Hu Yue

Multi-source separation method for partial discharge detection in substations

Partial discharge monitoring and location system at substation receives radio frequency (RF) signals emitted by partial discharge to position partial discharge using a set of omni-directional RF antenna array. As it is possible for the entire substation to have several partial discharge sources simultaneously, the basic requirement of partial discharge detection at the substation is to judge whether there are multi-source partial discharge signals, and then identify and isolate them. Based on different time domain and frequency characteristics of ultrahigh frequency (UHF) partial discharge signals generated by different partial discharge sources, we perform time-frequency analysis and fuzzy clustering over RF pulse waveforms generated by partial discharge which are continuously acquired and stored, and isolate partial discharge signals from different partial discharge sources or interferences. By this method, we can effectively isolate and accurately position multiple partial discharge sources. Our study lays the foundation for identification of partial discharge types. The feasibility of this method has been proved by field test.

Acoustic-electrical based detection system for partial discharge localization of GIS

A portal acoustic-electrical system for partial discharge localization of GIS is developed, which can be used for multi-channel acoustic-electrical detection of ultrahigh-frequency, ultrahigh-current and ultrasonic signals generated by partial discharge of GIS. The system program, developed based on LabVIEW, is able to continuously capture, sample and store pulse waveforms of partial discharge signals. It can also calculate the phase of discharge pulse, and analyze time-frequency & amplitude-phase distribution of partial discharge signals. By calculating time difference between ultrahigh frequency signals & ultrasonic signals of partial discharge or between ultrahigh frequency signal and ultrasonic signal by multiple-channel detection, this portable acoustic-electrical system can achieve localization of partial discharge. Our system has been applied in partial discharge localization at several 220kV and 500kV GIS substations. The detection results verified the performance of this portable acoustic-electrical detection system.

Pattern identification method of partial discharge based on the features of UHF envelope signals

To determine the relationship between partial discharge type and envelope signal of partial discharge is important to evaluate the insulation state of gas-insulated switchgear (GIS). In this paper, discretization and differential matrix reduction is first conducted over the feature matrix composed of feature vectors characterizing UHF PD envelope signals using rough set theory for dimensionality reduction. Then the reduced feature vectors are used for pattern identification of four different types of UHF PD envelope signals in combination with BP neural network classifier. The results show that this method has a high identification rate.

Finite Element simulation based sensitivity analysis of UHF sensing modes for PD detection

Ultra-High-Frequency (UHF) method is an effective approach to PD detection. Currently the UHF couplers applied for partial discharge monitoring are classified into three types according to their installation places: internal couplers, external couplers and dielectric window couplers. Sensitivity analysis of these three sensing modes is carried out based on Finite Element Method simulation by using real GIS model. Simulation results indicate that the sensitivity of dielectric window couplers are nearly equivalent with that of internal couplers, while it apparently exceeds the sensitivity of external ones. Considering some other advantages of dielectric window sensors such as easy dismantling, having no impact on electrical equipment operating, convenience for maintenance, the article proposes to popularize dielectric window sensing mode for UHF PD detection.

Phase angle calculation for UHF partial discharge detection in segment acquisition mode

Phase angle of PD pulse is an important parameter of PD pattern, and its accuracy determines the effectiveness of PD pattern recognition based on phase-resolved distribution analysis. In this paper, a novel phase angle calculation method for UHF PD detecting systems in segment acquisition mode is proposed. This method use a synchronous circuit to output UHF marker signals with certain frequency characteristics to represent the rising edge of zero-crossing points of power frequency signal. The UHF PD detection system receives both marker signals and PD pulses, and records the occurrence time of all of the UHF pulse signals. Those two types of signals could be separated and marked by calculating their different time-frequency characteristic parameters using fuzzy clustering method. Then the phase angle of PD pulses could be calculated from the time difference between the UHF PD pulses and the neighboring zero-crossing points of power frequency signal. This method realizes the phase angle calculation of UHF PD signals without sampling power frequency signal, which greatly simplifies the structure of the system and reduces its requirements for data processing. The validity of the method is verified by experimental application in Laboratory.

Adaptive Wavelet de-noising method based on segmented processing technology for PD location

It is possible to carry out PD location using the time difference of PD pulses received by a UHF antenna array. In practical applications, the electromagnetic interference within the substation will pollute these received signals and make it difficult to calculate the time difference of the PD pulses, which, thus, reduces the reliability of the PD source locating. Based on the segmented processing technology, an adaptive Wavelet de-noising method is proposed in this paper. Considering that the level and the time-frequency characteristic of the background noise differ little in microsecond time period of sampling, each waveform acquired by PD detecting system are divided into two sections, the one without PD pulse is used for the calculation of the wavelet coefficients which are used to estimate the de-noising thresholds of the signal. The influence of this adaptive wavelet de-noising process on the PD pulse time delay calculation is analyzed in details. The results of simulation and application show that this method can suppress noise effectively and extract the PD pulse accurately, which contribute to the reliability of the PD source locating.