The digital adaptive control of a linear process modulated by random noise

A procedure for the design of a digital controller to compensate a certain class of linear time-varying processes is presented. The process time variation may be rapid compared to the input signals and is assumed to be caused by modulation of a plant with known parameters by a number of correlated random disturbances. Correlated additive noise may also be present. Plant identification is performed using state variable estimation techniques, while correlation of the random disturbances allows future plant behavior to be predicted. The adaptive controller is designed to optimize the predicted plant behavior over the near future, with controller redesign taking place as new information becomes available from the estimator. Test signals are not required if the plant statistics are known.