基于自适应Voronoi检测器的故障检测算法  

FAULT DETECTION ALGORITHM BASED ON ADAPTIVE VORONOI DETECTOR

作　　者：尹中川 徐遵义[1] 韩绍超 王俊雪 Yin Zhongchuan;Xu Zunyi;Han Shaochao;Wang Junxue(College of Computer Science and Technology,Shandong Jianzhu University,Jinan 250101,Shandong,China)

机构地区：[1]山东建筑大学计算机科学与技术学院,山东济南250101

出　　处：《计算机应用与软件》2018年第3期257-261,共5页Computer Applications and Software

基　　金：山东省重点研发计划项目(2015GGX101047;2016GGX101024)

摘　　要：否定选择算法在单分类算法中具有良好特性,但在故障检测中,传统的否定选择算法训练时间过长,实际的检测精度不高。针对这些问题,提出一种基于自适应Voronoi检测器的否定选择算法。算法利用自体空间的内外边界样本生成检测器,弥补了实值检测器存在孔洞的缺陷,提高了检测器的覆盖率,且检测器仅需一次训练,减少了训练时间。通过对Iris数据和华北某电厂真实数据进行实验,将传统否定选择算法同V-Detector算法进行对比。实验证明该算法相对传统否定选择算法减少了检测器的生成时间,提高了算法整体的检测精度,避免了检测器间孔洞的发生。Negative selection algorithm has good characteristics in the single classification algorithm,but in the fault detection,the traditional negative selection algorithm training time is too long,and the actual detection accuracy is not high.In response to these problems,this paper presented a negative selection algorithm based on adaptive Voronoi detector.The algorithm used the inner and outer boundary samples of the self-space to generate the detector,which made up for the defects of the hole in the real value detector and improved the coverage of the detector.The detector only need train once and reduced the training time.Finally,we compared the traditional negative selection algorithm with the V-Detector algorithm by experimenting with Iris data and real data from a power plant in North China.Experimental results showed that the proposed algorithm reduced the generation time of the detector compared with the traditional negative selection algorithm,improved the overall detection accuracy of the algorithm and avoided the occurrence of holes between detectors.

关 键 词：密度聚类 否定选择算法 人工免疫 故障检测 冯洛诺伊图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]