非线性滤波方法及其在故障诊断中的应用

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。

Non - linear filtering method and its application in fault diagnosis

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,char-acteristics 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 applica-tion in fault diagnosis are introduced.