Shock Event Registration Using EMD Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer systems face challenges in registering and tracking overstress events due to limited dynamic range of embedded accelerometers, noise from internal cooling fans, and limited storage/bandwidth capabilities, making it difficult to differentiate between normal and shock events.
Innovation Solution
The method employs empirical mode decomposition (EMD) for feature extraction and an R-cloud pattern classifier to process accelerometer data, incrementing a shock event counter and generating an indication when a specified count is reached, allowing for cumulative tracking and registration of shock events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If embedded accelerometers are used to register shock events, then shock event detection is enabled, but the dynamic range is limited and low frequency cutoff occurs
Solution Approach 1:
The patent segments the accelerometer signal into multiple frequency bands using wavelet transform, allowing separate analysis of different frequency components. This overcomes the limited dynamic range by processing different frequency ranges independently and combining the results to achieve comprehensive shock detection across the full frequency spectrum.
Solution Approach 2:
The patent introduces wavelet transform as an intermediary processing step between the accelerometer sensor and the shock detection algorithm. This intermediary transforms the time-domain signal into time-frequency domain representation, enabling accurate detection of shock events across various frequency ranges despite the accelerometer's frequency limitations.
2Loss of information
If waveform time-domain monitoring is implemented for overstress event registration, then complete shock data is captured, but storage and bandwidth capabilities are exceeded
Solution Approach 1:
The patent extracts only the essential shock event features from the complete waveform data using wavelet transform and energy analysis. Instead of storing and transmitting the entire time-domain waveform, only the extracted shock features and energy metrics are retained, dramatically reducing storage and bandwidth requirements while preserving the critical shock information needed for fault diagnosis.
Solution Approach 2:
The patent changes the representation parameters of the shock data from time-domain waveform samples to frequency-domain energy metrics through wavelet transform. This parameter transformation compresses the data representation from many time samples to a few energy values, achieving efficient storage and transmission while maintaining the ability to detect and analyze shock events.
3Loss of information
If time domain data is monitored for shock detection, then complete shock information is available, but extreme noisiness from cooling fans obscures shock events
Solution Approach 1:
The patent segments the time-domain signal into different frequency components using wavelet transform, separating the shock event signals from the cooling fan noise. By analyzing the energy distribution across different frequency bands, the method can identify shock events in frequency ranges where fan noise is minimal, effectively filtering out the harmful noise while preserving shock detection capability.
Solution Approach 2:
The patent applies partial action by focusing analysis only on specific frequency bands and time windows where shock events are likely to occur, rather than processing the entire time-domain signal. This selective approach reduces the impact of continuous fan noise by concentrating computational resources on detecting transient shock events in relevant frequency ranges.
Data Source
AI summary
An electronic system includes an accelerometer. A method for excessive mechanical shock feature extraction for overstress event registration and cumulative tracking includes obtaining a sample from the accelerometer. Feature extraction is performed on the sample using empirical mode decomposition (EMD) to produce a plurality of modes. A pattern classifier is utilized for processing the plurality of modes to determine if the sample classifies as a shock event. If the sample classifies as a shock event, a shock event counter is incremented. If the shock event counter reaches a specified count, an indication to a user is generated.