AI Malfunction Diagnosis Using EMD Signal Image Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies can only predict the failure of automated apparatuses but fail to distinguish the type of malfunction, leading to ineffective preventive maintenance.
Innovation Solution
A method involving signal decomposition using empirical mode decomposition (EMD) to generate sub-signals, transforming them into grayscale images, and inputting these images into a neural network model for malfunction reason classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual analysis is used for malfunction diagnosis, then practical judgment experience can be applied, but the experience is difficult to pass on to others and requires heavy reliance on individual expertise
Solution Approach 1:
The patent replaces manual analysis with automatic identification using artificial intelligence and signal processing systems. The mechanical/manual process of expert judgment is substituted with automated algorithms including signal decomposition, feature extraction, and classification models that can objectively diagnose malfunctions without relying on individual expert experience.
Solution Approach 2:
The patent creates a digital model or virtual representation of the malfunction diagnosis process through training AI systems on historical data. This copying approach allows the system to learn from past cases and replicate expert-level diagnostic capabilities, making the knowledge transferable and reusable across different operators and situations.
2Reliability
If existing prediction technology is used, then apparatus failure can be predicted, but the type of malfunction cannot be distinguished
Solution Approach 1:
The patent segments the diagnosis process into multiple stages: signal decomposition into constituent components, feature extraction from each component, and classification of different malfunction types. This segmentation allows the system to not only predict failure but also identify specific malfunction categories by analyzing different aspects of the signal separately.
Solution Approach 2:
The patent transforms one-dimensional signal data into multi-dimensional feature space through signal decomposition and feature extraction. By adding dimensions such as frequency domain characteristics, time-frequency representations, and statistical features, the system gains the ability to distinguish between different malfunction types that would be indistinguishable in the original signal domain.
3Reliability
If generic maintenance is performed on apparatus about to fail, then basic maintenance can be carried out, but targeted preventive maintenance for specific problems cannot be implemented
Solution Approach 1:
The patent enables local quality in maintenance by providing different maintenance recommendations based on the specific malfunction type identified. Instead of applying uniform maintenance procedures to all failing apparatus, the system analyzes the specific characteristics of each malfunction (such as bearing defects, gear issues, or motor problems) and prescribes targeted maintenance actions appropriate to each local condition.
Data Source
AI summary
A method for diagnosing a reason of a malfunction is provided. The method includes: receiving a signal to be diagnosed; decomposing the signal to be diagnosed into a plurality of sub-signals; transforming each of the plurality of sub-signals into a corresponding grayscale image; and inputting the corresponding grayscale images to a neural network model, and outputting a malfunction reason classification result through the neural network model. Accordingly, the method can be used for diagnosing the reason of the malfunction and solves the problem of incapable of diagnosing the reason of the malfunction. In addition, a device and a computer-readable recording medium for diagnosing the reason of the malfunction are also provided.