Neural Model-Based Advanced Control of Chylla-Haase Reactor

Doctoral thesis English OPEN
Nabi, M

The objective of this thesis is to develop advanced control method and to design advanced control system for the polymerization reactor (Chylla-Haase) to maintain the high accurate reactor temperature. The first stage of this research start with the development of mathematical model of the process. The sub-models for monomer concentration, polymerization rate, reactor temperature and jacket outlet/inlet temperature are developed and implemented in Matlab/Simulink.\ud Four conventional control methods were applied to the reactor: a Proportional –Integral-Derivative (PID), Cascade control (CCs), Linear-Quadratic-Regulator (LQR), and Linear model predictive control (LMPC). The simulation results show that the PID controller is unable to perform satisfactorily due to the change of physical properties unless constant re-tuning takes place. Also, Cascade Control the most common control method used in such processes cannot guarantee a robust performance under varying disturbance and system uncertainty. In addition, LQR and linear MPC methods lead to better results compared with the previous two methods. But it is still under an assumption of the linearized plant.\ud Three advanced neural network based control schemes are also proposed in this thesis: radial basis function RBF neural network inverse model based feedforward-feedback control scheme, RBF based model predictive control and multi-layer perception (MLP) based model predictive control. The major objective of these control schemes is to maintain the reactor temperature within its tolerance range under disturbances and system uncertainty. Satisfactory control performance in terms of effective regulation and robustness to disturbance have been achieved.\ud In the feedforward-feedback control scheme, a neural network model is used to predict reactor temperature. Then, a neural network inverse model is used to estimate the valve position of the reactor, the manipulated variable. This method can identify the\ud controlled system with the RBF neural network identifier. A PID controller is used in the feedback control to regulate the actual temperature by compensating the neural network inverse model output. Simulation results show that the proposed control has strong adaptability, robustness and satisfactory control performance. These advanced methods achieved the much improved control performance compared with conventional control schemes.\ud The main contribution of this research lies in the following aspects. The MPC theory is realised to control Chylla-Haase polymerization reactor. Two adaptive reactor models including the RBF network model and MLP model are developed to predict the multiple-step-ahead values of the reactor output. Their modelling ability is compared with that of the models with fixed parameters and proven to be better. The RBF neural network and the MLP is trained by the recursive Least Squares (RLS) algorithm and is used to model parameter uncertainty in nonlinear dynamics of the Chylla-Haase reactor. The predictive control strategy based on the RBF neural network is applied to achieve set-point tracking of the reactor output against disturbances. The result shows that the RBF based model predictive control gives reliable result in the presence of some disturbances and keeps the reactor temperature within a tight tolerance range around the specified reaction temperature.\ud Moreover, RBF neural network based model predictive control strategy has also been used to reduce the batch time in order to shorten the reaction period. RBF neural network is considered as a prediction model for control purpose which is based to minimize a cost function in order to determine an optimal sequence of control moves. The result shows that the RBF based model predictive control gives reliable result in the presence of variation of monomer and presence of some disturbances for keeping the reactor temperature within a tight tolerance range around the specified reaction temperature without harming the quality of the temperature control.
  • References (30)
    30 references, page 1 of 3

    2.1. Introduction .................................................................................................... 8

    2.2. Polymerization reactors ................................................................................. 8

    2.3. Process modelling ........................................................................................... 9

    2.4. Control of a polymerization process ........................................................... 13

    2.4.1 Feed-forward feedback control………………………….…………..14

    2.4.2 Model predictive control……………………………………….……14

    2.4.3 Volterra model-based MPC…………………………………......…...21

    2.5. Estimation of parameters ............................................................................ 22

    2.6. Existing research of the Chylla-Haase reactor ............................................ 23

    2.7. Summary ..................................................................................................... 26 3. Chylla-Haase dynamics and simulation .......................................................... 27

  • Similar Research Results (2)
  • Metrics
    0
    views in OpenAIRE
    0
    views in local repository
    145
    downloads in local repository

    The information is available from the following content providers:

    From Number Of Views Number Of Downloads
    LJMU Research Online - IRUS-UK 0 145
Share - Bookmark