Electromagnetic Interference and Suppression in On-line Monitoring Signals of Partial Discharge in Transformers

1. Introduction

In addition to the interference that enters the monitoring system through the current sensor together with the partial discharge signal, the broad electromagnetic interference also includes the interference that affects the monitoring system itself, such as grounding, shielding, and interference caused by improper circuit processing. The latter can be solved by improving the system design, reasonably selecting circuits and components, and improving the system manufacturing level. On-site electromagnetic interference refers specifically to the former and is the focus of research. It can be divided into continuous periodic interference, pulse interference and white noise. Periodic interference includes system high-order harmonics, carrier communication and radio communication. Pulse interference is divided into periodic pulse interference and random pulse interference. Periodic pulse interference is mainly caused by high-frequency surge current generated by the action of power electronic devices. Random pulse interference includes corona discharge on high-voltage lines, partial discharge generated by other electrical equipment, discharge generated by tap changer action, arc discharge generated by motor operation, and suspended potential discharge generated by poor contact. White noise includes coil thermal noise, ground network noise, power supply line, and various noises coupled into the transformer relay protection signal line.

Electromagnetic interference generally enters the measurement point through two ways: direct spatial coupling and line conduction. Different measuring points will have different interference coupling paths and different effects on the measurement; different measuring points will have different types and intensities of interference.

The principle of selecting transformer partial discharge monitoring points is that the partial discharge signal strength is large, the signal-to-noise ratio is high, and the measurement is simple. There are mainly shell grounding wires and bushing end screen grounding wires, and some also choose neutral point grounding wires, core grounding wires and high-voltage outlet terminals. Sometimes, in order to suppress interference, the reference interference signal is also measured from the transformer power supply line. Since it is inconvenient to install sensors at the neutral point and high-voltage outlet terminal, and some transformer cores are grounded internally, the monitoring system often chooses the shell and bushing end screen grounding wires as measurement points.

2. Commonly used suppression methods

Interference suppression is always considered from three aspects: interference source, interference path, and signal post-processing. Finding the interference source and directly eliminating or cutting off the corresponding interference path is the most effective and fundamental way to solve interference, but it requires a detailed analysis of the interference source and interference path, and generally does not allow changes to the original transformer operation mode, so the measures that can be taken in these two aspects are always very limited. For various interferences that enter the monitoring system through current sensor coupling, various signal processing technologies are used to suppress them. Generally, the following aspects are used to distinguish partial discharge signals from interference signals: power frequency phase, spectrum, pulse amplitude and amplitude distribution, signal polarity, repetition rate and physical location, etc., and a large number of anti-interference technologies are proposed based on this.

There are two different ideas in anti-interference technology: one is based on narrowband signals (frequency band is generally 10kHz to several 10kHz). It picks up signals through narrowband current sensors of suitable frequency bands and bandpass filter circuits, avoids various continuous periodic interferences, and improves the signal-to-noise ratio of the measured signal. This method is only suitable for a specific substation and is inconvenient to use. In addition, since the partial discharge signal is a wide-band pulse, narrowband measurement will cause distortion of the signal waveform, which is not conducive to subsequent digital processing. The other is a processing method based on broadband (frequency band is generally 10 to 1000kHz) signals. The detection signal contains most of the energy of the partial discharge and a large amount of interference, but the signal-to-noise ratio is low. The processing steps for these interferences are generally: a. Suppress continuous periodic interference; b. Suppress periodic pulse interference; c. Suppress random pulse interference. With the development of digital technology and the application of pattern recognition methods in partial discharge, this processing method can often achieve better results.

Based on the above two ideas, detection signals with different signal-to-noise ratios can be obtained. In the post-processing, many processing methods are consistent. It can be summarized as frequency domain processing and time domain processing methods. The frequency domain method uses the discrete characteristics of periodic interference in the frequency domain to process it; while the time domain processing method is based on the discrete characteristics of pulse interference in the time domain. There are two implementation methods: hardware and software. The following is an introduction.

3. Suppression of periodic interference

Periodic interference is also called narrowband interference. It accounts for a large proportion of all types of interference, and the suppression and elimination of interference should also start from this. Because it has high intensity and fixed phase distribution, most of them are processed by frequency domain methods. It mainly includes FFT threshold filter, adaptive filter, fixed coefficient filter and ideal multi-band digital filter (IMDF). There

are many algorithms for suppressing narrowband interference, and they are also more mature. From the application effect, fixed coefficient filter and ideal multi-band filter are more ideal. Since IMDF needs to perform multiple FFT and IFFT when processing data, it will take a lot of computing time and is not conducive to real-time processing. However, according to the best monitoring frequency band found by IMDF, a finite impulse response (FIR) digital filter with a fixed coefficient can be formed to directly process in the time domain, which simplifies the operation and speeds up the processing speed.

The above methods can all be implemented through software or hardware circuits. Although the hardware filter adjustment is not flexible, after selecting the best frequency band through field tests, narrowband interference can be effectively suppressed. Although the software method is more flexible to adjust, it has the disadvantage of slow real-time operation speed.

4. Suppression of periodic pulse interference

When the signal removes the periodic interference, other interferences become the main contradiction. There are two main processing methods for suppressing periodic pulse interference: analog method and digital method. Analog methods include differential balance method, directional coupling method and reference signal method; the first two methods are also applicable to the suppression of random pulse interference, which will be introduced later. Select a distribution line that only contains pulse interference but not discharge pulses to measure the pulse interference signal, use the measured interference pulse as a control signal, and when the signal level exceeds the set threshold and is determined to be interference, stop the analog-to-digital converter (ADC) to eliminate the interference pulse from the distribution line.

The principle of the digital method is to use the different phase distribution characteristics of interference and partial discharge signals for processing. For example, KONIG.G. and KOPF.U. proposed a method that first records multiple cycles of signals, and then averages the data in the same phase of each cycle to form a template that is subtracted from the original signal to eliminate periodic interference signals. This method is more effective in removing interference when there are fewer partial discharge signals and their distribution characteristics are clear, but it is not effective when there are many and strong partial discharge signals.

V.Nagesh and BIGururaj of India proposed a method that draws on some achievements in biological signal processing. Its basic principle is that the partial discharge signal and the periodic interference signal have different shapes. First, the data is segmented to separate the pulse from the waveform signal to form a single pulse sequence. The FFT algorithm is used to calculate the cross-correlation of each pulse in the frequency domain, and the similarity is judged and grouped according to certain standards. The class signal template is obtained based on these group pulses, and then each class of signal is synthesized in the time domain. Analysis shows that the phase of the partial discharge signal is more dispersed, while the interference is very concentrated. Using this feature, the periodic pulse interference signal class is eliminated, and the remaining signal is reconstructed to obtain the signal after the periodic pulse interference is removed.

It can be seen that it is feasible to suppress interference by using the differences in waveform and phase between partial discharge and periodic pulse interference. This method can also be used for positioning. It identifies by analyzing the characteristics of the pulse waveform caused by different discharge points. The disadvantage of this method is that when the repetition rate is high, two adjacent pulses may be regarded as one, affecting the recognition effect; in addition, when there are many pulse waveforms, the calculation speed is affected. However, with the substantial improvement of microcomputer computing power, this influence will be increasingly ignored.

5. Suppression of random pulse interference

This type of interference is the most difficult to eliminate. Since the characteristics of interference and partial discharge signals in the frequency domain are similar, a large number of existing methods are considered from the time domain. Common methods include hardware circuit method, software waveform recognition method and artificial intelligence method.

1. Hardware circuit method

Its basic idea is to use the characteristics of the output signals of the two measurement points that the external pulse interference is in the same direction, while the internal discharge pulse is in the opposite direction to remove the pulse interference. It is specifically implemented as a hardware circuit. Common circuits include differential balance method, pulse polarity identification method and directional coupling method.

In practical applications, the effects of the first two are not ideal. This is because for the differential balance method, due to different propagation paths, the two signals that make up the differential often cannot correspond well, so the differential effect is not good. The concept of differential "balanced pair" was proposed to improve this, which can eliminate interference and obtain the amplitude and number of partial discharge pulses at the same time. The limitation of pulse polarity identification is that due to the analog delay and the polarity identifier being affected by external factors, it will cause electronic gating malfunction and reduce the accuracy of polarity identification.

The directional coupling method was proposed by Borsi H in Germany in 1987. See Figure 1 for the schematic diagram. It uses a specially wound Rogowski coil to couple the partial discharge signal at the bottom of the high-voltage bushing near the flange, and determines whether it is a partial discharge signal or external electromagnetic interference based on the voltage across the coil. This method connects the middle tap of the Rogowski coil to the transformer bushing end screen measurement terminal. At this time, a small resistor is connected in series with the end screen measurement terminal to ground, which can be regarded as the end screen and the end screen to ground capacitance forming the low-voltage arm of the capacitive voltage divider. After being grounded with a small resistor, a high-pass filter is formed, and only high-frequency signals can pass through. The Rogowski coil is connected to the high-voltage bushing end screen measurement terminal to form a directional coupling circuit.

When the current I is in the direction shown in the figure, U(1)=Uc+U1, U(2)=Uc-U2=Uc-U1. At this time, U(1)>U(2); if the current I is reversed, U(1) is greater than U(2). In practical applications, people have made improvements to this, using two Rogowski coils to replace the original measuring coils and using the frequency selection method to improve the signal-to-noise ratio of the measuring signal. According to the paper, good results have been obtained.

2. Software waveform recognition method

With the development of computer technology and digital signal processing technology, the use of pulse signal characteristics for logical judgment can also suppress interference. Its premise is pulse recognition, that is, judging whether a pulse exists, the pulse duration and the corresponding starting and ending points, so as to more accurately determine the discharge phase and acoustic wave delay.

At present, pulse recognition mostly uses the threshold recognition method. However, the pulses measured on site are mostly attenuated oscillation waves. This method is easy to misjudge and cannot determine the pulse duration. A method for identifying oscillation pulses by combining pulse amplitude threshold and waveform characteristics is proposed, and good results have been achieved in practice.

3.

Application The essence of this method is still to use the phase characteristics of the signal for distinction. Although the amplitude of partial discharge signals varies greatly, their phases are concentrated around 45° and 225° respectively. For example, since the phase of arc discharge is different from that of partial discharge, the amplitude variation is small and the pulse shape is slightly different, based on these characteristics, an experienced expert can easily distinguish the interference of arc discharge signal. Pattern recognition method is the software implementation of expert experience, which has been confirmed in the CIGER report, and some corresponding software has also appeared. Common methods include fuzzy logic method, Kohonen network classification method, KLT transformation method and artificial neural network method based on minimum distance. In general, the difficulty of pattern recognition method lies in the need to accumulate a lot of prior knowledge and be able to find the specific differences between interference and partial discharge. In online measurement, it is difficult to find these differences in strong interference signals. Several methods are introduced below.

(1) Karhunen-Loeve-Transform method

Research has found that when the dimension of the input vector used for pattern recognition is high, classification is more difficult and the effect is not good; after reducing the dimension, the classification effect can be improved. In other words, in order to improve the recognition rate and highlight the characteristics of the signal, it is necessary to first remove the interference or noise information in the signal. The principle of KLT transformation is shown in Figure 2. As can be seen from the figure, if the x1-x2 coordinate system is used, the x1 and x2 coordinates must be used simultaneously for classification; if an orthogonal transformation is performed on this and transferred to the w1-w2 coordinate system. Only the w2 coordinate is required for classification. It can be seen that the interference can be removed through KLT transformation.

(2) Pulse sequence analysis method - Kohonen network

This algorithm is an unsupervised algorithm (as shown in Figure 3). Its principle is to find the node with the shortest Euclidean distance from the input vector to the output layer, and use it as the output. It can also perform adaptive classification through the self-organizing algorithm to distinguish between local discharge signals and interference signals, thereby achieving the purpose of interference elimination and suppression.

(3) Pulse sequence analysis method

According to the introduction, this method is simple, effective and has a high recognition rate: it is composed of the discharge voltage difference or phase difference between partial discharges to form an analysis sequence, and these characteristics are used to distinguish different discharge modes and interferences to achieve the purpose of interference suppression; in addition, fault point location can also be performed.

VI. Summary

A large number of research results show that with the improvement of A/D conversion rate and the development of computer technology, the transformer partial discharge online monitoring system using wide-band (10k-1000kHz) sensors combined with high-speed sampling has become the mainstream of development. Signal processing has developed from traditional spectrum analysis to time domain analysis of partial discharge waveforms.

Some achievements in the field of digital processing technology and artificial intelligence have been widely used in interference suppression in online monitoring, and are expected to achieve breakthrough results.

In order to further improve the effectiveness of anti-interference measures, the study of the propagation laws of interference and pulses should be strengthened, including the study of propagation in substations and inside transformers, which may reveal the differences in their characteristics in terms of waveform, phase and direction.