Knowledge Graph Anomaly Detection for Root Cause Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional monitoring systems rely on static alert thresholds and significant human intervention to address system anomalies, leading to numerous false alarms and inefficiencies in identifying root causes due to the complexity and interconnectivity of large systems.
Innovation Solution
A machine learning model utilizing a knowledge graph and multivariate sensor metrics to detect, diagnose, and mitigate anomalous system behaviors by dynamically adjusting monitoring thresholds and frequencies, leveraging system entity relationships and historical time-series data to identify upstream root causes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional monitoring systems use static alert thresholds, then system monitoring is simple to implement, but false alarms increase and detection precision deteriorates
Solution Approach 1:
The patent implements dynamic alert thresholds that automatically adjust based on learned system behavior patterns and historical data, replacing static thresholds with adaptive ones that reduce false alarms while maintaining detection precision
Solution Approach 2:
The system performs self-diagnosis and self-monitoring by automatically analyzing sensor data, identifying anomalies, and generating work orders without human intervention, reducing the need for complex manual monitoring configurations
2Productivity
If human intervention is used to address system anomalies, then diagnostic accuracy can be improved, but response time and productivity deteriorate
Solution Approach 1:
The patent replaces manual human diagnostic processes with an automated machine learning model that analyzes sensor data, identifies anomalies, determines root causes, and generates remediation work orders automatically, maintaining high accuracy while dramatically improving response speed
Solution Approach 2:
The system introduces a knowledge graph as an intermediary that encodes domain expertise and system relationships, enabling automated diagnostic reasoning that captures human expert knowledge without requiring continuous human intervention
3Reliability
If monitoring coverage is expanded to capture more system data, then detection capability is improved, but data processing complexity and energy consumption increase
Solution Approach 1:
The patent extracts only the most relevant features and metrics from the comprehensive sensor data using the knowledge graph, which identifies key system components and relationships, allowing focused analysis on critical data points rather than processing all available data
Solution Approach 2:
The system segments the monitoring task by dividing the system into logical components represented in the knowledge graph, enabling distributed and parallel processing of data from different system segments, reducing overall processing complexity and energy requirements
Data Source
AI summary
Detecting and mitigating anomalous system behavior by providing a machine learning model comprising a knowledge graph depicting system entity relationships, and modeling behavioral correlations among system entities according to historical time-series data, receiving real-time time-series data for the system, detecting an anomalous system behavior in a system locale, according to the real-time time-series data, according to the machine learning model and multivariate sensor metrics, diagnosing the anomalous system behavior according to an upstream portion of the knowledge graph and a statistical behavior model for the system locale, and mitigating the anomalous behavior by deriving a recommended action according to the anomalous behavior and generating a work order to implement the recommended action.