Aeroengine Anomaly Detection Using Multi-Band Wavelet Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods for aeroengines struggle with noise resistance and robustness to changes in working conditions, leading to misidentification of normal signals as abnormal due to complex signal interference and varying operating states, which traditional and deep learning-based methods fail to address effectively.
Innovation Solution
A multi-band wavelet network is introduced, combining wavelet transform, threshold denoising, and neural network fitting to adaptively learn wavelet filters and thresholds, enhancing noise resistance and robustness to working condition changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional anomaly detection methods model statistical characteristics under single working condition, then detection accuracy is improved for stable conditions, but misidentification occurs under variable working conditions
Solution Approach 1:
The patent segments the frequency domain into multiple bands using wavelet transform, allowing the model to capture different statistical characteristics in different frequency ranges. This segmentation enables the anomaly detection system to adapt to varying working conditions by analyzing localized frequency components rather than treating the entire spectrum uniformly.
Solution Approach 2:
The patent introduces working condition parameters as an additional dimension to the modeling process. By incorporating these parameters into the statistical characteristic modeling, the system can distinguish between normal variations due to working condition changes and actual anomalies, thereby improving adaptability while maintaining detection accuracy.
2Productivity
If deep learning networks perform end-to-end automated feature extraction, then processing efficiency is improved, but noise resistance deteriorates leading to overfitting
Solution Approach 1:
The patent applies wavelet threshold denoising as a preliminary action before feeding signals into the deep learning network. By removing noise and preserving edge information in advance, the input data quality is improved, which prevents the network from overfitting to noise patterns while maintaining the benefits of automated feature extraction.
Solution Approach 2:
The patent introduces wavelet transform and threshold denoising as intermediary processing steps between the raw signal and the deep learning network. These intermediaries preprocess the signal to remove noise and highlight important features, enabling the network to focus on meaningful patterns rather than noise, thus improving noise resistance while maintaining automation.
3Reliability
If wavelet threshold denoising is applied to remove noise, then noise resistance is improved, but edge information loss occurs
Solution Approach 1:
The patent dynamically adjusts wavelet threshold parameters based on signal characteristics and working conditions. By adapting the threshold parameters rather than using fixed values, the denoising process becomes more selective, preserving edge information that would otherwise be removed while still effectively suppressing noise. This parameter adaptation resolves the contradiction between noise removal and edge preservation.
Data Source
AI summary
The present disclosure discloses a method for detecting anomalies in acroengine variable working conditions based on a multi-band wavelet network, including the steps of: collecting monitoring signals of an acroengine in a healthy state under a plurality of working conditions and corresponding working condition information; constructing and training a multi-band wavelet network using the signals and working conditions; calculating a state monitoring index of the signal using the trained network; determining a monitoring threshold according to the state monitoring index; and determining whether or not the engine is abnormal by comparing the state monitoring indexes of the monitoring signals of the aeroengine under the plurality of working conditions to the monitoring threshold. The present disclosure fully considers the advantages and disadvantages of the signal processing technique with the deep learning network for fault diagnosis.