Abstract:In modern industrial systems, components have complex interactions with each other, which makes it become a challenging task to identify the operational conditions of industrial systems. Considering that an industrial system, the embedded components and their interactions can be expressed as nodes and edges in a graph, respectively. Therefore, graph representation algorithms are powerful tools for fault diagnosis of industrial systems. As one of the most commonly used graph representation algorithms, Graph Neural Networks (GNN) mainly follow the law of “learning to attend”. GNN extract training data features, learn the statistical correlations between features and labels, resulting in the attended graph favoring for accessing non-causal features as a shortcut for prediction. This shortcut feature is unstable and depends on the data distribution characteristics in the training dataset, which reduces the generalization ability of the classifier. By performing the causal analysis of GNN modeling for graph representation, the results show that shortcut features act as confounding factors between causal features and predictions, causing classifiers to learn wrong correlations. Therefore, to discover patterns of causality and weaken the confounding effects of shortcut features, a Causal-Trivial Attention Graph Neural Network (CTA-GNN) strategy is proposed. Firstly, node and edge representations are given by estimating soft masks. Secondly, through disentanglement, both causal features and shortcut features are obtained from the graph. Thirdly, the backdoor adjustment of the causal theory is parameterized to combine each causal feature with a variety of shortcut features. Finally, comparative experiments on the Three-Phase Flow Facility (TFF) dataset illustrate the effectiveness of the proposed method.

Neural Network Weight Comparison for Industrial Causality Discovering and Its Soft Sensing Application

A Hybrid Deep Neural Network for Nonlinear Causality Analysis in Complex Industrial Control System

Spatial-Temporal Causality Modeling for Industrial Processes with a Knowledge-Data Guided Reinforcement Learning

Causality Enhanced Global-Local Graph Neural Network for Bioprocess Factor Forecasting

Long-short Memory with Feature Representation by Latent-variable Model for Industrial Soft Sensor

Feature Based Causality Analysis and Its Applications in Soft Sensor Modeling

Causal-Trivial Attention Graph Neural Network for Fault Diagnosis of Complex Industrial Processes

Weighted Nonlinear Dynamic System for Deep Extraction of Nonlinear Dynamic Latent Variables and Industrial Application

Causal Variable Selection for Industrial Process Quality Prediction Via Attention-Based GRU Network

Weighted Linear Dynamic System for Feature Representation and Soft Sensor Application in Nonlinear Dynamic Industrial Processes.

Multi-lag and multi-type temporal causality inference and analysis for industrial process fault diagnosis

Causal Network Construction Based on Convergent Cross Mapping (CCM) for Alarm System Root Cause Tracing of Nonlinear Industrial Process

Self-adaptive Graph Convolution Networks with Application to Industrial Soft Sensor Modeling

Data-driven dynamic causality analysis of industrial systems using interpretable machine learning and process mining

Soft Sensor Modeling of Nonlinear Industrial Processes Based on Weighted Probabilistic Projection Regression.

A Novel Label-Aware Global Graph Construction Method and Spiking-Coded Graph Neural Network for Intelligent Process Fault Diagnosis

A Multi-level Bayesian Network Based on Causality Analysis for Fault Diagnosis of Nonstationary Processes

Two-stage Approach to Causality Analysis-Based Quality Problem Solving for Discrete Manufacturing Systems

Sparse Causality Analysis Approach with Time-varying Parameters for Root Cause Localization of Nonstationary Process

Data-driven Dynamic Inferential Sensors Based on Causality Analysis

Information-based Gradient enhanced Causal Learning Graph Neural Network for fault diagnosis of complex industrial processes