Abstract: In order to satisfy logistics distribution demand under electronic commerce,we modify the traditional vehcile scheduling model,amend objective function based on expense minimization,and add time restriction,goods capacity restriction,maximum vehicle working time,multi-types of vehicles,load capacity restriction,maximum running distance and so on into the rest riction couditions,so as toim prove the applicability and universality of model Since vechile scheduling problem is NP puzzle,we get the optimal solution by improved two-phase algorithm That is,in the first phase,the customer group is divided into some regions through fuzzy hierarchy clustering analysis method,and each region is also divided into some small scale sub-groups satisfying some restiction conditions through scanning algorithm. In the second phase,we optimize the line of each single TSPTW model according to to cust omer dot in each group. Therefore,improved hybrid genetic algorithm is used to get the optimal solution. Finally,computational tests demonstrate the efficiency and feasibility of our algorithm.

Key words: VRPTN(vehicle routing problem with time window), fuzzy hierarchy clustering method, hybrid genetic algorithm, improved two phase algorithm