Abstract: The stochastic vehicle routing problems with uncertain demand and travel time and with simultaneous pickups and deliveries(SVRPSPD) is a  typical combinatorial optimization problem. It is known that the basic differential evolution algorithm(DE)  is not suitable for solving combinatorial optimization problem. To overcome this drawback, a novel discrete differential evolution algorithm(DDE) is proposed by designing new mutation by introducing two bitwise operators of computer language. Then, the proposed algorithm is applied to SVRPSPD and the benchmark problem of VRP to validate the effectiveness of the proposed DDE algorithm. The simulation results are compared with the basic differential evolution algorithm and the existing genetic algorithm. Simulation results show that the DDE algorithm outperforms the others．Not only DDE algorithm obtains better results, but also it converges much faster.

Key words: stochastic programming model, differential evolution algorithm, discrete differential evolution algorithm, vehicle routing problems(VRP)