Compressed Sensing Signal Extraction for PHM Data Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional signal sampling methods require significantly more data than necessary, leading to substantial storage and processing demands in Prognostic and Health Management (PHM) and Condition-based maintenance (CBM) applications, particularly for signals with high-frequency components like those in vibration analysis.
Innovation Solution
A method that randomly samples a signal at a lower rate, using a subset of measurements (10-15% of the full set) and applies a discrete Fourier transform to extract feature variables, minimizing data storage and processing requirements by solving an underdetermined system of equations with non-linear optimization techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Nyquist-Shannon sampling theorem is applied to capture frequency components, then measurement precision is improved, but quantity of substance (data volume) increases substantially
Solution Approach 1:
The patent extracts only the essential frequency components from the full signal by applying compressed sensing techniques. Instead of storing and processing the complete signal dataset, the system selects and processes only the critical frequency information through sparse representation and iterative algorithms, thereby reducing data volume while preserving measurement precision for frequency analysis
Solution Approach 2:
The patent changes the sampling parameter from conventional Nyquist rate to a reduced rate enabled by compressed sensing. By transforming the signal representation from time-domain sampling to frequency-domain sparse representation, the system achieves accurate frequency component capture with significantly fewer samples, changing the fundamental parameter of data acquisition requirements
2Measurement precision
If high sampling rate is used to capture high-frequency components, then measurement precision is improved, but device complexity (processing requirements) increases
Solution Approach 1:
The patent extracts and processes only the essential frequency information through compressed sensing algorithms. The system identifies and processes only the significant frequency components rather than processing the entire high-rate sampled signal, thereby reducing processing power requirements while maintaining detection accuracy for high-frequency components
Solution Approach 2:
The patent replaces conventional mechanical signal processing (requiring high processing power for high-rate data analysis) with compressed sensing algorithms that operate on sparse representations. This substitution enables high-frequency component detection with significantly reduced processing requirements by changing the mathematical approach from brute-force processing to intelligent sparse reconstruction
3Measurement precision
If substantial amounts of data are stored and processed, then measurement precision is improved, but loss of energy increases
Solution Approach 1:
The patent extracts and processes only the essential frequency components rather than handling the complete signal dataset. By applying compressed sensing to identify and process only the critical information, the system reduces the amount of data requiring storage and processing, thereby decreasing energy consumption while maintaining measurement precision for signal feature extraction
Solution Approach 2:
The patent applies partial action by processing only a subset of the data (compressed representation) rather than the full dataset. This partial processing approach achieves sufficient measurement precision for maintenance decisions while significantly reducing the energy consumption associated with processing and storing the complete signal data
Data Source
AI summary
A method and device for extracting information from data representing a signal is disclosed. A set of data comprising a plurality of measurements of the signal generated at a first sampling rate is received from a sensor. A subset of the plurality of measurements is selected. A plurality of feature variables, each of which corresponds to a particular feature in a set of features that may be present in the signal are determined by deriving an underdetermined system of equations based on a selected basis function, the subset of the plurality of measurements, and the plurality of feature variables and corresponding features. The underdetermined system of equations is solved to determine a value for each feature variable using a non-linear optimization technique to minimize an L1 norm of the set of features. Feature information is stored in a storage medium.