AI Noise Source Identification for Irregular Powertrain Sounds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Diagnosing the source of problematic noise in complex vehicle systems, such as powertrains, is challenging due to the irregular nature of the noise and the difficulty in determining which component is faulty, often requiring extensive time and expertise.
Innovation Solution
A noise data artificial intelligence apparatus and method that employs pre-conditioning techniques like Log Mel Filter analysis, Bidirectional RNN, Deep Neural Networks, Attention Mechanism, and Early stage ensemble algorithms to identify the source of problematic noise by extracting feature parameters and applying them to a probabilistic diagnostic model, enabling accurate and efficient noise source identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If noise experts use sensor or test conditions to diagnose problematic noise, then diagnostic accuracy may be improved, but diagnostic time increases significantly
Solution Approach 1:
The system performs preliminary action by pre-processing noise data through segmentation into unit frames, applying Log Mel Filter for frequency analysis, and extracting feature parameters before actual diagnosis. This preparation work is done in advance to enable faster real-time diagnosis without compromising accuracy
Solution Approach 2:
The patent replaces the mechanical/expert-based diagnostic system with an artificial intelligence system using Bidirectional RNN and Deep Neural Network. The AI model automatically learns patterns from pre-processed noise features, eliminating the need for expert manual analysis while maintaining or improving diagnostic accuracy and significantly reducing time
2Measurement precision
If conventional noise analysis methods are used, then simplicity is maintained, but the ability to accurately diagnose irregular noise sources is insufficient
Solution Approach 1:
The noise data is segmented into unit frames with overlapping time windows, and each frame is divided into N segments for frequency analysis. This segmentation transforms the complex irregular noise signal into manageable discrete units that can be processed by the AI system, improving accuracy without overwhelming complexity
Solution Approach 2:
The patent introduces an intermediary processing layer between raw noise data and the AI diagnosis system. This layer includes Log Mel Filter for frequency transformation and feature parameter extraction, which converts complex noise signals into standardized features that the Bidirectional RNN can effectively process
3Productivity
If noise data is collected and analyzed without pre-conditioning, then data processing is simpler, but the effectiveness of artificial intelligence learning is reduced
Solution Approach 1:
The system performs preliminary action by pre-processing noise data through segmentation into unit frames, applying Log Mel Filter for frequency analysis, and extracting feature parameters before actual diagnosis. This preparation work is done in advance to enable faster real-time diagnosis without compromising accuracy
Solution Approach 2:
The patent applies parameter changes by transforming noise data through Log Mel Filter which converts linear frequency scale to Mel scale, and by extracting specific feature parameters from the frequency spectrum. These parameter transformations make the data more suitable for AI learning, improving both speed and accuracy
Data Source
AI summary
A noise data artificial intelligence learning method for identifying the source of problematic noise may include a noise data pre-conditioning method for identifying the source of problematic noise including: selecting a unit frame for the problematic noise among noises sampled with time; dividing the unit frame into N segments; analyzing frequency characteristic for each segment of the N segments and extracting a frequency component of each segment by applying Log Mel Filter; and outputting a feature parameter as one representative frame by averaging information on the N segments, wherein an artificial intelligence learning by the feature parameter extracted according to a change in time by the noise data pre-conditioning method applies Bidirectional RNN.