Continuous Dynamics Of The System Parameters Model Identification And Industrial Test

With the industry application background in the large-scale metallurgy equipment dynamical model identification, the thesis has chiefly carried on research of the following several aspects:(1) The scale dilation principle that construct Gauss modulating function is put forward. The structure and the frequency spectrum property of Gauss wavelet modulating filter is investigated, and the identified parametric model of continuous time system based on modulating function has been established The least square(LS) estimation algorithm has been developed, and the statistical characteristics of LS are investigated. It was proved that the LS algorithm is biased even with white measurement noise.(2) The auxiliary variable system was constituted, and the recursive algorithm that can reduce the measurement noise interference has been developed. The bias property of LS with input/output measurement noise has been derived, and the self-adaptive deviation compensation principle of measurement noise variance has been proposed, and the corresponding algorithm is designed. The influence of parameters of wavelet modulating on estimation precision is investigated with simulation examples.(3) The correlative filtering principle of frequency response estimation under white noise input has been studied. The non-parameter frequency response model is estimated with spectrum estimation technique. The auto-correlative function and power spectrum of pseudo-random binary signals(PBRS) and inverse repeating PBRS are investigated, and some tips about the parameter selection of PBRS are drawn out.(4) The dynamic model of pressure control loop of hydraulic AGC of tandem mill is analytically derived. The model can be used as the guidance of selection of model structure in experiment.(5) The frequency response model of HAGC is estimated with correlative identification theory. The dynamical parameter model HAGC is estimated with LS, auxiliary variable LS, and bias-compensating LS respectively. It is proved with experiment that the methods proposed in the paper are accurate and practical.The modulating function method of continuous time system identification has not attracted much attention. The author fully uses the results of discrete time model identification, carry out the research of continuous time model identification systematically, and propose the series algorithms independently, which are applied to a large-scale industry equipment. The results of the thesis enrich the achievements of system identification...