Algorithms for the Adjustment of Technological Process Parameters in Power Installations.

Abstract

The article considers the possibility of developing algorithms for adapting the control systems of
technological plant parameters based on a neural network approach and designing this control system on
the basis of modern automatic control means. The development of algorithms and software and
hardware for real-time automatic control of power plants was carried out on the basis of the following
tasks: development of the structure of the automatic system and the structure of the control system;
implementation of the software and hardware implementation of the control system, namely the
implementation of the software structure and the software implementation of the proportional-integraldifferential controller. The system of adaptive training of the neural network was implemented.
To solve the tasks, the following methods were used: a neural network approach to solving
problems of finding the optimal parameters of a process controller in power plants; object-oriented
programming for the implementation of a control system. To search for the parameters of a
proportional-integral-differential controller, it is proposed to apply a neural network approach, in
which a neural network is embedded in the control system. The proportional-integral-differential
controller has the widest possibilities for imparting the necessary properties to the control system. It
is used in cases where it is necessary to obtain a high-quality automatic control system without high
costs for research on the synthesis of a more complex control law.
Based on the research results, we analyze the technological process taking place in a steam
boiler, generalize the algorithms for controlling parameters, and select an artificial neural network
apparatus for calculating the coefficients of a proportional-integral-differential controller.