Breaking the Decolorization Recovery Trade-off: Hybrid Adsorbent System Coupled With Response Surface Methodology and...

To overcome the long-standing "decolorization-recovery trade-off" in pullulan purification, this study innovatively developed a synergistic hybrid decolorization system integrating D392 resin and 325-mesh activated carbon, guided by a hybrid framework combining response surface methodology (RSM) and backpropagation artificial neural network (BP-ANN) for mechanistic exploration and high-precision prediction. Initial single-adsorbent optimization using RSM achieved high decolorization rates of 92.5% (1.8% resin) and 92.0% (1.4% activated carbon), but suffered from poor pullulan recoveries of only 80.14% and 77.14%, respectively. For the first time, mechanistic analysis revealed that pigment adsorption from complex pullulan fermentation broth (mixed pigment fractions) follows a monolayer Langmuir model for D392 resin and a multilayer Freundlich model for activated carbon, both conforming to pseudo-second-order kinetics, providing a theoretical foundation for synergistic system design. Considering the intrinsic limitation of quadratic response surface models for describing highly nonlinear multi-adsorbent interactions, a BP-ANN network was further introduced, and an independent test subset was applied to objectively compare predictive performance. Leveraging these distinct adsorption properties, sequential adsorption was optimized using RSM and BP-ANN, where the ANN demonstrated superior prediction accuracy. Under ANN-optimized conditions (0.17% resin and 0.88% activated carbon), the hybrid system achieved 92% decolorization and 90.4% recovery, drastically reducing resin consumption by 90.76% and activated carbon by 37.14% compared with single-adsorbent processes. Comprehensive characterization (ultraviolet-visible spectroscopy, Fourier transform infrared spectroscopy, X-ray diffraction, nuclear magnetic resonance, and scanning electron microscopy) confirmed that the native molecular structure and functional properties of pullulan were fully preserved. This work not only breaks the conventional trade-off between decolorization efficiency and pullulan yield but also provides a novel, cost-effective methodological paradigm for polysaccharide purification.