| Abstract [eng] |
In this thesis, the problem of pH regulation in fed batch biotechnological processes is addressed, where the process dynamics are nonlinear, time variant, and highly uncertain. The modeling of such systems is often complex, and the usual model-based control techniques are not very effective. The goal of this work was to design and test a model-free adaptive control strategy to keep the pH at the desired pH without the use of an explicit mathematical model of the process. Based on literature, a fed-batch process model was implemented in MATLAB/Simulink and firstly, a classical fixed PI controller and adaptive gain scheduling PI controller was implemented as a reference approach. Later, an adaptive (model-free) controller (MFA) was designed that modifies the control actions as per the process of input-output data. Each one of the three controllers were tested under various conditions, with a sampling time of 1s. The simulation results show that the model-free adaptive controller achieves improved performance in pH regulation compared to the classical fixed PI controller and adaptive gain scheduling PI controller, particularly in terms of tracking accuracy, and adaptability to disturbances and modeling uncertainties. The MFA approach has proven to be effective in achieving stable pH control without the need to know the system dynamics accurately. The greatest value of this work is the proof of an effective model-free control strategy for fed-batch bioprocesses for pH regulation. It can be used in systems where the modeling approach is complicated or plants with a lower amount of available data. Future studies can be done for experimental validation of the designed MFA controller and optimization of the control strategy for practical applications in different engineering applications. |
