Skip to content

The project focused on predicting polymer composite thermal conductivity using SVR algorithm, PSO optimization, and LOOCV.

Notifications You must be signed in to change notification settings

rishaviitd/thermal_conductivity_prediction_ML

Repository files navigation

Project Overview

In collaboration with Professor Divya Nayar, this project aimed to develop a robust model for predicting the thermal conductivity of polymer-based composites. The focus was on understanding how varying mass fractions of fillers, particularly polyethylene and polystyrene, influence the thermal conductivity of the composite material.

Contributions

  • 📅Situation: The accuracy of existing theoretical models for predicting the thermal conductivity of polymer-based composites was found to be insufficient. The project commenced in February 2023 as part of a course project under the guidance of Prof. Divya Nayar.
  • 🎯Task: Our task was to establish a predictive model capable of predicting more accurately than theoritical models, determining the thermal conductivity of polymer-based composites across different mass fractions of fillers.
  • ⚙️Action:
    • Utilized support vector regression (SVR) as the primary modeling technique.
    • Employed particle swarm optimization (PSO) for parameter tuning to enhance the model's performance and predictive accuracy.
  • 📈Result:
    • Achieved a superior generalization ability with the SVR model, showcasing a remarkable R-squared accuracy of 0.954.
    • The model outperformed existing theoretical frameworks, demonstrating its effectiveness in predicting the thermal conductivity of polymer-based composites.
    • Presented the findings to a group of PhD researchers, professors, and colleagues, receiving positive feedback and valuable insights.

About

The project focused on predicting polymer composite thermal conductivity using SVR algorithm, PSO optimization, and LOOCV.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published