Abstract
A computational algorithm is suggested in this paper to solve optimal control problems for linear lumped paramter systems with inequality constraints.
A penalty function is introduced, taking into account the inequality constraints and the system dynamics, so as to obtain the solutions of optimal control problems without explicitly solving the system dynamic equations.
Since the dynamic equations are regarded as equality constraints, the introduced penalty function is static and non-dynamic.
It is shown that this penalty function attains the minimum where the gradient is zero, and that the optimum of the original constrained problem is approached sequentially by solving unconstrained and non-dynamic new problems, each of which has a solution interior to the constraint set.
