Back to Search
Start Over
Adaptive differential search algorithm with multi-strategies for global optimization problems
- Source :
- Neural Computing and Applications. 31:8423-8440
- Publication Year :
- 2019
- Publisher :
- Springer Science and Business Media LLC, 2019.
-
Abstract
- Differential search (DSA) is a recently proposed evolutionary algorithm mimicking the Brownian motion-like random movement existing in living beings. Though it has displayed promise for global optimization, the original DSA suffers from relatively poor search capability, especially for exploitation. In this study, an augmented DSA (ADSA) is proposed by integrating memetic framework with multiple strategies. In ADSA, a sub-gradient strategy is combined to improve local exploitation, and the dynamic Levy flight technique is developed to strengthen the global exploration. Moreover, a mutation operator based on differential search is used to increase swarm diversity. An intelligent selection method is implemented to adaptively adjust the strategies based on historical performance. To fully benchmark the proposed algorithm, 35 test functions of various properties in 30-D and 100-D are adopted in numerical experiments. Seven canonical optimization algorithms are involved for experimental comparison. In addition, two real-world problems are also tested to verify ADSA’s practical applicability. Numerical results reveal the efficiency and effectiveness of ADSA.
- Subjects :
- 0209 industrial biotechnology
Mutation operator
Mathematical optimization
Computer science
Evolutionary algorithm
Swarm behaviour
02 engineering and technology
020901 industrial engineering & automation
Artificial Intelligence
Differential search algorithm
0202 electrical engineering, electronic engineering, information engineering
Benchmark (computing)
020201 artificial intelligence & image processing
Differential (infinitesimal)
Global optimization
Software
Global optimization problem
Subjects
Details
- ISSN :
- 14333058 and 09410643
- Volume :
- 31
- Database :
- OpenAIRE
- Journal :
- Neural Computing and Applications
- Accession number :
- edsair.doi...........2572eb623f1707148b6b7a21bb6d2a8e