Separation identification approach for the Hammerstein‐Wiener nonlinear systems with process noise using correlation analysis

Abstract

This article develops a novel separation identification approach for the Hammerstein‐Wiener nonlinear systems with process noise using correlation analysis technique. The Hammerstein‐Wiener nonlinear systems have three parts, namely, an input nonlinear block, a linear block, and an output nonlinear block. The designed hybrid signals that consist of separable signal and random signal are devoted to achieving parameters separation identification of the Hammerstein‐Wiener nonlinear system, that is, the three blocks are identified independently. First, the characteristics of separable signals under the action of static nonlinear block are analyzed, and two groups of separable signals with amplitude relation are utilized to estimate parameters of output nonlinear block. Moreover, the linear block parameters are identified by using correlation analysis approach, which deals with effectively immeasurable problem of internal variable information. Finally, the data filtering technique is implemented to weaken the influence of noises, and filtering‐based recursive extended least squares algorithm is developed for figuring out the parameters of nonlinear block and colored noise model. The validity and accuracy of the proposed scheme are verified by two simulations, and simulation results exhibit that the proposed method can obtain higher identification precision and better robustness than the existing identification algorithms.

WITHDRAWN: Identification Methodology for MIMO Hammerstein Nonlinear Model with Process Noise

“…For better analysis and designing of control systems, identification and control methods of nonlinear systems are becoming an extremely broad topic and attracting a lot of research interest. Block-oriented models, comprising of combination of nonlinear static blocks and linear dynamics blocks, is widely employed to describe the nonlinear systems [4][5][6].…”

Estimation of wiener nonlinear systems with measurement noises utilizing correlation analysis and Kalman filter

“…The static nonlinear block f (·) is modeled using a four‐layer NFM, 1 here we assume that it is invertible. The output of the NFM can be written as $$$$ \omega (t)=f\left(\gamma (t)\right)=\sum \limits_{l=1}^L{\zeta}_l\left(\gamma (t)\right){w}_l $$$$ where $$$ {\zeta}_l\left(\gamma (t)\right)=\frac{\mu_l\left(\gamma (t)\right)}{\sum \limits_{l=1}^L{\mu}_l\left(\gamma (t)\right)} $$$, $$$ {\mu}_l\left(\gamma (t)\right)=\exp \left(-\frac{{\left(\gamma (t)-{c}_l\right)}^2}{\sigma_l^2}\right) $$$ is the Gaussian function, where center and width can be represented by $$$ {c}_l $$$ and $$$ {\sigma}_l $$$, respectively, L is fuzzy rule.…”

“…For practical systems in industrial processes, each industrial object has the characteristics of nonlinearity more or less in nature 1 . It is well known that model prediction and control methodologies mainly depend on physical systems model, thus the nonlinear systems modeling and identification have attracted substantial interest in system identification and control fields.…”