VIP20
本课程属于：027-CAE Data Analytics & Crash Safety Engineering 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

Machine Learning Surrogate Models for CAE Design Optimization

30 Chapters
View Course Samples Online
01
Introduction to CAE design optimization
role of simulation, computational bottlenecks, and motivation for surrogate modeling.
CAEoptimization
02
Surrogate modeling fundamentals
definition, workflow, key terminology, and taxonomy of machine learning surrogates.
Surrogatemodeling
03
Design of experiments for CAE
sampling strategies (Latin hypercube, Sobol, factorial), space-filling criteria, and sample size considerations.
experimentsCAE
04
Data generation from CAE solvers
automation, solver coupling, batch execution, and handling of simulation failures.
generationCAE
05
Data preprocessing for surrogate models
cleaning, normalization, feature scaling, and outlier detection in simulation datasets.
preprocessingsurrogate
06
Feature engineering for CAE
parameterization of geometry, boundary conditions, and material properties for ML input.
CAEparameterization
07
Dimensionality reduction
PCA, autoencoders, and active subspaces for high-dimensional CAE design spaces.
Dimensionalityreduction
08
Regression surrogates I
polynomial response surfaces, radial basis functions, and Gaussian process regression.
Regressionsurrogates
09
Regression surrogates II
support vector regression, decision trees, and ensemble methods (random forests, gradient boosting).
Regressionsurrogates
10
Neural network surrogates
MLP, deep learning architectures, and training strategies for CAE regression tasks.
surrogatesMLP
11
Advanced neural surrogates
convolutional neural networks for field prediction and physics-informed neural networks.
Advancedsurrogates
12
Surrogate model validation
cross-validation, error metrics (RMSE, R², MAE), and generalization assessment.
Surrogatevalidation
13
Uncertainty quantification in surrogates
Gaussian process confidence intervals, Bayesian neural networks, and ensemble variance.
Uncertaintyquantification
14
Active learning and adaptive sampling
infill criteria, expected improvement, and efficient design space exploration.
adaptivesampling
15
Multi-fidelity surrogate modeling
combining low- and high-fidelity CAE simulations, co-kriging, and transfer learning.
Multi-fidelitysurrogate
16
Surrogate-assisted optimization
coupling surrogates with genetic algorithms, gradient-based optimizers, and Pareto methods.
Surrogate-assistedoptimization
17
Constraint handling in surrogate-based optimization
feasibility classification, penalty methods, and constrained expected improvement.
Constrainthandling
18
Surrogate modeling for transient and dynamic CAE
time-series prediction, recurrent networks, and temporal convolutional networks.
Surrogatemodeling
19
Surrogate modeling for nonlinear and failure analysis
classification surrogates for crash, fracture, and buckling predictions.
Surrogatemodeling
20
Surrogate modeling for multidisciplinary optimization
coupling multiple physics surrogates and managing coupled constraints.
Surrogatemodeling
21
Case study
aerodynamic shape optimization using ML surrogates for drag and lift prediction.
aerodynamicoptimization
22
Case study
structural topology optimization with surrogate-assisted evolutionary algorithms.
structuraltopology
23
Case study
thermal management design using Gaussian process surrogates for heat sink performance.
managementGaussian
24
Case study
crashworthiness design optimization with classification and regression surrogates.
crashworthinessoptimization
25
Deployment of surrogate models
integration with CAD/CAE workflows, real-time prediction, and API development.
Deploymentsurrogate
26
Surrogate model lifecycle management
versioning, retraining, drift detection, and maintenance in production.
Surrogatelifecycle
27
Scalability and HPC
parallel data generation, distributed training, and surrogate deployment on cloud platforms.
ScalabilityHPC
28
Best practices and pitfalls
overfitting, extrapolation risks, data leakage, and interpretability of surrogates.
practicespitfalls
29
Software tools and frameworks
Python libraries (scikit-learn, GPy, PyTorch), surrogate modeling toolkits, and CAE integration.
Softwareframeworks
30
Future trends
deep generative surrogates, digital twins, and integration of surrogates with real-time optimization.
generativesurrogates