Temporal Data Storage for Time-Based Sensor Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing time-based data face inefficiencies and inaccuracies due to time alignment issues, particularly when data recording frequencies change, leading to problems with missing data and the need for truncation or rounding, which complicates the analysis of sensor data from sources like aircraft.
Innovation Solution
The system converts sensor data into a temporal format, allowing for normalization of data over time periods, enabling efficient storage and analysis without time alignment issues, and allowing for accurate combination of data from multiple flights and sensors using native temporal SQL extensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is stored in traditional row-based format with fixed time intervals, then data storage is simple, but time alignment problems occur when recording frequency changes and missing data causes analysis issues
Solution Approach 1:
The patent applies dynamics by transitioning from static fixed-time-point storage to dynamic time-period storage. Each data row represents a time period during which sensor values remained constant, allowing the system to adapt to varying recording frequencies and handle missing data naturally through period overlap detection, thereby improving reliability without excessive complexity
Solution Approach 2:
The patent changes the fundamental parameter representation from discrete time points to continuous time periods. By storing start and end timestamps along with the sensor value valid during that period, the system transforms how temporal data is structured, enabling accurate time-based analysis even when recording frequencies change or data is missing
2Measurement precision
If data is truncated or rounded to lower frequency for joining, then joining speed improves, but time alignment accuracy is lost
Solution Approach 1:
The patent performs preliminary normalization of data into time periods during the data loading phase, organizing sensor readings into continuous periods where values remain constant. This preliminary structuring enables efficient joining operations later by allowing direct comparison of time periods without requiring truncation or rounding, thus maintaining precision while improving joining speed
3Productivity
If traditional point-in-time data storage is used, then storage space is minimized, but complex temporal analysis requires significant computational resources and time
Solution Approach 1:
The patent merges multiple consecutive data points with the same sensor value into a single time period row. Instead of storing individual readings at each timestamp, the system consolidates them into one record representing the entire duration where the value remained constant, reducing the quantity of data to process and significantly improving analysis speed while using less storage space
Data Source
AI summary
A system and method includes a data sensor providing sensor data. A database is configured to store the sensor data over time in rows of a table. A processor is configured to convert the table into a temporal format and collapse the rows in a temporal table. The temporal data can be normalized, and the normalized temporal data can be natively analyzed over normalized time periods, e.g., to improve an efficiency and/or accuracy of the analysis. The systems and methods can also be used to efficiently inter-operate with non-temporal data as needed.