间隙三明治系统的改进卡尔曼状态估计
Modified Kalman filter-based state estimation for sandwich systems with backlash

控制工程中许多实际系统都可以描述为带间隙的三明治系统, 由于间隙具有非光滑、局部记忆性和多值 映射等复杂非线性特性, 使得整个三明治系统的内部状态估计工作具有很大挑战性. 首先根据间隙三明治系统的 特性引入了几个自动切换函数, 采用关键项分离原理, 建立了随机噪声干扰下间隙三明治系统的非光滑整体伪线性 状态空间模型. 针对该系统提出了一种非光滑的改进卡尔曼滤波算法以估计系统状态, 其工作机制能够随系统当 前工作区间的转变而自动切换模式. 仿真和实验结果表明, 针对含噪声的间隙三明治系统, 非光滑的改进卡尔曼滤 波算法对系统状态的估计准确度要高于传统卡尔曼滤波算法.
Many practical systems in control engineering can be described as the so-called sandwich systems with backlash. As the embedded backlash is a complicated non-smooth nonlinear function with local memory and multi-valued mapping, the estimation of internal states for the whole sandwich systems becomes a challenge. Based on the separation principle for key terms, a non-smooth pseudo-linear state space model for the whole sandwich systems with backlash disturbed by random noises is built by introducing several embedded switch functions to handle the effect of backlash. Then, a non-smooth modified Kalman filtering (MKF) method is proposed to achieve the state estimation for the obtained non-smooth state space model. The operating mechanism of this filtering method makes the mode automatically switchable according to the transformation of operation zone of the system. Simulation and experimental results demonstrate that the proposed non-smooth MKF method achieves higher estimation accuracy for such sandwich systems with backlash affected by random noises than the conventional KF method.