点击次数：

论文类型：期刊论文

发表时间：2014-04-01

发表刊物：COMMUNICATIONS IN NONLINEAR SCIENCE AND NUMERICAL SIMULATION

收录刊物：SCIE、EI、Scopus

卷号：19

期号：4

页面范围：1229-1246

ISSN号：1007-5704

关键字：Chaos optimization algorithms; Chaotic sequences; Probability distribution; Search speed

摘要：Chaos optimization algorithms (COAs) usually utilize the chaotic map like Logistic map to generate the pseudo-random numbers mapped as the design variables for global optimization. Many existing researches indicated that COA can more easily escape from the local minima than classical stochastic optimization algorithms. This paper reveals the inherent mechanism of high efficiency and superior performance of COA, from a new perspective of both the probability distribution property and search speed of chaotic sequences generated by different chaotic maps. The statistical property and search speed of chaotic sequences are represented by the probability density function (PDF) and the Lyapunov exponent, respectively. Meanwhile, the computational performances of hybrid chaos-BFGS algorithms based on eight one-dimensional chaotic maps with different PDF and Lyapunov exponents are compared, in which BFGS is a quasi-Newton method for local optimization. Moreover, several multimodal benchmark examples illustrate that, the probability distribution property and search speed of chaotic sequences from different chaotic maps significantly affect the global searching capability and optimization efficiency of COA. To achieve the high efficiency of COA, it is recommended to adopt the appropriate chaotic map generating the desired chaotic sequences with uniform or nearly uniform probability distribution and large Lyapunov exponent. (C) 2013 Elsevier B.V. All rights reserved.