Authors: Anand J. Kulkarni; Neha S. Patankar; Kang Tai
Addresses: Odette School of Business, University of Windsor, 401 Sunset Avenue, Windsor, Ontario, N9B 3P4, Canada; School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore ' Department of Operations Research, North Carolina State University, 2366, Champion Court, Raleigh, NC 27606, USA ' School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore
Abstract: Almost all existing heuristic techniques are unconstrained optimisation methods and treat the system in centralised way. A distributed and decentralised optimisation technique in the framework of collective intelligence referred to as probability collectives (PCs) decomposes the entire system into subsystems and treats them as a multi-agent system. Similar to other contemporary heuristic techniques, its performance is significantly affected when constraints are involved. In order to handle constraints, a modified feasibility-based rule is incorporated into the PC algorithm. The approach is validated by solving a variety of constrained test problems. A tension/compression spring design problem, welded beam design problem and pressure vessel design problem are also solved. The approach is shown to be sufficiently robust and other strengths and weaknesses are also discussed. The solution to these problems proves that the constrained PC approach can be applied to a variety of practical/real world problems.
Keywords: probability collectives; multi-agent systems; MAS; agent-based systems; collective intelligence; feasibility-based rules; constrained test problems; constraint handling; tension-compression spring design; welded beam design; pressure vessel design.
International Journal of Computational Science and Engineering, 2016 Vol.13 No.4, pp.303 - 321
Received: 30 Jun 2014
Accepted: 24 Sep 2014
Published online: 04 Nov 2016 *