Font Size: a A A

Multivariable Process Identification Based On Closed-loop Step Test

Posted on:2013-02-15Degree:MasterType:Thesis
Country:ChinaCandidate:C J WangFull Text:PDF
GTID:2218330362967520Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The research of multivariable identification based on closed-loop testhas been the hotspots in industrial process control area since closed-loopidentification was advocated in1980s, which is important for bothmultivariable design and parameters adjustment of controllers. In recentyears, there have been some research achievements in this domain.However, they are usually difficult to be accomplished in industry processdue to rigid experiment conditions or complicated calculation. Generally,Step test under closed-loop condition is easy to realized in industry. Onsuch background, decentralized multivariable system identification basedon closed-loop step test is researched in this thesis, both time domainmethod and frequency domain method are proposed, moreover, theapplication of proposed algorithm is realized in practice.The main topics of this thesis are:Decentralized multivariable process identification is considered, atime-domain approach based on closed-loop step test for decentralizedMIMO process is proposed. The approach transforms MIMO identificationissue into equivalent SISO issues by combining input and output signals.Multiple convolution is used and a discrete convolution algorithm isproposed in the thesis. To solve the SISO issues, a parameter estimationapproach for FOPDT model is given by taking derivative of transferfunction, and the issue is finally transformed into a linear least-squareproblem. The algorithm can avoid errors generated by approximation ofthe model's delay term and it is easy to be solved due to no iteration isrequired. frequency domain perspective, and a parameterization method forfrequency model is given by step response reconstruction and FOPDTparameter estimation.There are always nonlinear factors and all sorts of noise withinindustrial process, and identification algorithm might become invalid inextreme environment. In this thesis, Decentralized control system is designfor heat exchanging process, and the proposed method is verified byapplication in that control system.
Keywords/Search Tags:Decentralized control system, Closed-loop step test, Multivariable process identification, FOPDT, Convolution algorithm, Least square algorithm, Heatexchanging process
PDF Full Text Request
Related items