Approaches to Computationally Efficient Implementation of Gain Governors For Nonlinear Systems With Pointwise-in-Time Constraints

The gain governors use receding horizon optimization to adjust parameters (such as gains) in the nominal control laws. The parameters are optimized at each time instant to minimize a cost function subject to pointwise-in-time constraints and subject to the condition that the parameter values are constant over the horizon. The gain governors may be viewed as a special class of Model Predictive Control (MPC) algorithms. They provide guaranteed stability properties without terminal set conditions as well as a large degree of flexibility in accommodating the on-line computational effort. The paper reviews the properties of the gain governors and discusses different implementations allowed by the general theory with a view towards effectively accommodating the computational effort involved with the on-line optimization.