Page 72 - IJAMD-2-3

International Journal of AI for
Materials and Design Optimization of membrane shrinkage and stability

Figure 1. Scattered plot of measured shrinkage ratios in the transverse direction and rotational direction of each membrane in the literature, with clusters
11
highlighted and inner confidence contours
However, current AI applications largely focus on achieving a data-driven and uncertainty-aware paradigm, offering
performance targets, often neglecting stability, especially both theoretical and practical guidance for designing
when data are limited. tunable and stimulus-responsive electrospun membranes.
To address these challenges, this study proposes a hybrid 2. Methodology
approach that integrates machine learning with Monte
Carlo simulation to model and analyze the shrinkage This study presents a data-driven framework for optimizing
behavior and stability of electrospun membranes based electrospinning processes, which focuses on shrinkage
on a limited experimental dataset. A supervised learning behavior and stability. The proposed methodology consists
model is first developed using experimental data (from of three main components: (i) dataset construction
literature; Figure 1) to capture the non-linear relationships based on experimental measurements under controlled
11
between processing parameters and shrinkage ratios under parameter variations; (ii) development and interpretation
multifactorial conditions. On the other hand, a shrinkage of machine learning models for predicting shrinkage ratios
stability coefficient is introduced to quantify the sensitivity and their stability; and (iii) a Monte Carlo simulation-based
of shrinkage to parameter perturbations, and the Monte strategy for identifying process conditions that satisfy target
Carlo simulation is employed to characterize its statistical shrinkage values, while ensuring minimal variability.
distribution. This framework enables the identification of a
controllable processing parameter space that ensures both 2.1. Dataset construction
target biaxial shrinkage performance and robustness in To develop a predictive and robust model for shrinkage
biaxial shrinkage. The proposed methodology establishes behavior in electrospun membranes, we constructed an

Volume 2 Issue 3 (2025) 66 doi: 10.36922/IJAMD025260022