带有停滞检测的蚁群算法在2D HP格点模型中的应用  

APPLICATION OF ANT COLONY OPTIMISATION ALGORITHM WITH STAGNATION DETECTION IN 2D HP LATTICE MODEL

作　　者：刘羽[1] 熊壬浩 

机构地区：[1]桂林理工大学机械与控制工程学院,广西桂林541004 [2]桂林理工大学信息科学与工程学院,广西桂林541004

出　　处：《计算机应用与软件》2016年第2期227-231,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(41264005);广西教育厅科研项目(201102ZD018)

摘　　要：为了提高蛋白质折叠结构预测的求解效率,针对2D HP格点模型,研究蚁群ACO(Ant Colony Optimization)算法在该问题上的应用。采用四元组表示绝对的折叠方向,并建立构象和解的一一对应关系。通过实验对算法各阶段的常用策略、方法进行比较分析。为了防止搜索陷入停滞,引入位置信息素停滞比和序列信息素停滞比两个参数,使用一种新的停滞检测机制。实验结果表明,改进的算法在保证预测质量的前提下,显著地提升了收敛速度。Aiming at 2D HP lattice model we studied the application of ant colony optimisation algorithm on protein folding structure prediction in order to improve the efficiency of its solution. We used the quadruple to express absolute folding direction, and established the one-to-one correspondence between conformation and solution. Through experiment we made the comparative analyses on common strategies and methods in each stage of the algorithm. To prevent the search from going to stagnation, we introduced two parameters, the position pheromone stagnation ratio and the sequence pheromone stagnation ratio, and applied a new stagnation detection mechanism as well. Experimental results showed that the improved algorithm remarkably accelerated the convergence speed on the premise of ensuring prediction quality.

关 键 词：HP模型 蛋白质折叠 蚁群算法 停滞检测 遗传算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301.6[自动化与计算机技术—控制科学与工程]