Intelligent Monitoring Equipment in Power Plants using IEC61850 Sample Values, Paraconsistent Artificial Neural Networks and Condition Based Maintenance
Antonio Bernardo de Vasconcellos Praxedes1; Alberto José Alvares2
1 Universidade de Brasilia - UnB; 2 University of Brasília - UnB
doi:10.20906/CPS/COB-2015-0294
Resumo
The high availability requirements and reliability in power generation required by the National Electric System Operator (ONS) and the Regulatory Agency (ANEEL) require the use of innovative techniques to aid the maintenance and operation in power plants. This work proposes to demonstrate that it is possible to develop a methodology for diagnosis and fault prognosis in systems, subsystems, and equipment of the plant in real time using Paraconsistent Artificial Neural Networks and the concept of Condition Based Maintenance on along with a new communication protocol for monitoring and control based on ethernet network. To achieve this goal, the following assumptions are considered: 1. High speed of ethernet available along with use of the IEC61850 standard Sample Values communication service (SV), lets you send high speed measurement information from the sensors to the pattern recognition software. This enables real-time analysis and also the monitoring of a large number of systems, subsystems and devices in a centralized manner, avoiding the use of separate systems for diagnosis and fault prognosis; 2. Use of Artificial Neural Network Paraconsistent (RNAP) as the supervisory information analysis feature. This type of network has the characteristics of tolerance to noise, multiple patterns, conditioning, plasticity, memory, processing inconsistencies, fast and efficient software implementation, being a new Artificial Intelligence feature. Thus, the RNAP is a resource, that besides making the recognition of failure patterns can also treat properly supervisory information from two devices operating in redundant mode.
Palavras-chave: IEC61850 Standard; Paraconsistent Neural Network; Condition Based Network