Machine learning based approach for shape memory polymer behavioural characterization

Abstract

In this article we aim to combine video data analysis techniques, scalable machine learning, and Shape memory polymers (SMPs) materials to develop a model-based architecture for the advancement of rapid characterization of a novel material. Although artificially intelligent machines, e.g. soft robotics systems, with high flexibility have conquered the production line and other controlled, predictable environments, their use in complex real-world scenarios has to date remained limited. Newly discovered and experimented SMPs are increasingly being used for application solutions in automotive, aerospace, construction and commercial field. But being a nascent field there is little knowledge on the shape recovery behaviour of laminates with a SMP film and there are only methods reported in literature for quantifying the material behaviour. Through various experimental data gathering and predictive modelling it was established that proposed methodology can rapidly characterize novel materials. The proposed modelling workflow showed accuracy of 90% with 92% sensitivity and 94% specificity while predicting recovery behaviour of SMP body, showcasing high potential for data driven rapid characterisation of shape memory materials.

Keywords

• Shape memory polymer
• Machine learning
• Data driven modelling
• Data science