Application of iterative dynamic programming to problems with time-invariant parameters
Andrea Pietro Reverberi, DOVI', VINCENZO · CINECA IRIS Institutial Research Information System (University of Genoa) · 1996
Iterative dynamic programming has proved to be a powerful tool in the determination of optimal controls. However, the presence of time-invariant parameters to be optimized requires some modification of the basic theory. This paper considers a generalization of the original algorithm to include this class of parameters (typically equipment characteristics). A two stage algorithm based on sensitivity equations and iterative dynamic programming is proposed. The parameters contained in both algorithms (step-length and reduction-factor) are adjusted so as to improve both computational efficiency and convergence to the global optimum. An example is considered for illustration purposes.