Predicting the surface elastic parameters of soft solids using multi-output decision tree regressor

The research presented here suggests a novel technique for forecasting the surface elastic properties (parameters) of soft solids, specifically σo (surface tension), μs (lame second parameters), and λs (lame first parameter), using machine learning approaches. Soft solids, distinguished by their distinct mechanical properties, are essential in a wide range of applications, from biomedical devices to wearable electronics. Understanding and forecasting surface characteristics is critical for progressing in material design and engineering. While previous research has looked into components of surface elasticity, this study stands out by using machine learning algorithms to predict these properties. Data is generated using Abaqus Finite Element Analysis (FEA) software, resulting in a comprehensive dataset (340875 instances) for training and evaluation. The major goal is to create accurate predictive models that can estimate the surface parameters σo, μs, and λs which govern the material’s reaction to external pressures and strains. These surface properties are critical in characterising the behaviour of soft solids, making their prediction an important step towards improved material design and innovation. Considering the multi parameter nature of the prediction work, the paper adopts Multi-Output Decision Tree Regression (MODTR) as the machine learning algorithm of the choice. Various measures of performance, such as Mean Squared Error (MSE), Mean Absolute Error (MAE), Coefficient of Determination (R2 Score), and Root Mean Square Error (RMSE), are used to analyse results. While previous research has looked into surface elasticity, this paper is the first to use machine learning and finite element analysis data to predict surface characteristics. The results have the potential to advance the uses of soft materials by providing significant insights into their behaviour and permitting creative designs in domains ranging from electronics to biology.