Description

Electric motors are everywhere around us , fans, washing machines, electric cars, fume extractors, grinders ... . Motors and electric generators are just as ubiquitous in industry, just much bigger. Overall, electric motors are reliable, and are thus kept in use for many years, often decades. With wear and tear however, problems can develop over time and this can lead to major damage as shown below from the Bold Power Station after a turbine explosion.

One solutions is to perform some regular maintenance to detect problems before they occur. Unfortunately this is costly, not only because of the maintenance cost, but also because this requires shutting down equipment for a substantial length of time, which leads to loss of earnings. To tackle that problem, a small Cambridge company, Faraday Predictive , has developed a method to monitor the current drawn by electric motors to detect problems early and well before any damage is done.

The method consists in continuously comparing the current drawn by the motor to what it is when operating normally. Small deviations can then be indicative of a developing fault. The problem then consists in analysing these deviations and try to infer from them the importance and the nature of the fault.

The aim of the project will be to analyse the signal measured on some motors and model them using various techniques which can easily be implemented in a programming language such as python. Most of the signals can be described as a combination of an amplitude and phase modulation, and the idea is to extract the modulation functions.

No physics knowledge are required for this project as the problem consists in analysing time series, but good programming skills are necessary. Mathematically, the main tool will be convolutions and Fourier series, both of which which will be reviewed at the beginning of the project, as well as similar transformations which will be studied as part of the project itself.

Having recently established contacts with Faraday Predictive I have already received some signals to analyse and I am also expecting to receive some more from them as well as establish a mutual collaboration during the second part of the project.

Prerequisites

• Programming and Dynamics (MATH1041)
• Mathematical Modelling is strongly recommended (MATH2637)

Resources

• The Hilbert transform. (Mathias Johansson)
• Hilbert transform: Mathematical theory and applications to signal processing Mans Klingspor.
• Signal Processing (1) Signal Processing (2) Signal Processing (3) Signal Processing (4) Signal Processing (5) Signal Processing (6) Signal Processing (7) Signal Processing (8)
• Introduction to Digital Filters, (Julius O. Smith III)
• Digital Filters (3rd edition) R.W. Hamming. Dover Publication (1989)
• Digital Signal Processing (M.H. Hayes, Schaum's outlines).