Solution And Implementation Of Logistics Vehicle Routing Problem Based On Memetic Algorithm

Logistics industry is a basic and strategic industry supporting the development of national economy.However,the logistics cost is still at a relatively high level,and the efficiency of logistics distribution still has much room for improvement.Effective solution of vehicle path planning is the key to improve delivery efficiency and customer service experience.This dissertation studies the problem of vehicle routing planning related to logistics,and the main work is as follows:(1)For one to many to one vehicle routing problem,a multi-objective memetic algorithm based on sorting priority first(MOMASPF)is proposed,which mainly solves two objective in terms of route length and workload.Six different service strategies are adopted to simulate the service situations of the logistics distribution process in real life.The comparision study with four different algorithms on six benchmark problems shows the efficiency of MOMASPF.(2)In order to further optimize the vehicle routing problem,this dissertation proposes a multi-objective memetic algorithm based on three-dimensional dynamic request prediction(3DPMOMA).Based on dynamic vehicle path planning,three objectives,namely route length,served time and workload,are optimized simultaneously.Dyanmic requests are predicted in three dimension including two space coordinates and time based on the statistical distribution of historical data.Based on the accurate prediction of dynamic requests,the route is planned in advance.When the dynamic requests appear,the dynamic requests can be quickly responsed based on the predicted information,which can highly improve customer service experience.(3)Finally,to further improve the planning efficiency of static nodes and improve the performance of the algorithm.According to transfer learning,this dissertation proposes similarity matching based muliti-objective memetic algorithm(SMMOMA).Extract useful information in similar historical issues to help solve existing problems.The historical information database is established and features are extracted by matching the features between current problem and historical problems.The most similar historical problem is found out to study the planning rules to promote current problems.This dissertation improvs the solving of one-to-many-to-one vehicle routing problems by using new multi-objective memetic algorithms and historical information,and the method of MOMASPF is proposed to reduce the path length effectively.In order to improve the customer experience and shorten the waiting time of customers,the idea of learning new information based on historical information is proposed to effectively,which can provide experience and reference for later technologies and related research.