Solving capacitated vehicle routing problems by modified differential evolution

To solve capacitated vehicle routing problems (CVRP), a modified differential evolution (MDE) was employed, which was derived from the differential evolution (DE) and incorporated with the genetic reproduction mechanisms, namely crossover and mutation. The Greedy Subtour Crossover (GSX) was employed to generate an offspring to denote the difference of the parents. A modified ordered crossover (MOX) was employed to perform mutation to generate trial vector with user defined parameters, the parameters were used to control the rates of the target vector components and the mutated vector components in the trial vector. Moreover, the 2-opt local search was implemented to optimize the sub routings. MDE was implemented to the well-known CVRP benchmarks. The simulation results have shown the feasibility and the effectiveness of the approach.