摘要
异常检测技术在故障预测与健康状态管理(PHM)领域有着重要的作用,通过识别一个系统反常的运行状况,可监控卫星性能、检测故障、识别故障的根本起因,也可用于预测剩余使用寿命(RUL)以提高在轨卫星的安全性并减少其故障时间;提出了一种基于最小二乘支持向量机(LS-SVM)的用于检测在轨卫星的异常状态的方法,该方法具有良好的学习、分类和概括能力,具体过程包括数据采集和预处理、特征提取、特征选择、异常检测等步骤;利用带有故障信息的卫星电源系统真实遥测数据来对提出的方法的可行性和性能进行验证,试验中选择LS-SVM的最佳参数(γ,σ)是(120,0.05),伴随着的交叉验证率是90.6%,验证结果证明了在卫星异常检测中应用基于最小二乘支持向量机检测方法的有效性。
Anomaly detection plays an important role in Prognostics and System Health Management （PHM） . By identifying abnormal behaviors of a system, it can be applied to monitor performance of satellite, detect faults, identify the root cause of the fault, and predict the remaining useful life （RUL） in order to increase safety and reduce downtime of satellite in--orbit. This paper proposes a method to detect anomaly of satellite in--orbit based on Least Squares Support Vector Machine （LS--SVM）, which has excellent learning, classification and generalization ability. The process includes several steps： data collection and preprocessing, feature extraction, feature selection, anomaly detection. Finally , experimental telemetry data from satellite power system which has fault information are used to test the performance and the feasibility of the algorithm . The experiment select the optimal parameter （γ,δ） of LS--SVM is （120, 0. 05） with cross--validation rate 90. 6% and the results show that the method based on LS--SVM achieves perfect efficiency in anomaly detection of the satellite.
出处
《计算机测量与控制》
北大核心
2014年第3期690-692,696,共4页
Computer Measurement &Control
