Time-Series Signal Similarity for Abnormal Event Cause Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current information processing systems lack the capability to effectively detect and analyze abnormal events from abnormal time-series data, which includes multiple signal data pieces, making it difficult to identify the cause of anomalies in complex systems like robot control, engine control, and smart grid management.
Innovation Solution
An information processing system comprising an acquisition unit to gather time-series data, a data extraction unit to differentiate between normal and abnormal data, and an examination unit that determines similarity between signal data pieces to identify abnormal events and specify their causes, utilizing techniques such as partial time-series generation, parameter calculation, ID extraction, similarity calculation, timing specification, and signal specification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual analysis methods are used to detect abnormal events in time-series data, then domain expertise can be applied to identify anomalies, but the process requires excessive man-hours and cannot handle complex multi-signal data efficiently
Solution Approach 1:
The system enables automated self-analysis of abnormal events by having the computer automatically perform similarity comparisons between abnormal time-series data and normal time-series data, extract cause signals, and identify abnormal events without requiring manual domain expert intervention
Solution Approach 2:
The patent replaces manual mechanical analysis processes with automated computer-based processing, using algorithms to calculate similarity between time-series data pieces and automatically identify cause signals, thereby eliminating the need for manual domain expert analysis
2Productivity
If automated analysis methods are implemented to reduce manual work, then processing speed increases, but the system lacks the capability to accurately identify cause signals in multi-signal time-series data
Solution Approach 1:
The system segments the analysis process into distinct steps: dividing time-series data into time sections, comparing each section separately, identifying similar normal time-series data pieces, and extracting cause signals from specific segments, thereby maintaining accuracy through systematic breakdown
Solution Approach 2:
The patent introduces similarity calculation as an intermediary mechanism that bridges automated processing and accurate identification, using similarity scores to mediate between abnormal and normal time-series data to reliably identify cause signals without manual intervention
3Reliability
If comprehensive analysis of all signal data pieces is performed to ensure accurate cause identification, then detection reliability improves, but the complexity of the analysis process increases significantly
Solution Approach 1:
The system extracts only the essential cause signals from the time-series data by comparing abnormal data with normal data and identifying similar patterns, thereby simplifying the analysis process while maintaining reliable cause identification by focusing on key discriminative features
Solution Approach 2:
The patent performs preliminary comparison between abnormal and normal time-series data to identify similar patterns before detailed cause analysis, pre-filtering the data to focus subsequent analysis on relevant time sections and signal pieces, thereby reducing overall process complexity
4Measurement precision
If detailed comparison of each signal data piece is conducted to identify specific cause signals, then the precision of cause identification improves, but the time required for analysis increases
Solution Approach 1:
The system applies periodic action by dividing the time-series data into discrete time sections and systematically comparing each section with normal data in a structured sequence, enabling efficient processing through regular, methodical comparison cycles that maintain precision without excessive time consumption
Data Source
AI summary
An information processing system, an information processing method, and a program that can examine an abnormal event based on abnormal time-series data including a plurality of signal data pieces are provided. An information processing device includes an acquisition unit configured to acquire a plurality of time-series data pieces, each time-series data piece including a plurality of signal data pieces, and the plurality of time-series data pieces including a plurality of normal time-series data pieces and abnormal time-series data, and an examination unit configured to determine for each signal data piece whether the abnormal time-series data is similar to each normal time-series data piece in each time section and examine an abnormal event based on a result of the determination.