A simplex algorithm for on-line computation of time optimal controls

The minimum-time regulator problem is solved computationally for general linear-discrete systems by a modification of the simplex algorithm of linear programming (LP). The algorithm presented is faster than the LP solutions devised previously, taking but a single application to solve for the optimal control. An extension of the LP bounded-variable technique further reduces computer time and storage requirements. Application to a sixth-order process with two control inputs shows that the algorithm is sufficiently economical of running time and storage to be implemented on-line with a small process-control computer.