Software implementation of mathematical models for production-rule knowledge bases in power system dispatch control

Abstract

This paper analyzes approaches to knowledge-base-driven intelligent information computer systems designed to support the dispatch control of electric power systems under time and resource constraints. Significant results were obtained through concepts of situational management, simulation modeling, and heuristic analysis, as well as the application of pattern recognition methods. The core idea for solving the problem of ensuring power system mode quality is that the bulk of the initial topology and state data is pre-processed, while the direct control actions are determined using relatively simple algorithms based on data concerning the previous normal state and the vector of disturbance parameters.

Success in solving power system mode control tasks is largely determined by the representativeness of the mode-quality characteristic parameters.

Thus, the search for optimal control actions entails solving the following tasks: recognition and classification of emergency situations; determination of the locations and types of control actions depending on the specific emergency context; and determination of the dosage of dispatch control actions.

The study highlights the specific application of the production-rule form of knowledge representation, which is based on the implementation of conjunctive-disjunctive logical operations. The task of building production-rule expert systems is considered on the theoretical basis of Petri nets, which enables the generation of consistent logical decision-making models while accounting for parallelism. The paper examines the task of ensuring the quality of the electric power system in both normal and post-emergency modes by utilizing heuristic methods for situational management and dispatcher decision support. It is proposed to construct the knowledge base using parameter training sets generated according to factorial design matrices. The study proposes an integrated knowledge representation model for the dispatch control domain based on semantic and production networks.

The research results can be utilized both at the higher level of dispatch control and in regional power grids during the planning of emergency control measures.