VIP20
本课程属于：009-AI and Machine Learning for Automotive CAE 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

Learning-Based Crash Test to CAE Model Discrepancy Analysis

30 Chapters
View Course Samples Online
01
Course overview
scope, learning objectives, and the role of crash test to CAE model discrepancy analysis in vehicle safety engineering.
objectivesCAE
02
Crash test data landscape
physical test setups, sensor channels, accelerometers, load cells, high-speed imaging, and data acquisition metadata.
landscapephysical
03
CAE model fundamentals for crashworthiness
explicit finite element formulation, LS-DYNA and PAM-CRASH workflows, material models, and contact definitions.
CAEcrashworthiness
04
Sources of discrepancy
modeling assumptions, material strain-rate effects, mesh discretization, boundary conditions, and test fixture variability.
discrepancymodeling
05
Data preprocessing for discrepancy analysis
signal filtering, time synchronization, resampling, and alignment of test and simulation time histories.
preprocessingdiscrepancy
06
Feature engineering for crash signals
peak force, intrusion, acceleration pulse, energy absorption, and derived injury criteria such as HIC and chest deflection.
intrusionacceleration
07
Statistical foundations for discrepancy quantification
bias, variance, correlation, confidence intervals, and hypothesis testing on paired test-CAE data.
Statisticalfoundations
08
Metrics for time-history comparison
CORA, ISO/TS 18571, magnitude-phase error, and objective rating methods for crash signals.
time-historycomparison
09
Introduction to machine learning for discrepancy analysis
supervised regression, classification, and anomaly detection paradigms.
discrepancyanalysis
10
Data-driven surrogate modeling
Gaussian process regression, polynomial chaos expansion, and Kriging for CAE response approximation.
Data-drivensurrogate
11
Neural network architectures for time-series discrepancy
MLP, CNN, LSTM, and temporal convolutional networks applied to crash pulses.
architecturestime-series
12
Physics-informed machine learning
embedding conservation laws, constitutive constraints, and boundary conditions into discrepancy models.
Physics-informedembedding
13
Transfer learning and domain adaptation
bridging test rig data and full-vehicle CAE models under limited physical test samples.
Transferadaptation
14
Uncertainty quantification in crash simulation
aleatory versus epistemic uncertainty, Monte Carlo sampling, and polynomial chaos.
Uncertaintyquantification
15
Bayesian calibration of CAE models
prior elicitation, likelihood construction from test data, and posterior inference for model parameters.
Bayesiancalibration
16
Sensitivity analysis and parameter screening
Sobol indices, Morris screening, and Shapley values for identifying dominant discrepancy drivers.
Sensitivityanalysis
17
Discrepancy localization in space and time
identifying critical regions and time windows where CAE deviates most from test results.
Discrepancylocalization
18
Case study
frontal impact discrepancy analysis using learning-based metrics and CAE model updating.
discrepancyanalysis
19
Case study
side impact and pole crash discrepancy analysis with sensor fusion and neural network classifiers.
discrepancyanalysis
20
Case study
rear impact and whiplash CAE correlation using Gaussian process and Bayesian methods.
whiplashCAE
21
Pedestrian protection and head impact discrepancy analysis
test vs. CAE correlation for HIC and head acceleration.
Pedestrianprotection
22
Battery electric vehicle crashworthiness
learning-based discrepancy analysis for battery pack intrusion and thermal runaway risk.
electriccrashworthiness
23
Material model calibration from test data
using machine learning to update Johnson-Cook, GISSMO, and crushable foam parameters.
Materialcalibration
24
Joint test-CAE data fusion
combining physical test measurements with simulation outputs for improved discrepancy prediction.
test-CAEcombining
25
Active learning and design of experiments
selecting informative test configurations to reduce CAE discrepancy efficiently.
experimentsselecting
26
Real-time discrepancy monitoring
online learning and streaming data approaches for crash test and simulation pipelines.
Real-timediscrepancy
27
Validation and verification of learning-based discrepancy models
cross-validation, holdout test sets, and robustness checks.
Validationverification
28
Regulatory and industrial standards
Euro NCAP, NHTSA, and IIHS correlation requirements and acceptance criteria.
Regulatoryindustrial
29
Software tools and implementation
Python, scikit-learn, TensorFlow, PyTorch, LS-DYNA, and CAE post-processing APIs.
Softwareimplementation
30
Capstone project
end-to-end learning-based discrepancy analysis workflow from raw crash test data to CAE model improvement recommendations.
Capstoneend-to-end