Control Mechanisms for Large Company's Branch Production with Supervised Learning

We are considering a model of a large company with a subsidiary with a production base. The company is headed by top manager (the Boss), who is assisted by the Adviser. The head of the branch (Manager) organizes production, including at a subordinate mill headed by the Director. Production is affected by random external influences. When making decisions, the Manager knows exactly the production capacity at the branch level. But he does not know the maximum production capacities (limits) of the mill at random external influences. The Director knows exactly these limits when making decisions. Neither the Adviser nor the Boss knows the capabilities of the branch and the limits of the mill. Thus, the Manager can manipulate the branch's production volumes to influence the Boss's decisions in order to increase his own remuneration. Therefore, the Boss needs to learn how to control the Manager so that the production volumes in the branch are maximized under random external influences. Likewise, the Director can manipulate the performance of the mill in order to influence the Manager's decisions about his own remuneration. Therefore, the Manager needs to manage the Director in such a way as to increase the production of the mill. As a result of building and studying this model, a company management mechanism was found, in which the branch's production is maximized under random external influences. This mechanism encourages the Manager to use an effective mill control mechanism and other opportunities to increase production in the branch. The application of this approach is illustrated by the example of freight wagon repair management.