UDE: Differential Evolution with Uniform Design

Differential evolution (DE) is significantly faster and robust for solving numerical optimization problem and is more likely to find true global optimum of functions. It has solved many real-world optimization problems. However, DE has sometimes been shown slow convergence and low accuracy of soluti...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2010 3rd International Symposium on Parallel Architectures, Algorithms and Programming s. 239 - 246
Hlavní autoři: Lei Peng, Yuanzhen Wang, Guangming Dai
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.12.2010
Témata:
Accuracy
Algorithm design and analysis
Benchmark testing
Chaos
Convergence
Design methodology
differential evolution
global optimization
Optimization
orthogonal design method
uniform design method
ISBN:1424494826, 9781424494828
ISSN:2168-3034
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Differential evolution (DE) is significantly faster and robust for solving numerical optimization problem and is more likely to find true global optimum of functions. It has solved many real-world optimization problems. However, DE has sometimes been shown slow convergence and low accuracy of solutions when the solution space is hard to explore. Population initialization is very important to the performance of differential evolution. A good initialization method can help in finding better solutions and improving convergence rate. In this paper, a uniform-differential evolution algorithm (UDE) is proposed. It incorporates uniform design initialization method into differential evolution to accelerate its convergence speed and improve the stability. UDE is compared with other two algorithms of standard differential evolution (SDE) and orthogonal differential evolution (ODE). Experiments have been conducted on 23 benchmark problems of diverse complexities. The results indicate that our approach has the stronger ability and higher calculation accuracy to find better solutions than other two algorithms.
ISBN:1424494826
9781424494828
ISSN:2168-3034
DOI:10.1109/PAAP.2010.61