基于SVDD优化的南极望远镜驱动非预期故障研究  

Unanticipated Fault of Driving Antarctic Telescope Based on SVDD Optimization

作　　者：邓壮壮 杨世海[2,3] 朱节中[1] 李运[2,3] 高志文[1] DENG Zhuangzhuang;YANG Shihai;ZHU Jiezhong;LI Yun;GAO Zhiwen(College of Electronic Information,Nanjing University of Information Science&Technology,Nanjing JiangSu 210044,China;National Astronomical Observatories/Nanking Institute f Astronomical Optics&Technology,Chinese Academy of Sciences,Nanjing JiangSu 210042,China;CAS Key Laboratory of Astronomical Optics&Technology,Nanking Institute of Astronomical Optics&Technology,Nanjing JiangSu 210042,China)

机构地区：[1]南京信息工程大学电子信息学院,南京210044 [2]中国科学院国家天文台南京天文光学技术研究所,南京210042 [3]中国科学院天文光学技术重点实验室(南京天文光学技术研究所),南京210042

出　　处：《自动化与仪器仪表》2022年第1期18-22,共5页Automation & Instrumentation

基　　金：国家自然科学基金项目资助(11973065);天文联合基金重点项目资助(U1931207)。

摘　　要：支持向量数据描述SVDD(Support Vector Data Description,SVDD)算法,在小样本故障检测中表现出良好的学习能力。然而,现有的SVDD故障检测方法由于对实验数据和参数过于依赖,从而限制了故障检测的精度。针对上述问题,提出了一种基于SVDD的智能寻优非预期故障检测算法。实验数据采取核主成分分析法(Kernel Principal Component Analysis,KPCA)降维去噪处理;并对SVDD算法与粒子群优化(Particle Swarm Optimization,PSO)算法结合,对算法参数优化。待检测样本点的球心距与超球体半径之差为衡量标准,实现非预期故障的检测。数据采集于南极望远镜的驱动系统,实验结果表明,优化后的SVDD算法检测精度提高。基于半实物仿真平台验证该算法,在非预期故障检测中具有很高的应用价值。Support vector data description(SVDD)algorithm shows good learning ability in small sample fault detection.However,the existing SVDD fault detection methods are too much rely on experimental data and parameters,which limits the accuracy of fault detection.Aiming at the above problems,an intelligent optimization algorithm for unanticipated fault detection based on SVDD is proposed.Kernel principal component analysis(KPCA)is used to reduce dimensionality and denoising of the experimental data;and the SVDD algorithm is combined with particle swarm optimization(PSO)algorithm to optimize the algorithm parameters.The difference between the center distance of the sample point and the radius of the hypersphere is taken as the measurement standard,and realize unanticipated fault detection.The data was collected in the driving system of Antarctic telescope,The experimental results show that the detection accuracy of the optimized SVDD algorithm has been improved.the algorithm verifies based on hardware in the loop simulation platform and has high application value in the detection of unanticipated faults.

关 键 词：南极望远镜 驱动系统 SVDD 非预期故障 参数优化 KPCA降维 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置] TH743[自动化与计算机技术—控制科学与工程]