Search speed control strategies for the solution of optimal control problems

Implementation of the Maximum Principle by a continuous gradient search for the extremals of a restrained Hamiltonian function, allows gross solution errors to occur from changes in the search velocity vector. The nature of these errors is outlined and two search speed control strategies are presented for overcoming the solution error problem. Experimental results from an analogue computer study demonstrate the relative values of the alternative speed control techniques in illustrative examples with linear and non-linear system performance functions.