Research On The Optimization Of Stochastic Location-Routing Problem

Location-routing problem, LRP for short, is one 6f the most important research projects with the rapid development of integrated logistic system theory. LRP is concerned with location of the facilities, allocation of suppliers and customers to the facilities and the vehicle routing around the depots. Hereby, LRP is a subject about the optimization between location, allocation and vehicle routing structure. At present most classical LRP research is based on determinate information, but in practice, information is often stochastic based on statistic. Consequently, classical model and algorithm of LRP is not often working effectively. It is indispensable to research characteristics of stochastic LRP and design effective models and algorithms. Until now, stochastic LRP research is scant and many items need amelioration and modification. So this paper, in which a multi-objective stochastic LRP is researched, takes on important theoretical and practical value.A multi-objective LRP based on arriving on time, meeting the orders and lowering the total cost is studied in this paper. This paper uses chance-constrained programming and dependent-chance programming to build models of stochastic location-routing problem considering the uncertain facts in reality and the constraint of route. Because LRP is NP hard problem, the paper designs a hybrid artificial algorithm, which is consisted of genetic algorithm and artificial neural network based on stochastic simulation. The computer simulation shows that the model is feasible and the algorithm is effective. And the solution analysis is provided at the end.