Real-Time Parameter Estimation Method Applied to a MIMO Process and its Comparison with an Offline Identification Method


International Multiconference of Engineers and Computer Scientists, Hong Kong, PEOPLES R CHINA, 19 - 21 Mart 2008, cilt.1089, ss.35-36 identifier identifier

  • Cilt numarası: 1089
  • Basıldığı Şehir: Hong Kong
  • Basıldığı Ülke: PEOPLES R CHINA
  • Sayfa Sayıları: ss.35-36


An experiment based method is proposed for parameter estimation of a class of linear multivariable systems. The method was applied to a pressure-level control process. Experimental time domain input/output data was utilized in a gray-box modeling approach. Prior knowledge of the form of the system transfer function matrix elements is assumed to be known. Continuous-time system transfer function matrix parameters were estimated in real-time by the least-squares method. Simulation results of experimentally determined system transfer function matrix compare very well with the experimental results. For comparison and as an alternative to the proposed real-time estimation method, we also implemented an offline identification method using artificial neural networks and obtained fairly good results. The proposed methods can be implemented conveniently on a desktop PC equipped with a data acquisition board for parameter estimation of moderately complex linear multivariable systems.