A Combined Generative and Selective Hyper-heuristic for the Vehicle Routing Problem

Hyper-heuristic methods for solving vehicle routing problems (VRP) have proved promising on a range of data. The vast majority of approaches apply selective hyper-heuristic methods that iteratively choose appropriate heuristics from a fixed set of pre-defined low-level heuristics to either build or perturb a candidate solution. We propose a novel hyperheuristic called GP-MHH that operates in two stages. The first stage uses a novel Genetic Programming (GP) approach to evolve new high quality constructive heuristics; these can be used with any existing method that relies on a candidate solution(s) as its starting point. In the second stage, a perturbative hyper-heuristic is applied to candidate solutions created from the new heuristics. The new constructive heuristics are shown to outperform existing low-level heuristics. When combined with a naive perturbative hyperheuristic they provide results which are both competitive with known optimal values and outperform a recent method that also designs new heuristics on some standard benchmarks. Finally, we provide results on a set of rich VRPs, showing the generality of the approach.