Near ideal digital predictive compensation

The problem of realizing a near ideal predictor for use in compensation of error sampled linear systems is considered and a solution using fast-time simulation is proposed. Prediction theory is extended to include the near ideal predictor and analytical results are presented. A comparison is made between ideal and near ideal prediction as applied to the compensation of example systems. Near ideal predictive compensation is compared with conventional compensation; their combination provides a new and versatile approach to compensation. Errors in the fast-time simulation model are considered, and it appears that the fundamental advantages of using near ideal predictive compensation are retained even when the model is inexact. The ability of this new technique to provide, in effect, a zero without a pole in the z plane offers unique advantages in the compensation of systems containing a time delay.