A dynamic system is represented by n linear first different ial equations with constant coefficients. It is assumed, that the m input variables are linearly dependent on the n state variables. The feedback gains are considered to be optimal, if the infinite time integral of a positive definite quadratic functional of the state variables and the input variables are minimized, for a given initial state of the system. Equations were derived for optimal feedback. These equations are valid if all feedback gains were chosen arbitrarily and if one or more feedback gains are subjected to constraints. If one or more state variables cannot be fed back, the solution of the equations is subject to the constraint that the corresponding feedback gains be equal to zero. In this case the optimum depends on the initial condition of the system. Two iteration methods are suggested to solve the equations. (Author)
展开▼