改进人工鱼群算法在外汇预测和投资组合中的应用  

Application of the improved artificial fish swarm algorithm in foreign exchange forecast and portfolio

作　　者：马骊[1] 李阳[1] 樊锁海[1] 

机构地区：[1]暨南大学信息科学技术学院数学系,广州510632

出　　处：《系统工程理论与实践》2015年第5期1256-1266,共11页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(11071089);广东省自然科学基金(10151063201000005);中央高校基本科研业务费专项基金(21609602)

摘　　要：人工鱼群算法具有良好的全局搜索能力和自适应能力,在解决投资组合问题上有较好的应用前景.本文通过改进人工鱼群算法,分别对汇率预测和外汇投资组合双目标优化两部分进行研究.首先利用基于平均距离视野的人工鱼群优化的支持向量回归机算法对汇率进行短期预测,提高了外汇预期收益率的准确性.然后建立外汇投资组合双目标模型,通过借鉴带精英策略的快速非支配排序遗传算法(non-dominated sorting genetic algorithm-Ⅱ,NSGA-Ⅱ)的思想,提出基于Pareto排序理论的双目标非支配排序人工鱼群算法(non-dominated sorting artificial fish swarm algorithm,NSAFSA).实证分析表明该算法在求解外汇投资组合方案时,获得的Pareto前沿比NSGA-II的结果分布更均匀,多样性更好.最后对NSAFSA算法进一步改进,通过两次剪枝策略提高了解的质量,并给出了可供选择的最优外汇投资组合方案.研究结果表明人工鱼群算法可以对汇率预测和外汇投资组合提供重要参考,在外汇市场中具有较大的应用潜力.As the artificial fish swarm algorithm has a good global search capability and self-adaptability,there is a good application prospect to deal with the portfolio problem.In this paper,the improved artificial fish swarm algorithm is presented to be applied in foreign exchange forecast and portfolio multi-objective optimization.Firstly,the algorithm based on the average distance view of support vector regression machine optimized by artificial fish swarm algorithm is used to predict short-term exchange rate,so that the accuracy of expected yield is improved.Then foreign portfolio model with double objective is built.Non dominated sorting artificial fish swarm algorithm(NSAFSA) with double objective based on the Pareto sorting theory is put forward through using the thought of non dominated sorting genetic algorithm(NSGA-Ⅱ).The empirical analysis shows that the Pareto front of the algorithm has a better distribution as well as better diversity compared to NSGA-Ⅱ when solving the foreign exchange portfolio model.Last but not least,the NSAFSA is further improved.The quality of solutions is improved through using pruning strategy twice and the optimal portfolio selections are finally proposed.The research above shows that the artificial fish swarm algorithm offers an important reference to exchange rate forecast and portfolio,it has great application potential in foreign exchange market.

关 键 词：人工鱼群算法 汇率预测 外汇投资组合 支持向量回归机 Pareto排序理论 剪枝策略 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] F830.59[自动化与计算机技术—控制科学与工程]