Electric drive diagnosis (11 faults)
Classify data based on different types of faults in an electric drive.
Electric drives are used for various applications and are becoming increasingly performant. They can be monitored in a very precise way thanks to the data they provide during operation. This data can also be used to enhance the application using Predictive Maintenance techniques.
Predictive maintenance consists in optimizing maintenance strategies by automatically detecting aging or predicting anomalies. Machine learning translates the data generated by the system into meaningful data. We have added AI solution directly next to the Motor Control algorithm to run both anomaly detection & classification and motor control on the same microcontroller, reducing cost of system and optimizing resources. This approach an easily be adapted to many motors and for various applications
Predictive maintenance consists in optimizing maintenance strategies by automatically detecting aging or predicting anomalies. Machine learning translates the data generated by the system into meaningful data. We have added AI solution directly next to the Motor Control algorithm to run both anomaly detection & classification and motor control on the same microcontroller, reducing cost of system and optimizing resources. This approach an easily be adapted to many motors and for various applications
Approach
The basic components are the drive motor to be tested (permanent magnet synchronous motor), a torque measuring shaft, the test modules and a load motor (synchronous servomotor).
The tests are performed at various bearing loads, torque loads and speeds.
The different combinations of defects, loads and speeds result in 11 classes.
The signal is current based.
The tests are performed at various bearing loads, torque loads and speeds.
The different combinations of defects, loads and speeds result in 11 classes.
The signal is current based.
Sensor
Generic current sensor
Data
>> Download the dataset
11 classes of data 11 different combinations of defects, loads and speeds
Signal length 48 (1 axis), 5300 signals per class
11 classes of data 11 different combinations of defects, loads and speeds
Signal length 48 (1 axis), 5300 signals per class
Results
11 classes classification:
98.56% Balanced accuracy, 0.5 KB RAM, 140.6 KB Flash
Green points represent well classified signals. Red points represent misclassified signals. The classes are on the abscissa and the confidence of the prediction is on the ordinate
98.56% Balanced accuracy, 0.5 KB RAM, 140.6 KB Flash
Resources
Model created with NanoEdge AI Studio
A free AutoML software for adding AI to embedded projects, guiding users step by step to easily find the optimal AI model for their requirements.
Compatible with STM32
The STM32 family of 32-bit microcontrollers based on the Arm Cortex®-M processor is designed to offer new degrees of freedom to MCU users. It offers products combining very high performance, real-time capabilities, digital signal processing, low-power / low-voltage operation, and connectivity, while maintaining full integration and ease of development.
