Abstract:Machine learning (ML) has been extensively adopted for the online sensing-based monitoring in advanced manufacturing systems. However, the sensor data collected under abnormal states are usually insufficient, leading to significant data imbalanced issue for supervised machine learning. A common solution for this issue is to incorporate data augmentation technique, i.e., augmenting the available abnormal states data (i.e., minority samples) via synthetic generation. To generate the high-quality minority samples effectively, it is vital to learn the underlying distribution of the abnormal states data. In recent years, the generative adversarial network (GAN)-based approaches become popular to learn data distribution as well as perform data augmentation. However, in practice, the quality of generated samples from GAN-based data augmentation may vary drastically. In addition, the sensor signals are collected sequentially by time from the manufacturing systems, which means the consideration of sequential information is also very important in data augmentation. To address these limitations, inspired by the multi-head attention mechanism, this paper proposed an attention-stacked GAN (AS-GAN) architecture for the sensor data augmentation of online monitoring in advanced manufacturing. In this proposed AS-GAN, a new attention-stacked framework is incorporated to strengthen the generator in GAN with the learning capability of considering sequential information. Furthermore, the developed attention-stacked framework also greatly helps to improve the quality of generated sensor signals. The case studies conducted in additive manufacturing also successfully validate the effectiveness of AS-GAN to augment high-quality artificial multi-channel sensor signals for online monitoring of manufacturing systems.

Virtual sensor-based imputed graph attention network for anomaly detection of equipment with incomplete data

Data anomaly detection for structural health monitoring using a combination network of GANomaly and CNN

Accurate Anomaly Detection Leveraging Knowledge-enhanced GAT

A Graph Embedded In Graph Framework With Dual-sequence Input For Efficient Anomaly Detection of Complex Equipment Under Insufficient Samples

Anomaly Detection for Data from Unmanned Systems via Improved Graph Neural Networks with Attention Mechanism

Hybrid Anomaly Detection via Multihead Dynamic Graph Attention Networks for Multivariate Time Series

Anomaly Detection with Graph Attention Network for Multimodal IoT Data Monitoring

AddAG-AE: Anomaly Detection in Dynamic Attributed Graph Based on Graph Attention Network and LSTM Autoencoder

Anomaly detection by using a combination of generative adversarial networks and convolutional autoencoders

Self-Supervised Variational Graph Autoencoder for System-Level Anomaly Detection

AD-Graph: Weakly Supervised Anomaly Detection Graph Neural Network

MST-GAT: A Multimodal Spatial-Temporal Graph Attention Network for Time Series Anomaly Detection

Anomaly Detection for Internet of Things Time Series Data Using Generative Adversarial Networks With Attention Mechanism in Smart Agriculture

MFGAN: Multimodal Fusion for Industrial Anomaly Detection Using Attention-Based Autoencoder and Generative Adversarial Network

GTAD: Graph and Temporal Neural Network for Multivariate Time Series Anomaly Detection

AddGraph: Anomaly Detection in Dynamic Graph Using Attention-based Temporal GCN

Attention-stacked Generative Adversarial Network (AS-GAN)-empowered Sensor Data Augmentation for Online Monitoring of Manufacturing System

DAGAD: Data Augmentation for Graph Anomaly Detection

Anomaly-GAN: A data augmentation method for train surface anomaly detection

Graph neural network-based anomaly detection for river network systems

Graph attention U-Net to fuse multi-sensor signals for long-tailed distribution fault diagnosis