Abnormal Sound Determination Using Frequency Band Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining abnormal sounds in vehicles after assembly face challenges in accuracy due to interference from nearby frequency bands and inability to handle unknown sounds, especially when using low-pass filters, which impairs the detection of instantaneous abnormalities like falling objects.
Innovation Solution
An apparatus and method that record sound data, resolve it into multiple frequency bands using time-frequency techniques like STFT, calculate correlation coefficients between these bands, and determine abnormal sound occurrence based on these coefficients, allowing for accurate identification of abnormal sounds without preselecting frequency components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If low-pass filters are used to extract envelope of second order powers, then the processing is simplified, but accurate determination of instantaneous abnormal sounds becomes impossible
Solution Approach 1:
The patent extracts only the necessary frequency components related to abnormal sounds using bandpass filters targeted at specific frequency bands where abnormal sounds occur, rather than processing the entire frequency spectrum. This extraction approach maintains determination accuracy for instantaneous abnormal sounds while reducing processing complexity by focusing only on relevant frequency ranges.
2Ease of manufacture
If multiple preselected bandpass filters are used to resolve sound into frequency components, then the determination process is structured, but accuracy is impaired when other components near the frequency band get mixed in
Solution Approach 1:
The patent dynamically adjusts the center frequencies and bandwidths of bandpass filters based on the actual sound signal characteristics. Instead of using fixed preselected filters, the system adapts the filter parameters to match the specific frequency components of abnormal sounds detected in the signal, thereby improving separation accuracy while maintaining structured processing.
Solution Approach 2:
The patent changes the parameters of bandpass filters (center frequency, bandwidth) based on the analyzed sound signal properties. By adjusting these parameters dynamically, the system optimizes the separation of frequency components and prevents mixing of adjacent frequency bands, thereby improving determination accuracy while keeping the process structured.
3Ease of operation
If human inspectors perform abnormal sound inspection, then flexible judgment is possible, but the workload is heavy and hearing acuity varies among inspectors
Solution Approach 1:
The patent replaces the mechanical system of human hearing and judgment with an automated electronic system that uses microphones, signal processing algorithms, and abnormal sound determination logic. This substitution eliminates variations in human hearing acuity and removes the need for heavy inspector workload while maintaining flexible abnormal sound detection through adaptive processing.
4Productivity
If frequency components are preselected for analysis, then the determination process is efficient, but unknown abnormal sounds cannot be properly detected
Solution Approach 1:
The patent employs dynamic frequency analysis where the system first performs a broad frequency spectrum analysis to identify abnormal sound components, then adapts the bandpass filter settings based on the detected abnormal frequency ranges. This dynamic approach allows the system to efficiently detect both known and unknown abnormal sounds by adjusting its analysis focus based on actual signal characteristics rather than relying on preselected fixed frequency components.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables precise detection of abnormal sounds, including those from unknown sources, by extracting correlations in frequency bands, enhancing accuracy and eliminating the need for human intervention, and effectively differentiates between normal and abnormal vehicle noises.
Implementation Method 1
a sound data time-frequency resolution unit configured to resolve the recorded sound data of the test object into multiple frequency bands by time unit
Implementation Method 2
a correlation coefficient calculation unit configured to compare the sound data resolved into multiple frequency bands by time unit among the multiple frequency bands and calculate correlation coefficients indicating strength of correlation between the multiple frequency bands
Data Source
AI summary
In determining presence/absence of abnormal sound occurrence in a test object such as a vehicle immediately after completion of vehicle assembly, sound data of the test object during running on a rough test track is recorded and resolved into multiple frequency bands by time unit. The sound data resolved into multiple frequency bands by time unit is compared among the multiple frequency bands and correlation coefficients matrix values indicating strength of correlation between the multiple frequency bands n are calculated. Presence/absence of the abnormal sound occurrence is finally determined based on the calculated correlation coefficients matrix values.