A two-level decomposition-aggregation approach for large-scale optimal control problems (Q1903655)

A new approach is proposed for the solution of large-scale constrained optimal control problems governed by ordinary differential equations. The method is intended for problems where the main difficulty is the large number of controls and hard control constraints. A typical example of such a system is a dynamical resource allocation problem. The method is based on aggregation-disaggregation ideas and is within the class of feasible decomposition methods. The aggregation is applied to the controls. It is shown that the problem of updating the disaggregation in each iteration decomposes into independent subproblems of lower dimension. If the original problem has block- or block-separable structure, then the subproblems are formulated according to the blocks. Numerical examples to illustrate the method are given.