含未知时延的多输入系统梯度迭代辨识算法

Gradient Iterative Identification Algorithm for Multi-input Systems with Unknown Time Delays

针对含有未知输入时延的多输入动态调节系统,基于过参数化后系统模型具有的稀疏特性,将压缩感知思想、梯度搜索和递阶辨识原理相结合,提出一种梯度追踪迭代辨识算法。该算法能够从过参数化的高维稀疏参数向量中获取少数关键非零参数位置,实现系统降维,并采用有限的输入输出测量数据,实现对未知参数和时延的联合估计,提高了辨识效率。最后,采用仿真示例验证了所提算法的有效性,仿真结果表明所提方法能够有效估计系统参数和时延。

For multiple-input dynamic adjustment systems with unknown input time delays,the idea of compressed sensing,gradient search and hierarchical identification principle are combined based on the sparse characteristics of the over-parameterization system model,and a gradient pursuit iterative identification algorithm is proposed.The presented algorithm can obtain a few key non-zero parameter locations from the over-parameterization high-dimensional sparse parameter vector to achieve system dimensionality reduction.The presented algorithm can realize the joint estimation of unknown parameters and time delays by using limited input and output measurement data to improve the identification efficiency.Finally,a simulation example is given to verify the effectiveness of the proposed algorithm.The simulation results show that the proposed algorithm can effectively estimate the system parameters and time delays.