基于加速收敛与递阶迭代的多变量系统辨识算法

Identification method based on accelerated convergence and hierarchical iterations for multivariable systems

根据递阶辨识原理和迭代辨识原理,可以实现多变量系统参数的准确辨识,但是其参数的收敛速度有待提高。该文针对参数的收敛性进行研究,提出了基于加速收敛技术的辨识方法。该方法根据递阶辨识原理将多变量系统分解成2个子系统,使其分别含有参数向量和参数矩阵,再根据迭代辨识原理得到参数的迭代解,并应用加速收敛技术进行加速。仿真实验表明:该方法可以得到很好的辨识结果,加速收敛技术的应用显著提高了参数的收敛速度。

The parameters in multivariable systems can be obtained using hierarchical identification and iterative identification, but the convergence rate needs to be improved. This paper presents a faster identification method. According to the hierarchical identification principle, a multivariable system can be decomposed into two subsystems containing a parameter vector and a parameter matrix. Then, iterative solutions are used to accelerate the convergence. Simulations indicate that the algorithm works quite well and the convergence rate is greatly improved.