一类离散时变系统的在线无限脉冲响应滤波逆控制

Online infinite-duration impulse response filtering inverse control for a class of discrete time-varying systems

中文关键词  离散时变系统  自适应逆控制  无限脉冲响应滤波器  粒子群优化算法

英文关键词  discrete time-varying system  adaptive inverse control  infinite impulse response filter  particle swarm optimization algorithm

中文摘要

      针对一类离散时变系统, 提出了一种基于自适应惯性权重合作粒子群(AIW--CPSO)算法的在线无限脉冲响应(IIR)滤波自适应系统辨识方法, 实现零极点实时跟踪的全匹配控制. IIR滤波器可解决有限脉冲响应(FIR)滤波器在辨识时变系统时因其相关矩阵的特征值会无规律变大而被迫离线训练的问题. 同时又降低了在线训练所需的权值向量长度, 提升了优化与建模效率. 本文设计的自适应惯性权重合作粒子群(AIW--CPSO)算法可在传统粒子群优化(PSO)算法的基础上更好地解决因选用IIR滤波器所带来的全局优化问题. 通过仿真分析可以看出, 对于此类离散时变系统, 基于在线AIW--CPSO--IIR滤波器的自适应逆控制方法可以快速有效的实现未知对象的在线建模, 同时实时跟踪时变系统的特征值变化.

英文摘要

      For a class of discrete time-varying systems, we proposed an online infinite-duration impulse response(IIR) filtering adaptive system identification method based on the adaptive inertia-weighted cooperated particle swarm optimization(AIW--CPSO) algorithm. This approach achieves the perfect matching of zero-pole for the real-time tracking control. The IIR filter avoids the enforced off-line training problem induced by the irregular variation in the eigenvalues of the correlation matrix in finite-duration impulse response(FIR) filter when identifying a time-varying system. It also reduces the weighting vector length in the online training process, and improves the efficiency of optimization and modeling. On the basis of the traditional standard particle swarm optimization(PSO) algorithm, the designed AIW--CPSO algorithm provides a better solution to the global optimization problem in selecting the proper IIR than the traditional standard PSO algorithm. Simulation analysis shows that, for discrete time-varying systems, the adaptive inverse control method based on the online AIW--CPSO--IIR filter can realize the fast online modeling of unknown plants effectively, and track the real-time variation of eigenvalues of time-varying systems.