The development of automation and microcontroller equipment allows software implementation and industrial application of adaptive and optimal control systems. The issue of tuning the PI controllers of the zones of a furnace for heating steel before rolling at the Oskol Electrometallurgical Plant is resolved. The tuning of the PI controllers is described, and the choice of an adaptive control system based on a switching table is justified. The adaptive system allows improving the quality of control and keeping the temperature within the required operating range.

This paper discusses whether laser technology holds promise for use in marking steel billets in a continuous-production environment. A laser-marking experiment was performed on “dull metal” and “bright metal,” which have different oxidation layers. The experiment was performed on rolled-shape samples produced using a wide variety of carbon steels and differing scale produced by using different cooling rates during heat treatment (in water, or in a furnace). A QR code or bar code was applied to the heat-treated samples using laser markers where the laser exposure and laser power used for application of the code were varied. The experimental results revealed significant technical difficulties preventing use of this technology with “dull metal” (application of the code to areas coated with scale causes the corresponding areas to change color — gray instead of bronze resulting in the code being impossible to read using standard equipment). Delamination of the oxide coating on the billet will lead to loss of code, and make the code impossible to subsequently read. Successful laser marking of “dull metal” requires initial preparation of the surface for marking in order to remove scale inclusions, so that the code will have uniform color and contrast. A clean surface requires less time to laser mark and lower laser-marker power because of the high contrast between the code and the surface. The code remained intact and readable after long storage (on the order of 9 months) under standard conditions.

The article proposes a method for constructing and correcting a multistep forecast for the year ahead (with a monthly breakdown) of prices for raw materials and products of industrial enterprises. The proposed approach consists in the formation of a price forecast taking into account 1) the price of the predicted indicator, the prices of goods participating in the product value chain, and macro indicators (time series); 2) information about the strength and direction of environmental factors affecting the market. Structured information about the effects of the external environment is the result of processing expert knowledge and hypotheses from heterogeneous information sources, through analysis and modeling on a cognitive map of the situation (CCS). We form a forecast by constructing an ensemble of time series models, each of which reflects the dependence of the target indicator on its past values and the prices of related products, the composition of which is determined by the results of cognitive modeling and time series analysis. Based on the results of monitoring on the cognitive map of the situation, conducting in order to analyze possible changes in the external environment and digital monitoring of prices, to identify changes in prices modes, we perform a forecast correction. The results obtained in this study show that the use of cognitive modeling and monitoring of changes improve the accuracy of predictions.

The paper investigates the problem of adaptive planning and control of a generalized computational experiment in mathematical modeling of real physical processes. A generalized computational experiment implies a multiple solution of the numerical simulation problem for different sets of values of defining model parameters. A general structure of a control model for a generalized computational experiment is proposed, the concept of the state of experiment is introduced, and a mechanism for evaluating its effectiveness is considered. A description of the concepts of reliability, representativeness, and efficiency for a generalized computational experiment is introduced. A detailed description of the algorithm for calculating the effectiveness of a generalized computational experiment based on the analysis of series of approximating curves built on subsets of modeling parameters is given. An example is given of determining the effectiveness of the state of a generalized computational experiment when assessing the accuracy of numerical models of the OpenFOAM software platform for a three-dimensional problem of inviscid flow around a cone.

One of the main factors that have a direct impact on the crop yield is correct fertilization for achieving an optimal soil composition. The yield, in turn, is one of the main indicators of the agroeconomic substantiation of intra-farmland use projects. There are many factors to consider when choosing a fertilization schedule. These include, in particular, the planned yield, the fore crop previously cultivated on the sown area, the composition of the soil, the region and the crop planned for sowing, etc. A scientifically substantiated solution to such a problem is impossible without the use of economic and mathematical methods and optimization models. Methods of linear optimization have been widely used in land utilization by now. The article proposes an algorithm for calculating the doses of mineral fertilizers, which consists of two stages: calculating the amount of active substance for three main macroelements and directly calculating the dosage of fertilizers applied, based on a linear optimization model.