改进的多模态遗传算法及其在投资组合中的应用  

Improved multi-modal genetic algorithms and its application on portfolio investment model

作　　者：刘洪杰[1] 王秀峰[1] 王治宝[1] 

机构地区：[1]南开大学信息技术科学学院,天津300071

出　　处：《控制与决策》2003年第2期173-176,共4页Control and Decision

基　　金：国家重大自然科学基金资助项目 (79790 130 )

摘　　要：提出用多模态遗传算法求解投资组合的新思路。首先 ,针对小生境遗传算法搜索结果不稳定的缺点 ,提出具有“迁徙操作”的新多模态遗传算法 ,不仅有效地找到了全部优质解 ,而且无需峰间距的信息。然后 ,针对传统的投资组合模型存在不能满足不同风险偏好投资者需要的缺点 ,提出符合中国国情的证券投资组合模型 ,并给出利用改进的多模态遗传算法求解的方法。最后进行了实证研究 ,得到了满意的结果。A new approach to find investing scheme with multi-modal genetic algorithm is proposed. In view of the fact that the traditional portfolio investment model does not meet the demands of different venture capitalist, a new model that fits the circumstances of Chinese securities business is developed. A new approach is presented to find the solution that incorporates Niche GA with 'Migrating Operator' which can overcome the unstability of Niche GA and find effectively all good solutions without the information of the distances between the good solutions. The method is tested and the expected results are obtained.

关 键 词：遗传算法 多模态 小生境 投资组合 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]