多模函数的一种全局优化策略研究

Global Optimization Strategy for Multimodal Functions

Animat理论是人工生命的一种表现形式,具有自治性、适应性、涌现性等复杂系统所体现的特点。基于animat理论,采用面向对象的分析方法建立其视觉模型和反应行为机制,通过模拟动物勘查周围复杂、陌生的环境以寻找路径的方式,提出了一种灵活稳定的优化策略,尝试用其解决多模函数全局优化的问题。利用标准测试函数对此优化策略进行测试,仿真结果表明其充分发挥了animat理论特点,具有良好的收敛性和计算精度,可以克服现有智能优化算法在求解优化问题时的早熟、求解精度不高等问题,对各种优化问题均具有很强自适应性。

Animat is one of the artificial life theories. It has properties of autonomy, adaptability and emergence which are the possessions of complex systems. An optimization strategy based on animat theory was proposed and was used to deal with multimodal function optimization problems. It simulated animal＇s path seeking process in complex and unknown environment. An animat structure mainly consisted of the vision model and reacting behavior mechanism of animals was implemented via object-oriented analysis method, which provides us a flexible and stable optimization strategy. Experiments on four typical test functions were carried out, and the results show that this optimization strategy has nice convergence ability and high precision. Autonomous and adaptive ability of animat theory are also demonstrated, And it has no prematurity that other intelligent algorithms have suffered.