Robust optimal control of material flows in demand-driven supply networks

We develop a model based on stochastic discrete-time controlleddynamical systems in order to derive optimal policies for controllingthe material flow in supply networks. Each node in the network isdescribed as a transducer such that the dynamics of the material andinformation flows within the entire network can be expressed by asystem of first-order difference equations, where some inputs to thesystem act as external disturbances. We apply methods from constrainedrobust optimal control to compute the explicit control law as afunction of the current state. For the numerical examples considered,these control laws correspond to a certain classes of optimalordering policies from inventory management while avoiding, however,any a priori assumptions about the general form of the policy