TY - GEN

T1 - Weightless Swarm Algorithm (WSA) for dynamic optimization problems

AU - Ting, T. O.

AU - Man, Ka Lok

AU - Guan, Sheng Uei

AU - Nayel, Mohamed

AU - Wan, Kaiyu

PY - 2012

Y1 - 2012

N2 - In this work the well-known Particle Swarm Optimization (PSO) algorithm is applied to some Dynamic Optimization Problems (DOPs). The PSO algorithm is improved by simplification instead of introducing additional strategies into the algorithm as done by many other researchers in the aim of improving an algorithm. Several parameters (w, Vmax, Vmin and c 2) are being excluded from the conventional PSO. This algorithm is called Weightless Swarm Algorithm (WSA) as the prominent parameter, inertia weight w does not exist in this proposed algorithm. Interestingly, WSA still works effectively via swapping strategy found from countless trials and errors. We then incorporate the proven clustering technique from literature into the framework of the algorithm to solve the six dynamic problems in literature. From the series of tabulated results, we proved that WSA is competitive as compared to PSO. As only one parameter exists in WSA, it is feasible to carry out parameter sensitivity to find the optimal acceleration coefficient, c 1 for each problem set.

AB - In this work the well-known Particle Swarm Optimization (PSO) algorithm is applied to some Dynamic Optimization Problems (DOPs). The PSO algorithm is improved by simplification instead of introducing additional strategies into the algorithm as done by many other researchers in the aim of improving an algorithm. Several parameters (w, Vmax, Vmin and c 2) are being excluded from the conventional PSO. This algorithm is called Weightless Swarm Algorithm (WSA) as the prominent parameter, inertia weight w does not exist in this proposed algorithm. Interestingly, WSA still works effectively via swapping strategy found from countless trials and errors. We then incorporate the proven clustering technique from literature into the framework of the algorithm to solve the six dynamic problems in literature. From the series of tabulated results, we proved that WSA is competitive as compared to PSO. As only one parameter exists in WSA, it is feasible to carry out parameter sensitivity to find the optimal acceleration coefficient, c 1 for each problem set.

KW - Dynamic optimization

KW - Swapping

KW - Weightless swarm

UR - http://www.scopus.com/inward/record.url?scp=84871545734&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-35606-3_60

DO - 10.1007/978-3-642-35606-3_60

M3 - Conference Proceeding

AN - SCOPUS:84871545734

SN - 9783642356056

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 508

EP - 515

BT - Network and Parallel Computing - 9th IFIP International Conference, NPC 2012, Proceedings

T2 - 9th IFIP International Conference on Network and Parallel Computing, NPC 2012

Y2 - 6 September 2012 through 8 September 2012

ER -