Adaptive implementation of one-step-ahead optimal control via input matching

Directly adapting feedback control parameters to match an optimal input avoids the current limitations of indirect adaptive control and direct adaptive model matching. This original concept, termed input matching, allows combination of stable parameter adjustment algorithms and optimal performance considerations. Furthermore, the emphasis on input matching permits consistent control despite inconsistent parameter identification obviating efforts such as test or probing inputs required to ensure consistent identification. The control signal is reduced to a constant weighted sum of the measurable information-state vector components by the use of a one-step-ahead quadratic cost function to govern the behavior of a linear, time-invariant multivariable plant. The control effort from this linear combination proves globally estimable by a vector equation error formulation since the one-step-ahead cost function permits simple a posteriori input error calculation. Several simulations demonstrate the behavior of this new multivariable adaptive input matching control method.