A Particle Swarm Optimization with diversity-guided convergence acceleration and stagnation avoidance
Conference proceedings article
ผู้เขียน/บรรณาธิการ
กลุ่มสาขาการวิจัยเชิงกลยุทธ์
ไม่พบข้อมูลที่เกี่ยวข้อง
รายละเอียดสำหรับงานพิมพ์
รายชื่อผู้แต่ง: Worasucheep C.
ปีที่เผยแพร่ (ค.ศ.): 2012
หน้าแรก: 733
หน้าสุดท้าย: 738
จำนวนหน้า: 6
ISBN: 9781457721311
นอก: 2157-9555
eISSN: 2157-9555
ภาษา: English-Great Britain (EN-GB)
บทคัดย่อ
This paper proposes an enhanced Particle Swarm Optimization (PSO) algorithm by using the swarm diversity as a main guidance in both convergence acceleration and stagnation avoidance. This proposed algorithm, namely Diversity-Guided PSO (DGPSO), includes three features that employ swarm diversity at each generation. First, the inertia weight is adapted using a feedback from diversity. Second, DGPSO operations include a perturbation, whose distance is controlled with the diversity information, significantly accelerating the convergence. Third, the diversity-guided mechanism prevents the swarm from being trapped in local optima. DGPSO is evaluated using 10 well-known benchmarks of non-linear functions with various characteristics. The test results at 20 and 50 dimensions are compared with those from Standard PSO 2007 (SPSO07) [19] and Ratnaweera's MPSO-TVAC (RPSO) [6]. The experiment demonstrates that DGPSO outperforms both SPSO07 and RPSO in most cases with statistical significance. ฉ 2012 IEEE.
คำสำคัญ
Diversity, Stagnation