Research on Cable Online Monitoring Technology Based on Traveling Wave

0 Introduction
The 3-35kV power supply network of coal mines mostly adopts the operation mode of ungrounded neutral point or grounding (resonant grounding) through arc suppression coil, which is generally called small current grounding mode. For a long time, there has been a lack of reliable fault identification methods for single-phase grounding faults of cables in small current grounding power grids. The main reason why it is difficult to detect cable grounding faults in small current systems is that the fault current is small and the arc at the grounding point is unstable. Especially in the system grounded through the arc suppression coil, the steady-state current flowing through the cable fault line is very weak, even smaller than the current change felt by the sound line. The unstable arc at the fault point will cause the fault voltage and current signals to be seriously distorted. Since the online detection methods that have been developed mainly use various steady-state signals, the actual use effects are not ideal due to this. Therefore, it is necessary to study the cable online monitoring system that uses transient signals to analyze and judge cable faults, discover and warn in advance, and realize real safety prediction.

1 Basic Principles
The cables used in the 3-35kV power supply network of coal mines are all cross-linked polyethylene (XLPE) cables. For XLPE cables, water tree aging is the main reason for the breakdown of cables during operation. Under the premise that there are defects, micropores and moisture in the cable insulation, due to the electric field distortion at the defects or micropores, water trees will be induced under the operating voltage. The growth of water trees is relatively slow, but with the growth of water trees, the electric field at the tip of the water tree will become more concentrated. The local high electric field strength will eventually cause electric trees to form at the tip of the water tree, causing instantaneous arcing of the operating cable and the decrease of cable insulation. As the water trees continue to grow and accumulate, the cable insulation is eventually damaged.
The instantaneous arcing caused by water trees has the characteristics of recoverability, which we call "recoverable fault". As the number of such faults accumulates, after reaching a certain level, the arc is continuous and irrecoverable, which is manifested as a ground fault in the operating system, and the cable is damaged and needs to be repaired. The duration of instantaneous recoverable faults is generally 3-5ms (within a quarter cycle). For short-duration instantaneous recoverable faults, conventional power grid insulation monitoring devices (including small current ground fault detection devices) generally do not have time to act and give alarm information.
Therefore, we study ultra-high-speed data acquisition technology that can capture and record instantaneous recoverable ground faults, as well as digital identification methods that can effectively determine their characteristics. At the same time, we study the effective accumulation of the number of times such faults eventually cause cable damage. Through simulation and simulation experiments, we establish cable fault warning criteria to provide an effective basis for field operation.
For the instantaneous recoverable ground fault data obtained by the ultra-high-speed data acquisition system, due to its transient process characteristics, it also contains the traveling wave process information of the instantaneous recoverable ground fault. Using this traveling wave information, the fault distance can be further calculated.
The basic principle of the XLPE power cable fault online early warning system is shown in Figure 1. A high-frequency current sensor is used to extract the high-frequency signal of "recoverable fault" from the grounding current at one end of the cable. The high-speed signal acquisition unit captures and records the "instantaneous recoverable grounding fault" information. It can detect and record one or more XLPE power cables under test in real time online. When a "recoverable fault" occurs in the tested cable, an instantaneous electromagnetic transient signal of the grounding conductor such as the metal shielding layer or steel armor will be generated. The fault warning of the tested power cable can be carried out by using specific criteria or manual analysis methods, and the distance to the fault point can be indicated.

During normal operation, the internal hardware logic circuit of the high-speed data acquisition and processing unit automatically performs high-speed sampling and A/D conversion on each channel signal according to the set sampling order and sampling frequency, and automatically writes the A/D conversion results into the cyclic SRAM at high speed.

2 Algorithm Design
2.1 Early
Warning The function of the early warning algorithm is to ensure that the device can reliably warn when a "recoverable fault" occurs in the internal line of the monitored range, and the device can reliably not warn when there is a transient disturbance in the power grid, a fault occurs in the external line of the monitored range, or a switch operation.
The high-frequency acquisition system collects the signals transmitted by the sensor at high speed. Let iset be the set threshold (the minimum amplitude of a recoverable fault). When a certain signal collected meets

When (the waveform amplitude reaches above the threshold, realized by high-speed comparator). Integrate the current sampling value in a specific time period before and after the trigger moment:

Where J0 is the sampling number of the current at the initial moment of each integration period; △J is the number of sampling points contained in each integration period. As shown in Figure 2.

If the numerical integral is greater than or equal to the set value, a "recoverable fault" may have occurred, that is, it satisfies:

Where Aset is the preset threshold, that is, the minimum absolute value of the numerical integral of a given "recoverable fault".
In summary, the conditions for the system to automatically store the waveform recorded in the SRAM can be determined as follows :

When the above conditions are met, the system automatically extracts the waveform of this period and the surrounding periods from the recorded waveform of SRAM and stores it, and performs correlation test on the stored multiple harmonics. The formula used in the test can be expressed as:

a[j+i+32] is the signal function after filtering, b[i] is the selected convolution window function, Cj is the convolution result, n is the number of sampling points of the convolution window, and 32 means that when a recoverable fault occurs, the system will automatically record 32 sampling data before that moment.
The larger the convolution result, the higher the degree of correlation. When the degree of correlation reaches a certain critical value, it is determined that a recoverable fault has occurred. The system automatically records the number once, that is, the system considers that a "recoverable fault" has occurred. Then the waveform is measured.
When the number of recoverable faults recorded by the system within a certain period of time reaches a certain amount, the system begins to issue warnings to the user, and issues yellow, orange, and red warnings in turn according to the frequency of the faults.
2.2 Fault Distance Measurement
For the "instantaneous recoverable ground fault" data obtained by the ultra-high-speed data acquisition system, due to its transient process characteristics, it also contains the traveling wave process information of the "instantaneous recoverable ground fault". Using this traveling wave fault distance measurement technology, we have overcome the problem of measuring the fault distance. While providing fault warning, the distance to the fault point can be clearly indicated through the automatic algorithm of traveling wave distance measurement.
Using the single-ended traveling wave principle, the distance between the measurement point at this end and the fault point is calculated by using the time delay between the first positive traveling wave surge extracted at the measurement point at one end of the line (device installation end) after the line fault and its reflected wave at the fault point. In the system shown in Figure 1, when the initial traveling wave surge of the fault reaches the busbars at both ends of the fault line, reflection and transmission phenomena will occur, as shown in Figure 3.

Assume that the M end is the measuring end, and the propagation direction of the traveling wave from the local bus to the fault point is the positive direction. When the initial traveling wave surge of the fault (taking the current traveling wave as an example) reaches the M end, it forms the first reverse traveling wave surge of this end, denoted as. The reflected waveform of this traveling wave surge on the bus at the M end is the first forward traveling wave surge of this end, denoted as, and it will propagate toward the fault point. When the traveling wave surge reaches the fault point, reflection and transmission will occur, where the reflected wave from the fault point returns to the M end as a reverse traveling wave surge, denoted as, and the forward traveling wave surge that continues to be reflected from the measuring point to the fault point is denoted as, and so on. and respectively represent all transient current traveling waves from the fault direction (reverse traveling wave surge) and all transient current traveling waves propagating toward the fault direction (forward traveling wave surge), which can prove that each current traveling wave surge from the fault direction and its reflected wave surge on the bus always have the same polarity relationship.
When a "recoverable fault" occurs in the cable, the first forward current wave surge emitted by the fault point and its reflected wave surge at the fault point have a similar polarity in waveform and a time delay of 2τ (τ is the propagation time of the traveling wave from the measurement point to the fault point), so a matching filter can be designed to detect the reflected wave surge at the fault point.
The first forward current wave surge emitted by the fault point is t∈[0, △T], where △T is slightly larger than the time width occupied by the wave surge, and the impulse response of the matched filter can be expressed as:

△t is the sampling interval.
When k=int(2τ／△t), the reflected wave from the fault point reaches the local measurement point, and the corresponding matched filter output reaches the maximum value when k=int[(△T+2τ)／△t). Conversely, if the matched filter output reaches the maximum value when k=kmax, the time when the reflected wave from the fault point reaches the local measurement point is t=kmax△t-△T. The distance DL from the fault point to the local measurement point can be expressed as:

In the formula, v is the wave velocity.
In order to improve the accuracy of distance measurement, we use the method of multiple distance measurement and comprehensive analysis to perform fault distance measurement, namely the "dot method".

As shown in Figure 4, the horizontal axis represents the measured fault distance, and the vertical axis represents the fault traveling wave energy. For each fault traveling wave obtained by the system, there is a fault distance measurement result and the energy value of the traveling wave. A point is made on the coordinate axis to represent this traveling wave. Theoretically, if the number of points is close to infinity, the distribution of these points is close to normal distribution. The center of the area with the densest points, that is, the symmetry axis of the normal distribution, can be taken as the final distance measurement result.

3 Conclusion

At present, this cable fault online monitoring project has been implemented in the Jier Mine Power Plant of Shandong Huaju Energy Company. The online monitoring device has realized online fault monitoring of the 10 most important 6kV cables in the power plant, and has monitored more than 600 water tree discharges, and issued an early warning for the cable with the highest discharge frequency, and accurately measured the fault point at 210m.
Through theoretical research and practical application, this cable fault online detection technology has matured, and can fully realize online dynamic monitoring of operating cables, advance prediction, and predict cable fault trend information. It has great social, safety and economic benefits for mine safety production and safe and reliable power supply in the power industry, and has good promotion prospects.