| Chain management is an effective method to improve retail management ability. It has a 100-history, which is a regular successful retail management policy and organization style in advanced counties. Logistics distribution is necessary in chain management, which connects chain enterprises' selling and purchasing. It ensures chain enterprise goods circulating regularly, primary chain stores selling normally, and meets market demands. Many of enterprises in our country are in experience dependent stage of planning distribution. In order to improve distribution effect, efficiency and deduce distribution cost, it is necessary to study how quantities of goods distribution to be scientific, routs of goods distribution to be optimized and assorting goods in vehicles to be equitable.In this paper, the author introduces and analyses distribution of chain stores, regards of multi-depot, fleet sizes, vehicle capacity, back goods and time-windows constraints. The author advances two methods. The one is cost based policy of chain stores distribution with single depot in city. The other is cost based policy of chain stores distribution with multi-depot in region. The first model fulfills self-adapted hybrid genetic approach in Matlab6.5 and reaches good result. The second method is extended of the first model. The author advances three-stage policy and uses self-adapted hybrid genetic approach policy.On the genetic approach policy, the author uses distinct integer chromosome and advances a mix policy of best-retain, worst-eliminate and climb search. The author improves the traditional self-adapted crossover probability and mutation probability with evolutionary generation related policy and proves it to be effective through testing.On the base of traditional two-stage policy of solving multi-depot vehicle routing problem, the author adds critical points' part adjustment to the third stage. |
