Utilization of Machine Learning Algorithms in Optimizing Finite Element Modeling and Analysis

Today, an engineer’s success is largely based on their skill at employing Finite Element Analysis (FEA), a standard engineering modeling tool to assess the behavior of structures numerically. So far, improvements in computational power and FEA element formulations have supported significant advancements in this field. FEA can be used to provide accurate and fast answers to many engineering problems. However, FEA models are still associated with prohibitive time costs when solving complex scenarios. The research applies machine learning computing to FEM simulations for improved prediction outcomes through analysis of a database exceeding 1.6 million data points. The analysis uses step-by-step procedures which start with preprocessing and feature selection and normalization and move into model training before evaluation. The research evaluates four ML models, including Artificial Neural Networks (ANN), Deep Neural Networks (DNN), Random Forest (RF), and Light Gradient Boosting Machine (LGBM), by analyzing the performance metrics R-squared ($\mathbf{R}^{\mathbf{2}}$), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The ANN model demonstrates superior performance than other methods based on experimental results, which produce $\mathbf{R}^{\mathbf{2}}$ of $\mathbf{9 9. 9 4 \%}$ and MAE of 2.87 and RMSE of 10.58 showing accurate prediction with minimal error outputs. Experimental results show that ANN produces superior optimization of FEM than DNN, RF and LGBM. The model succeeds well in data prediction yet it encounters challenges regarding computation demands and its ability to address unfamiliar situations.