非均匀采样Hammerstein系统的梯度迭代辨识算法

Gradient based iterative identification algorithm for non-uniformly sampled Hammerstein systems

为了解决Hammerstein非线性系统在非均匀采样条件下的辨识问题,该文提出了1种能够用于在线参数估计的梯度迭代算法。通过引入时变后移算子,推导了非均匀采样Hammerstein系统的离散时间模型。采用关键项分离技术将系统参数化为1个线性回归模型。基于辅助模型辨识思想对未知中间变量进行重构,并利用负梯度搜索原理获得模型参数的迭代估计。仿真结果表明,该文方法是有效的,且比辅助模型随机梯度算法具有更快的收敛速度,参数估计精度提高近40倍。

A gradient based iterative algorithm for online parameter estimation is proposed to solve the identification problem of Hammerstein nonlinear systems with non-uniform sampling. A discretetime model of non-uniformly sampled Hammerstein systems is derived by introducing a time-varying backward shift operator.The system is parameterized into a linear regression model by applying the key-term separation technique. The unknown intermediate variables are reconstructed based on the auxiliary model identification idea,and the iterative estimates of model parameters are obtainedthrough the negative gradient search principle. The simulation results indicate that,the proposed method is effective and has a faster convergence rate than the auxiliary model based stochastic gradient algorithm,and the estimation accuracy is improved by nearly 40 times.