Elitism Mechanism in the Genetic Optimization of Networked Distribution Systems

This paper addresses the inventory control problem in distribution systems with a non-trivial, networked connectivity structure and goods relocation with delay. The distributed (r, Q) inventory management policy is applied to control the replenishment process at network nodes in response to uncertain market demand. The non-dominated sorting genetic algorithm (NSGA) is used to solve a nonlinear three-objective optimization problem focused on reducing operational costs and providing high customer satisfaction. The performed numerical studies confirmed the effectiveness of the NSGA in adjusting the parameters of the (r, Q) policy in the considered class of logistic systems. The consequences of applying the elitism mechanism are discussed via observation of various performance indicators commonly considered in multi-objective optimization problems. The obtained Pareto set approximations have been compared in terms of cardinality, convergence, and diversity of potential solutions.