A New Multivariable Self-tuning Control Algorithm and Its Applications

This paper presents a new multivariable self-tuning algorithm which adopts the techniques of updating weighting polynomial matrices of the cost function and eliminating the steady-state errors without using an integrator.The method of updating weighting matrices does not affect the control-law estimation,so makes the algorithm simple and gives nice performance.The results show that the self-tuning algorithm proposed has better performance than that of Koivo in the experiment oncontrol of a double-input-double-output electrically heated furnace.