Carbon production is extremely resource- and energy-intensive because of the significant shortage of output and the long duration of energy-consuming production. Increasing the efficiency of this production even to a small level leads to significant financial savings, and therefore is an urgent scientific and practical problem, especially in the current conditions of constant increase in the cost of energy.
Increasing the efficiency of the carbon product baking process is based on identifying and implementing in real time the optimum operating modes of the furnace. This task can be solved by implementing effective control systems for this technological object based on its mathematical model.
In this regard, the urgent scientific and practical task of developing a system of optimal control of the process of baking carbon products that meets the modern requirements of resource and energy conservation needs to be solved.
The application of the mathematical modeling method as a virtually non-alternative method of investigation of the process of baking graphite production and the objectives of this study are substantiated.
As a result of the analysis of the mathematical models of the baking process known to date, and similar processes, the model was used, on the basis of which further study of the modes of operation of the multi-chamber baking furnace is based.
For mathematical modeling of the campaign of baking carbon products, a mathematical model in distributed parameters is implemented based on the Reinolds-averaged equation of continuity, conservation of the amount of motion, transport of chemical components of the combustion reaction, energy in enthalpy, turbulent kinetic energy and velocity of its energy.
The methodology and procedure for conducting research into the influence of the main technological parameters of the carbon baking process on the campaign progress have been established.
It has been shown that a mathematical model of the carbon-baking process used to investigate the modes of operation of a given technological object requires a considerable calculation time - from several hours to several tens of hours - and therefore cannot be used in real-time control systems.
The known methods of simplification of mathematical models are analyzed, their advantages and disadvantages are given. Based on the analysis, it is reasonable to use a method based on the approximate representation of exact solutions. A Fourier variable method was used to construct a simplified mathematical model of the baking process. Simplified mathematical models have been developed for the three modes of operation of the chamber - cooling, heating with flue gases and "chamber under fire".
The conducted study of the accuracy and adequacy of the developed models showed their high efficiency.
An analysis of the existing quality criteria for finished products has shown that it is impossible to use them as the criteria for the finished product readiness. In this paper, it is proposed to consider entropy as a product readiness criterion, on the basis that the change in the state of carbonaceous materials undergoing heat treatment is accompanied by a change in their thermodynamic properties.
The analysis of existing control systems of the baking process has revealed the advantages and disadvantages of the proposed control systems, and also showed that the main difficulties in the development of the control system of this process is the failure to take into account the air intake, combustion air temperature, inertia of the process and its overregulation.
Ways to solve these difficulties are presented: presenting the control object as a sequence of related stages of the baking campaign, and the expediency of using MPC control. The influence of the horizons of forecasting and control during the development of MPC-controller is investigated.
To study the effectiveness of the developed control system with the MPC controller, a comparison of the latter with a classical PID-controller, tuned by the Ziegler - Nichols method, and a selective PID-controller was performed.
The feasibility of synthesis of control system based on artificial neural networks to solve the problem of determining the defect probability in the blanks is substantiated. The problem of limited necessity for training of neural network of data by application of special structure - autoencoder is solved.
Compared to the control system with the classic PID-controller, the developed system provides less fuel consumption to achieve the products of the required quality. The proposed controller provides smaller variations and temperature increases throughout the baking process.
