An Evolutionary Algorithm for Decisional Assistance to Project Management

Abstract

Manufacturers must always develop products faster and better to satisfy their client’s requirements. To help them, we have developed and experimented with a methodology to improve the management process by connecting it with the design process. An issue for the project manager is to select an organization from among the possible ones to schedule the project tasks in order to reach the targets in terms of costs, duration, quality, and so forth. This constitutes a tricky operation, because many options defined during the design can be associated to each task of the project. Choosing and optimizing the schedules is a combinatorial problem which can be solved by heuristic. This document explores the use of evolutionary algorithms to help the decision maker. It introduces the industrial context and presents our methodology to connect the design and project management processes, expressing the problem as a multi-objective optimization one. We detail the scenarios selection process and demonstrate which performances are obtained.

Hybridation of Bayesian networks and evolutionary algorithms for multi-objective optimization in an integrated product design and project management context

“…The multi-objective aspect invites to provide the decision maker with a panel of good solutions which represents various compromises between identified objectives (Pareto front). This kind of problem is often addressed using metaheuristic optimization methods (see Baron, 2006 andChelouah et al, 2009) for studies of different metaheuristic applied to the scenarios selection problem). Among those, this study relies on evolutionary algorithms (EA) (Holland, 1975), as illustrated on the left part of Fig.…”