Segmentação automática de registros de oscilografia fasorial com ênfase em sistemas de geração

Abstract

With the ever increasing need to obtain the reliability of electric power generation systems, the study of disturbances that occur in these systems has become a key point in preventing the supply stops. With the development of devices able to monitor these disturbances such as Digital Fault Recorders (DFRs), digital oscillography emerged, which are signs that record the magnitudes of the electric system (current and voltage) on waveforms over time or in phasor form. Upon the occurrence of a system fault, these oscillography record the disturbance. In the command centers of the generation systems, oscillography database are networked with DFRs, which sends sequentially records. A large flow of information is sent, however, in most cases, that information cannot be analyzed. The segmentation is part of the automatic analysis oscillography, which has the function of automatically analyzing oscillographys which require detailed analysis, discarding irrelevant records. This work presents a program which implements a tool called standard deviation to identify abnormal amplitude variations of the signal and has ability to identify multiple levels of segmentation. From the segmentation can begin a process of thorough analysis, generating a stream of lower records and providing agility in the study of disturbances.