APPLICATIONS OF SELF-TUNING PID-CONTROLLERS

Because of its simplicity and robustness conventional PID control still predominates in process industry and a wide range of applications. However, the classical methods for tuning PID-controllers have the disadvantages of being time consuming and requiring open-loop experiments. Other proposed approaches require using an intentional step response to determine the controller parameters. This dissertation applies the recently developed adaptive control concept to investigate the self-tuning PID-controller which can adjust the controller parameters on-line and is especially useful for the applications for which the use of an intentional step test is undesirable.; For systems represented by a second-order model, a recursive parameter estimation algorithm is used to estimate the system parameters on-line and the estimated system parameters are used to compute the PID-controller parameters based on pole-placement design method. Thus, by updating the system parameters the controller can then adjust its parameters in response to changes of operating conditions.; Since PID-control is usually designed for a class of control systems to track constant setpoints, during the steady-state the lack of persistent excitation of plant inputs and outputs could pose a problem for parameter estimation. Remedies like the dead-zone scheme are discussed. Due to the PID-control structure and the pole-placement design method the self-tuning technique theoretically is limited to only second-order system models. Thus, another problem involves reduced-order modelling. In this dissertation, a technique of information filtering is proposed in which the plant inputs and outputs are filtered prior to their use in determining the adapted system and controller parameters. This modification can then provide an improvement in the robustness of the adaptive controller to unmodelled plant dynamics.; Finally, applications of self-tuning PID-controllers implemented on a microcomputer for controlling the left ventricular assist device and the total artificial heart are studied. The performance of the developed system has been demonstrated by in vitro experiments on a mock circulatory system and promising results have been obtained.