Research On Multi-objective Optimization Of Green Logistics With Stochastic Demands

Vehicle routing problem(VRP)is an important research direction in operations research and is widely used in many practical scenarios.With the rapid development of social economy,it is of great significance to solve the VRP efficiently for the green and healthy development of enterprise economy.And logistics transportation also ushered in changes.The vehicle routing problem with deterministic demand can no longer meet the actual market demand.Then,the vehicle routing problem with random demand(VRPSD)became the research focus of scholars.At the same time,only considering the optimization cost can no longer meet the requirements of the new logistics transportation,and ensuring high quality services has also become a major criterion for logistics transportation.Therefore,VRPSD needs to be solved by multi-objective optimization algorithm to get a more balanced solution to meet multiple obj ectives.The multi-objective optimization strategy based on genetic algorithm has been widely used in various fields,including VRP.But the traditional multi-objective genetic algorithm has many defects.The main drawback is that the evolutionary process is too random and the evolutionary direction of the population can not be regulated,which leads to a lot of time spent in meaningless computing.It is always a difficult problem in multi-objective genetic algorithm that which method is used to keep good individuals and ensure the diversity of population.This paper proposes a membrane inspired multi-objective algorithm with clustering strategy(MIMOA)to solve VRPSD more efficiently.This paper takes the bi-objective vehicle routing problem with random demand as the application point.The research results include the following aspects:1)After analyzing the requirements of new-type logistics and transportation,a logistics model combining the two needs of green economy and high-quality service was constructed to conform to the current market development trend.2)A guidance strategy combined with multi-objective genetic algorithm is designed.The traditional genetic algorithm uses a single population,but the evolution direction of the population is not always correct,and the algorithm usually converges prematurely and falls into local optimum.MIMOA contains multiple subsystems(populations)and uses guidance strategies to control the development direction of the subsystems.The design of the guidance strategy refers to the skin membrane control strategy[1].Through the guidance strategy,the control subsystem collects high-quality solution sets of the operation subsystem and reasonably feeds back to each subsystem,thereby adjusting the search direction of the operation subsystem.In this way,a part of high-quality solution sets can be exchanged between different subsystems,while low-quality solution sets are eliminated.Multi-group co-evolution makes the algorithm converge faster and can jump out of the local optimal solution with a certain probability3)In the subsystem,a clustering strategy is designed to regulate the evolution direction.In traditional genetic algorithms,invalid crossover and mutation operations cost a lot of computing power,that is,the quality of the solution set of the offspring is worse than that of the parent.The clustering strategy reduces the probability of invalid operations by increasing the restriction on crossover and mutation,and accelerates the convergence speed of the algorithmThe MIMOA proposed in this paper uses real data for experiments,which confirms its practical significance.Research on algorithms at this stage has achieved some results,but the potential ability to develop parallel mechanisms of genetic algorithms is still being considered,so that the computational efficiency of the algorithm is further improved...