Synthesizing a Multivariable Predictive Algorithm for Noninteracting Control

A simple and effective multivariable digital control algorithm has been derived from open-loop responses. The algorithm inherently compensates for process dead-time and interactions. Process modeling is not required and the order of the process is not at all important. Inverse response behavior poses no special problems for the algorithm although an open-loop unstable plant must be stabilized with conventional feedback control first before the new algorithm can be applied. Constraints on inputs or outputs can be accommodated and robustness issues can be addressed during the design phase. The memory requirements of the algorithm are modest enough that it can be implemented on a personal computer. Simulation results are presented which show excellent capability of the algorithm.