| HVDC links can transfer heavy capacity to far distance and become the only technical solution for long links. These features turn HVDC links to the preferred choice in China, where the load centre usually is far from energy centre. Nowadays,±800kV HVDC system is building in South China.The aging and partial discharge (PD) of insulation is inevitable at long term HVDC. Although PD at DC voltage can not be the predominant degradation factor. However, PD can spotlight insulation flaws and prevent unexpected failures. The harm degree is different for different flaws in insulation; therefore the identification of PD type is important. The other major problem is to distinguish noise from PD signals. These two problems are researched thoroughly in this dissertation:(1) The setting up of measurement system of PD at DC voltage. Considerating the characteristic of PD at DC voltage, the PD measurement system is set up based on that of AC device. The artificial models of three typical discharges (corona, internal discharge and surface discharge) are built.(2) Lots of HVDC PD tests are done in laboratory. To identify PD source, the database of three typical PDs at DC voltage is necessary. The waveforms of negative and positive PDs at different voltage, including corona, internal discharge and surface discharge, are recorded. To avoid the effect of noise, the tests are detected in sheilding room.(3) New PD patterns for PD identification at DC voltage. Under DC voltage the concept of phase-resolved PD pattern loses meaning and conventional techniques for PD source identification are no longer applicable. The PD amplitude, PD number and time difference between subsequent PD events are selected as basical parameters at DC volatge. By the statistical correlation analysis between above basical paramters, the PD amplitude VS. its number and time difference between subsequent PD events VS. its number are selected as the partterns to identify PD type at DC voltage.(4) Extraction of PD characteristic parameters at DC volatge. After the comparison and statistics analysis of large number of PD tests at DC volatge, the statistical characteristic parameters, including skewness (Sk), kurtosis (Ku), and the shape parameterβof duble parameters Weibull distribution of the PD amplitude VS. its number, are selected as the PD characteristic parameters at DC voltage.(5) Fuzzy identification of PD type at DC voltage. Based on the PD characteristic parameters of Sk, Ku andβ, the strategy of identification of the PD type is proposed here: Firstly, identify corona from other discharges byβof PD amplitude VS. its number using fuzzy technic and fuzzy logical;Secondly, fuzzy identify internal discharge from surface ones by the other statistical characteristic parameters.After the fuzzy PD identification of tests at DC voltage, results show harmony between the accuracy and the compication of PD identification.(6 ) Large number of PD tests at DC volatge in noisy surrounding. The three typical PD at DC voltage are detected in noisy laboratory, the waveforms of PD and noise are recorded together.(7) the strategy of noise rejection of PD at DC voltage. Considerring the sign consistency of PD signals and applied voltage, the strategy of noise rejection of PD at DC voltage is proposed based on correlation analysis. After noise rejection, result shows the validity of the proposed strategy for the noise with positive and negative ones.Based on large number of PD tests at DC voltage, the PD amplitude VS. its number and time difference between subsequent PD events VS. its number are selected as PD pattern to identify its type. Sk, Ku andβof the PD pattern are extracted as the PD characteristic parameters. Based on the distribution of characteristic parameters of different PD, the fuzzy indentification of PD sources is proposed in this paper, which considers both the accuracy and the compication of PD identification. Furthermore, the strategy of noise rejection is proposed in this paper, the result after denoise shows its validity.
|
PDF Full Text Request