Build a Motor Defect Detector Solution

This solution covers the basic implementation of fast Fourier transforms (FFT), logistic regression, K-Means clustering, and Gaussian Mixture Model (GMM). It also shows how FFT is helpful in feature engineering of the vibrational data of a machine.

Target Operating System Ubuntu* 16.04 LTS
Programming Language Python*
Time to Complete 30 minutes

View on GitHub*

What It Does

Learn to build and run an application that:

Detects irregularities in machine vibration

Local processing using the sklearn library

Sends data to the cloud for aggregation and viewing

There are several methods that do not require training of a neural network to be able to detect failures, starting with the most basic fast Fourier transforms (FFT) to the most complex Gaussian Mixture Model (GMM). These have the advantage of being able to be reused with minor modifications on different data streams. They also do not require a lot of known previously classified data (unlike neural nets). In fact, some of these methods can be used to classify data in order to train deep neural networks (DNN).

What We Used

Hardware Requirements

Tested on the IEI TANK* AIoT Developer Kit

Software Requirements

Ubuntu* 16.04 LTS

Python* 3.5 with the following libraries:

  • NumPy
  • pandas
  • matplotlib.pyplot
  • sklearn
  • SciPY Stack