Application of a Novel Discrete Differential Evolution Algorithm to SVRP

In order to propose a novel discrete differential evolution algorithm for stochastic vehicle routing problems (SVRP), the two bit wise operators of the computer language are introduced. In the algorithm, the individuals are represented as natural numbers, and new mutation and revised operators are developed for this representation. Computational simulations and comparisons based on Benchmark Problem of VRP and a larger-sized problem of SVRP. Results demonstrate that not only the proposed algorithm obtains better results than the basic differential evolution algorithm (DE) and the existing genetic algorithm (GA), but also it converges much faster.