We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Machine Learning‐Driven Discovery of Thermoset Shape Memory Polymers With High Glass Transition Temperature Using Variational Autoencoders
Summary
A machine learning framework was developed to discover thermoset shape memory polymers with high glass transition temperatures, enabling their use in extreme high-temperature applications such as geothermal energy and aerospace. The approach rapidly identified candidate polymer formulations that outperformed traditionally synthesized materials.
ABSTRACT The discovery of high‐performance shape memory polymers (SMPs) with enhanced glass transition temperatures (Tg) is of paramount importance in fields such as geothermal energy, oil and gas, aerospace, and other high‐temperature applications, where materials are required to exhibit shape memory effect at extremely high‐temperature conditions. Here, we employ a novel machine learning framework that integrates transfer learning with variational autoencoders (VAE) to efficiently explore the chemical design space of SMPs and identify new candidates with high Tg values. We systematically investigate the effect of different latent space dimensions on the VAE model performance. Several machine learning models are then trained to predict Tg. We find that the SVM model demonstrates the highest predictive accuracy, with R 2 values exceeding 0.87 and a mean absolute percentage error as low as 6.43% on the test set. Through systematic molar ratio adjustments and VAE‐based fingerprinting, we discover novel SMP candidates with Tg values between 190°C and 200°C, suitable for high‐temperature applications. These findings underscore the effectiveness of combining VAEs and SVM for SMP discovery, offering a scalable and efficient method for identifying polymers with tailored thermal properties.
Sign in to start a discussion.
More Papers Like This
Leveraging Artificial Intelligence for Accelerated Polymer Synthesis and Design
This review examines AI-enabled advances in polymer informatics, focusing on machine learning and deep learning approaches for accelerating the design of application-specific polymeric materials across energy storage, production, and sustainable economy applications including recyclable and biodegradable polymers. The review highlights how AI-powered workflows are shortening the design-to-discovery cycle for next-generation polymer materials.
A data-driven approach for the assessment of the thermal stratification of reservoirs based on readily available data
Researchers used a data-driven machine learning approach to assess the thermal structural integrity of materials under variable conditions, providing predictive models that can reduce reliance on costly physical testing. The methodology has broader applications for materials used in environments with high thermal stress.
Recent Advances in Polymer Design through Machine Learning: A Short Review
This review examines recent advances in applying machine learning — including supervised learning, unsupervised learning, and artificial neural networks — to polymer informatics, covering property prediction, synthesis optimisation, and polymer classification across diverse applications from medicine to aerospace. The authors highlight how growing datasets and improving ML techniques are enabling more systematic and effective polymer design compared to traditional trial-and-error approaches.
Applied machine learning as a driver for polymeric biomaterials design
Researchers reviewed how machine learning could accelerate the design of new medical-grade polymers by predicting properties like biodegradability and biocompatibility, bypassing slow trial-and-error lab work. The main obstacle identified is the lack of standardized, publicly available data on medically relevant polymer characteristics needed to train reliable AI models.
Feasibility of Utilizing Machine Learning to Identify a More Sustainable Alternative to Polyester in Textiles
Researchers used machine learning trained on the PI1M polymer dataset of over one million compounds to identify sustainable alternatives to polyester in textiles, predicting key material properties including glass transition temperature, density, melting temperature, oxygen permeability, and bulk modulus. The study found that artificial neural networks could screen polymer candidates for environmental compatibility while maintaining performance characteristics comparable to polyester.