A generalized self-tuning feedforward controller and multivariable application

This paper describes a new generalized self-tuning feedforward controller which combines the classical strategy of pole-zero placement with the optimal strategy of self-tuning control and possesses the major benefits of both. The techniques of updating the weighting polynomials of the cost function on line and eliminating the steady-state error without using an integrator are proposed. The derived controller not only cancels the effect of measurable disturbances exactly but also is applied to a multivariable system to realize decoupling control. The simulation example and successful application to a multivariable electrcheated system demonstrate the effectiveness of this controller.