Complexity and the Aerospace Industry: Understanding Emergence by Relating Structure to Performance using Multi-Agent Systems

Engineering Doctorate Thesis Abstract. The aerospace industry is at the forefront of technological innovation, both at product level and manufacturing and support levels. Not only are new manufacturing systems computer controlled, they are controlled by networked computers, which, increasingly, are globally connected by public or private internets. Such advances in communication and information systems technology are causing global changes to market places. Drawing upon experience and research in this sector, the increasing challenges faced by large scale complex organisations, exemplified by this sector, are highlighted. I discuss why traditional methodologies are no longer globally appropriate as issues of emergence and complexity come to the fore.

Complex distributed systems often can exhibit behaviour that is not easily predictable when looking at the individual components of the system. The exhibited behaviours are sometimes beneficial and sometimes not. Such behaviour is observable in many natural systems from the swirling patterns of bird flocks to the purposeful social activities of insect colonies. Man-made systems, products and the organisations for their creation and maintenance, can also display such emergent behaviour – often unintended and detrimental. Conversely, benefits can be had when simple systems interact to produce desired complex behaviours where the whole is greater than the sum of its parts. This is particularly evident in biological systems. A study of flocking is presented where novel predator avoidance is developed and explained.

Examining and contrasting the trends and requirements of the aerospace industry in relation to research trends in emergence and multi-agent systems leads to the observation that more research into organisational structure and its quantifiable relationship with organisational behaviour is required. A full set of generic organisational structural metrics are therefore developed, along with charting and agent interaction recording methodologies.

In order to test these metrics a multi-agent system simulation approximating a simple unmanned air vehicle group task is developed. Organisational behaviour such as performance, cost and robustness to failure are recorded alongside organisational structural metrics. These metrics are used to successfully explain key organisational traits in a quantitative manner.

Download PDF Download J M E Gabbai EngD Thesis.pdf

{ 0 comments… add one }

Leave a Comment