Incremental Time Series Pattern Detection for Real-Time Anomaly Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting patterns in time series data, such as system loads, require substantial data collection and explicit modeling of each time cycle, making them inefficient for real-time prediction and anomaly detection.
Innovation Solution
A flexible technique that incrementally detects daily, weekly, and monthly patterns in time series data, allowing for real-time prediction and anomaly detection by matching time series data with existing patterns and automatically expunging unused patterns, using a system that includes a Time Series Pattern Manager, Bookkeeper, and Time Series Pattern Recurrence Recognizer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If explicit modeling of each time cycle is performed, then pattern recognition accuracy is improved, but data storage requirements and processing time increase substantially
Solution Approach 1:
The patent segments time cycle data into representative daily patterns that capture essential characteristics without storing complete cycle data. Instead of explicit modeling of each time cycle, the system divides and stores only the necessary pattern features, reducing storage requirements while maintaining recognition accuracy.
Solution Approach 2:
The patent creates simplified copies or representations of time cycle patterns rather than storing complete original data. These pattern representations capture the essential characteristics needed for recognition while occupying minimal storage space, enabling efficient pattern matching without bulk data storage.
2Measurement precision
If explicit modeling of each time cycle is performed, then pattern recognition accuracy is improved, but processing time increases substantially
Solution Approach 1:
The patent performs preliminary extraction of pattern characteristics from time cycle data before actual pattern recognition is needed. By pre-processing and storing only essential pattern features, the system eliminates the need for extensive processing during real-time recognition, significantly reducing processing time while maintaining accuracy.
Solution Approach 2:
The patent extracts only the essential characteristics and features from complete time cycle data, separating the necessary pattern information from redundant details. This extraction process creates compact pattern representations that can be quickly processed and compared during recognition without requiring analysis of entire time cycles.
3Reliability
If bulk data collection and storage is performed, then comprehensive pattern detection is improved, but storage costs and system complexity increase
Solution Approach 1:
The patent implements a dynamic pattern management system that automatically adapts the number and characteristics of stored patterns based on observed system behavior. Rather than maintaining a fixed large set of patterns, the system dynamically creates, updates, and removes patterns as needed, reducing system complexity while maintaining comprehensive detection capability.
Solution Approach 2:
The patent implements automatic removal of patterns that are no longer relevant or useful for pattern detection. By discarding obsolete patterns and retaining only those that provide value, the system maintains comprehensive detection capability with a minimized set of patterns, reducing storage requirements and system complexity.
Data Source
AI summary
The present disclosure describes a flexible technique to learn patterns in time series data that recur over time. The patterns may be used for simulation, predicting future behavior, or detecting anomalies in a system in which the data is collected. The technique incrementally detects daily, weekly, monthly, and yearly patterns. Each pattern is built over time instead of requiring all the data to be available at the beginning of the analysis. Instead of modeling each pattern explicitly, each pattern is described in the context of a day and formed based on time series data collected over an entire day. An example use of the technique is detecting load patterns in a computer system. A metric of system load such as CPU utilization may be collected periodically over a day. The techniques presented herein capture multiple daily models, each representing a different load pattern.