Enhancing O-Ring Performance Prediction Using Machine Learning

<div class="section abstract"><div class="htmlview paragraph">O-rings are essential components in engineering products as they ensure leak-proof sealing and hinders amalgamation of various fluids in the system. O-rings in general have lot of factors that go into deciding the right design for a system. With the help of FEA, O-ring design is varied to ensure optimal results. However, this process is time and resource consuming. Considering this situation, an alternative approach to predict the outcome with the help of DOE study is chosen in this paper. It leverages the Machine Learning models to predict the output parameters effectively with less resources. With the help of performance parameters, this paper proposes a comparison of various native ML models like Linear Regression, Random Forrest, SVM, KNN, Boosting, Artificial Neural Networks and Kriging [<span class="xref">7</span>]. The Goal is to systematically compare the prediction performance of various models based on bootstrapping and hypothesis testing techniques to identify the most effective approach. This research aims to enhance the efficiency and reliability of predictive modelling and help in virtual validation of O-rings.</div></div>