Self-learning and Controlling of Production in a Large-scale Company

The article discusses the problem of controlling production in a large-scale vertically integrated company with representative offices and enterprises located in different regions, in conditions of uncertainty. Controlling production in such a company based on the concept of Industry 4.0 in the era of artificial intelligence includes machine learning. At the same time, need to consider activity of managers of underlying company units associated with the presence of their own targets, which do not necessarily coincide with the target of the company. To achieve own targets in the face of uncertainty, they can manipulate metrics in subordinate units to influence machine learning outcomes in their favor. The article examines the problem of using both the theory of organizational control and machine learning to increase production in domain. The manager of underlying unit of vertically integrated company usually knows the own capability of production better than the manager of superior unit. Therefore, such manager can manipulate own volume of production to get more stimuli from the superior unit as a result of machine learning. Such unwanted activity can result in decline in total company production. To avoid this, a controlling mechanism is proposed that includes two-level self-learning and stimulation procedures. Sufficient conditions have been found for the synthesis of such a controlling, in which stochastic possibilities of increasing the company’s production are used. An example of such controlling is considered when managing the repair of freight carriages in a large-scale carriage repair company.