摘要
针对工业过程中数据维数高,导致SVDD算法在建立不同类别数据的超球面时会产生混叠区域的问题,提出了基于近邻密度-支持向量数据描述(LD-SVDD)的数据分类方法。结合局部密度信息在判别数据相似性和SVDD在数据分类的优势,使用SVDD算法对数据进行分类,对分布在混叠区域中的样本采用密度信息进一步判断其类别,通过随机产生的数据集进行仿真,并与SVDD分类结果进行比较,结果表明LD-SVDD的分类准确率提高到了93%。
In order to solve the problem that SVDD algorithm will produce aliasing regions when building hyperspheres of different types of data due to high data dimension in industrial process,a data classification method based on local density support vector data description(LD-SVDD)is proposed.Combining the advantages of local density information in discriminating data similarity and SVDD in data classification,the SVDD algorithm is used to classify the data,and the density information is used to further judge the classification of samples distributed in the aliasing area.The simulation results of randomly generated data sets are compared with the SVDD classification results.The results show that the classification accuracy of LD-SVDD is improved to 93%.
作者
白蕾
胡平
苑易伟
BAI Lei;HU Ping;YUAN Yiwei(Electrical Engineering School,Shaanxi Polytechnic Institute;Xianyang Key Laboratory of New Power and Intelligent Microgrid System,Shaanxi Xianyang 712000,China;Faculty of Automation and Information Engineering,Xi'an University of Technology,Shaanxi Xi'an 710048,China)
出处
《工业仪表与自动化装置》
2021年第3期80-83,共4页
Industrial Instrumentation & Automation
基金
陕西省教育厅科研计划项目(20JK0802)
咸阳科技局科研攻关项目(2018k02-10)
陕西工业职业技术学院院级项目“基于数据挖掘技术的制粉系统故障诊断方法研究与设计”(2020YKYB-051)
陕西工业职业技术学院院级项目(ZK19-11)。
