In this article, we consider the solution of adaptive control problems by developing a unified algorithm for regulator tuning. The task of adaptive control of dynamic processes in real time, today plays a key role in the development of modern control systems. Existing adaptive management development assumes, as a rule, an unchanged structure of the technological process, where only the parameters of the process under investigation are subject to change. To create such systems, detailed modeling of the technological process is necessary, while the typical management model after identification becomes unique for a particular enterprise due to the parameters found. This system is designed for operating objects that allows you to take into account the dynamic change of parameters in the process of object identification while the stage of mathematical modeling is replaced by a standard approximating model.

Today the industry is facing a shortage of skilled workers and an aging workforce, which will eventually lead to a loss of knowledge. But augmented reality technology can connect field workers with experts who are able to provide remote guidance in real time. The subsequent advantage is that the information obtained by AR devices can be used as accumulated successful experience in the future, which facilitates decision-making in specific business processes of the company. With AR, employees gain experience and skills much faster. This paper shows the application of augmented reality technology in the maintenance of electromechanical equipment. The main functions of the augmented reality system for servicing electrical equipment are presented, the solution to the problem of integrating an augmented reality software application with existing automation systems is shown, and the methods of interaction of the developed software module with third-party modules, for example, various analytical modules, etc. are described.

The paper presents a methodology for training neural networks for vision tasks on synthe-sized data on the example of steel defect recognition in automated production control systems. The article describes the process of dataset procedural generation of steel slab defects with a symmetrical distribution. The results of training two neural networks Unet and Xception on a generated data grid and testing them on real data are presented. The performance of these neural networks was assessed using real data from the Severstal: Steel Defect Detection set. In both cases, the neural networks showed good results in the classification and segmentation of surface defects of steel workpieces in the image. Dice score on synthetic data reaches 0.62, and accuracy—0.81.