Characterising Code Clustering for Fast Physical System Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern systems, particularly in aerospace contexts, face challenges in extracting meaningful insights from vast amounts of data due to the large volume of usage data generated, leading to inefficient maintenance and operational inefficiencies.
Innovation Solution
A computer-implemented method is introduced to create a database of characterising codes by clustering parameters associated with physical systems, generating unique labels for each cluster, and storing these codes, allowing for efficient management and comparison of systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If data from multiple parameters is collected for each system example, then the information content and insight quality improve, but the data volume and storage requirements become prohibitively large
Solution Approach 1:
The patent segments the continuous parameter space into discrete clusters (e.g., dividing engine temperature ranges into multiple clusters). Each parameter's continuous values are partitioned into distinct groups, allowing complex multi-parameter data to be represented as combinations of discrete cluster labels rather than storing all raw continuous values, thus reducing data volume while preserving information structure
Solution Approach 2:
The patent transforms the original continuous parameter values into discrete cluster identifiers. By changing the parameter representation from continuous numerical values to discrete categorical labels (cluster indices), the system maintains the essential information about parameter characteristics while dramatically reducing the storage requirements for large datasets
2Measurement precision
If all parameter values are stored in detail, then measurement precision and analysis accuracy improve, but the complexity of data management and exploration increases
Solution Approach 1:
The patent segments the data management task into two stages: (1) preprocessing where continuous parameter ranges are divided into discrete clusters with defined boundaries, and (2) storage where only cluster labels are retained. This segmentation simplifies subsequent data exploration and comparison operations while maintaining sufficient precision for practical analysis through the structured cluster definitions
Solution Approach 2:
The patent creates a simplified copy of the original data structure by replacing detailed continuous parameter values with discrete cluster identifiers. This copied representation (using cluster labels instead of raw values) maintains the essential information needed for comparison and analysis while significantly reducing management complexity
3Productivity
If characterising codes are generated from clustered data, then data exploration efficiency improves, but the time and computational resources required for clustering increase
Solution Approach 1:
The patent performs preliminary clustering action during data ingestion or batch processing, organizing raw parameter values into predefined clusters and generating characterising codes in advance. This preliminary organization creates a structured database of cluster labels that enables rapid subsequent exploration and comparison operations, trading upfront computational effort for long-term efficiency gains in data analysis
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A computer-implemented method. The method comprising: (a) obtaining a plurality of unique identifiers, each unique identifier being associated with a respective characterising code, each characterising code being indicative of character of an example of a same physical system; (b) accessing a database of characterising codes, and retrieving each respective characterising code; (c) determining a degree of similarity between each retrieved characterising code; and (d) providing, as an output, the determined degrees of similarity.