Root Cause Anomaly Detection in Time Series via GNN Dependency Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current analysis methods are unable to automatically identify root cause anomalies or sequences of anomalies within a group of time series, which hinders effective maintenance and troubleshooting in complex systems.
Innovation Solution
The use of a graph neural network (GNN) to identify anomalies in time series, followed by the construction of a dependency graph and application of a root cause analysis algorithm to determine root cause trees and identify sequences of anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If current analysis methods are used to monitor time series data, then basic anomaly detection can be performed, but automatic identification of root cause anomalies and their sequences cannot be achieved
Solution Approach 1:
The patent segments the complex root cause analysis problem into distinct components: (1) anomaly detection in individual time series, (2) construction of dependency graphs representing relationships between anomalies, and (3) application of root cause analysis algorithms to identify root causes and sequences. This segmentation enables automatic identification while managing system complexity through modular processing steps.
Solution Approach 2:
The patent introduces dependency graphs as an intermediary data structure that captures relationships between anomalies across different time series. These graphs serve as a bridge between raw anomaly detection results and root cause analysis, enabling the system to automatically identify root cause anomalies and their sequences without requiring direct complex analysis of all time series relationships.
2Productivity
If manual analysis methods are used to identify root cause anomalies, then diagnostic accuracy can be maintained, but diagnostic time and computational resources increase
Solution Approach 1:
The patent performs preliminary actions by first detecting anomalies in individual time series and constructing dependency graphs before applying root cause analysis algorithms. This preliminary processing organizes the data in advance, enabling faster automatic identification of root cause anomalies and their sequences, thereby reducing overall diagnostic time while maintaining accuracy.
Solution Approach 2:
The system uses feedback from anomaly detection results to dynamically construct and update dependency graphs, which then guide the root cause analysis process. This feedback mechanism enables the system to learn from detected patterns and automatically identify root cause anomalies more efficiently, reducing diagnostic time while maintaining high accuracy through iterative refinement.
Data Source
AI summary
Techniques are described for identifying root cause anomalies in time series. Information to be used for root cause analysis (RCA) is obtained from a graph neural network (GNN) and is used to construct a dependency graph having nodes corresponding to each time series and directed edges corresponding to dependencies between the time series. Nodes corresponding to time series that do not contain anomalies may be removed from this dependency graph, as well as edges connected to these nodes. This edge and node removal may result in the creation of one or more sub-graphs from the dependency graph. A root cause analysis algorithm may be run on these one or more sub-graphs to create a root cause graph for each sub-graph. These root cause graphs may then be used to identify root cause anomalies within the multiple time series, as well as sequences of anomalies within the multiple time series.