Estrutura multiagentes para autorrecuperação de redes elétricas inteligentes

Abstract

Electricity distribution systems represent the final link between energy-generating companies and consumers. Having radial characteristics, they have lower reliability, so a failure in some components causes the interruption of the power supply. Even with the development of new technologies to improve the operation and maintenance of these systems, losses and interruptions are inevitable. When the power supply is out of the supply due to a failure, the system must be re-established to meet the largest number of loads safely and in the shortest possible time interval. The current distribution network is at a stage of development known as the smart grid, using advances in information and communication technologies. This thesis presents a contribution to implementing self-healing strategies of distribution systems in a decentralized way, which brings as main advantages the good cost-benefit ratio, the improvement in reliability, and the possibility of incremental expansion of the system. The proposed system is based on an intelligent multiagent structure, in which each agent has a logical structure that is classified as hierarchical and hybrid. Hierarchical, because the actions are divided into four logical levels: instinct, normal operation, abnormal operation, and optimization and prediction. And hybrid, because in addition to the logical part, the rules can trigger numerical routines that help in the agents' decision-making process. In addition to self healing, several demand response functions are part of the agent, such as overload, load cutting, load prioritization, and load evolution. These functions allow the self- healing squealing solution found to consider the current system load and its evolution in the coming hours. In the proposed multiagent structure, agents only communicate with agents adjacent to it, reducing the need for a more robust communication system covering long distances. It is also proposed a structure of communication and exchange of messages with redundancy and effectiveness, mitigating the risks of erroneous communication between agents. Even with this type of communication, agent action occurs locally but is based on data received from multiple (even non-adjacent) system agents. The agent also implements functions that allow it to interact with the traditional protection systems of urban distribution networks, monitoring the operation of fuses, reclosers, and sectionalizers, allowing the advantages of an intelligent electrical network, even in a still incomplete system. The proposed multiagent system is validated through several examples using energy distribution systems and the computational implementation of the proposed hybrid hierarchical intelligent agents working together.