Evolving Middlemen Strategies For Simple Supply Chains

This research is concerned with the evolution of middlemen strategies for simple, but non-trivial, supply chains. The work contributes a language for modelling simple supply chains, a framework for representing middlemen strategies and a platform for evolving these strategies. The work shows that for reasonably complex economic problems, that contain a great deal of uncertainty, the evolution of effective strategies provides a useful technique for investigation. In the course of this thesis the Simple Supply Chain Model (SSCM) is defined, allowing the modelling of these problems. The SSCM Strategy Framework (SSF) is introduced providing a framework capable of specifying strategies that are able to tackle SSCM problems. The SSCM Market Simulation System (SMSS) strategy evolution platform is then considered. The SMSS supplies a system based on Population Based Incremental Learning with Guided Mutation (PBIL+GM) that is capable of evolving effective SSF-based strategies within SSCM-defined environments. Experimentation using the SMSS demonstrates that effective middlemen strategies may be evolved for a variety of supply chain environments and that these strategies tend to be specialised to their home environments.