MATLAB based dashboard showing Tool health dial, Machine ON/OFF indicator and x,y & z accelerations
Irregular maintenance of industrial componenents costs money & time to company, especially in manufacturing sector. To takle this issue, I've been working on Predictive Maintenance of CNC tools. With this project, an attempt is made to predict whether the tool is going to function well in next specified duration or not. If not, the authorized person is alerted to change the tool.
- How? Neural Network (ML) model to predict anomaly based on standard fracture data of tool material
- Prediction duration: next 5 min
- Hardware: MPU6050 accelerometer gyroscope, Rpi, CNC machine
- Software: Matlab, Firebase
- Working: Raw Data -> Filtering -> Data Reduction -> Feature Extraction
- Dimentionality Reduction: Principal Component Analysis
- Filtering: Butterworth filter
- Detail report:
Link
Scope
This project is irrespective of type of tool and can be installed on any mechanical machine. Accelerometer will send the vibration data to firebase through Rpi and prediction can be performed based on respective material's standarad data.