Desenvolvimento de solução técnica com integração de software e hardware para medição, separação e identificação de descargas parciais em equipamentos elétricos

Abstract

The costs from electrical equipment failures and the unavailability of the power supply directly impact the utilities' operational results. Therefore, improving the reliability of the electricity supply is one of the technical reasons for investing in the electrical power system, especially power substations. Investments in measurement and monitoring systems allow a more assertive diagnosis of equipment insulation and are essential to improve condition-based maintenance protocols, increasing reliability and minimizing fault occurrence. In this context, diagnosing the equipment insulation quality with partial discharges (PD) measurements is a valuable resource that is widely used. However, there are difficulties in analyzing the measured signals since, during PD measurements, it is common for more than one phenomenon to cooccur, whether noise or multiple PD sources, compromising the analysis and subsequent identification and classification of the PD activities. In this light, to contribute to solving this problem, the objective is to develop a cost-effective solution that allows for assertively measuring, separating, and classifying the basic types of partial discharges found during PD measurements in electrical equipment: corona, superficial, and internal. The proposed solution consists of hardware integrated with software capable of measuring, separating, and classifying partial discharges. The solution has digitizing hardware suitable for PD measurements and flexibility to work with conventional (IEC 60270) and non-conventional methods. The software allows the control of measurement settings by applying noise mitigation techniques and PD pulse extraction; it also allows the separation of the measured phenomena in different PRPD (Phase Resolved Partial Discharges) patterns with a Time-Frequency Map (T-F Map). The measurement validation process, which includes measurements with a PD simulator, demonstrates the solution's reliability. The work evaluated the separation of partial discharges with the PD pulses measured in the stator of a generator and measurements of simultaneous partial discharge sources provided by the PD simulator. Finally, this work validates the results of the identification mechanism for PD categories, corona, surface, and internal, with a test dataset extracted from the general database developed. The results indicate that the proposed solution could adequately measure the partial discharge signals, producing PRPD patterns equivalent to those obtained with the commercial system, with deviations of less than 4% for the parameters evaluated, as well as for the measurement of apparent charge, in which the average error was 1.3%. Despite its limitations, the separation provided by the T-F Map was satisfactory and could separate overlapping PD. The identification mechanism classified the three basic PD types with an accuracy of 95.4%.