Resumo:
Power system operation is made from a control center. Through a complex
telemetry system, that supplies the dispatcher the current values measured into the
electric system, he/she should make the decisions of how to supply the demand of the
load. The decision-making process is supported by dispatcher's experience and by of a
available computational programs.
Há algumas décadas, as ferramentas disponíveis nos centros de controle de
sistemas elétricos vêm se beneficiando da evolução das técnicas inteligentes. Este
trabalho é mais uma contribuição neste sentido. Ele visa a aplicação de novas técnicas
inteligentes para a solução de problemas ainda em aberto nestes centros.
There are some decades, control center computational tools has used intelligent
systems. This work is one more contribution in this sense. It applies new intelligent
techniques to solve open problems in these centers.
Initially, an integrated solution for alarm processing is presented, applying the
intelligent multi-agent techniques. This solution tries to determine the main equipments
reached in a disturbance of the system.
Following, the output of the previous processing is used in order to find the
corrective measures that the dispatcher should apply into system for. These corrective
measures are sought by an orderly search in the Instructions of Operation. And then, a
solution using intelligent multi-agents is presented. This technique has been chosen for
support of the two above processing due to its flexibility, modularity and capacity of
encapsulating several techniques. The main module of this system is a ordered search by
the graph dynamic technique.
Finally, this thesis presents a tool for operative rule extraction from control
center databases. These databases possess a great number of relevant information that
the operators don’t have accessed, due to the frequency and the number of measured
points. This tool allows the extraction of rules in large databases. It has been built based
on granular computation technique. This technique was chosen because it is the best
way to represent the human reasoning.
