Time-Series Data Processing for Inertial Sensor Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for processing data from inertial sensors to detect dynamical system states are computationally intensive and do not provide readily interpretable results, making it difficult to identify malfunctions in periodic or quasi-periodic systems.
Innovation Solution
A data processing method that applies transformations such as high-pass filtering, delay embedding, and projection into a new coordinate space to simplify data analysis, allowing for the extraction of characteristic signatures that can be used for classification and anomaly detection with reduced computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional Fast Fourier Transform domain processing is applied to sensor data, then information about the underlying dynamical system can be obtained, but the computations become heavy and cumbersome and the results are not readily interpretable
Solution Approach 1:
The patent extracts only the essential features needed for dynamical system characterization by projecting sensor data onto specific planes and computing compact signatures, rather than performing complete FFT analysis. This extraction approach obtains the necessary information while avoiding heavy computations.
Solution Approach 2:
The patent creates simplified representations (projections and signatures) that copy the essential characteristics of the original sensor data in a more compact and interpretable form, enabling analysis without processing the full complex dataset.
2Ease of operation
If point cloud data from sensors is processed using conventional techniques, then analysis of the data can be performed, but the analysis is hardly straightforward and requires significant computational resources
Solution Approach 1:
The patent transforms the analysis by projecting data onto 2D planes and computing compact signatures, changing the dimensional representation to make the data more straightforward to analyze while improving processing efficiency through reduced computational requirements.
3Measurement precision
If heavy computational processing is applied to sensor data to detect system state, then detailed analysis can be performed, but the processing is cumbersome and does not provide readily interpretable results
Solution Approach 1:
The patent extracts characteristic signatures that capture the essential system state information in a compact form, providing both accurate detection and interpretability without requiring heavy computational processing.
Data Source
AI summary
An embodiment method of processing at least one sensing signal comprising a time-series of signal samples comprises high-pass filtering the time series of signal samples to produce a filtered time series; applying delay embedding processing to the filtered time series; producing a first matrix by storing the set of time-shifted time series as an ordered list of entries in the first matrix; applying a first truncation to produce a second matrix by truncating the entries in the ordered list of entries at one end of the first matrix to remove a number of items equal to the product of the first delay embedding parameter decreased by one times the second delay embedding parameter; applying entry-wise processing to the second matrix, and forwarding a set of estimated kernel densities and/or a set of images generated as a function of the set of estimated kernel densities to a user circuit.