Convolutional Dictionary Learning for Label-Consistent Fault Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine fault diagnosis methods face challenges with limited labeled data, high computational cost, and shift-variant behavior in patch-based dictionary learning, leading to ineffective representation learning and redundancy in signal processing.
Innovation Solution
A method and system utilizing label-consistent convolutional dictionary learning to optimize dictionary atoms, coefficients, and weights through joint optimization, incorporating label consistency terms to learn class-discriminative dictionaries for improved fault classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If patch-based dictionary learning is used to reduce computational cost, then computational efficiency is improved, but shift-variant behavior causes loss of signal structure information
Solution Approach 1:
The patent merges multiple patch-based dictionaries into a single global dictionary that captures shift-invariant features across the entire signal. This combines the computational efficiency of patch-based methods with the structural integrity of global signal representation, eliminating the shift-variant behavior that causes information loss.
2Quantity of substance
If conventional dictionary learning is used to learn from limited data, then data efficiency is improved, but class-discriminative information is lost due to lack of label utilization
Solution Approach 1:
The patent incorporates label consistency feedback into the dictionary learning process. The label consistency term provides feedback during optimization to ensure that learned dictionary representations maintain class-discriminative information, allowing effective learning even from limited labeled data.
3Adaptability or versatility
If Deep Neural Networks are used to learn meaningful representations, then representation learning capability is improved, but large amounts of labeled data are required
Solution Approach 1:
The patent changes the fundamental parameters of the learning approach by using dictionary learning with label consistency constraints instead of deep neural networks. This alternative parameterization achieves representation learning with significantly reduced data requirements by leveraging the structured nature of signal dictionaries and incorporating label information directly into the learning objective.
Data Source
AI summary
Existing Convolutional Dictionary Learning (CDL) based machine fault classification do not utilize label information while learning the dictionary, hence the representation learned are not class-discriminative. Method and system disclosed herein provide a label-consistent convolutional dictionary learning approach for machine fault classification. The approach involves generating a training data for a classifier, wherein coefficients forming a plurality of class-discriminative features form the training data. The training data is then used to train a classifier, which is then used to perform machine fault classification for a given test data.