Dynamical System Approach for Optimal Control Problems with Equilibrium Constraints Using Gap-Constraint-Based Reformulation

Abstract: Optimal control problems for nonsmooth dynamical systems governed by differential variational inequalities (DVI) are called optimal control problems with equilibrium constraints (OCPEC). It provides a general formalism for nonsmooth optimal control. However, solving OCPEC using the direct method (i.e., first-discretize-then-optimize) is challenging owing to the lack of correct sensitivity and constraint regularity. This study uses the direct method to solve OCPEC and overcomes the numerical difficulties from two aspects: In the discretization step, we propose a class of novel approaches using gap functions to smooth the DVI, where gap functions are initially proposed for solving variational inequalities. The generated smoothing approximations of discretized OCPEC are called gap-constraint-based reformulations, which have a concise and semismoothly differentiable constraint system; In the optimization step, we propose an efficient dynamical system approach to solve the discretized OCPEC, where a sequence of gap-constraint-based reformulations is solved approximately. This dynamical system approach involves a semismooth Newton flow and achieves local exponential convergence under standard assumptions. The benchmark test shows that the proposed method is computationally tractable and achieves fast local convergence.