关键词: 免疫优化算法, 社会网络模型, Kleinberg网络模型, 克隆选择, 数值优化

Abstract:

Based on the effective information searching and transmission ability of the social network, a novel immune optimization algorithm, named the Immune Optimization Algorithm based on the Social Network Searching Model (SNSIA), is proposed. The new algorithm considers the settling of optimization problem as the process of information transmission. It constructs the evolutionary process by the modeling method of classical social network searching model known as the Kleinberg network model. The new individual is introduced into the antibody population via short-range and long-range connections with the aid of the network's structure growth mechanism. The probability of long-range connection would adjust adaptively when the searching reaches a certain degree, which would avoid the local optimum effectively and find the global optimum at last. The new algorithm introduces the short-range and long-range connections which take full advantage of the population's structure info, and it quickens the speed of convergence. At the same time, it brings down the probability of getting in the local optima. In experiments, the SNSIA is tested on the complex functions and compared with the particle swarm algorithm, the clonal selection algorithm and other optimization methods. Theoretical analysis and experimental results indicate that the SNSIA could maintain the solution's diversity better with a high convergence rate, and that it is also an effective and robust technique for optimization.

Key words: Immune optimization algorithm, social network model, Kleinberg network model, clonal selection, numerical optimization