A co-evolutionary design methodology for complex AGV system

Autor: Hong-Fei Teng, Xiaojun Zheng, Zhuang-Cheng Liu, Yanjun Shi, Luyang Hou
Rok vydání: 2016
Předmět:
Zdroj: Neural Computing and Applications. 29:959-974
ISSN: 1433-3058
0941-0643
DOI: 10.1007/s00521-016-2495-1
Popis: Our design for AGV system integrates machine assignment, machine layout, transfer station setting and loop arrangement and other issues and become a complex combinatorial design problem in the manufacturing. In previous studies, researches tried to address these issues in several steps, which may separate the coupling relations in these issues. We herein propose a co-evolutionary methodology to design one complex AGV system that includes two tandem AGV systems (workshops) synchronously. Our method gives an overall consideration for the aforementioned issues, which overcomes the defects of solving these issues in sequence by the previous studies. Moreover, the corresponding mathematical model is built for this design of complex AGV system. The proposed co-evolutionary methodology has two optimization parts, part A and part B for optimizing two workshops synchronously. Workshop 1 divides the four aforementioned issues into two classes, machine assignment and loop layout. The machines are assigned to different loops, and the exact layout of machines is optimized in each loop; meanwhile, the transfer station is set and loops are arranged in workshop. Workshop 1 has a re-optimization step using part B. For another, workshop 2 only optimizes the machine layout by part B. An improved fuzzy IWO (f-IWO) is proposed to execute the optimization for parts A and B in the methodology. Therein, a synthetic evolution mechanism with a fuzzy number has efficiently improved the quality of f-IWO. At last, a numerical experiment of design for complex AGV system validates the co-evolutionary methodology and f-IWO comparing to the contrastive methods.
Databáze: OpenAIRE
Externí odkaz: