Software for Partial Discharge and Localization

Transcription

1 48 PIERS Proceedings, Taipei, March 25 28, 2013 Software for Partial Discharge and Localization M. Cap, P. Drexler, P. Fiala, and R. Myska Department of Theoretical and Experimental Electrical Engineering Brno University of Technology, Kolejni 2906/4, Brno , Czech Republic Abstract This article deals with possibilities of localization of the partial discharges (PD) in oil power transformers. Localization can be performed on the basis of measured UHF waveforms analysis during activity of the partial discharges. The time-shifts of the waveforms related to transient process occurrence in the signals are the main input parameters for localization methods. In order to estimate the position of the signal source in the 3D space a minimum of four antennas has to be used, since the time of the PD is unknown. Designed application uses a numerical method for detection of the discharge spatial location. Diagnosis of the partial discharge is split into four Gross. At first, time graphs of detected signal are displayed for its visual verification. Each group of displayed signals include time stamp of the actual trigger. This time stamp is used as information for visualization of the position of the signal group in time range of the power voltage period. Third part is focused on the spatial detection of measured discharges in volume of the transformer. At least, statistical results are shown. 1. INTRODUCTION Partial discharges (PD) in the high voltage transformer could cause risk of the transformer damage. Detection of the discharge presence is possible by several methods based on sound or electromagnetic signal measurement. This work is focused on measurement and evaluation of the UHF (Ultra High Frequencies) electromagnetic signal. Partial discharge is detected as a signal in range from hundreds of megahertz to units of gigahertz. Propagation delay of the signal from the place of the PD to the each detector can be used for spatial localization of the PD. Goal of this work to design software for detection, analysis and localization of the partial discharges. 2. MEASUREMENT SYSTEM AND LOCALIZATION METHOD 2.1. Diagnostic System PD signal is detected by the special measuring system. Whole system contains 4 specially designed sensing heads (Fig. 1), central unit and software for PD analysis and localization. Sensing heads are mounted in to the front wall of the transformer. Heads are connected by triaxial cables. This type was chosen for higher EMC robustness. These cables are simultaneously used for RF signal transmission and DC powering of the sensing heads. Signal preprocessing part of each head includes conical antenna, controllable attenuators, amplifiers and high pass filter. Data are acquired by Agilent data acquisition system which uses a four-channel, 10 bit high speed cpci digitizer. Acquired signal is preprocessed by sensing head and digitalized by each channel with preset parameters, see Table 1. Because of strong electromagnetic interference, diagnostic system is mounted in a shielded box. This two stage box contains all required powering [6], data acquisition and communication parts, Fig. 3. Figure 1: Sensing head. Figure 2: Head mounted on the transformer.

2 Progress In Electromagnetics Research Symposium Proceedings, Taipei, March 25 28, Table 1: Data acquisition parameters. Number samples 2000 Time to be sampled 1000 ns Sample interval 0.5 ns Delay time 50 ns Channel full scale 0.5 V Offset 0 V Trigger level 10% of full scale Trigger source All input channels 2.2. Localization Method Method of the partial discharge location is based on the signal time differences of arrival TDOA. Signal arrival time is determined from energy accumulation curve EAC given by equation [1, 2]: w i = t s Z 0 i u 2 k, k = 1... N (1) k=0 Waveform point of the signal arriving time was set to 10% from maximal value of the EAC. Due to given antenna arrangement shown in Fig. 2 the localization results lies on circle which plane is perpendicular to the straight line join of antennas. Therefore, solution can be found only in 2D space as a intersection of the circle with relevant part of tested transformer. The initial equation system defines source position towards three antennas x 2 + y 2 = v 2 t 2 0, (2) x 2 + (y y 2 ) 2 = v 2 (t 0 + t 12 ) 2, (3) x 2 + (y y 3 ) 2 = v 2 (t 0 + t 13 ) 2, (4) Solution of Equations (2) and (3) leads to definition solution for t 0 (, x, y as t 0 = v2 y 2 t 2 13 y 3t 2 ) 12 + y3 y2 2 y 2y3 2 2v 2, (5) (y 3 t 12 y 2 t 13 ) y = y ( 2 2 v2 y2 t 2 13 y 3t 2 ) ( 12 + y3 y2 2 y 2y3) 2 t 12 v2 t 2 12, (6) 2y 2 (y 3 t 12 y 2 t 13 ) 2y 2 x = v 2( ( v 2 y 2 t 2 13 y 3t12) 2 +y3 y2 2 y ) 2 ( 2y3 2 y 2 ( 2v 2 (y 3 t 12 y 2 t 13 ) 2 v2 y2 t 2 13 y ) ( 3t y3 y2 2 y ) 2 2y3) 2 t 12 v2 t (7) 2y 2 (y 3 t 12 y 2 t 13 ) 2y 2 Results of PD localization are calculated as a mean value from three partial solution obtained from different combination of channels. 3. PARTIAL DISCHARGE DIAGNOSTIC AND LOCALIZATION SOFTWARE APPLICATION PD localization software application is designed as a multithread application. Basic function concept of this application is to split measurement and visualization into separate threads in order to achieve maximal possible computer performance for both application parts. Because of very short times which are acquired and theoretically large amount of discharges per one 50 Hz period is necessary to set up the acquisition hardware to the maximal performance Data Structures Type of the data variable has to be chosen with respect to data transfer times and requirement of maximal number of acquisition in one 50 Hz period. Data which need to be transferred after one trigger depend on the type of variable used to store the data. Comparison for int, int16 and double are shown in Table 2. With respect to the maximal performance was the data variable type set to basic integer.

3 50 PIERS Proceedings, Taipei, March 25 28, 2013 Table 2: Comparison of bit rate for different data types. bits per second Number of data packets per second bits of variable 1 50 (One per period) 2500 (50 per period) Figure 3: Assembly of the diagnostic system in shielded box. Figure 4: Analysis of the 50 Hz period time Trigger Detection of the partial discharge is provided on all digitizer channels. Detected signal energy depends on the place of origin of the PD. If the PD will occur on the left or right side of the transformer winding the energy of detected signal will be biggest on the first or fourth channel. Therefore, trigger input is configured as a multichannel trigger so the trigger input is used all four input channels Hz Detection Diagnosis of the partial discharge requires information about time position of the actually acquired data in 20 ms time period. This value is necessary for the PD visualization in phase graph. Cross zero detection algorithm uses special hardware and external trigger input to detect position of the supply voltage. Stability of the 50 Hz frequency is in Czech Republic defined as 50 Hz ±1% in 99,5% the year and 50 Hz ± 6% in 100% in the year. With requirements of the precise visualization of the PD in phase chart is necessary to detect actual frequency [5]. In case of one percent difference detected and real frequency can this cause an error 3.6 in one period or 180 in one second. This was solved by calculation of actual period time. Unfortunately, actual period is not stabile and visualization error suppression require to refresh the period time after a several seconds. Fig. 4 shows fluctuation of the time period in time range of 1000 seconds ( periods). Detected time position of the zero value of the supply voltage and period time are calculated as a mean value from ten measured values with regard to eliminate influence of electromagnetic interference Acquisition Settings and Attenuators Control Basic acqusition setting is shown in Table 1. These parameter are setted up with regard to signal parameters. Possibillity of changing of acqusition parameters is crucial especially in case of first examination of signals in continual aqusition mode. Controable attenuators regulate transmition path output power. Attenuator attenuation depend on controll voltage which is set by usb controlled hardware in range 3 40 db, see Fig Application Modes Diagnostic of partial discharge could be split into the two modes. Firs mode works in loop in order to acquire and visualize data continuously. This mode is to a certain extent similar to the

4 Progress In Electromagnetics Research Symposium Proceedings, Taipei, March 25 28, Figure 5: Setting. Figure 6: Main window with phase chart. Figure 7: Time graphs and scheme of transformed for visualization of PD localization algorithm results. basic function of the traditional oscilloscope. Second mode could be described as acquire and wait. In this case program will acquire 300 of data packets and stops acquisition. Both modes use two types of acquired signal visualization, phase chart and time graphs of all four channels which shows position of the acquired data to the zero supply voltage point. Chart uses two values, first is angle calculated as the position of data to the time of the supply voltage zero time, Fig. 6. Actual angle need to be additionally corrected with power and transformer angle difference. Second parameter is mean value calculated from maximal voltage values in all channels [3]. Measurement of the partial discharges starts with examination of the signals in continual mode. Continual mode works similarly to the traditional oscilloscope, Fig. 7. User defines acquisition parameters to achieve the best visualization of various signals in transformer. Optimal attenuators and trigger settings have a crucial influence on selection of required signal. Second mode saves 300 data sets to memory and stops acquisition. This mode is designed for offline review of acquired data. Data could be played continually in loop or selected manually and examined one by one Partial Discharge Location Partial discharge location uses TDOA method. Localization is because of large amount of processed data in online mode available only in memory mode. Use of four antennas allows locating position of partial discharge in three dimensions. Unfortunately, transformers in nuclear power plant Dukovany has all antenna plug-in points placed on the front side in line (see Fig. 2). Hence, localization is possible only in two dimensions. 4. CONCLUSION Designed application uses TDOA method for partial discharge localization. Special requirements lead to online and offline function modes. Basics of the partial discharge analysis are shape of the wave and its time position examination. For this purpose, first two tabs were designed to show

5 52 PIERS Proceedings, Taipei, March 25 28, 2013 this information clearly. PD location itself lies on the top of the application usage. Localization method allows locating of the PD origin in 3D space. Unfortunately, 3D location requires nonlinear placing of the sensing heads on the transformer wall. Therefore, application is not at this time able to locate PD origin in 3D space. REFERENCES 1. Fiala, P., T. Jirku, P. Drexler, and P. Dohnal, Detection of partial discharge inside of HV transformer, modeling, sensors and measurement, PIERS Proceedings, , Cambridge, USA, July 5 8, Myška, R. and P. Drexler, Simulation and verification of methods for partial discharge source localization, PIERS Proceedings, , Kuala Lumpur, Malaysia, March 27 30, Polivka, J., P. Fiala, and J. Machac, Microwave noise field behaves like white light, Progress In Electromagnetics Research, Vol. 111, , Fiala, P., Pulse-powered virtual cathode oscillator, IEEE Trans. Dielectr. Electr. Insul., Vol. 18, No. 4, , Szabó, Z., J. Sedlácek, and M. Hadinec, Optimization method of EMI power filters and its measurement, PIERS Proceedings, , Hangzhou, China, March 24 28, Fiala, P. and P. Drexler, Power supply sources based on resonant energy harvesting, Microsystem Technologies-Micro-and Nanosystems-Information Storage and Processing Systems, Vol. 18, Nos. 7 8, , Szabó, Z. and P. Fiala, Characterisation and testing shielding fabrics, PIERS Proceedings, , Moscow, Russia, August 18 21, Marcon, P., P. Fiala, M. Steinbauer, and M. Cap, Special high voltage function generator, PIERS Proceedings, , Suzhou, China, September 12 16, 2011.