Scholarlink

<p>The viscosity of carbon dioxide-saturated polyethylene glycol (PEG) polymers plays a critical role in optimizing their performance across various industrial applications, underscoring the need for precise predictive models. This research leverages a Gradient Boosting Machine (GBM) model, enhanced by four sophisticated optimization techniques: Batch Bayesian Optimization (BBO), Evolution Strategies (ES), Bayesian Probability Improvement (BPI), and Gaussian Processes Optimization (GPO). The model is developed using a dataset of 183 experimental samples, with 90% designated for training and 10% for testing, incorporating essential input parameters such as temperature, pressure, and PEG molecular weight to predict viscosity. K-fold cross-validation is implemented during training to mitigate overfitting. The efficacy of each optimization technique is evaluated through computational runtime and performance metrics, including R-squared (R<sup>2</sup>), mean squared error (MSE), and average absolute relative error (AARE%). Correlation analysis indicates that pressure has the strongest negative correlation with viscosity (correlation coefficient: −&#xa0;0.72), followed by temperature (−&#xa0;0.49) and molecular weight (−&#xa0;0.47). Among the optimization approaches, GBM-ES achieves the highest precision, with an R<sup>2</sup> of 0.99956 for the training set and 0.9854365 for the test set, surpassing other methods. In terms of computational efficiency, GPO is the fastest, requiring 151.9 s, while BBO is the slowest at 216.9 s. Sensitivity analysis further clarifies the impact of each input parameter on viscosity, highlighting the robustness of data-driven approaches in tackling complex systems. These models provide dependable tools for viscosity prediction in carbon dioxide-saturated PEG polymers, minimizing the need for costly, time-intensive, and labor-heavy experimental procedures.</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

Modeling viscosity of carbon dioxide-saturated polyethylene glycol polymer

  • 期刊论文

  • 发表日期：

    2025年11月4日

查看全文
  • Haosheng Chen,
  • Ayat Hussein Adhab,
  • Bhavesh Kanabar,
  • Anupam Yadav,
  • M K Ranganathaswamy,
  • Rishabh Thakur,
  • Parveen Kumar,
  • Braj Krishna,
  • Morug Salih Mahdi,
  • Aseel Salah Mansoor,
  • Usama Kadem Radi,
  • Nasr Saadoun Abd,
  • Samim Sherzod,
  • Aseel Smerat

摘要

The viscosity of carbon dioxide-saturated polyethylene glycol (PEG) polymers plays a critical role in optimizing their performance across various industrial applications, underscoring the need for precise predictive models. This research leverages a Gradient Boosting Machine (GBM) model, enhanced by four sophisticated optimization techniques: Batch Bayesian Optimization (BBO), Evolution Strategies (ES), Bayesian Probability Improvement (BPI), and Gaussian Processes Optimization (GPO). The model is developed using a dataset of 183 experimental samples, with 90% designated for training and 10% for testing, incorporating essential input parameters such as temperature, pressure, and PEG molecular weight to predict viscosity. K-fold cross-validation is implemented during training to mitigate overfitting. The efficacy of each optimization technique is evaluated through computational runtime and performance metrics, including R-squared (R2), mean squared error (MSE), and average absolute relative error (AARE%). Correlation analysis indicates that pressure has the strongest negative correlation with viscosity (correlation coefficient: − 0.72), followed by temperature (− 0.49) and molecular weight (− 0.47). Among the optimization approaches, GBM-ES achieves the highest precision, with an R2 of 0.99956 for the training set and 0.9854365 for the test set, surpassing other methods. In terms of computational efficiency, GPO is the fastest, requiring 151.9 s, while BBO is the slowest at 216.9 s. Sensitivity analysis further clarifies the impact of each input parameter on viscosity, highlighting the robustness of data-driven approaches in tackling complex systems. These models provide dependable tools for viscosity prediction in carbon dioxide-saturated PEG polymers, minimizing the need for costly, time-intensive, and labor-heavy experimental procedures.