Self-tuning control for systems employing feedforward

The paper concerns itself with the application of self tuning to systems incorporating feedforward control. The self-tuning controller considered minimises a cost function incorporating system output, control input, disturbance input and set-point variations. Techniques are presented which allow desired steady state and dynamic performance to be achieved by automatic adjustment of the disturbance weighting polynomial of the cost function. In the case of dynamic performance, the concept of applying frequency-response criteria to the determination of the weighting polynomial is introduced, providing dynamic filtering in the feedforward loop. The capabilities of self-tuning feedforward control and the effectiveness of the techniques proposed are demonstrated by their successful application to the control of a nuclear reactor.