Product planning with reinforcement learning

In this thesis, an approach has been developed that minimizes production time and cost losses in factories engaged in mass production. With this developed approach, it arranges the order of the products in order to produce the list created according to the requested product types and quantities in a way that eliminates the model change time losses in production. Thus, it is aimed that factories can produce more products with the same time and with the same workforce. The most important innovation in this study is that if the requested machines come for different products at different times, iterative learning uses a method that does not need to be retrained. In such a case, the structure trained by iterative learning can update the results according to the minimized loss of unordered products.