Making Finite Element Modeling Choices Using Decision-Tree-Based Fuzzy Inference System

This work presents a decision-tree-based Fuzzy Inference System (FIS) for making optimal choices in the development of reduced-order finite element (FE) models, in our case, shell-solid multifidelity models. FE analysis is widely used to simulate the real-world response of complex engineering structures and requires a high level of expertise for making a priori modeling decisions. Many times, these decisions are quite subjective in nature and lead to significant analyst-to-analyst variability, which in turn leads to considerable differences in engineering solutions. An expert system that recommends optimal modeling choices would notably reduce such variability. Expert systems use a knowledge base, developed by a subject matter expert, which is not always easy for complex structures. This work assesses the potential of interpretable machine learning (decision trees) to create data-driven rules that could be used by a FIS to make modeling choices for a multifidelity T-joint model. Specifically, the FIS takes the structural geometry and desired accuracy as inputs and infers the optimal two-dimensional/three-dimensional topology distribution. Once developed, the FIS is able to provide real-time optimal choices along with interpretability that fosters analysts' confidence. Potential improvements to the presented framework that can enable its application to complex and nonlinear problems are discussed.