Multimodal Molecular Spectroscopy (UV–Vis, ATR-MIR, SERS) Coupled with Chemometrics for Lipopeptide Fermentation Monitoring

Lipopeptide fermentation faces challenges of high costs and low yields, requiring advanced monitoring solutions. This study presents an integrated methodology combining ultraviolet–visible (UV–vis), attenuated total reflection mid infrared (ATR-MIR), and surface enhanced Raman scattering (SERS) spectroscopy with chemometrics. Fermentation samples were used to build partial least-squares regression models for predicting reducing sugars, cell concentration, and lipopeptide yield. The stability competitive adaptive reweighted sampling (sCARS) algorithm outperformed principal component analysis in feature extraction. Evaluation of four fusion strategies revealed that full feature fusion yielded the highest predictive accuracy. The combined UV–vis, ATR-MIR, and SERS approach achieved optimal performance for lipopeptide yield (determination coefficient of prediction, R 2 p = 0.986; root-mean-square error of prediction, RMSEP = 0.357) and cell concentration ( R 2 p = 0.977, RMSEP = 0.010). This spectroscopic strategy enables efficient, non-destructive fermentation monitoring and shows considerable promise for industrial bioprocess optimization.