Download
slide1 n.
Skip this Video
Loading SlideShow in 5 Seconds..
UHF detection of partial discharge in oil–insulated power transformers PowerPoint Presentation
Download Presentation
UHF detection of partial discharge in oil–insulated power transformers

UHF detection of partial discharge in oil–insulated power transformers

169 Vues Download Presentation
Télécharger la présentation

UHF detection of partial discharge in oil–insulated power transformers

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. UHF detection of partial discharge in oil–insulated power transformers 10 Author – Alan Pak Lon Ao (z3235241)Supervisor – Dr. ToanPhung Assessor: Dr. JayashriRavishankar • All power equipments need good insulating systems to prevent dielectric breakdown and unwanted current flow between conductors. Partial Discharge (PD) is a one of the usual disasters which partially bridges the insulating material between conductors. Online monitoring of PD is significant for preventing power transformer failure that may lead to huge economic loses. One way of PD detection is via capturing the emitted electromagnetic (EM) waves during discharge by ultra high frequency (UHF) antennas. • The problem statement is defined as: • Detecting and localizing PD inside oil-insulated transformers using UHF method. • The high-level structure of UHF detection system is presented by a four-step approach as following: • Electromagnetic waves generated by PD • Signals detected by UHF Antennas • Waveforms stored by oscilloscope • Data analysed by computer program • Hence, the two main objectives are classified as designing a type of UHF antenna and developing a series of MATLAB® functions which helps to localizePD. • 10cm-diameter log spiral Antenna • Micro strip type antenna is preferable since it provides good uniformity. The size of antenna should be limited as one or more of them will be installed inside transformer tanks. The physical looking of log spiral is shown in the following figure. • Simulations of varies design parameters have been completed to obtain the best possible response with the size constraint. More simulations for other micro strip type antenna have also been carried out. Log spiral (middle in figure above) is concluded as the best among log periodic and spiral. • A balun (left in figure above) is needed for connecting antenna with an unbalanced transmission line and also transforming the impedance of antenna to match up with the impedance of transmission line. Its design is based on a Chebyshevmulti-section impedance transformer circuit. The overall performance of built antenna has been tested for verifying the consistence with simulated result. • Experiment setup • Background and problem statement • Four log spirals are placed coplanar on an antenna array (left in figure below) which is expected to overcome the issues of reflection and obstruction inside transformer tanks. An artificial corona partial discharge (middle in figure below) has been set up and immersed in oil during experiment. The antenna array is installed in air gap half way between the oil level and the lid. • The model of UHF detection system is shown in the circuit diagram below. • Wavelet de-noisingThe signal and noise manifest differently in the post-decomposition results. The noise components are hence possible to be distinguished as they have lower energy level. • PD arrival time estimationCorrelative first-peak energy method estimates the time shift of first arrived PD energy between antennas. • Two level space-grid-search • Light of sight propagation is assumed and different propagation speeds in respective media have been counted in. The best possible PD location is estimated by minimizing the differences between actual and measured time delay. • The experiment results have shown that this UHF detection system is able to identify the existence of partial discharge. The arrival order of PD signals can be successfully determined by correlative first-peak energy method. However, the noticeable errors in arrival time could not made estimating PD location completely reliable (S=57.1%). De-noised waveforms are proven to be more effective for localization (S=71.4%). • Future works such as employing more sensors in antennaarrayandahighersamplingrateoscilloscope are expected to improve accuracy in localization. The application can be further expanded to identify Pd types. • Approaches and objectives • Data analysis • Antenna construction • Results discussion and future works