Multi-agent decision system for solving dynamic location in manufacturing cells

Abstract

This study proposes a new approach to multi-criteria decision support systems (MDSS) using an agent-based approach. The dynamic location /distribution problem is formulated as a mixed integer programming model which forms the basis of the MDSS and whose objectives are to minimize total weighted delivery time, purchasing cost, total weighting cost, or total number of vehicles, by using multi-agent negotiation and cooperation The proposed models and methods are applied interactively to determine the weight of each criterion. The multiple objectives are then transformed into a single objective. A prototype system for delivering goods from multiple suppliers to customers is presented. Since the proposed problem is NP-hard, business software does not work well as the number of suppliers or customers becomes a large (over 100). Thus, a heuristic algorithm is developed and adopted. Compared to the lower bound of the optimal solution, the heuristic algorithm performs well and outputs an adequate solution. An agent-based approach is presented to determine favorable alternatives in changing market environments