Fault detection and classification in linear actuators
Using AI to detect and diagnose faults in linear actuators.
Linear actuators play a critical role in industrial automation, powering a wide range of machinery and processes. However faults and malfunctions in these actuators can lead to costly downtime production delays and compromised operational efficiency.
Optimized Artificial Intelligence (AI) algorithms enable predictive maintenance and can diagnose faults in real-time, enhancing reliability and productivity by minimizing unplanned downtime. AI-based diagnosis systems go beyond simple detection and can offer insights into the specific fault type and the root cause of the problem, enabling targeted and efficient troubleshooting by engineers and maintenance teams.
Optimized Artificial Intelligence (AI) algorithms enable predictive maintenance and can diagnose faults in real-time, enhancing reliability and productivity by minimizing unplanned downtime. AI-based diagnosis systems go beyond simple detection and can offer insights into the specific fault type and the root cause of the problem, enabling targeted and efficient troubleshooting by engineers and maintenance teams.
Approach
This use case is based on the "Detection and Diagnosis of Faults in Linear Actuators" dataset from Cranfield University. The goal was to detect and classify 4 states of a linear actuator: normal, backlash, lack of lubrication, and spalling.
Using data collected from a linear actuator, the dataset contains several .mat files that can be converted to .csv files (using the SciPy library). All files corresponding to the same behavior were concatenated to have only four files in the end: Normal.csv, Backlash.csv, LackOfLubrication.csv and Spalling.csv. As the dataset contains a lot of spalling data compared to the other classes, only half of the spalling data was used to ensure a more balanced dataset for the training.
We then used NanoEdge AI Studio to create an N-class classification project based on these inputs, capable of classifying the state of the linear actuator.
Using data collected from a linear actuator, the dataset contains several .mat files that can be converted to .csv files (using the SciPy library). All files corresponding to the same behavior were concatenated to have only four files in the end: Normal.csv, Backlash.csv, LackOfLubrication.csv and Spalling.csv. As the dataset contains a lot of spalling data compared to the other classes, only half of the spalling data was used to ensure a more balanced dataset for the training.
We then used NanoEdge AI Studio to create an N-class classification project based on these inputs, capable of classifying the state of the linear actuator.
Sensor
Current sensor and Accelerometer.
Data
4 classes of classification Normal, LackOfLubrication, Backlash, and Spalling
Signal length 6000
Data rate 25 Hz
Signal length 6000
Data rate 25 Hz
Results
N-class classification:
90.65% accuracy, 20.3 Kbytes of RAM, 184 Kbytes of Flash memory
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
90.65% accuracy, 20.3 Kbytes of RAM, 184 Kbytes of Flash memory
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 Any STM32 MCU
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.
