A list,
In the broad sense, electromagnetic interference not only includes the interference that enters the monitoring system through the current sensor together with the local discharge signal, but also includes the interference that affects the monitoring system itself, such as grounding, shielding, and dry S disturbance caused by improper circuit processing. The latter can be solved by improving system design, rationally selecting circuits and components, and improving system production level. Field electromagnetic interference refers to the former, which is the focus of research. It can be divided into continuous periodic interference, pulse interference and white noise. Periodic interference includes high harmonics, carrier communication and radio communication. Pulse interference is divided into periodic pulse interference and random pulse interference. The periodic pulse interference is mainly caused by the high frequency inrush current generated by the operation of power electronic devices. Random pulse type interference includes corona discharge on high voltage line, partial discharge generated by other electrical equipment, discharge generated by tap switch action, arc discharge generated by motor operation, suspension potential discharge generated by poor contact, etc. White noise includes coil thermal noise, ground grid noise, power line and transformer relay protection signal line coupled into all kinds of noise.
Electromagnetic interference generally enters the measuring point through space direct coupling and line conduction. Different measuring points lead to different interference coupling paths and different influences on measurement. The type and intensity of interference are different with different measuring points.
The principle of selecting the transformer local discharge monitoring point is that the local discharge signal intensity is high, the signal-to-noise ratio is high, and the measurement is simple. There are mainly shell grounding wire and casing end screen grounding wire, some also choose neutral grounding wire, iron core grounding wire and high voltage outlet wire. Sometimes in order to suppress the interference, the reference interference signal is also measured from the power line of the transformer. Since it is inconvenient to install sensors at the neutral point and high voltage outlet, and some transformer cores are internally grounded, shell and grounding wire at the end of casing are often selected as measuring points in the monitoring system.
Two, commonly used inhibition methods
Interference suppression is always considered from interference source, interference path and signal post-processing. To find the interference source and directly eliminate or cut off the corresponding interference path is the most effective and fundamental method to solve the interference. However, detailed analysis of the interference source and interference path is required, and it is generally not allowed to change the original transformer operation mode. Therefore, the measures that can be taken in these two aspects are always very limited. Various signal processing techniques are used to suppress the interference that enters the monitoring system through current sensor coupling. Generally from the following aspects of the district branch signals and interference signals; The phase of power frequency, frequency spectrum, pulse amplitude and amplitude distribution, signal polarity, repetition rate and physical position, and a lot of anti-jamming techniques are put forward.
There are two different ideas in anti-jamming technology: one is based on narrowband (frequency band is generally 10kHz to several 10kHz) signal. It can avoid all kinds of continuous periodic interference and improve the signal-to-noise ratio of the measured signal through the narrow-band current sensor and bandpass filter circuit. This method is only suitable for a specific substation and is not convenient to use. In addition, because pd signal is a wide band pulse, narrowband measurement will cause distortion of signal waveform, which is not conducive to the subsequent digital processing. The other is a signal processing method based on broadband (frequency band is generally 10 to 1000kHz). The detection signal contains most of the local emission energy and a lot of interference, but the SNR is low. The processing steps for these interference are generally as follows: a. Suppress continuous periodic interference; B. Suppression of periodic pulse interference; C. Suppress random pulse interference. With the development of digital technology and the application of pattern recognition in local broadcasting, this processing method can often achieve better results.
According to the above two ideas, detection signals with different SNR can be obtained. In the latter stage, much of the processing is consistent. It can be summarized as frequency domain processing and time domain processing. The frequency-domain method is based on the characteristic of periodic interference discrete in frequency domain. The time domain processing method is based on the characteristic of pulse interference discrete in time domain. There are hardware and software two ways to achieve. The following are introduced respectively.
Third, the suppression of periodic interference
Periodic interference, also known as narrowband interference, occupies a large proportion in all kinds of interference, and the suppression and elimination of interference should be started from this. Because of its high intensity and fixed phase distribution, frequency domain method is used to deal with it. It mainly includes FFT threshold filter, adaptive filter, fixed coefficient filter and ideal multipassband digital filter (IMDF).
Narrowband interference suppression algorithms are more mature. From the application effect, the fixed coefficient filter and the ideal multiband pass filter are ideal. Because IMDF needs to perform FFT and IFFT for many times when processing data, it will cost a lot of calculation time, which is not good for real-time processing. However, according to the optimal monitoring frequency band found by IMDF, finite impulse response (FIR) digital filter with fixed coefficient can be directly processed in the time domain, simplifying the operation and speeding up the processing speed.
All of the above methods can be implemented through software or hardware circuits. Although the hardware filter is not flexible, the narrowband interference can be effectively suppressed after selecting the best frequency band through field test. Although the software method is flexible in adjustment, it has the shortcoming of slow real-time operation speed.
4. Suppression of periodic pulse interference
When the signal removes the periodic interference, the other interference becomes the main contradiction. For the suppression of periodic pulse interference, there are two main processing methods: analog method and digital method. The simulation methods include differential balance method, directional coupling method and reference signal method. The first two methods are also suitable for the suppression of random pulse interference and will be introduced later. The distribution line containing pulse interference but not discharge pulse is selected to measure the pulse interference signal, and the measured interference pulse is used as the control signal. When the signal level exceeds the set threshold and is judged to be interference, the analog-to-digital converter (ADC) is stopped to eliminate the interference pulse from the distribution line.
The principle of digital method is to deal with the different characteristics of interference and phase distribution of local release signal. For example, KONIG. G. And KOPF.U. Proposed a method that firstly recorded the signals of multiple periods, and then averaged the data at the same phase of each period to form a template subtracted from the original signal, thus eliminating the interference signals of periodic type. In this method, the effect of interference removal is better when the signals emitted by authorities are less and the distribution characteristics are relatively clear, while the effect is not good when the signals emitted by authorities are more and stronger.
V.Nagesh and B.I.Gururaj in India proposed a method, which used some achievements of biological signal processing for reference. Its basic principle is that the local emission signal has different shapes from the periodic interference signal. Firstly, data segmentation is carried out to separate the pulse from the waveform signal and form a single pulse sequence. The FFT algorithm is used to calculate the cross-correlation of each pulse in the frequency domain, judge its similarity and group it according to a certain standard. The class signal template is obtained according to these groups of pulses, and then the signal of each class is synthesized in the time domain. It is found that the phase of local emission signal is scattered, but the interference is very concentrated. Using this characteristic to eliminate the periodic pulse interference signal class, the rest of the signal reconstruction, can get the signal after the removal of the periodic pulse interference. It can be seen that it is feasible to use the difference between local discharge and periodic pulse interference in waveform and phase to suppress interference. The method can also be used for localization. It can be identified by analyzing the characteristics of pulse waveforms caused by different discharge points. The disadvantages of this method are: when the repetition rate is high, the two adjacent pulses may be regarded as one, which affects the recognition effect; In addition, when the pulse waveform is large, the speed of operation has an impact, but with the improvement of computer computing ability, this impact will be more and more ignored.
5. Suppression of random pulse type interference
These kinds of distractions are the hardest to weed out. Because the characteristics of interference and local emission signals are similar in the frequency domain, a large number of existing methods are considered in the time domain. The common methods include hardware circuit method, software waveform recognition method and artificial intelligence method.
1. Hardware circuit method
The basic idea is to remove the pulse interference by taking advantage of the characteristic that the external pulse interference in the output signals of two measuring points is in the same direction while the internal discharge pulse is in opposite direction. Specific implementation for hardware circuit, commonly used circuits include differential balance method, pulse polarity identification method and directional coupling method.
In practical application, the effect of the first two is not ideal. This is because for the differential balance method, because of the different propagation path, the two signals that constitute the differential often can not correspond well, so the differential effect is not good. The concept of differential "balance pair" is proposed to improve it, which can eliminate the interference and obtain the amplitude and number of local discharge pulses simultaneously. The limitation of pulse polarity identification is that the simulation delay and polarity discriminator are affected by many external factors, which will cause the electronic gate misoperation and reduce the accuracy of polarity identification.
The directional coupling method was proposed by Borsi H equals of Germany in 1987. Schematic diagram is shown in Figure 1. It uses a specially wound Rogowski coil to couple the local discharge signal at the bottom of the hV casing near the flange, and determines whether it is partial discharge signal or external electromagnetic interference according to the voltage at both ends of the coil. This method connects the middle tap of the Rogowski coil to the measuring terminal at the end of the transformer bushing. At this time, the measuring terminal of the end screen is grounded with a small resistance, which can be regarded as the low-voltage arm of the capacitor divider composed of the ground capacitance of the end screen and the end screen. After grounding with a small resistance, a high-pass filter is formed, and only high-frequency signals can pass through. The Rogowski coil is connected with the measuring terminal at the end of the hV casing to form a directional coupling circuit. Current I in the direction shown,U (1) =Uc U1,U (2) = UC-u 2 = UC-U1. U (1)>U (2); If the current I is reversed, then U (1)
In practical application, people have made some improvements, using two Roche coils to replace the original measurement coils and adopting the method of frequency selection to improve the signal-to-noise ratio of the measurement 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, using pulse signal characteristics for logical judgment can also suppress interference. The premise of pulse recognition is to judge whether the pulse exists, the duration of pulse and the corresponding starting point and end point, so as to accurately determine the discharge phase and acoustic delay.
At present, threshold recognition method is mostly used in pulse recognition. However, the pulse measured in the field is mostly attenuated oscillating wave, so the method is easy to misjudge and cannot determine the pulse duration. A method combining pulse amplitude threshold and waveform characteristics to identify oscillating pulse is proposed, and good results are obtained in practice.
3. Application of pattern recognition
The essence of this method is still to make use of the phase characteristics of signals to distinguish. Although the amplitude of local emission signals varies greatly, their phases are concentrated around 45° and 225° respectively. For example, an experienced expert can easily distinguish the interference from the arc discharge signal because the arc discharge occurs in a different phase from the local discharge, the amplitude variation is small, and the pulse shape is slightly different. The pattern recognition approach, which is a software implementation of expert experience, has been identified in CIGER's report and some corresponding software has appeared. The common methods include fuzzy logic, Kohonen network classification, KLT transformation and artificial neural network based on minimum distance. Generally speaking, the difficulty of pattern recognition method lies in the accumulation of a large amount of prior knowledge and the ability to find out the specific differences between interference and local emission. However, in online measurement, it is difficult to find these differences in strong interference signals. Here are a few of them.
(1) Karhunen-Loeve-Transform method
It is found that when the dimension of the input vector is high, the classification is difficult and the effect is not good. The classification effect can be improved by reducing the dimension. In other words, in order to improve the recognition rate and highlight the characteristics of the signal, the interference or noise information in the signal should be removed first. The principle of KLT transformation is shown in Figure 2. As can be seen from the figure, if the x1-X2 coordinate system is adopted, both X1 and X2 coordinates must be adopted for classification. So if YOU take the orthogonal transformation of this, let's go to w1-W2. W 2 coordinates are required for classification. Thus, interference can be removed by KLT transformation.
(2) Pulse sequence analysis -- Kohonen network
The algorithm is unsupervised (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 take it as the output, and through the self-organization algorithm algorithm can carry out self-adaptive classification, distinguish the partial discharge signal and interference signal, so as to achieve the purpose of interference elimination and suppression.
(3) Pulse sequence analysis
It is introduced that the method is simple, effective and has high recognition rate. It consists of discharge voltage difference or phase difference between partial discharges to constitute analysis sequence, and can distinguish different discharge modes and interference by these characteristics, so as to achieve the purpose of interference suppression. In addition, you can locate faults.
Six, summarized
A large number of research results show that with the improvement of A/D conversion rate and the development of computer technology, the transformer local line monitoring system using broadband (10K-1000khz) sensors combined with high-speed sampling has become the mainstream of development. Signal processing has developed from the traditional spectrum analysis to the local emission waveform analysis in time domain.
Some achievements in the field of digital processing technology and artificial intelligence have been widely used in interference suppression in on-line monitoring and are expected to achieve breakthrough results.
In order to further improve the effectiveness of anti-jamming measures, it is necessary to strengthen the research on the propagation law of interference and pulse, including the propagation in transformer substation and the internal propagation of transformer, from which it is possible to find their differences in waveform, phase and direction.
