Research On Vehicle-Cargos Matching Problem With Three-Dimensional Loading Constraints

With the rapid development of e-commerce,traditional logistics enterprises recognize the importance of logistics process optimization.In order to schedule reasonable vehicle routes saving logistics distribution costs,this paper studies Vehicle-Cargos Matching problem,and finds that the algorithms of VCM are the simplest matching model,ignoring the actual factors such as the distance between the customers,the time window of customers,and the volume of the loaded cargos.Therefore,this paper researches the joint optimization problem of ThreeDimensional Loading Problem with Vehicle Routing Problem with Time Windows,and the following work and two main innovations are list as follow:1.A literature survey of 3DLP and VRPTW was carried out,and the two were inseparable in actual distribution,so the research work of 3L-CVRPTW was carried out.Finally,this paper selected a two-stage algorithm to optimize.2.This paper establishes a problem optimization model for 3L-CVRPTW,defining four objective functions that minimize the total distance,minimize the number of vehicles,maximize the minimum vehicle load rate,and maximize the minimum vehicle volume utilization,and normalize each objective function to reduce computational complexity.3.In this paper,an improved partial-random key genetic algorithm(IBRKGA)is proposed to improve the utilization of the residual space of the algorithm.The residual space is divided,updated,merged and the box placement strategy is improved.Finally,BRKGA is used to optimize the placement order and placement direction of the boxes to improve the customer loading rate.Finally,the results were tested on the Martello dataset,and the results were excellent.There are 68.75% cases exceeded the similar algorithms.4.For 3L-CVRPTW,this paper proposes a multi-stage hybrid algorithm(MSHA),which is divided into four stages: loading,request merging,routing optimization and route integration.The loading phase uses IBRKGA to independently pack each customer.The request merge phase is added,and the customer who does not satisfy the fill rate threshold after independent packing need to be merged.The customers merge according to the similarity,so that the goods of the two customers are uniformly boxed to increase the packing rate.In comparison with P1R2 in the GI series cases,the number of vehicles has achieved good results,and the parameter analysis of the similarity threshold confirms the necessity of the customer request phase.Finally,the paper combines MSHA with the logistics big data platform to solve the real logistics distribution,and proves the practicability of the algorithm,which provides guidance for the operation strategy of logistics distribution enterprises.