Abstract:The efficient operation of industrial robots relies on reliable anomaly detection systems, but the problem of missing data caused by sensor failures, data transmission errors, and system maintenance or updates in the industrial field poses a serious challenge to their accuracy and may lead to delays in recognizing critical faults, which in turn threatens production safety and causes economic losses. To address this problem, this paper constructs an Anomaly Detection Method with Combined Data Imputation and Feature Collaboration (AD-CIFC), which firstly employs a comprehensive imputation strategy to impute the missing data in order to better adapt to the natural flow of time series and complex patterns; then two graph attention mechanisms are proposed to be used in parallel to capture the relationship between nodes more comprehensively by introducing synergies, thus effectively utilizing the correlation between features to make the model more sensitive; finally, two collaborative learning models are proposed to be used in parallel to more comprehensively take into account the strong correlation between features in the data, and a collaborative loss function computation method is proposed, thus improving the overall detection accuracy and discrimination ability. Through anomaly detection experiments on industrial robots with different levels of missing joint data, it is demonstrated that the AD-CIFC model demonstrates a performance that matches or even outperforms other mainstream anomaly detection models in complete fault data tests, while the model performance using the comprehensive imputation strategy to impute in the joints is about 25 % higher on average compared to the model performance of traditional linear interpolation methods, and meanwhile, in the case of missing data at the minor to severe levels, the model still demonstrates robustness beyond traditional methods, which is very suitable for industrial robot production lines.

Unsupervised Anomaly Detection of Industrial Robots Using Sliding-Window Convolutional Variational Autoencoder

Industrial Robot Vibration Anomaly Detection Based on Sliding Window One-Dimensional Convolution Autoencoder

The Study on Multivariate Time Series Anomaly Detection Via Convolutional Variational Autoencoder

Unsupervised Machine Anomaly Detection Using Autoencoder and Temporal Convolutional Network.

Multivariate Time Series Anomaly Detection: Missing Data Handling and Feature Collaborative Analysis in Robot Joint Data

Cognitive visual anomaly detection with constrained latent representations for industrial inspection robot

Anomaly Detection Using Autoencoder Reconstruction upon Industrial Motors

Unsupervised Deep Anomaly Detection for Industrial Multivariate Time Series Data

An Unsupervised Method for Industrial Image Anomaly Detection with Vision Transformer-Based Autoencoder

An Unsupervised Anomaly Detection Approach Based On Industrial Big Data

Adaptive Anomaly Detection in Industrial Systems: an EVT-DTS Approach with LSTM Autoencoders

An Anomaly Detection Method for Nonlinear Industrial Process Using Sparse Stacked Denoising Autoencoder.

Hybrid robust convolutional autoencoder for unsupervised anomaly detection of machine tools under noises

Robust Unsupervised Anomaly Detection Via Multi-Time Scale DCGANs with Forgetting Mechanism for Industrial Multivariate Time Series

A deep learning approach for anomaly detection based on SAE and LSTM in mechanical equipment

Memory-Enhanced Spatial-Temporal Encoding Framework for Industrial Anomaly Detection System

Unsupervised Learning for Machinery Adaptive Fault Detection Using Wide-Deep Convolutional Autoencoder with Kernelized Attention Mechanism

Deep Anomaly Detection for Industrial Systems: a Case Study

Lightweight Long Short-Term Memory Variational Auto-Encoder for Multivariate Time Series Anomaly Detection in Industrial Control Systems

Data-driven Unsupervised Anomaly Detection of Manufacturing Processes with Multi-Scale Prototype Augmentation and Multi-Sensor Data

An Attention-Based ConvLSTM Autoencoder with Dynamic Thresholding for Unsupervised Anomaly Detection in Multivariate Time Series