Study On On-line Insulation Monitoring System Of Phase-shifting Capacitors

Power compensation equipments have been widely used in electrical power system in the past decades. As the most widely used power compensation equipment, it is important to build an on-line monitoring system to comprehend the insulating state of phase-shifting capacitors. Dielectric loss factor plays a key role in power instrument on-line insulation monitoring, however, due to the influence of fluctuations in power grids and high-frequency harmonic interference and other factors, the current stability of the measurement results of dielectric loss factor is unstable and inaccurate which makes the analysis of equipment insulation status incredible. Aiming at improving the measurement accuracy of dielectric loss and the analytical level of equipment insulation status, this article focus on the use of high-precision dielectric loss measurement algorithms and data mining technology in on-line insulation monitoring system, by this way, the correlation factors of equipment insulation can be studyed in depth. The research work in this paper are orgnized as follows:A new spectrum correction algorithm, based on Blackman-Harris window, for dielectric loss measurement is introduced. The simulation result shows that this algorithm can decrease the influence of some factors such as harmonic interference, which will meet the requirement of high-precision measurement in dielectric loss signal.This paper presents the overall design of an on-line insulation monitoring system for phase-shifting capacitors. The entire monitoring terminal modules are connected by a fieldbus network based on CAN bus and detail material on the hardware and software design, as well as the communication protocol design are given. System software are developed on LabVIEW, the basic database operations such as query, store, table create, can be conducted in this software and the high-precision dielectric loss measurement algorithm are programmed on"G"language too. As to the long-term trend extraction of dielectric loss, multi-scale wavelet analysis is used; and a support vector machine regression (SVR) technique is introduced to establish the model of power equipment insulation status change.As an open, modular design method is introduced in the insulation on-line monitoring system design, it can be upgraded to monitor other electrical power equipments easily.