Preference based task allocation in holonic manufacturing

Abstract

This paper presents an agent-based preference model for task allocation in holonic manufacturing systems (HMS). In such systems agents/holons interact to assemble products, using preferences on things like assemblers, end times and so on. We show how to derive a preference-based solution in presence of contention where awarding preferences to the solution of one task can only be done by removing preferences from that of another task. The model can trade preferences in a "fairly competitive" market, enforce timely termination and produce a "reasonable" solution that satisfies the preferences. We have implemented the preference model using a manufacturing example under Java. In this paper we present the results obtained from this implementation.