Knowledge Graph GNN Monitoring for Industrial Process Anomalies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale and complex industrial systems/processes are difficult to monitor and optimize due to their highly interconnected nature, limiting traditional modeling approaches in capturing interrelations between entities and requiring improved performance and anomaly detection.
Innovation Solution
Modeling dependencies in a knowledge graph as a graph neural network (GNN) to generate a reference graph model, allowing for real-time monitoring and anomaly detection by learning entity characteristics and interrelations, predicting links, and scoring anomalies based on the model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional modeling approaches are used to monitor complex industrial systems, then the system structure is simpler to implement, but the ability to capture interrelations between entities deteriorates
Solution Approach 1:
The patent introduces graph neural networks as an intermediary computational framework that mediates between traditional modeling approaches and complex industrial systems. The GNN models dependencies between entities as graphs, where nodes represent entities and edges represent relationships, enabling precise capture of interrelations while providing a structured approach to manage the complexity of large-scale systems with thousands of entities.
2Reliability
If graph neural networks are used to model dependencies in knowledge graphs, then anomaly detection capability is improved, but computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the knowledge graph into manageable components for processing. The system processes graph data through structured pipelines that break down the complex computational task of analyzing thousands of entities into smaller, more tractable operations, enabling anomaly detection while managing computational complexity through systematic data processing stages.
3Productivity
If real-time monitoring is implemented for large-scale industrial systems, then system performance optimization is improved, but data processing requirements increase
Solution Approach 1:
The patent implements preliminary action by pre-processing and structuring data into knowledge graphs before real-time analysis. Dependencies between entities are modeled in advance as graph structures, and reference graphs are pre-computed, enabling efficient real-time anomaly detection without requiring intensive processing of raw data during operational monitoring, thus optimizing system performance while managing data processing volumes.
Data Source
AI summary
Embodiments are provided for providing increased performance of various industrial systems and processes in a computing system by a processor. Each of a plurality of dependencies of a plurality of entities in a knowledge graph are modeled as a graph neural network (“GNN”). A reference graph model is generated based on the modeling. One or more anomalies are monitored and detected for a plurality of process based on the reference graph model.