(202h) Artificial Neural Network Based Control of Bioreactor With Input Multiplicities | AIChE

(202h) Artificial Neural Network Based Control of Bioreactor With Input Multiplicities

Authors 

Prabhaker Reddy, G. - Presenter, UNIVERSITY COLLEGE OF TECHNOLOGY , OSMANIA UNIVERSITY
Radha Krishna, P., UNIVERSITY COLLEGE OF TECHNOLOGY, OSMANIA UNIVERSITY
Swetha, C., UNIVERSITY COLLEGE OF TECHNOLOGY, OSMANIA UNIVERSITY



In this paper, the Neural network based NARMA-L2 controller  is analyzed to a continuous bioreactor which exhibits  input multiplicities [1-2] in dilution rate on productivity. i.e., two values of dilution rate will give the same value of productivity. In the first step the neural network model of bioreactor is obtained by Levenburg- Marquard [3-6]training.The data for the training of the network is generated using mathematical model of bioreactor.The neural network model of bioreactor is used in NARMA-L2 controller design. The performance of neural network based NARMA-L2 controller is evaluated using MATLAB. The neural network controller performance is evaluated in set point change in productivity and in the presence of disturbance in feed substrate concentration. The Performance of Neural network based NARMA-L2 controller and conventional PI controller have been evaluated  through simulation studies. As the NARMA-L2 controller provides always the two values of Dilution rate for control action and by selecting the value nearer to the operating point, it is found to give stable and better responses than conventional PI controller.  The PI controller results in wash out condition or switch over from initial lower input dilution rate to higher input dilution rate or vice versa.  Thus, NARMA-L2  controller  is found to overcome the control problems of  PI controller due to the input multiplicities.

References

[1]. Chidambaram. M and Reddy G.P. (1994), “Nonlinear control of bioreactors with input multiplicities”,

         Computers & Chemical Engineering, Vol. 11, pp. 97-100.

[2]. Chidambaram and G.P.Reddy (1995), “Non Linear IMC of Systems with Input Multiplicities, Indian Journal

           of Chemical Technology, Vol. 21, pp.115-118.

[3]. Dauda Olurotimi ARAROMI, Tinuade Jolaade AFOLABI, Duncan ALOKO (2007), “Neural Network Control of CSTR 

      for Reversible Reaction Using Reverence Model Approach”, Leonardo Journal of Sciences, ISSN 1583-0233,

      pp. 25-40.

[4]. Dauda Olurotimi Araromi Aremu, Mujidat Omolara Adedeji, Oluwaseun Ezekiel PID (2000) “Control of aerobic

      waste water reactor using relay feedback”, Trans. Journal of Science and Technology,Vol. 2, pp. 108-124.

[5]. D.B.Anuradha, G.Prabhaker Reddy*, J.S.N.Murthy (2009), “Direct Inverse Neural Network Control of A

     Continuous Stirred Tank Reactor”, IMECS, Vol.2, pp. 236-241.

[6]. Farzad Hourfar , Karim Salahshoor, Alireza Paivar (2006), “A Neural Adaptive Feedback Linearization Control 

      for CSTR Using NARMA-L2 Model”, Leonardo Journal of Sciences, Vol. 5, pp. 285-290.