Scholarlink

<p>The utilization of lignocellulosic biomass, such as spent coffee grounds (SCG), for bioprocesses such as fermentation presents challenges due to the complex structure of its components. Effective pretreatment methods are needed to overcome these difficulties. This study investigates the use of artificial neural networks (ANN’s) to predict the behavior of the hydrolysis process (pretreatment), with a focus on efficient sugar extraction from SCG. By developing a feedforward neural network with input parameters such as temperature (130–190&#xa0;°C), sulfuric acid concentration (0.5-2.0% v/v), solid/liquid ratio (1:4 − 1:40), and reaction time (20–120&#xa0;min), the research aims to estimate the hydrolysis process using the Levenberg-Marquardt backpropagation algorithm in MATLAB R2024b software. With a neural model structure of four neurons in a single hidden layer, the model successfully predicts the amount of hemicellulosic sugars obtained from the hemicellulose fraction based on the input variables. The results demonstrate the effectiveness of the model in identifying the optimal conditions for converting polysaccharides in coffee waste into simple sugars, with an R² of 0.99 for validation, training, and test. The model showed an average percentage error of 9.20% (calculated by comparing experimental data with the values obtained with the ANN’s). This innovative approach uses the power of artificial intelligence, specifically machine learning, to accurately measure the hydrolysis behavior of spent coffee grounds (SCG).</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

Artificial Neural Networks To Predict the Behavior of Sugars Obtained by Acid Hydrolysis Process from Spent Coffee Grounds

  • 期刊论文

  • 发表日期：

    2025年5月26日

查看全文
  • Matheus Costa Monteiro dos Santos,
  • Henrique Maziero Fogarin,
  • Sarha Lucia Murillo-Franco,
  • Jonas Paulino de Souza,
  • Erica Regina Filletti,
  • Kelly Johana Dussán

摘要

The utilization of lignocellulosic biomass, such as spent coffee grounds (SCG), for bioprocesses such as fermentation presents challenges due to the complex structure of its components. Effective pretreatment methods are needed to overcome these difficulties. This study investigates the use of artificial neural networks (ANN’s) to predict the behavior of the hydrolysis process (pretreatment), with a focus on efficient sugar extraction from SCG. By developing a feedforward neural network with input parameters such as temperature (130–190 °C), sulfuric acid concentration (0.5-2.0% v/v), solid/liquid ratio (1:4 − 1:40), and reaction time (20–120 min), the research aims to estimate the hydrolysis process using the Levenberg-Marquardt backpropagation algorithm in MATLAB R2024b software. With a neural model structure of four neurons in a single hidden layer, the model successfully predicts the amount of hemicellulosic sugars obtained from the hemicellulose fraction based on the input variables. The results demonstrate the effectiveness of the model in identifying the optimal conditions for converting polysaccharides in coffee waste into simple sugars, with an R² of 0.99 for validation, training, and test. The model showed an average percentage error of 9.20% (calculated by comparing experimental data with the values obtained with the ANN’s). This innovative approach uses the power of artificial intelligence, specifically machine learning, to accurately measure the hydrolysis behavior of spent coffee grounds (SCG).