Dynamic Data Storage Allocation for Condition Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data storage methods for dynamic data in condition-based monitoring are inefficient, leading to excessive storage requirements due to the bulkiness of dynamic data, with most collected data being repetitive and not crucial for users, thus occupying valuable storage space.
Innovation Solution
A computer-implemented method and system that utilizes a dynamic data storage allocation module to determine the 'interestingness' of data samples based on their percentage of danger value, storing only samples that exceed a threshold and adjusting the threshold based on collection rates to optimize storage efficiency, thereby selecting and storing only the most important data without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional fixed interval data collection is used, then complete data coverage is achieved, but storage requirements become excessive (many terabytes per year)
Solution Approach 1:
The system dynamically changes the data collection parameter (interestingness threshold) based on machine state. When the machine is in steady state, the threshold is set high to filter out repetitive data. When exceptions occur, the threshold adjusts to capture relevant data, thus reducing storage volume while maintaining data coverage for critical events
Solution Approach 2:
The data collection strategy transitions from static fixed-interval sampling to dynamic exception-based sampling. The system continuously monitors machine parameters and adapts its collection behavior based on detected exceptions, collecting data only when meaningful changes occur, thereby reducing overall storage requirements while preserving reliability for critical events
2Loss of information
If all dynamic data is collected and stored, then no important information is missed, but most stored data is repetitive and not important to users
Solution Approach 1:
The system extracts and stores only the essential portion of dynamic data - specifically data associated with exceptions and meaningful changes. By identifying and isolating these critical data points from the bulk repetitive data, the system maintains information completeness for important events while dramatically reducing the total data volume stored
Solution Approach 2:
Different quality standards are applied to different portions of the data stream. Critical data points (exceptions, transients) are captured with high fidelity and stored, while steady-state repetitive data is filtered out. This local differentiation ensures important information is preserved without storing volumes of irrelevant data
3Quantity of substance
If exception-based data collection is used, then storage efficiency improves, but data collection complexity increases
Solution Approach 1:
The system performs self-service by automatically detecting exceptions and determining which data to collect without human intervention. The exception-based collection mechanism is self-regulating, automatically adjusting to capture relevant data while filtering out repetitive information, thus improving storage efficiency without requiring complex manual management
Solution Approach 2:
The system performs preliminary analysis of incoming data to identify exceptions before full data collection occurs. By pre-processing data streams and detecting meaningful changes in advance, the system can trigger targeted data collection only when necessary, simplifying the overall collection process while maintaining storage efficiency
Data Source
AI summary
The storage allocation system reduces the amount of storage needed to store data, and also to store the most interesting dynamic data samples (e.g., waveforms). Dynamic data is relevant if it has changed from a previous sample. Changes need to be detected for different machinery conditions. The storage allocation approach normalizes static data to a percentage of danger alarm value that indicates how much the data is changing. The difference of successive percent of danger alarm values provides an instantaneous change value indicating how interesting the data is. To manage how much data is collected; the system maintains an interestingness threshold and stores data samples having an interestingness value above the threshold. The threshold is automatically adjusted if too much/little data is being saved. This approach optimizes the amount and value of data stored and data storage requirements for improved data storage efficiency.