VIP20

本课程属于：035-AI for CAE Simulation GNN and VVUQ English Technical Courses · 共 20 门相关课程，组合学习更完整。

查看所属VIP全部课程 →

← 返回所属目录咨询 / 购买

Active Learning and Adaptive LS-DYNA DOE Optimization

30 Chapters

View Course Samples Online

Course overview

scope, active learning and adaptive DOE concepts, LS-DYNA integration context, and engineering motivation.

LS-DYNA DOE fundamentals

design variables, responses, keywords *PARAMETER and *DEFINE_CURVE, and simulation workflow for optimization.

Design of experiments for LS-DYNA

full factorial, Latin hypercube, and space-filling designs; sampling constraints and run matrix generation.

experimentsLS-DYNA

Surrogate modeling for crashworthiness and impact simulations

Kriging, polynomial response surface, and radial basis functions with LS-DYNA data.

Surrogatemodeling

Adaptive sampling strategies

infill criteria such as expected improvement, probability of improvement, and lower confidence bound for LS-DYNA responses.

Adaptivesampling

Active learning fundamentals

query strategies, uncertainty quantification, and batch vs sequential selection for simulation-based optimization.

strategiesuncertainty

LS-DYNA run automation

parameter substitution, job submission, and result extraction using LS-DYNA solver scripts and *DATABASE cards.

LS-DYNAautomation

Response extraction and post-processing

parsing d3plot, binout, and ASCII files; computing metrics like HIC, intrusion, and acceleration peaks.

Responseextraction

Building a closed-loop adaptive DOE workflow

Python orchestration, LS-DYNA execution, surrogate update, and convergence monitoring.

Buildingclosed-loop

Initial DOE design for LS-DYNA

choosing sample size, variable ranges, and design quality metrics (maximin, discrepancy) for crash models.

Surrogate model validation

cross-validation, error metrics (RMSE, R2), and residual analysis for LS-DYNA response surfaces.

Surrogatevalidation

Adaptive refinement with Kriging

fitting variograms, hyperparameter tuning, and infill for global accuracy in impact simulations.

Adaptiverefinement

Constrained optimization with active learning

handling LS-DYNA constraints (mass, intrusion) via feasibility-aware infill criteria.

Constrainedoptimization

Multi-response adaptive DOE

Pareto front discovery, desirability functions, and trade-off analysis for crashworthiness metrics.

Multi-responseadaptive

adaptive DOE for frontal crash pulse optimization using LS-DYNA and active learning.

side impact intrusion minimization with adaptive sampling and LS-DYNA surrogate models.

intrusionminimization

pedestrian protection head impact HIC optimization using active learning and LS-DYNA DOE.

pedestrianprotection

battery pack crash safety optimization with adaptive LS-DYNA DOE and multi-physics responses.

optimizationadaptive

Noise and robustness in LS-DYNA DOE

stochastic simulations, variability quantification, and robust design criteria.

robustnessLS-DYNA

Adaptive DOE for transient responses

time-series surrogate modeling and infill for LS-DYNA time-history outputs.

Efficient global optimization (EGO) with LS-DYNA

expected improvement maximization and convergence criteria.

Efficientoptimization

Multi-fidelity adaptive DOE

combining LS-DYNA coarse and fine meshes for cost-effective optimization.

Multi-fidelityadaptive

Batch adaptive sampling for high-performance computing

parallel LS-DYNA runs and batch infill strategies.

adaptivesampling

Handling failed LS-DYNA runs

error detection, classification, and adaptive sampling with failure-aware surrogates.

HandlingLS-DYNA

Sensitivity analysis in adaptive DOE

variance-based methods and screening with LS-DYNA surrogate models.

Sensitivityanalysis

Adaptive DOE for shape and topology optimization

morphing, *DEFINE_TRANSFORMATION, and variable parameterization in LS-DYNA.

Machine learning surrogates for LS-DYNA

neural networks and Gaussian processes in adaptive DOE loops.

surrogatesLS-DYNA

Workflow integration and reproducibility

version control, experiment tracking, and automated reporting for LS-DYNA DOE.

Workflowintegration

Scalability and cloud deployment

containerized LS-DYNA runs, job schedulers, and adaptive DOE at scale.

Scalabilitydeployment

Capstone project

end-to-end adaptive LS-DYNA DOE optimization for a real-world crashworthiness problem with active learning.

Capstoneend-to-end