PLC Control Sequence Anomaly Detection Using Graph Autoencoders
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automation equipment in industrial processes often experiences anomalies that go undetected, leading to production delays and decreased operational efficiency due to the lack of effective monitoring and analysis of programmable logic controller (PLC) control logic patterns.
Innovation Solution
A method is developed to generate graph data from log data by classifying contact point changes, identifying major states, and converting them into node and edge index data, which is then used to train a graph neural network (GNN) AutoEncoder to detect anomalies by calculating edge prediction loss and setting reference thresholds for anomaly detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional PLC control logic monitoring is used, then the system operates with simple control logic, but anomalies cannot be detected and production delays occur
Solution Approach 1:
The patent replaces traditional mechanical/logic-based PLC monitoring with a data-driven machine learning approach. Graph neural networks process PLC log data to automatically detect anomalies, substituting rule-based systems with intelligent algorithms that learn normal operation patterns and identify deviations without explicit programming of anomaly conditions.
Solution Approach 2:
The patent introduces graph neural networks as an intermediary layer between raw PLC log data and anomaly detection. The GNN model transforms complex control logic data into graph representations, enabling the system to interpret and detect anomalies in PLC operations without directly analyzing the complex control logic structure.
2Productivity
If PLC control logic is monitored continuously, then operational efficiency improves, but the complexity of analyzing control logic patterns increases
Solution Approach 1:
The patent substitutes manual or rule-based control logic pattern analysis with machine learning models. The graph neural network automatically learns normal operation patterns from historical PLC data, replacing the need for explicit pattern definition and simplifying the detection process while improving operational efficiency through automated anomaly identification.
Solution Approach 2:
The patent performs preliminary training of the graph neural network model using normal operation data before deployment. This preliminary action enables the system to establish baseline patterns of normal PLC operations, so that during continuous monitoring, anomalies can be detected by comparing against these pre-established patterns rather than analyzing complex logic in real-time.
3Loss of time
If anomalies are detected early, then production delays are reduced, but the complexity of implementing advanced detection methods increases
Solution Approach 1:
The patent replaces simple threshold-based or rule-based detection systems with graph neural networks that can identify subtle anomalies in PLC control logic. This substitution enables early detection of deviations from normal operation patterns, reducing production delays by identifying issues before they cause interruptions, despite the increased computational complexity of the detection model.
Data Source
AI summary
Disclosed is a method of analyzing a programmable logic controller (PLC) logic to detect whether an anomaly that deviates from a standard pattern occurs in a repeated cycle. After modeling and patterning an operation pattern of automation equipment and processes with a graph, an anomaly detecting model capable of detecting whether a pattern is abnormal may be constructed as a graph AutoEncoder model. By detecting the change in the process pattern, it is possible to early detect the anomaly of the equipment and processes.