response-specific objective. Range analysis showed that fermentation time was the dominant factor for several fiber-related indices. TOPSIS selected A4B3C6D2E6F1 as the preferred multi-indicator treatment, corresponding to 15% glucose, 10% urea, 65% moisture, 14 days, a bacterial ratio of Enterococcus faecium: Bacillus licheniformis: Bacillus amyloliquefaciens = 15:10:5, and an enzyme ratio of cellulase: laccase: lignin peroxidase = 5:10:15. Relative to the raw residue, NDF, cellulose, hemicellulose, and ADL contents decreased by 25.83%, 43.54%, 74.40%, and 45.65%, respectively. In the subsequent NDF-oriented RSM stage, PBD identified fermentation time, urea, and glucose as the three quantitative factors taken forward to BBD. The model predicted an NDF degradability of 39.054% at 20.449 days, 9.865% urea, and 9.901% glucose; a rounded operating point of 20 days, 10% urea, and 10% glucose yielded a mean of 37.03% in three replicate runs. Thus, TOPSIS and RSM provided complementary results for different optimization objectives rather than a single global optimum.