Decomposition Based Multi-Objective Variable Neighborhood Descent Algorithm for Logistics Dispatching
- Authors
- Lan, Yu-Lin; Liu, Fagui; Ng, Wing W. Y.; Zhang, Jun; Gui, Mengke
- Issue Date
- Oct-2021
- Publisher
- Institute of Electrical and Electronics Engineers Inc.
- Keywords
- Decomposition; multi-objective optimization; variable neighborhood descent; vehicle routing problem with time window (VRPTW)
- Citation
- IEEE Transactions on Emerging Topics in Computational Intelligence, v.5, no.5, pp 826 - 829
- Pages
- 4
- Indexed
- SCIE
SCOPUS
- Journal Title
- IEEE Transactions on Emerging Topics in Computational Intelligence
- Volume
- 5
- Number
- 5
- Start Page
- 826
- End Page
- 829
- URI
- https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/115400
- DOI
- 10.1109/TETCI.2020.3002228
- ISSN
- 2471-285X
- Abstract
- Vehicle routing problem with time window (VRPTW) is the most representative optimization problem in logistics researches, which appears to be a multi-objective optimization problem in essence. However, most previous studies transform it into a single objective optimization problem to deal with. Also, the vehicle routing models in related studies focus on minimizing the cost. However, the efficiency (e.g., the waiting time due to early arrivals) and customer satisfaction equally important as the cost for the long-term development of logistics companies. In this article, a corresponding tri-objective VRPTW model for logistics dispatching is introduced, and a decomposition-based multi-objective variable neighborhood descent algorithm (D-VND) is proposed to solve the tri-objective VRPTW. In D-VND, the tri-objective VRPTW is decomposed into multiple single-objective sub-problems via a set of uniformly distributed weight vectors and then the route-based evolutionary operator is used to generate the offspring. Meanwhile, the objective-wise neighborhood operators are incorporated into the variable neighborhood descent method as local searches to improve qualities of sub-problems. Moreover, we adopt a heuristic initialization strategy and an external archive mechanism based on fast sorting and crowding distance methods to improve the performance of D-VND. Experimental results on benchmarking datasets with different scales show that D-VND outperforms two other representative algorithms regarding convergence and diversity and yields better solutions. © 2017 IEEE.
- Files in This Item
-
Go to Link
- Appears in
Collections - COLLEGE OF ENGINEERING SCIENCES > SCHOOL OF ELECTRICAL ENGINEERING > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/115400)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.