Development of Intelligent Controller with Virtual Sensing

In  many  industrial  plants,  some  key  variables  cannot  always  be measured on-line and for the purpose of control, an alternative of sensing system is  required.  This  paper  is  concerned  with  a  development  of  an  alternative intelligent  control  strategy,  which  is  an  integration  between  the  neuro-fuzzy based  controller  and  virtual  sensing  system.  This  allows  an  immeasurable variable to be inferred and used for control. The  virtual sensor is  composed of the  Diagonal  Recurrent  Neural  Network  (DRNN)  for  plant  modeling  and  the Extended  Kalman  Filter  (EKF)  as  the  estimator  with  inputs  from  DRNN.  The integration between virtual sensor and the controller enables a development of an on-line  control  scheme  involving  the  immeasurable  variable.  The  real -time implementation  demonstrates  the  applicability  and  the  performance  of  the proposed  intelligent  control  scheme,  especially  in  dealing  with  nonlinear processes.