Recurrent Neural Network for Accelerated Finite Element Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional finite element analysis (FEA) for vehicle component design is time-consuming and computationally intensive, limiting its ability to rapidly consider dynamic and variable load conditions.
Innovation Solution
The method involves providing a graphical representation of a finite element mesh as input to a recurrent neural network (RNN), which generates a time series mesh (TSM) to facilitate accelerated, variable load FEA, allowing for iterative design adaptations based on the output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional finite element analysis is used for vehicle component design, then accurate structural analysis can be achieved, but the analysis process is time-consuming and computationally intensive
Solution Approach 1:
The patent creates a neural network model that learns from traditional FEA results and creates a computational copy that can predict structural responses without running full FEA simulations. The neural network is trained on FEA data and then used to generate predictions that approximate FEA accuracy at fraction of the computational cost and time
Solution Approach 2:
The patent performs preliminary action by pre-training the neural network model on comprehensive FEA data covering various loading conditions and structural configurations. This preliminary training phase stores learned patterns and relationships that enable rapid prediction during actual design iterations, avoiding the need to run time-consuming FEA analyses for every design evaluation
2Adaptability or versatility
If traditional finite element analysis is used for vehicle component design, then comprehensive load condition analysis can be achieved, but intensive computational resources are required
Solution Approach 1:
The neural network creates a computational model that copies the essential physics and structural behavior patterns learned from extensive FEA simulations. Once trained, this copied model can evaluate any load condition within its training domain using minimal computational resources, maintaining versatility while dramatically reducing energy and computational requirements
Solution Approach 2:
The patent transforms the computational approach by changing from direct numerical solution of partial differential equations (FEA) to statistical inference based on learned parameter relationships. The neural network learns mappings between input parameters (geometry, material properties, loading conditions) and output responses (stresses, displacements, natural frequencies), enabling rapid evaluation across diverse load conditions
3Loss of information
If traditional finite element analysis is used, then detailed structural insights can be obtained, but the design cycle time is extended
Solution Approach 1:
The neural network model copies the essential structural insight capabilities of FEA by learning to predict key output parameters including stress distributions, displacement fields, and natural frequencies. This copied knowledge enables designers to obtain detailed structural insights during rapid design iterations without waiting for time-consuming FEA analyses
Solution Approach 2:
The patent performs preliminary action by pre-computing and storing structural response patterns through FEA simulations during the training phase. The neural network learns from this preliminary comprehensive analysis and then applies the learned patterns to generate rapid predictions, maintaining structural insight quality while enabling fast design cycle iterations
Data Source
AI summary
Devices, systems, and methods accelerated, variable load finite element analysis for vehicle design can provide a graphical representation of a finite element mesh and enter the graphical representation as an input to a recurrent neural network (RNN). Such solutions can develop a time series mesh (TSM) as an output from the RNN. Design of a component of the vehicle can be adapted based on the output.