Tony Bagnall, Research

[ Tony's Home Page | Rotor Machines | Agents]

Information Systems




Mathematical Algorithms Group


  • Personal Background
  • MSc by Research Background
  • PhD Background
  • Publications
  • Modelling the Market in Electricity using Artificial Adaptive Agents
  • The Applications of Genetic Algorithms in Cryptanalysis
  • Links
  • MSc Bibliography
  • Contact Information





    Personal Background

    I completed my first degree in Mathematics and Statistics (1st class) at the University of Hertfordshire in 1994. By fortunate circumstance I ended up in Norwich, taking a job as a part time teaching assistant and part time MSc student in the department of Computer Science. I finished my MSc by research on the application of genetic algorithms in cryptanalysis in January 1996, and graduated in May. The postgrad life appealed to me and I am now doing a PhD using agent modelling techniques (particularly classifier systems) to model the market in electricity, sponsored by a CASE award from the National Grid Company.

    Back to top.

    PhD Background

    I started a PhD in Computer Science in Febuary 1996. I am sponsored by a CASE award from the National Grid Company (NGC). The (provisional) title of my PhD is "The Use of Adaptive Agents to Develop Strategies in the UK Power Pool". My supervisor is Dr George Smith.

    On the privatisation of the electricity industry the government set up a complex market in electricty to facilitate the buying and selling of electricity from generator to supplier. The NGC, which owns the electricity transmission network of England and Wales (the Grid) was entrusted with the administration of this market. The generating companies, dominated by National Power and PowerGen, submit "offer bids" each day to produce electricity. NGC has an interest in understanding how the generators arrive at their bids and how possible changes in the operation of the market effect their bidding strategies. The goal of this project is to produce an adaptive agent simulation of the market, where the "agent" generators evolve bidding strategies in the face of a changing environment, which in some way emulates the actual behaviour of the generating companies. Such a model would allow experimentation with the effect of altering the market rules and may even highlight existing weaknesses in the present market structure.

    Graph - Generator's Market Share

    More information about our approach to this problem is found here.

    Back to top.

    Publications