非线性滤波方法及其在故障诊断中的应用  

Non-linear filtering method and its application in fault diagnosis

作　　者：张卫峰[1] 惠俊军 

机构地区：[1]兰州工业学院,兰州730050 [2]陕西宝鸡,721013

出　　处：《工业仪表与自动化装置》2017年第1期20-25,共6页Industrial Instrumentation & Automation

摘　　要：传统的滤波方法一般基于线性化和高斯假设,在一定程度上影响了滤波精度和非线性系统故障诊断的准确率。该文从"近似非线性"和"近似概率"的方法入手,分析3种常用的非线性滤波算法:扩展卡尔曼滤波器(EKF)、U-卡尔曼滤波器(UKF)以及粒子滤波器(PF)的原理、方法及特点并介绍其在非线性故障诊断中的应用价值。Traditional filtering methods are generally based on linearization or Gaussian hypothesis,which may influence the filtering precision and lead to low diagnosis precision to a certain extent.In this paper,from ＂approximate nonlinearity＂and ＂approximate probability＂,the principles,methods,characteristics of three widely used methods for estimation of nonlinear system,i.e.,Extended Kalman Filter（EKF）,Unscented Kalman Filter（UKF）,and Particle Filter（PF） are analyzed,and finally the application in fault diagnosis are introduced.

关 键 词：非线性状态估计 扩展卡尔曼滤波 UNSCENTED卡尔曼滤波 粒子滤波 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] TN713[电子电信—电路与系统]