A CLASS OF NONSMOOTH DISCRETE-TIME CONSTRAINED OPTIMAL-CONTROL PROBLEMS WITH APPLICATION TO HYDROTHERMAL POWER-SYSTEMS

We consider a class of discrete-time constrained optimal control probl ems in which the cost function is nonsmooth. By using a smoothing tech nique, the cost function is approximated by a smooth function. Further more, the all-time-step inequality constraints on the state variables are transcribed and appended to the cost function. This gives rise to a sequence of discrete-time optimal control problems. Each of these is solvable by existing software packages such as DMISER3. Important con vergence properties are established. For illustration, an example invo lving the long-term planning of hydrothermal power systems is solved u sing the proposed technique.