Learning Mechanisms in Digital Control of Large-Scale Industrial Systems

The prerequisites for digital control of a large-scale industrial system based on INDUSTRY 4.0 with the help of learning are considered. It is shown that along with the use of machine learning, big data, and artificial intelligence, it is necessary to take into account the human factor or the activity of each element of this system. This activity relates to the existence of the own goal of the element, which do not necessarily coincide with the goal of the system as a whole. The problem of constructing a learning correct mechanism of functioning of such an active system is considered. Such a mechanism should ensure both the aspiration of its element to fulfill plans and the disclosure of its potential in the interests of the whole system. This allows one to identify the parameters of the element based on the data mining and self-learning. The formal task of synthesizing the learning correct mechanism of functioning of a two-level active system is formulated. In this mechanism planning and stimulating the plan execution are combined with the identification of a constraint model of the active element based on the processing of big data about its inputs and outputs and self-learning. Constructive sufficient conditions for the correctness of learning mechanism are found. The obtained results are illustrated by the example of constructing the learning correct mechanism for increasing energy efficiency within the digital control of the holding “Russian Railways”.