Predictive adaptive feedforward control with input constraints

This paper is concerned with predictive adaptive feedforward control from accessible disturbances, incorporating constraints in the manipulated variable variance. An algorithm, resulting from the modification of the MUSMAR controller, is presented. Its possible convergence points are characterized, using the ODE method for analysing stochastic recursive algorithms, as the local minima of the steady-state quadratic cost taking into account the a priori chosen regulator structure and the maximum allowed value of the manipulated variable variance. Simulations confirming the actual convergence to these points are presented.