Gradient Optimization Algorithms for Systems with Terminal Constraints.

摘要

The application of optimization techniques to the derivation of predictive information for flight displays is being investigated in connection with operational situations involving large disturbances and manoeuvres. This note surveys current gradient methods for computing extremal solutions of optimal control problems for systems having boundary conditions but without state or control constraints. It is concluded that a two stage procedure is required, with the first stage using a first order gradient, while the second stage would use a higher order variable metric method. Proposals are advanced for further research on optimization algorithms. (Author)