Scholarlink

<p>Adaptive neuro-fuzzy is a hybrid learning algorithm used for function approximation problems; this technique consolidates both artificial neural networks and response surface methodology to optimize the individual and interactive effects of independent parameters. The parameters under consideration include percentage of sulfidity, Kappa number, and refining intensity. These are utilized as inputs for the development of models based on artificial intelligence. The proposed response surface scheme is employed in the context of mechanical and optical properties of paper pulp. The properties of interest include tear index, opacity, and permeability, which are the outputs of the process. The chemical structure of Kraft pulp was characterized through the utilization of Fourier transform infrared spectroscopy and X-ray diffraction. Subsequently, a dynamometer, tear meter, chip counter, micrometer, and densometer were utilized to assess the quality of pulp during the refining process. The findings indicate that the adaptive neuro-fuzzy model provides optimal conditions, characterized by a % sulfidity of 20%, a kappa number of 16.20, and a refining intensity of 33.00°SR. It is predicted that these conditions will yield specific responses, including a tear index of 43.9&#xa0;g/m<sup>2</sup>, an opacity of 84.8%, and a permeability of 4.6 cm<sup>3</sup>/s&#xa0;Pa&#xa0;m<sup>2</sup>. The experimental results obtained were consistent with the theoretical findings, as evidenced by the following parameters: The T.I. was found to be 43&#xa0;g/Nm<sup>2</sup>, the Op was 84.5%, and the A.P. was 4.25 cm<sup>3</sup>/s·Pa·m<sup>2</sup>. The findings of this study indicated that artificial intelligence exhibited efficacy in predicting these properties, as evidenced by the low residual sum of squares and mean square error for the predicted values. Finally, the structural characterization facilitated the identification of the primary characteristic bands of the refined pulp and the crystalline index of the pulp, which was determined to be 51.00%.</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

Harnessing AI-RSM to improve the mechanical and optical properties of Kraft pulp during the refining process

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Youssef Elrhayam,
  • Fatima Ezzahra Bennani,
  • Mohamed Allaoui,
  • Omar Berradi,
  • Brahim Sallek,
  • Abderrahim El Bachiri

摘要

Adaptive neuro-fuzzy is a hybrid learning algorithm used for function approximation problems; this technique consolidates both artificial neural networks and response surface methodology to optimize the individual and interactive effects of independent parameters. The parameters under consideration include percentage of sulfidity, Kappa number, and refining intensity. These are utilized as inputs for the development of models based on artificial intelligence. The proposed response surface scheme is employed in the context of mechanical and optical properties of paper pulp. The properties of interest include tear index, opacity, and permeability, which are the outputs of the process. The chemical structure of Kraft pulp was characterized through the utilization of Fourier transform infrared spectroscopy and X-ray diffraction. Subsequently, a dynamometer, tear meter, chip counter, micrometer, and densometer were utilized to assess the quality of pulp during the refining process. The findings indicate that the adaptive neuro-fuzzy model provides optimal conditions, characterized by a % sulfidity of 20%, a kappa number of 16.20, and a refining intensity of 33.00°SR. It is predicted that these conditions will yield specific responses, including a tear index of 43.9 g/m2, an opacity of 84.8%, and a permeability of 4.6 cm3/s Pa m2. The experimental results obtained were consistent with the theoretical findings, as evidenced by the following parameters: The T.I. was found to be 43 g/Nm2, the Op was 84.5%, and the A.P. was 4.25 cm3/s·Pa·m2. The findings of this study indicated that artificial intelligence exhibited efficacy in predicting these properties, as evidenced by the low residual sum of squares and mean square error for the predicted values. Finally, the structural characterization facilitated the identification of the primary characteristic bands of the refined pulp and the crystalline index of the pulp, which was determined to be 51.00%.