The complex and compact structure of straw represents the primary obstacle to its high-value conversion. This research integrated ball milling depolymerization, microwave hydrothermal conversion and machine learning methodology to gradually optimize and enhance the conversion potential of straw to 5-hydroxymethylfurfural (HMF). The findings indicated that the HMF yield of from straw, following ball milling depolymerization and direct microwave hydrothermal conversion, ranges between 0.10% and 0.45%. However, under acid-catalyzed microwave hydrothermal conditions, the HMF yield markedly increased to 2.35%–2.95%. It was noteworthy that AlCl₃, Ca-zeolite, and Mg-zeolite increased the maximum HMF yield to 9.47%, 7.67%, and 9.04%, respectively. Based on these findings, introducing organic solvents to form a biphasic system could further increase the HMF yield to 10.63% (75% MIBK addition scenario). Finally, machine learning prediction indicated that the HMF yield could be optimized to 11.13 wt% in the biphasic system with 0.08 g AlCl₃ per 0.2 g reaction material and 7.5 mL MIBK per 10 mL reaction solution and pH 1.25 under microwave hydrothermal at 190 °C for 0.75 min. The findings of this study provide a valuable reference point for the upcycling of straw and other biomass wastes, thereby facilitating the development and utilization of renewable resources.

中文翻译：

---

机器学习辅助微波水热转化优化秸秆 5-羟甲基糠醛产量


秸秆复杂而紧凑的结构是其高价值转化的主要障碍。本研究整合了球磨解聚、微波水热转化和机器学习方法，以逐步优化和增强秸秆向 5-羟甲基糠醛 （HMF） 的转化潜力。结果表明，球磨解聚和直接微波水热转化后，秸秆的 HMF 产量在 0.10% 至 0.45% 之间。然而，在酸催化微波水热条件下，HMF 产率显著提高至 2.35%–2.95%。值得注意的是，AlCl₃、Ca-沸石和 Mg-沸石将最大 HMF 产率分别提高到 9.47%、7.67% 和 9.04%。基于这些发现，引入有机溶剂形成双相系统可以进一步将 HMF 产率提高到 10.63%（75% MIBK 添加情况）。最后，机器学习预测表明，在 190 °C 微波水热下，0.75 min，每0.2 g反应材料中0.08 g AlCl₃，每10 mL反应液中7.5 mL MIBK，pH 1.25，双相系统中的HMF产率可以优化至11.13 wt%。本研究结果为秸秆和其他生物质废弃物的升级再造提供了有价值的参考点，从而促进了可再生资源的开发和利用。