Comparative Analysis of Machine Learning and Numerical Modeling for Combined Heat Transfer in Polymethylmethacrylate

Area of Science:

• Materials Science
• Thermal Engineering
• Computational Physics

Background:

• Polymethylmethacrylate (PMMA) is a polymer used in sensors and actuators, requiring accurate thermal analysis.
• Understanding combined conductive and radiative heat transfer is crucial for optimizing PMMA applications.
• Existing numerical methods can be computationally intensive for complex heat transfer problems.

Purpose of the Study:

• To compare the predictive accuracy and efficiency of a novel deep neural network (DNN) method against traditional numerical techniques for heat transfer in PMMA.
• To investigate the simultaneous effects of conductive and radiative heat transfer within a PMMA sample.
• To validate the performance of the DNN model using established metrics.

Main Methods:

• One-dimensional combined heat transfer was modeled using the implicit finite difference method.
• Kirchhoff transformation was applied to handle non-linear conductive heat transfer equations.
• A Long Short-Term Memory (LSTM) based deep neural network (DNN) was developed and implemented.
• Model performance was evaluated using Receiver Operating Characteristic (ROC) curves and confusion matrices.

Main Results:

• The LSTM-based DNN method demonstrated high accuracy and significantly reduced processing time compared to numerical solutions.
• Numerical analysis revealed similar gradient behaviors for conductive and radiative heat flux, with radiative flux being approximately twice as high.
• The total heat flux approached a constant value under approximated steady-state conditions.
• The DNN model accurately predicted transient temperature profiles, validated against existing studies.

Conclusions:

• The novel DNN (LSTM) method provides an accurate and efficient alternative for predicting combined heat transfer in PMMA.
• Artificial intelligence offers a powerful tool for accelerating thermal analysis in materials science and engineering.
• Further research can explore the application of this AI approach to more complex geometries and heat transfer scenarios.