Vehicle Noise Source Degradation Prediction From Vibration Spectra
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods fail to accurately predict the degradation level of vehicle components, such as noise sources, leading to inefficiencies in maintenance and potential discomfort for drivers due to unaddressed noise and vibration issues.
Innovation Solution
A degradation level prediction system using deep learning to estimate the state of noise sources by converting vibration signals into the frequency domain and employing state estimation and prediction models to forecast future degradation levels, thereby calculating the remaining useful life of vehicle components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning models are used to predict degradation levels, then prediction accuracy is improved, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent segments the degradation prediction process into two distinct phases: an offline learning phase where deep learning models are trained using historical vibration data and degradation information, and an online prediction phase where the trained models rapidly predict future degradation levels. This segmentation allows computationally intensive model training to be performed separately from real-time prediction operations, reducing the computational burden during actual deployment while maintaining high prediction accuracy.
Solution Approach 2:
The patent performs preliminary actions by pre-training the deep learning models offline using accumulated historical data before actual degradation prediction is needed. The models learn patterns and relationships between vibration characteristics and degradation levels in advance, so that during operation, only lightweight inference is required. This preliminary training phase captures complex nonlinear relationships without requiring heavy computation during real-time prediction.
2Measurement precision
If vibration signals are converted to frequency domain for analysis, then degradation detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent extracts only the essential frequency domain features that are most indicative of degradation, rather than processing the entire frequency spectrum. By identifying and extracting specific frequency components and characteristics that correlate with degradation patterns, the system achieves high detection accuracy while minimizing processing time. This selective extraction avoids unnecessary computational overhead from analyzing all frequency components.
Solution Approach 2:
The patent transforms the vibration signal from time domain to frequency domain using Fast Fourier Transform (FFT), changing the representation parameters to make degradation patterns more apparent. This parameter transformation reveals frequency-specific characteristics that are harder to detect in the time domain, improving degradation detection accuracy. The FFT algorithm efficiently performs this transformation with computational complexity of O(N log N), balancing accuracy improvement with acceptable processing time.
3Loss of time
If remaining useful life is calculated using time series degradation data, then maintenance timing accuracy is improved, but data requirements and system complexity increase
Solution Approach 1:
The patent implements a feedback mechanism where calculated remaining useful life predictions are fed back into the system to update and refine the degradation models. Historical predictions are compared with actual degradation outcomes, and the deep learning models are retrained with this feedback information. This continuous feedback loop improves prediction accuracy over time while the system learns to require fewer data points for accurate RUL estimation, as the models become more efficient at extracting predictive information.
Data Source
AI summary
A degradation level of a noise source is predicted by performing deep learning using frequency data indicating a vibration of the noise source during a time period and degradation data indicating degradation of the noise source during the predetermined period to determine a state estimation model. First vibration signals are measured at the noise point during a monitoring period. The first vibration signals are converted into a frequency domain, and the frequency domain converted first vibration signals are input into the state estimation model to estimate time series degradation data of degradation of the noise source during the monitoring period. A degradation prediction model is determined by deep learning using the time series degradation data. Degradation data indicating predicted degradation of the noise source during a prediction period are predicted via the degradation prediction model. A remaining useful life of the noise source may be calculated.