Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 94
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY
Edited by:
Paper 19

Agent Swarm Optimization: A Platform to Solve Complex Optimization Problems

I. Montalvo, J. Izquierdo, M. Herrera and R. Pérez

Department of Hydraulic Engineering and Environment, Universidad Politécnica de Valencia, Spain

Full Bibliographic Reference for this paper
, "Agent Swarm Optimization: A Platform to Solve Complex Optimization Problems", in , (Editors), "Proceedings of the Seventh International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 19, 2010. doi:10.4203/ccp.94.19
Keywords: water supply networks, optimal design, multiobjective optimization, swarm intelligence, parallel computing, distributed computing.

Summary
Agent swarm optimization (ASO) is a common framework for population-based algorithms in general. It is aimed at supporting decision-making processes by solving either single or multi-objective optimization problems. In this direction, good results were obtained after being applied to water distribution system design. Among the most important capabilities of the algorithm it should be noted the concept of including a high human interaction during the solution process. Users can focus the search in some desired part of the Pareto front and also they can influence the behaviour of the agents by proposing a potential solution to the algorithm.

In ASO rule-based agents could be also added to the solution process. The use of rule-based agents, increases the probability of finding good solutions for a problem because those agents are closer to the essence of the problem. ASO makes it possible to have rule-based agents and evolutionary algorithms working together to solve the same problem.

The use of parallel and distributed computing makes it possible that ASO could be run on different computers involving several swarms at the same time. There could be a simile to a cloud of agents but it would not be taking into account the amount of internal dynamics: it is not a cloud, it is a swarm. The hierarchical mechanism used for constructing the approximated Pareto front helps to make this process more efficient. Additionally, the algorithms could increase automatically its own population when needed in order to have enough agents to cover the whole Pareto front. These characteristics help to obtain a good approximation of the Pareto front considering the problem being solved.

Integrating the search capacity of algorithms and the ability of specialists to redirect the search towards specific interest points, based on their experience in solving problems, results in a powerful collaborative system for finding solutions to engineering problems. Most of the artificial intelligent works try to substitute humans in some of their tasks; ASO is not intended to be a substitute for a work team but to be integrated with it. Artificial agents can profit from the creativity and ideas of human experts to improve their own solutions; in their turn, human experts can profit from the speed and search capabilities of artificial agents to explore broader solution spaces.

To run ASO successfully it will be always necessary to have informatics applications with a proper interface in order to facilitate the work for final users. Applications should be also oriented to use as much as possible the informatics resources of enterprises, ASO and WaterIng, the software developed and used in this work, are just some steps in that direction.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £125 +P&P)