Two agent-based approaches for solving multi-objective multi-constraint optimization problems

Abstract

In this thesis, we propose two different agent-based approaches, the Coarse Grained Agent System (CGAS) and Fine Grained Agent System (FGAS) to solve Multi-Objective Multi-Constraint Problems (MOMCP), which represent the nature of many real life problems. CGAS gives a generic agent-model for multi-objective multi-constraint problem, while FGAS caters more for distributed multi-objective multi-constraint problem like most multiagent systems. We apply our approaches to solve the Inventory Routing Problem with Time Window (IRPTW). Experimental results indicate CGAS achieves a much better results than previous work, while FGAS runs very fast, whose run time is about one-sixth of CGAS with solution quality less than 10% poorer, which is still better than previous work.