Motor Fault Diagnosis Using Edge Autoencoder and Cloud Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional predictive maintenance for motor drive systems requires large hardware resources, complex algorithms, and is limited in diagnosing fault types, with high network bandwidth consumption and dependence on labeled data.
Innovation Solution
A fault diagnosis method using an autoencoder for unsupervised learning to identify fault data, trained at the cloud or edge, reducing network bandwidth by uploading only fault data and minimizing hardware resource usage, with user feedback for model refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning methods with neural network models are used for fault diagnosis, then fault diagnosis accuracy is improved, but hardware resource consumption increases
Solution Approach 1:
The system segments the fault diagnosis process into two parts: local autoencoder for anomaly detection and cloud-based fault diagnosis model for classification. The motor driver only performs lightweight anomaly detection locally, while complex fault diagnosis is performed in the cloud, reducing onboard hardware requirements while maintaining high diagnostic accuracy.
Solution Approach 2:
The patent introduces an intermediary cloud server that receives anomaly data from the motor driver, trains the fault diagnosis model, and returns results. This intermediary handles the computationally intensive tasks in the cloud, allowing the motor driver to use minimal hardware resources while achieving accurate fault diagnosis.
2Measurement precision
If all motor operating data is uploaded to the cloud for analysis, then comprehensive fault diagnosis is achieved, but network bandwidth consumption increases
Solution Approach 1:
The system extracts only the essential anomaly data from the complete motor operating dataset and transmits only this extracted information to the cloud server. The autoencoder identifies and separates fault-related patterns from normal operation data, ensuring that only necessary data is uploaded, thus reducing network bandwidth consumption while maintaining diagnostic comprehensiveness.
Solution Approach 2:
The autoencoder performs preliminary anomaly detection and data filtering before data is transmitted to the cloud. By pre-processing and identifying only the anomalous patterns that require further analysis, the system avoids transmitting the entire dataset, significantly reducing network bandwidth requirements while preserving complete fault diagnosis capability.
3Use of energy by moving object
If conventional signal processing algorithms are used, then fault diagnosis is achieved with limited hardware, but algorithm complexity increases and fault type coverage is limited
Solution Approach 1:
The patent changes the approach from fixed conventional algorithms to adaptive deep learning models. The autoencoder and fault diagnosis model are trained on actual motor data patterns, allowing the system to automatically adapt to different fault types and operating conditions without requiring complex manual algorithm design or multiple fixed algorithms for different fault scenarios.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A fault diagnosis method for a motor, comprising: acquiring a data group of a motor from a motor driver, and uploading the data group to a cloud server or an edge device, the data group comprising multiple motor operating parameters collected at the same moment (110); training an autoencoder on the cloud server or the edge device according to the data group, downloading the trained autoencoder to the motor driver, and performing, at the motor driver, loading at runtime on the autoencoder (120); after the loading at runtime of the autoencoder is completed, identifying fault data in the data group of the motor by using the autoencoder, and uploading the fault data to the cloud server or the edge device (130); and training a fault diagnosis model on the cloud server or the edge device according to the fault data, and performing fault diagnosis on the fault data in the operation of the motor by using the trained fault diagnosis model (140).