ABSTRACT
The piecewise constant control policy provides a reasonably good approximation to the continuous control policy, and the approximation can be improved by taking a larger number of time stages. The use of piecewise linear control does indeed give an excellent result with a relatively small number of time stages, and allows the optimal control policy for very high dimensional systems to be determined accurately. This chapter discusses the algorithm for iterative dynamic programming for piecewise linear control. It presents several examples, including non-differentiable system, linear system with quadratic performance index, and gas absorber with a large number of plates, to show the advantages of using piecewise linear control.
