基于混合蚁群关联规则挖掘的危险源分析算法  

A Hazard Analysis Algorithm Based on Mixed Ant Colony Association Rules Mining

作　　者：佘雅莉 周良[1] SHE Ya-li;ZHOU Liang(School of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics, Nanjing 210016,China)

机构地区：[1]南京航空航天大学计算机科学与技术学院,江苏南京210016

出　　处：《计算机技术与发展》2018年第11期89-93,共5页Computer Technology and Development

基　　金：江苏省产学研联合创新资金项目(SBY201320423)

摘　　要：针对民航危险源原因分析中存在人工参与较多缺乏客观性的问题,设计了一种基于混合蚁群关联规则挖掘的危险源原因分析算法(HA-MACR),利用关联规则挖掘来探索危险源原因。该算法摒弃了传统关联规则挖掘算法重复扫描数据库导致挖掘效率较低及产生大量候选集、容易出现"组合爆炸"现象等缺点,将改进后的蚁群算法用于挖掘最大频繁项集,并由此产生质量较好的强关联规则,从而找到导致危险源的不安全事件。同时,为了避免蚁群的盲目性,混合了粒子群,借助粒子群确定蚁群的初始信息素浓度。通过上述改进,有效增强了算法的搜索能力,提高了关联规则挖掘的效率,且避免了算法陷入局部最优,从而使危险源原因分析更加快速、准确。Aiming at the problem of much human participation and lack of objectivity in the hazard causes analysis of civil aviation,we design a hazard analysis algorithm based on mixed ant colony association rules mining which is used to explore the cause of hazard.This algorithm discards the disadvantages of repeated scanning database of traditional association rule mining algorithm,which leads to low mining efficiency,a large number of candidate sets and easy occurrence of“combined explosion”.It uses the improved ant colony algorithm to mine the maximal frequent item sets instead,and generates association rules with strong quality from them,thus finding unsafe incidents which lead to hazard by these rules.At the same time,in order to avoid the blindness of the ant colony,the particle swarm is mixed and the initial pheromone concentration of the ant colony is determined by the particle swarm.Through the above improvement,the search ability of the algorithm is effectively enhanced,the efficiency of the association rule mining is improved,and the algorithm is prevented from falling into a local optimum,so that the analysis of hazard cause is faster and more accurate.

关 键 词：危险源原因分析 关联规则挖掘 蚁群算法 粒子群 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]