Network Topology Based Hybrid Particle Swarm Optimization Algorithm And Logistics Optimization

Particle swarm optimization(PSO)provides a way to solve the problem of logistics optimization.However,the rapid development of e-commerce has put forward higher requirements for prompt response capabilities of the modern logistics industry.The scale of logistics system is becoming larger and larger,and the factors involved are increasing.As a result,logistics problems are becoming more and more complicated.Some classical logistics optimization problems,such as multimodal transportation problem and distribution center location problem,etc.,have become hot research topics due to their new requirements and challenges.The existing PSO algorithms show insufficient optimization ability for logistics problems that have lots of local optimum and complex constraints.They have the defects of slow convergence rate and easy falling into local minimum.Therefore,it is necessary to study the mechanism of the PSO algorithm to improve the optimization ability.It is also necessary to study how to combine PSO with other efficient local optimization algorithms adaptively,so as to realize the complementary advantages of both by judging the global-local search conversion window adaptivelyThis thesis focuses on the typical modern logistics problems,such as the multimodal transportation problem and distribution center locating problem,which are featured with nonlinear,multi-optimum and complex constraints.Firstly,theoretical research on the PSO algorithm is carried out to propose a new PSO algorithm with better optimization capabilities and higher optimization efficiency.Then,the proposed PSO algorithms are applied to solve the logistics models with complex constraints.The main contents and achievements are summarized as follows.1)The neighbor network topology in PSO is dynamically updated using the link prediction method with fitness-based weighted index.A network topology based PSO(NTPSO)is proposed.In NTPSO,a dimensional comprehensive selection strategy based on neighbor network topology is established for learning objects of individual previous best solution(pbest).NTPSO also uses dynamic multi-group individual learning strategies and leads to better search capability of PSO.The computational performance of the NTPSO algorithm is validated by numerical tests on the multi-dimension functions of benchmark.Comparisons are also made between NTPSO and other algorithms,and the results verify the superiority of NTPSO algorithm.2)Aiming at the inherent problem of slow search speed in the local convergence stage of PSO algorithm when solving multimodal problems,a network topology based PSO algorithm with adaptive local search(ALS-NTPSO)is developed by combining the NTPSO algorithm with the traditional local search(LS)method which has fast convergence ability.The adaptive coupling strategy of NTPSO and LS is proposed to solve the key issue of when to start LS.The local convergence rate of NTPSO and LS is compared and analyzed in theory,and the necessity of an adaptive coupling strategy is analyzed.The quasi-entropy index is proposed to identify when to adaptively trigger the LS method,and its performance is proved theoretically and numerically.The computational performance of the proposed hybrid algorithm has been verified by numerical tests on multi-dimensional problems of the benchmark functions.The experimental results compared with other algorithms further verify the efficiency and superiority of the ALS-NTPSO algorithm.3)Considering the demand for green logistics,mathematical models of multimodal transportation under different carbon regulatory policies are established.The NTPSO algorithm is applied to solve these optimization problems,which provides a new scheme for green multimodal transportation.A simple and efficient coding scheme to solve multimodal transportation models under different carbon regulatory policies by NTPSO algorithm is studied.The comparison results with other algorithms verify the effectiveness and performance advantages of NTPSO algorithm in solving this kind of optimization model with complex constraints.The impact of different carbon regulatory policies on the decision-making of multimodal transport is also analyzed.4)In view of the new requirements for the multi-distribution centers locating due to the rapid development of e-commerce,a robust optimization model for multi-distribution center location under uncertain conditions and a multi-distribution center location model considering uncertainty and survivability are established.The improved PSO algorithm is applied to solve the mixed integer programming problem with complex multimodal and NP hard characteristics,which verifies the application effect of ALS-NTPSO algorithm.The superiority of the proposed algorithm in solving the logistics optimization problem with NP-hard and multimodal characteristics are verified by numerical experiments and comparison with other algorithms.