The university has a large repository of electricity and gas readings dating back as far as 1996 from 1300 meters. A project is currently ongoing to centralise and automated the collection of this data so it can be made available to interested parties. An associated talk I gave on the subject is: here.
This project aims to construct tools for analysing this data. Initially this will take the form of software components (perhaps supported by a relational database) which can quickly and efficiently answer questions such as "what is the electricity consumption of the physics department?". This involves summing values from various meters (some of which will be an apportioned percentage rather than the entire consumption). Some meters are read at 30 minute intervals and some on a monthly interval - answers must combine these values correctly. Further work will then consider techniques for measuring or estimating the reliability and accuracy of the data by looking for anomalous readings or summarising different reading techniques.
Will suit a student interested in database design, indexing techniques, algorithmic performance and scaling, inference and probability.
Software development will be in Java and so prior programming experience is essential.
Project supervisor: Dr Andrew Rice