VIP20
本课程属于：005-Artificial Intelligence and Machine Learning 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

Graph Neural Networks for Industrial Fault Localization

30 Chapters
View Course Samples Online
01
Industrial fault localization landscape
challenges, conventional methods, and the role of graph neural networks.
Industriallocalization
02
Graph representation of industrial systems
nodes, edges, sensors, actuators, and process topology.
representationindustrial
03
Graph construction from time-series sensor data
correlation, causality, and physical connectivity.
constructiontime-series
04
Fundamentals of graph neural networks
message passing, aggregation, and readout operations.
networksaggregation
05
Spectral graph convolutions
from graph Fourier transform to ChebNet and GCN.
Spectralconvolutions
06
Spatial graph convolutions
GraphSAGE, GAT, and their suitability for fault localization.
convolutionsGraphSAGE
07
Graph autoencoders and variational graph autoencoders for unsupervised fault detection.
Graph autoencoders and variational graph autoencoders for unsupervised fault detection.
autoencodersvariational
08
Temporal graph neural networks
combining GNNs with RNNs, TCNs, and transformers for dynamic systems.
Temporalnetworks
09
Attention mechanisms in GNNs for identifying critical fault propagation paths.
Attention mechanisms in GNNs for identifying critical fault propagation paths.
Attentionmechanisms
10
Heterogeneous graph neural networks for multi-component industrial systems.
Heterogeneous graph neural networks for multi-component industrial systems.
Heterogeneousnetworks
11
Scalability of GNNs for large-scale industrial graphs
sampling, clustering, and distributed training.
ScalabilityGNNs
12
Data preprocessing and feature engineering for industrial graph-based fault localization.
Data preprocessing and feature engineering for industrial graph-based fault localization.
preprocessingindustrial
13
Handling class imbalance and noisy labels in industrial fault datasets.
Handling class imbalance and noisy labels in industrial fault datasets.
Handlingimbalance
14
Benchmark datasets and simulation environments for industrial fault localization.
Benchmark datasets and simulation environments for industrial fault localization.
Benchmarkdatasets
15
Evaluation metrics for fault localization
accuracy, precision, recall, F1, and localization error.
Evaluationlocalization
16
Case study
fault localization in chemical process plants using GNNs.
localizationchemical
17
Case study
fault localization in power grids with graph neural networks.
localizationnetworks
18
Case study
fault localization in manufacturing assembly lines using GNNs.
localizationmanufacturing
19
Case study
fault localization in water distribution networks with GNNs.
localizationdistribution
20
Case study
fault localization in HVAC systems using graph neural networks.
localizationHVAC
21
Integrating physical models with GNNs for physics-informed fault localization.
Integrating physical models with GNNs for physics-informed fault localization.
Integratingphysical
22
Explainability and interpretability of GNN-based fault localization
attention visualization and GNNExplainer.
Explainabilityinterpretability
23
Transfer learning and domain adaptation for GNN fault localization across industrial assets.
Transfer learning and domain adaptation for GNN fault localization across industrial assets.
Transferadaptation
24
Online and incremental learning for GNN-based fault localization in streaming data.
Online and incremental learning for GNN-based fault localization in streaming data.
incrementalGNN-based
25
Deployment considerations
edge computing, latency, and model compression for GNNs.
Deploymentconsiderations
26
Robustness of GNNs to adversarial attacks and sensor failures in industrial settings.
Robustness of GNNs to adversarial attacks and sensor failures in industrial settings.
RobustnessGNNs
27
Uncertainty quantification in GNN-based fault localization.
Uncertainty quantification in GNN-based fault localization.
Uncertaintyquantification
28
Human-in-the-loop systems
integrating operator feedback with GNN predictions.
Human-in-the-loopintegrating
29
Practical project
building an end-to-end GNN pipeline for a simulated industrial fault scenario.
buildingend-to-end
30
Future directions
graph foundation models, self-supervised learning, and digital twins for industrial fault localization.
directionsfoundation