A multi-population evolutionary algorithm with single-objective guide for many-objective optimization

Optimization of multistage timeliness transit consolidation problem using adaptive-weighted genetic algorithm

Cargo consolidation is becoming a crucial part of international transportation and changing the customer consumption patterns of the international community. Poor connections between different operations and the delay of international express have motivated sellers and logistics organizers to put timeliness first in international multimodal transport, especially during the COVID-19 epidemic. However, for cargo with small quality and multiple batches, designing an efficient consolidation network presents a set of unique challenges, including the coupling of multiple origins and destinations (ODs), and fully utilizing the capacity of the container. We defined a multistage timeliness transit consolidation problem to decouple the multiple ODs of the logistics resource. By solving this problem, we can increase the connectivity between different phases and make full use of the container. To make this systematic multistage transit consolidation more flexible, we proposed a two-stage adaptive-weighted genetic algorithm that mainly focuses on the edge area of the Pareto front space and the diversity of the population. Computational experiments indicate that the correlation between parameters has certain regular trends, and appropriate parameter settings can lead to more satisfactory results. We also confirm that the pandemic has a giant influence on the market share of different transportation modes. Moreover, the comparison with other approaches demonstrates the feasibility and effectiveness of the proposed method.

Cooperative multi-population Harris Hawks optimization for many-objective optimization

This paper presents an efficient cooperative multi-populations swarm intelligence algorithm based on the Harris Hawks optimization (HHO) algorithm, named CMPMO-HHO, to solve multi-/many-objective optimization problems. Specifically, this paper firstly proposes a novel cooperative multi-populations framework with dual elite selection named CMPMO/des. With four excellent strategies, namely the one-to-one correspondence framework between the optimization objectives and the subpopulations, the global archive for information exchange and cooperation among subpopulations, the logistic chaotic single-dimensional perturbation strategy, and the dual elite selection mechanism based on the fast non-dominated sorting and the reference point-based approach, CMPMO/des achieves considerably high performance on solutions convergence and diversity. Thereafter, in each subpopulation, HHO is used as the single objective optimizer for its impressive high performance. Notably, however, the proposed CMPMO/des framework can work with any other single objective optimizer without modification. We comprehensively evaluated the performance of CMPMO-HHO on 34 multi-objective and 19 many-objective benchmark problems and extensively compared it with 13 state-of-the-art multi/many-objective optimization algorithms, three variants of CMPMO-HHO, and a CMPMO/des based many-objective genetic algorithm named CMPMO-GA. The results show that by taking the advantages of the CMPMO/des framework, CMPMO-HHO achieves promising performance in solving multi/many-objective optimization problems.