Metaheuristics for a crop rotation problem

Abstract

This paper presents a mathematical model adapted from literature for the crop rotation problem with demand constraints (CRP-D). The main aim of the present work is to study metaheuristics and their performance in a real context. The proposed algorithms for solution of the CRP-D are a genetic algorithm, a simulated annealing and hybrid approaches: a genetic algorithm with simulated annealing and a genetic algorithm with local search algorithm. A new constructive heuristic was also developed to provide initial solutions for the metaheuristics. Computational experiments were performed using a real planting area and semi-randomly generated instances created by varying the number, positions and dimensions of the lots. The computational results showed that these algorithms determined good feasible solutions in a short computing time as compared with the time spent to get optimal solutions, thus proving their efficacy for dealing with this practical application of the CRP-D.