Scholarlink

<p>Rice straw hydrolysate, a sugar rich lignocellulosic biomass, offers a sustainable feedstock for bioethanol production, exploiting <i>Kluyveromyces marxianus</i> thermotolerance and C-5/C-6 co-fermentation capabilities. This study introduces a novel hybrid Long Short-Term Memory (LSTM) kinetic model, which is integrating LSTM neural networks with mechanistic kinetic modelling, for data driven predictions to optimize bioethanol production. Using Response Surface Methodology with the Central Composite Design (RSM CCD), the optimal conditions were determined. The hybrid-LSTM kinetic model performed efficiently, with R² &gt; 0.97 and RMSE of 0.018–0.030&#xa0;g/L, improving the prediction accuracy by 10–15%. Simulated real time monitoring with Internet of Things (IoT) sensor data enhanced ethanol yield by 5% through dynamic process control achieving a maximum ethanol yield of 7.9&#xa0;g/L sugar confirming after validation predictions (errors &lt; 4%, t- test <i>p</i> &gt; 0.2). This integrated framework offers a scalable, precise approach for sugar-based bioethanol production, addressing challenges in sustainable biofuel production.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic Modelling and Optimization of Bioethanol Production from Rice Straw Hydrolysate Using Kluyveromyces Marxianus with Hybrid LSTM Kinetic Model and RSM-CCD

  • 期刊论文

  • 发表日期：

    2026年1月6日

查看全文
  • Sonampreet Kaur,
  • Richa Srivastava,
  • Kumar Gaurav

摘要

Rice straw hydrolysate, a sugar rich lignocellulosic biomass, offers a sustainable feedstock for bioethanol production, exploiting Kluyveromyces marxianus thermotolerance and C-5/C-6 co-fermentation capabilities. This study introduces a novel hybrid Long Short-Term Memory (LSTM) kinetic model, which is integrating LSTM neural networks with mechanistic kinetic modelling, for data driven predictions to optimize bioethanol production. Using Response Surface Methodology with the Central Composite Design (RSM CCD), the optimal conditions were determined. The hybrid-LSTM kinetic model performed efficiently, with R² > 0.97 and RMSE of 0.018–0.030 g/L, improving the prediction accuracy by 10–15%. Simulated real time monitoring with Internet of Things (IoT) sensor data enhanced ethanol yield by 5% through dynamic process control achieving a maximum ethanol yield of 7.9 g/L sugar confirming after validation predictions (errors < 4%, t- test p > 0.2). This integrated framework offers a scalable, precise approach for sugar-based bioethanol production, addressing challenges in sustainable biofuel production.