Intelligent computer-aided automation of operating state control for mining and metallurgical power complexes

Abstract

This paper addresses the synthesis of the informational framework for specialized expert systems designed to guarantee nominal power grid performance standards in post-contingency states through the application of heuristic situational dispatch control algorithms and automated decision support. The aim of this study is to substantiate a conceptual approach to the architecture of a knowledge base designed to support dispatch personnel decisions during power system parameter stabilization. The proposed paradigm involves generating a dataset utilizing a sensitivity matrix that correlates controlled state variables of the network with the magnitudes of dispatch control actions. The methodological framework of this research is based on the fundamental principles of power network theory, automatic control theory, artificial intelligence tools, mathematical statistics, and formal logic. Polynomial regression models were synthesized through multi-factor experimental design and the acquisition of a representative data sample. These mathematical relationships map the connection between the characteristics of the normal pre-contingency state, the network configuration, the nature of the disturbance, and the optimal magnitudes of control actions. The scientific novelty of this work lies in the development of a method for the comprehensive integration of heterogeneous information into a unified knowledge base. Specifically, formalized emergency restoration dispatch operational guidelines are integrated with calculated sensitivity matrix coefficients. The latter are determined via simulation modeling tied to the current network configuration and physical parameters of its equipment. Combining logico-semantic rules and numerical characteristics ensures high system performance sufficient for real-time operation. Results: the structural model of the knowledge base was optimized; the experimental design algorithm for calculating sensitivity matrix elements in regression models was adapted, along with the general algorithmic framework of the information and control system. The proposed heuristic control system facilitates solving a wide range of tasks: maintaining steady-state stability margins, optimizing active and reactive power flows, monitoring power grid parameters, and verifying the actions of operational and technological personnel. The practical value of this study lies in establishing a foundation for the deep intellectualization of automated dispatch control systems. Integrating the developed expert subsystem into active automated dispatch control system suites can significantly minimize decision-making time and enhance power supply reliability. An additional application of these models is their use as a baseline for training simulators intended for the professional development of dispatch and technological personnel.