Prediction of Tensile Property of Polycarbonate/Acrylonitrile‐Butadiene‐Styrene (PC/ABS) Blend Under UV‐Accelerated Aging Based on Hybrid DT‐MLP Model

Abstract

Polycarbonate/acrylonitrile‐butadiene‐styrene (PC/ABS) blends are widely utilized engineering plastics known for their distinctive advantages in automotive, electronic, and electrical applications. However, their long‐term durability is significantly limited by their susceptibility to harsh environmental degradation. In this study, systematic UV‐accelerated aging tests were conducted on PC/ABS blends with five different compositions to establish a comprehensive dataset encompassing key tensile properties: tensile strength, elastic modulus, and elongation at break. To predict the evolution of tensile performance under environmental exposure, four machine learning (ML) models—Decision Tree (DT), Random Forest (RF), Multilayer Perceptron (MLP), and a novel hybrid MLP‐DT model—were constructed and optimized. Experimental results revealed that UV‐accelerated aging induced severe degradation in tensile strength and elongation at break, while the elastic modulus remained relatively stable. Notably, the proposed hybrid MLP‐DT model achieved the highest predictive accuracy for tensile strength (R 2 = 0.96) and elongation at break (R 2 = 0.75), outperforming individual ML algorithms. Furthermore, feature importance analysis identified aging duration, temperature, UV irradiance, and resin composition as the dominant factors governing the degradation behavior. This work offers a robust, data‐driven approach for predicting the durability of PC/ABS blends, providing valuable insights for formulation optimization and lifespan assessment in engineering applications.