Edge Computing for Predictive Maintenance

Project description

Edge Computing can be a useful tool for executing Predictive Maintenance. The biggest advantage is a low-power capability because of avoiding network transfers of regular sensor readings and making decision on the controller itself. This project will create a proof-of-concept for measuring sound and making decisions on the controller itself.
The second goal of the project is to compare two different edge computing platforms based on ease of use, sensor sensitivity and any other relevant features.

Context

This project is related to production. It explores the possibilities of using AI on the Arduino Nano BLE Sense Rev2 and the Thingy:53 for predictive maintenance on the edge. This could minimize unscheduled downtime due to machine failure and lead to improved yields and improved machine performance.

Results

In the end, a model for detecting irregularities in sound was developped and deployed on both devices, and guides for developing and deploying your own models were delivered.

This can be used by students in future projects to speed up their development and utilize it in their projects.