Automatic detection and elimination of periodic pulse shaped interferences in partial discharge measurements

Abstract

The interferences present in partial discharge (PD) measurement can he classified as narrow-band and broad-band, the latter being pulsed shaped. The pulse shaped interferences can be periodic or random with respect to power frequency, the former being very common and strong. The paper describes an algorithm for automatic detection and elimination of periodic pulse shaped interferences in PD measurements. The algorithm is developed on lines similar to that used in decomposing an electromyogram (EMG) into its individual motor unit potentials and is verified with practical and simulated PD measurements. In addition to interference pulse rejection, the algorithm can be used to process data from a digital PD detector for overcoming integration errors, obtaining pulse to pulse time interval histograms and distinguishing different PD pulses based on their waveshapes.

Application of Chaotic Oscillator to Partial Discharge Signal Detection

“…The simulation experiments of the chaotic system described in (2) is done in this paper by using C++ Builder 6.0. The simulating model is shown in Fig.…”

“…(1) can be rewritten in state equation as following dt dy = (x3 -cx5 -y + ycos (St)) c (2) dt~~d t Equation (2) is the chaotic model for detecting periodic interference. The basic thought utilizing this chaotic system to detect periodic interference is to make chaotic system at the critical state between chaos and periodic movement through adjusting the parameter y, then make the measuring signal as the disturbance of the periodic motive force or parameter c, according to the phase change of chaotic system, whether the detecting periodic signal is contained in the parameter disturbance or not will be judged, the periodic interference will be accurate estimated and suppressed through adjusting the relevant parameter in self-adapt.…”

“…In the experiment, we assumed the parameters c =1.0, 6= 314rad /s, k=0.5 and the calculation step is 0.1/ while using four rank Runge-Kutta algorithm to solve (2). When y = 0.86530, the chaotic system was at the critical state from chaos to periodic movement.…”

“…Inputting the difference into chaotic system and be disturbance of the parameter c shown in (1), as the done above, we can realize the periodic pulse reference estimation accurately through observing the change of the greatest Lyapunov index A and adjusting the phase and amplitude of the reference signal automatically. Before adopting chaotic (2) to process the PD online measured signal which containing periodic pulse interference, of course, this signal have been eliminated already the periodic narrowband interference and white interference, one must separate the signal into two single polarity time sequences according to the positive and negative pulse. After processed by chaotic system respectively and the periodic pulse interference has been suppressed, those two sequences can be composed the PD pulse signal.…”

“…At present, all kinds of interference suppressing algorithm such as the signal filters, wavelet analysis and fuzzy cluster method have certain deficiency to process the PD detecting pulses, some make PD pulse distortion though it can effective restrain interference, which loses part insulating defect characteristic message that PD signals carry, the monitoring system is difficult to realize the PD patternrecognition and assessment of the insulating state. For instance, infinite impulse response (IIR) digital filter is applied to eliminate the periodic narrowband interference [2]. But it is difficult to setup the filter parameters in practice, even cause the algorithm to be unstable sometimes.…”