Random Start Local Search and Tabu Search for a Discrete Lot-Sizing and Scheduling Problem

Abstract

In this paper we describe random start local search and tabu search for solving a multi-item, multi-machine discrete lot sizing and scheduling problem with sequence dependent changeover costs. We present two construction heuristics with a random component; one of them is purely random and another is based on the linear programming relaxation of the mixed integer programming model. They are used to generate initial solutions for random start local search and tabu search. We also propose two ways of exploring the neighborhoods, one based on a random subset of the neighborhood, and another based on exploring the whole neighborhood. Construction and improvement methods were combined on random start local search and tabu search, leading to a total of eight different methods. We present results of extensive computer experiments for analyzing the performance of all methods and their comparison with branch-and-bound, and conclude with some remarks on the different approaches to the problem.