الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
instant of application, is additively incorporated into the performance index. The dual of the regulator problem with cost associated with the application of a nonzero control is a least squares estimation proble~ in which there is a
cost for taking measurements. .
For the transfer problem with minimum total variation, functional analysis techniques are used to characterize the optimum control and to deduce some of its properties, such as its piecewise-constant nature and the number of times it switches. Some attention is given to computational matters, and detailed procedure for computing the optimum solution for systems with one or two outputs, where a reasonable amount of computational complexity is involved, are given.
The matrix minimum principle is used to derive neces¬sary conditions for optimality when the regulator problem with a cost for control application is modified to yield
a total cost indepedent of the initial state. Because these necessary conditions are in the form of a two point boundary value problem in nonlinear matrix difference or differential equations which are difficult to solve in general, some easily-computed suboptimal control policies are derived and some useful bounds are derived on the number of times the optimal control is nonzero. Con¬siderable attention is also given to special cases for which the computation of the true optimum control is relatively easy, notably the case when the system’s
output is a single linear combination of state and is costed at the final time only. A regulator problem in which a cost is made for changing the control is examined by reducing it to the problem involving a cost on applying a nonzero control.