Abstract:In this paper, we explore point-cloud based deep learning models to analyze numerical simulations arising from finite element analysis. The objective is to classify automatically the results of the simulations without tedious human intervention. Two models are here presented: the Point-Net classification model and the Dynamic Graph Convolutional Neural Net model. Both trained point-cloud deep learning models performed well on experiments with finite element analysis arising from automotive industry. The proposed models show promise in automatizing the analysis process of finite element simulations. An accuracy of 79.17% and 94.5% is obtained for the Point-Net and the Dynamic Graph Convolutional Neural Net model respectively.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to reduce human intervention by automatically analyzing the simulation results generated by Finite Element Analysis (FEM). Specifically, the author aims to develop a deep - learning model based on point clouds for automatically classifying the results of FEM simulations without cumbersome manual inspection and preliminary verification. ### Problem Background The Finite Element Method (FEM) is widely used to simulate physical phenomena in the real world, such as computational mechanics, fluid mechanics, and heat conduction. In FEM, mathematical equations are reformulated through variational formulas and discretized in time and space. The discretization is carried out on a grid, and the quality of the grid directly affects the quality of the approximate solution. However, the simulation results generated by FEM usually require manual inspection, which is a time - consuming process that requires human intervention. ### Research Objectives To automate this process, the paper explores two deep - learning models based on point clouds: 1. **Point - Net Classification Model** 2. **Dynamic Graph Convolutional Neural Net (DGCNN)** The goal of these models is to automatically classify "normal" and "abnormal" results from FEM simulation results. A specific application case is the finite - element simulation of acoustic noise inside a car, and the results are divided into two categories according to the high - sound - pressure areas near the car windows: high - noise level and low - noise level. ### Data Set The data set consists of 72 finite - element grids, and each grid is marked as "normal" or "abnormal" by manual visual inspection. The data set is divided into a training set (66.6%) and a test set (33.3%) for training and evaluating the models. ### Model Performance The experimental results show that both models perform well in the classification task: - The accuracy of the Point - Net model on the test set is 79.17%. - The accuracy of the DGCNN model on the test set is 94.5%. The DGCNN model outperforms the Point - Net model in the classification task because it can better capture local features. ### Conclusion This research shows that deep - learning models based on point clouds can effectively automate the analysis process of FEM simulation results. In particular, the DGCNN model performs excellently in the classification task and has high accuracy. This provides strong support for further improving the automation degree of FEM simulation result analysis in the future.

Point-Cloud-based Deep Learning Models for Finite Element Analysis

Automated Generation of FE Model for Digital Twin of Concrete Structures from Segmented 3D Point Cloud

Point Cloud-Based Dimensional Quality Assessment of Precast Concrete Components Using Deep Learning

DeepFEA: Deep Learning for Prediction of Transient Finite Element Analysis Solutions

Simulation of hyperelastic materials in real-time using Deep Learning

Bridging finite element and deep learning: High-resolution stress distribution prediction in structural components

Finite Element Network Analysis: A Machine Learning based Computational Framework for the Simulation of Physical Systems

Representing engineering design changes in finite element models using directed point cloud autoencoders

A Point-Cloud Deep Learning Framework for Prediction of Fluid Flow Fields on Irregular Geometries

Learning-Based Finite Element Methods Modeling for Complex Mechanical Systems

A finite-element-informed neural network for parametric simulation in structural mechanics

Detection of Plausibility and Error Reasons in Finite Element Simulations with Deep Learning Networks

Application of Machine Learning and Deep Learning in Finite Element Analysis: A Comprehensive Review

On point cloud failure criterion predictions

Neural network-augmented differentiable finite element method for boundary value problems

Data-Driven Finite Elements Methods: Machine Learning Acceleration of Goal-Oriented Computations

Deep Learning-Based Quality Assessment Of 3d Point Clouds Without Reference

Interfacing finite elements with deep neural operators for fast multiscale modeling of mechanics problems

Point Cloud Deep Learning Network Based on Local Domain Multi-Level Feature

A novel parallel classification network for classifying three-dimensional surface with point cloud data