The scope of this thesis is to address production planning in stochastic environments by proposing a cost function approximation solution for it. We use a master production schedule, extracted from point forecasts, to define the weekly production goals of a disaggregated production level. The low- level planning occurs for multiple products in a shrinking time horizon, where both demand and supply are stochastic. By combining cost function approximations with a deterministic lookahead simulator, we showcase that the overall profit over the course of the planning horizon can be increased, compared to the non-parametrized planning alternative. The results are directly applicable to real world production systems in the form of decision-making policies. Lastly, we compare the profit performance of our approach to the more conservative production planning approach of scenario optimization, where we consider various forecast scenario combinations of the stochastic demand and supply.     «

The scope of this thesis is to address production planning in stochastic environments by proposing a cost function approximation solution for it. We use a master production schedule, extracted from point forecasts, to define the weekly production goals of a disaggregated production level. The low- level planning occurs for multiple products in a shrinking time horizon, where both demand and supply are stochastic. By combining cost function approximations with a deterministic lookahead simulator,...     »