The procedure for identification of a cascade of technological facilities involved in the process of iron ore dressing at the mining and processing plant is carried out. The purpose of the development of neural network models of the units under study is the implementation of a predictive system, which is based on the analysis of current values of technological indicators. Such system is used to simulate the future trend of the production situation to present it to the operator. The expediency of using this apparatus is explained by the complexity of the ongoing iron ore material enrichment process. It includes multidimensional multi-connected objects with significantly non-stationary parameters due to changes in the physicochemical characteristics of the raw material. So, functioning modes of the units need to be changed by the operator depending on the class of ore being processed. In addition, the task is complicated by the nonlinearity of the characteristics of many objects, their considerable inertia and the presence of recycling flows. An additional difficulty is the discrete laboratory evaluation of the useful content (the percentage of iron) in the output product. It is produced at the end of the process chain and does not allow for prompt assessment of the result of a change in control signals. These characteristics significantly complicate the implementation of an effective management strategy by the operator. The primary task for the operator is to prevent emergencies while managing the process within the technological requirements of production. Therefore, operators are forced to form a so-called "stock from above", thereby reducing the specific energy efficiency of the technological process for its stability. In these conditions, it becomes urgent to develop a system that is able to predict the overall possible course of development of the production process with high accuracy on the basis of the analysis of technological indicators at each of the production line facilities,. For this, an artificial neural network apparatus was used Due to the possibility of approximating complex nonlinear dependencies, it will allow to develop models of technological objects distinguished by high quality of operation.
