OPC Data Anomaly Detection With AI Root Cause Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial control systems (ICS) and industrial internet of things (IIoT) environments face challenges in analyzing complex OPC data due to limited data historian analysis capabilities, performance issues, and difficulty in identifying anomalies and root causes, which can lead to operational disruptions and security threats.
Innovation Solution
A method and system utilizing artificial intelligence/machine learning (AI/ML) models to analyze historic OPC data, identify events, patterns, and root causes, and predict anomalies and faults, employing techniques such as supervised learning, unsupervised learning, and ensembling to enhance data analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional data historian analysis methods are used, then system simplicity is maintained, but anomaly detection capability and root cause identification are insufficient
Solution Approach 1:
The patent introduces AI/ML models as intermediary components between the data historian and analysis functions. These models preprocess and interpret complex OPC data, transforming raw historical data into actionable insights about anomalies and root causes, thereby enhancing detection capability without directly increasing system complexity
Solution Approach 2:
The patent replaces traditional mechanical/manual data analysis methods with AI/ML-based automated analysis. This substitution enables sophisticated anomaly detection and root cause identification capabilities while reducing the need for complex manual intervention systems
2Measurement precision
If AI/ML models are deployed for data analysis, then anomaly detection and root cause identification improve, but resource constraints (CPU, memory, I/O) are exacerbated
Solution Approach 1:
The patent applies AI/ML models selectively to specific data patterns and anomaly scenarios rather than processing all data uniformly. This partial action approach maintains high detection accuracy while reducing overall computational resource consumption by focusing analysis only where needed
Solution Approach 2:
The patent segments the data analysis process into multiple stages: initial data collection, pattern recognition, anomaly detection, and root cause analysis. Each stage uses appropriate AI/ML techniques tailored to its specific requirements, optimizing resource usage at each step rather than applying heavy computation throughout
3Loss of information
If comprehensive historical data analysis is performed, then root cause identification improves, but performance issues arise when retrieving large amounts of archived data
Solution Approach 1:
The patent performs preliminary processing and indexing of historical OPC data during the data collection phase. This preliminary action organizes data in advance for efficient retrieval, enabling comprehensive analysis without performance degradation when querying large archived datasets
Solution Approach 2:
The patent extracts only the relevant features and patterns from comprehensive historical data that are necessary for root cause identification. This extraction approach maintains complete analytical capability while reducing the volume of data that needs to be retrieved and processed from archives
4Measurement precision
If additional infrastructure and resources are deployed for data analysis, then analysis capability improves, but system cost and complexity increase
Solution Approach 1:
The patent designs the AI/ML analysis system to perform multiple functions: anomaly detection, pattern recognition, root cause identification, and predictive analysis. This multi-functionality consolidates what would otherwise require separate infrastructure components into a unified system, improving analysis capability without proportionally increasing complexity
Data Source
AI summary
A method and system for determining anomaly and fault in open platform communications (OPC) data is disclosed. Through the utilization of at least one processor, the method comprises receiving a historic data from one or more sources for a predefined time period, wherein the historic data corresponds to a historical open platform communications (OPC) data from the one or more sources and an input data from at least one OPC client; analyzing the historic data using artificial intelligence/machine learning (AI/ML) models to identify events in the historic data; identifying patterns associated with the identified events using the AI/ML models; identifying one or more root causes associated with each of the patterns using the AI/ML models; correlating the identified patterns with the identified one or more root causes; and predicting one or more anomalies and faults associated with historic data, based at least on the correlation.