USE OF ITERATIVE DYNAMIC-PROGRAMMING WITH VARIABLE STAGE LENGTHS AND FIXED FINAL TIME

The use sf time stages of varying lengths in iterative dynamic program ming (IDP) allows accurate switching in control policy, and therefore it is possible to obtain accurate optimal control policy with the use of a small number of time stages. In optimal control problems where th e final time is specified, this restriction can be handled with the us e of penalty functions. By using a variable shifting term in the quadr atic penalty function, the final time is reached accurately, and the p erformance index is extremised. Three examples, which are used to test the approach, show that very good results can be obtained.