Identification method for nonlinear system with disturbance

Abstract

The invention discloses an identification method for a nonlinear system with disturbance, and the method comprises the following steps: A, converting a to-be-identified industrial control system intoa nonlinear system model with disturbance, wherein the nonlinear system consist of a nonlinear link and a linear link, i.e., a Hammerstein system for outputting an error types; B, decomposing the nonlinear system model with disturbance into two sub-models: a system noise-free output sub-model and a disturbance sub-model; and C, updating system parameters to construct P [zeta] (k) and e [zeta] (k),updating the parameters to enable k to be equal to k + 1, returning to the step A until a cut-off condition D is met, and identifying the parameters and disturbance of the industrial control system.According to the invention, defects in the prior art can be overcome, the convergence speed is high, and the identification precision is high.

Description

technical field [0001] The invention relates to the technical field of industrial control, in particular to an identification method for nonlinear systems with disturbances. Background technique [0002] Nonlinear systems widely exist in industrial systems, and the identification and control of nonlinear systems have attracted more and more attention from scholars and engineers, and have become the focus of research. Nonlinear systems can be divided into Hammerstein systems, Wiener systems, and Hammerstein-Wiener systems, which generally consist of nonlinear links and linear links. Collection, etc., the linear part is mainly the output error model. Among them, the Hammerstein system of output error type is the most widely studied. In the industrial process, measurement noise is widespread. When the output error model is converted into a regression equation, white noise will be converted into colored noise, and the least squares identification algorithm will become a biased...

AI Technical Summary

Problems solved by technology

In industrial processes, measurement noise is widespread. When converting the output error model into a regression equation, white noise will be converted into colored noise, and the least squares identification algorithm will become a biased estimate, reducing the identification accuracy.

During the identification process, the disturbance will always pollute the output data and reduce the identification accuracy, so the influence of the disturbance must be eliminated

Few literatures and patents at home and abroad mention the nonlinear system identification method with disturbance, such as the scholar Y.Mao et al. in the literature "A novel parameter separation based identification algorithm for Hammerstein systems", (simplified translation: a new decoupling The application of the identification algorithm in the identification of the Hammerstein system was published in the international journal Applied Mathematics Letters, Vol.60, 21-27, 2016 in the field of control. Based on filtering technology and multiple innovation theory, a parameter separation method is proposed The stochastic gradient identification algorithm reduces the computational complexity and avoids the calculation of redundant parameters. At the same time, the introduction of multi-innovation theory improves the convergence speed and identification accuracy of the algorithm. However, this method does not consider the influence of disturbances and cannot eliminate disturbances. The impact on the identification algorithm, the identification accuracy is reduced

Scholars M, Pouliquen, et al. published in the international journal IEEE Transactions on Automatic Control,Vol.61,No.2,550-555,2016.) Assuming that the noise is bounded, an iterative identification algorithm is proposed, which can simultaneously identify the linear part and the nonlinear part of the Hammerstein system of the output error type Part of the parameters, but this method does not take into account the impact of disturbance on the identification algorithm. When the output is disturbed by disturbance, the identification accuracy will be reduced. At the same time, this algorithm is an offline algorithm and cannot be used online.

[0003] For the Hammerstein system with perturbed output error type, the existing methods have the following disadvantages: (1) The influence of measurement noise is not well handled. When the output error model is converted into a regression equation, the white noise will be converted into a colored Noise makes the identification problem more complicated; (2) The influence of the disturbance is not considered, so that the disturbance pollutes the output data during the identification process, which reduces the identification accuracy, or the system parameters and disturbances are identified at the same time without distinction; (3) The system parameters The identification adopts the single innovation identification method, which can only use the current data

Images