Automatic Link Prediction for Sensor Relationship Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In automated industrial and commercial environments, manually determining and labeling relationships between data points, such as sensors and actuators, is labor-intensive and error-prone, limiting scalability and automation speed.
Innovation Solution
The method involves retrieving data points from various sources, detecting events indicative of anomalies, generating a correlation matrix, normalizing it to suppress non-physical relations, and clustering data points based on the normalized matrix to identify relationships between points belonging to the same device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling of relationships between data points is performed, then accuracy of relationship identification can be maintained, but productivity and scalability deteriorate due to labor-intensive processes
Solution Approach 1:
The patent replaces the manual mechanical process of relationship labeling with an automated computational system. The system retrieves timeseries data from multiple sources, detects events indicating anomalies, generates correlation matrices, normalizes them to suppress non-physical relations, and clusters data points to automatically identify relationships between points belonging to the same device, eliminating manual intervention entirely
Solution Approach 2:
The system enables self-service by allowing data points to automatically identify their own relationships through the computational process. The algorithm autonomously processes the timeseries data, detects anomalies, computes correlations, and clusters points without requiring external manual labeling, making the system self-sufficient in identifying relationships
2Reliability
If manual determination of relationships is used, then error rates can be controlled, but loss of time increases due to labor-intensive processes
Solution Approach 1:
The patent replaces the slow manual process with a rapid automated computational system that processes timeseries data through event detection, correlation matrix generation, normalization, and clustering algorithms to quickly identify relationships without manual intervention, dramatically reducing time loss
Solution Approach 2:
The system performs preliminary actions by pre-processing the timeseries data through event detection and correlation analysis before final relationship identification. This preparatory computational work enables faster and more accurate relationship determination without requiring time-consuming manual review of raw data
3Productivity
If automated methods are implemented for relationship identification, then productivity and scalability improve, but device complexity increases
Solution Approach 1:
The patent segments the complex relationship identification process into distinct modular steps: retrieving timeseries data from multiple sources, detecting events indicating anomalies, generating correlation matrices, normalizing matrices to suppress non-physical relations, and clustering data points. This segmentation makes the complex system more manageable and scalable
Solution Approach 2:
The patent introduces intermediate computational structures to bridge the gap between raw data and relationship identification. The correlation matrix serves as an intermediary that captures relationships between data points, and normalization acts as an intermediary process to eliminate spurious correlations, simplifying the overall system architecture
4Measurement precision
If manual labeling processes are used, then data accuracy can be verified, but ease of operation deteriorates due to labor-intensive processes
Solution Approach 1:
The system enables self-service by automatically performing the entire relationship identification process through computational algorithms. The system retrieves data, detects events, computes correlations, normalizes results, and clusters points autonomously, making operation extremely easy while maintaining accuracy through systematic computational validation
Data Source
AI summary
Disclosed are methods and systems for predicting relationships between points in automated environments by retrieving a plurality of data points from a plurality of data sources associated with an automated environment, wherein the plurality of data points comprises timeseries data structure; detecting a plurality of events based on the plurality of data points, wherein an event of the plurality of data points is indicative of an anomaly reading of one or more states associated with the plurality of data sources; generating a correlation matrix of events across the plurality of data points; suppressing non-physical relation factors in the correlation matrix by normalization to generate a normalized matrix; and clustering the data points based on the normalized matrix, wherein the clustered data points represent the data points belonging to a same data source.