The fusion of particle swarm optimization (PSO) and interior point method (IPM) as cooperative movement control algorithm in Swarm Robotics / Dada Emmanuel Gbenga
Research in Particle Swarm Optimisation and its applications to real world problems has become a very interesting field in recent years. Particle Swarm Optimisation (PSO) despite its simplicity, ease of implementation and efficiency still has some flaws, which include its tendency to premature conve...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/6589/1/dada.pdf http://studentsrepo.um.edu.my/6589/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Research in Particle Swarm Optimisation and its applications to real world problems has become a very interesting field in recent years. Particle Swarm Optimisation (PSO) despite its simplicity, ease of implementation and efficiency still has some flaws, which include its tendency to premature convergence and inability to escape local minima. To address these weaknesses, many variants of PSO have been proposed in the literature. Also, many of these PSO algorithms employed hybrid methods that integrate other optimisation algorithms with the standard PSO. It is demonstrated in the literature that methods that hybridize PSO and some other optimisation algorithm have a better performance over the standard PSO algorithm. The Primal Dual method have been used to solve many optimisation problems. We proposed the Primal-Dual Particle Swarm Optimisation (pdPSO) and Primal-Dual Asynchronous Particle Swarm Optimisation (pdAPSO) to resolve the shortcomings of the standard PSO without the limitations of the IPM methods. To evaluate the performance of our new algorithms, we first compared the performance of pdPSO with IPM and PSO using nine (9) different dynamic benchmark functions. Our results revealed that pdPSO performed better than both the conventional PSO algorithm and the IPM method. The proposed algorithm is not susceptible to premature convergence, and can handle local minima avoidance better when compared to conventional PSO. Hence, pdPSO has the potential to perform better than many other PSO variants. Secondly, we compared the performance of our new algorithm pdAPSO with APSO, and PSO using 7 benchmark functions. Optimisation results reveal that pdAPSO offers similar (or in many test cases better) solutions than the other PSO variants to which we compared. Thirdly, we make a comparison between the performance of pdPSO and pdAPSO. Finally, we used our hybrid algorithms (pdPSO and pdAPSO) to solve the flocking and pattern formation problem in swarm robotics. Our simulation result
iv
provides a clear indication of the effectiveness of the algorithm. The hybrid algorithms perform better in terms of precision, rate of convergence, steadiness, robustness and flocking capability for homogenous set of swarm robots compared to some other variants of PSO.
We also compared the performance of pdAPSO and pdPSO with 9 state of the art PSO algorithms using 12 benchmark functions. Our proposed algorithms have mean dependability of 80.4% for pdAPSO and 69.69% for pdPSO. Also, pdAPSO and pdPSO is a better convergence speed compared to the other 9 algorithms. For instance, on Rosenbrock function, the mean FEs of 8938, 6786, 10,080, 9607, 11,680, 9287, 23,940, 6269 and 6198 are required by PSO-LDIW, CLPSO, pPSA, PSOrank, OLPSO-G, ELPSO, APSO-VI, DNSPSO and MSLPSO respectively to get to the global optima. However, pdPSO and pdAPSO only use 2997 and 2124 respectively which shows that pdAPSO is the fastest convergence speed and closely followed by pdPSO. In summary, pdPSO and pdAPSO uses the lowest number of FEs to arrive at acceptable solutions for all the 12 benchmark functions. |
|---|