应用混合模型蚁群算法解决连续多自变量问题  

Research on Ant Colony Algorithm with Mixed System Used for Continuous and Several Variables Questions

作　　者：王华桥[1] 尹成[1] 马洪艳[1] 丁峰[1] 

机构地区：[1]西南石油大学资源与环境学院,四川成都610500

出　　处：《西部探矿工程》2007年第1期70-72,76,共4页West-China Exploration Engineering

摘　　要：蚁群优化算法自提出的十几年来已广泛应用于解决旅行商问题(tralvel salesman problem,简称TSP)等离散化的组合优化问题,随着对算法的进一步研究,近两年一些学者提出用该算法解决连续问题已经初见成效。进一步探讨了如何将该方法用于解决连续多自变量问题,通过对比文献[1]发现在当用该文献中提到的方法解决连续多自变量问题时,计算所得的解并非最优解,计算所得的目标函数值不准确。分析了出现问题的主要原因在于以往应用蚁群算法解决问题时大多采用基本蚁群算法的Ant-cycle system模型,在面对连续多自变量问题上是失效的。通过研究基本蚁群算法的原理,发现将Ant-cycle system模型与Ant-quantity system模型相结合,在解决连续多自变量问题时有很大突破。因而提出了一种新的蚁群算法模型——Ant-cycle&quantity system(ACQS模型),在此简称为混合模型算法。用该算法进行了大量的试验,取得了很好的结果。同时为了加快收敛速度,算法中还提出了信息素的奖惩机制,取得了很好的试验效果。Ant colony algorithm has been used in assembled optimized questions such as TSP for more than ten years. With the deep research on this algorithm, some intestine researchers have fetched it in some simple continuous problems and got many optimistic results. The keystone of this paper is to resolve continuous and multi-variables questions. Comparing with the problem in reference and using the method provided by it we found that we couldn＇t get eximious results when facing the continuous and multi-variables questions. This paper believes that the main reason, which causes the former fault in continuous multi-variables, is that the ＂ant-cycle system＂ is mot suitable for the new problems. By studying the mechanism of basic ant colony algorithm, we put forward a new system of this algorithm Ant-cycle ＆ quantity system（ACQS）;e call it as mixed system. Through a lot If experiment with this system we got good results. We also bring forward the reward and punishment mechanism in the system in order to quicken the convergent speed, we obtain some excellent experimenting resuhs.

关 键 词：蚁群算法 优化 混合模型 连续问题 多自变量 

分 类 号：TE319[石油与天然气工程—油气田开发工程]