1. Introduction

Insulator string leakage current has an inseparable relationship with electricity security. When the leakage current of an insulator string is severe, flashover can occur, sometimes leading to a large area of power failure. Current methods for the measurement of leakage current measurement primarily include the effective equivalent salt deposit density (EESDD) method, infrared thermal image method and leakage current method [1–3]. Due to the poor equivalence between artificial pollution and natural pollution, there could be a large deviation in the measurement results of the EESDD method. The infrared thermal image method has strict requirements in terms of meteorological environment, beyond which the measurement result will be inaccurate. The leakage current method can be used to intuitively derive the characteristics of the leakage current, which is widely used because of its simple measurement structure, small error, and insensitivity to the meteorological environment.

The existing electronic-type sensors used to detect insulator string leakage current primarily depend on electronic measurement and wireless public network transmission [4–7]. As electronic-type sensors are commonly installed in complex electromagnetic fields and complex geographical conditions, they are prone to high failure rates, lack of reliability, and unsafe data transmission during long-term operation.

Fiber Bragg grating (FBG)-sensing technology is widely used [8,9] because of its reduced size, high sensitivity, passive component, and immunity to complex electromagnetic interference. The FBG sensor technology has been used in the fields of materials [10,11], chemistry [12], textiles [13] and meteorological monitoring [14–16].

The cross-sensitivity of temperature and strain in FBG sensing technology is a key issue [17]. The dual FBG pair structure was used in many different projects to achieve temperature insensitive measurement [18]. However, there is no report on the use of dual FBG pair structures to measure high-frequency variations in cantilever structures and increased FBG sensitivity.

This paper demonstrates a sensor based on an FBG, cantilever beam, and spiral coil for monitoring insulator string leakage current. Two FBGs are fixed on both sides of the cantilever beam, and the spiral coil is used to change the current measurement of the insulator string into the wavelength difference of the two FBGs, which can be obtained using an integrated demodulator. This method compensates for the temperature effect on the measured results and improves the stability and accuracy of the sensor. The cantilever beam, FBG, and spiral coil are passive components that do not require additional power. In addition, the monitoring data transmitted through the all-dielectric self-supporting (ADSS) optical fiber cable is stable. This system can be applied to safe monitoring at substations and is suitable for use as an insulator string flashover warning.

2. Principle and structure

2.1 Overall structure

The insulator string leakage current sensor in this paper is based on the FBG. A schematic diagram of the insulator string leakage current sensor is shown in Fig. 1. The entire system consists of two parts: the current acquisition part and the current conversion part. The current acquisition part is based on two conductive current collectors and wires; because the resistance of the spiral coil is much smaller than the internal resistance of the insulator, most of the current is introduced into the spiral coil by connecting the coil in parallel with the last insulator of the insulator string, thereby ‘short-circuiting’ the insulator. The current conversion part is based on the spiral coil, magnet, and two FBGs fixed at both sides of the cantilever beam. When most of the leakage current is introduced into the spiral coil, the coil generates a magnetic field. As the leakage current changes, the changes in the magnetic induction cause the repulsion force of the magnet on the cantilever beam to change, which causes the strain on the FBGs to change. The leakage current can be obtained by monitoring the changes in the strain on the FBGs.

 

Fig. 1 Schematic diagram of insulator string leakage current sensor.

Download Full Size | PDF

2.2 Current acquisition part

To collect the current, this paper demonstrates a method to ‘short-circuit’ the last insulator of the insulator string. In real-life engineering applications, the insulation performance of the power pole tower can be guaranteed by a “short-circuiting” an additional insulator. A schematic is shown in Fig. 2.

 

Fig. 2 Schematic of current acquisition part

Download Full Size | PDF

When the leakage current is lower than the warning value of 250 mA [4], the frequency of the leakage current is approximately 50 Hz and the current amplitude does not change drastically. The inductance of the spiral coil can be regarded as having no hindrance to amplitude-stabilized current; therefore, the inductance characteristics can be ignored. Thus, we consider the spiral coil as a resistor, and assume that there are no losses, and that all insulators have the same value of resistance$R_{insulator}$. According to Ohm’s law, the value of the leakage current is

Substituting Eq. (2) in Eq. (3), and then assuming the condition of Eq. (4)

$\begin{array}{l}{I}_{acqisition}\text{=}\frac{U}{(n-1)R_{insulator}\text{+}\frac{R_{spiralcoil}{R}_{insulator}}{R_{spiralcoil}+R_{insulator}}}\frac{R_{insulator}}{R_{spiralcoil}+R_{insulator}}\\ \approx \frac{U}{(n-1)R_{insulator}\text{+}{R}_{spiralcoil}}\approx \frac{U}{(n-1)R_{insulator}}.\end{array}$ (5)

Substituting Eq. (5) in Eq. (1)

$I_{leakage-1}\approx \frac{n-1}{n}{I}_{acqisition}.$ (6)

From Eq. (6), it is clear that the current that the sensor collects is a multiple of the original leakage current. As long as the sensor is installed and the number of insulators is determined. In general, the insulator string consists of six insulators. The original leakage current can be compensated for in the calculation formula to enable the measurement of the leakage current.

2.3 Spiral coil

A spiral coil is a device commonly used to measure current, it usually converts the current into a magnetic field generated by the spiral coil. When current is introduced into the coil, the distribution of the magnetic field on the axis passing through the center of the spiral coil and perpendicular to the plane of the spiral coil is shown in Fig. 3.According to Biot-Savart law, the magnetic induction at a point on the axis is

 

Fig. 3 Spiral coil and its magnetic field distribution on the axis

Download Full Size | PDF

In this paper, we connected four same spiral coils in series. Each spiral coil has resistance of $1.75\Omega$ and an inductance of $0.78mH$.The parameters$N$,$R$, and$x$ are 180, 16.5 mm, and 8 mm, respectively. As the amplitude of cantilever vibration is within 1 mm, the gap between the vibration amplitude and $x$ can be further increased by changing the spiral coil position and properties. We assume that $x$ is constant when the cantilever beam bends.

2.4 Two FBGs fixed on both sides of cantilever beam

2.4.1 Cantilever beam

Based on the principles of the mechanics of materials [19], the relative wavelength shift of single fiber grating and the axial strain are as follows:

In this paper, the material of the cantilever beam is titanium alloy. The parameters $L$, $E$, $h$, and $b$of the cantilever beam are 120 mm, $1.048\text{e+0}11N/m^2$, 2 mm, and 8 mm, respectively.

2.4.2 FBG temperature compensation theory

An FBG can be seen as a band-pass filter during operation; only a specific wavelength of light can be reflected back, while the wavelength of the fiber grating $\lambda$ is related to the grating period $\Lambda$ and the effective refractive index of fiber core $n_{eff}$.

The reflection wavelength is related to temperature and strain [20], and the specific impact formula is as follows,

This paper demonstrates an improved method to convert the physical quantity that needs to be measured into the strain on the FBG. Reducing the temperature effect on the Bragg wavelength is achieved by fixing two FBGs on both sides of the cantilever beam using 353ND glue and special thermal treatment, as shown in Fig. 4.

 

Fig. 4 Structure of cantilever beam: (a) structural sketch, (b) physical map.

Download Full Size | PDF

The strain on FBG-1 and FBG-2 is equal and opposite when the cantilever beam is subjected to an external force. FBG-1 and FBG-2 are placed under the same condition; therefore, the FBGs are affected by the same temperature.

Assume $\text{(}\alpha \text{+}\beta )$ and $p_e$ of FBGs are approximately equal.

In the communication band of 1525-1565 nm, we assume that the central wavelength of FBG-1 and FBG-2 are approximately equal, because there is only a maximum range of 40 nm.

Substituting Eqs. (11)-(14) into Eq. (15),

From Eq. (16), it is clear that the structure not only compensates for temperature, but also increases the amount of strain and increases the resolution of the measurements.

In this paper, the relationship between the wavelength shift difference of two FBGs and the leakage current is

3. Experimental results

3.1 FBG temperature compensation and strain linearity test

3.1.1 FBG temperature compensation test

We conduct a method to compensate for the effect of temperature on the measurement results using a double grating structure. We fixed two FBGs on opposite sides of the cantilever beam and placed the cantilever beam into a thermostat. The process of changing the temperature was as follows: the temperature was first increased from 0 °C to 50 °C, remaining at each 10 °C for 120 minutes, then decreased from 50 °C to 0 °C at the same speed, also remaining at each 10 °C for 120 minutes. Figure 5(a) shows the wavelength shift of the two FBGs during the process, and Fig. 5(b) shows the difference between the wavelength shifts of the two FBGs during the process.

 

Fig. 5 (a) Wavelength shift of two FBGs during the process. (b) Wavelength shift difference of two FBGs during the process

Download Full Size | PDF

During the process of increasing, stabilizing, and decreasing the temperature, we believe that the effect of temperature on the measurement results can be ignored, meaning that the FBGs’ wavelength shift is only related to the strain. In the actual measurement, the data fluctuates within $\pm 1\%$ in the range of 0-40 °C while maintaining the same current input.

3.1.2 FBG strain linearity test

To easily observe the changes in the wavelengths of the two FBGs, we applied an external force on the free end of the cantilever beam. During the test, we added 80 grams in weights to the free end of the cantilever beam four times, and after a certain period, we remove the weights at the same speed. The results are shown in Fig. 6.

 

Fig. 6 (a)Changes in wavelength shift of two FBGs with applied force. (b) Changes in wavelength shift difference of two FBGs with applied force

Download Full Size | PDF

During the process of increasing, stabilizing, and decreasing the force, the changes in the wavelength shift between FBG-1 and FBG-2 are opposite to each other, and the symmetry of the wavelength shift of the two FBGs is clear. Figure 6(b) shows that the cantilever beam has a great linear sensitivity to force.

3.2 Insulator string leakage current sensor test

We conducted a series of tests in the laboratory before the accuracy test; the testing environment is shown in Fig. 7(b). The spectrum of the FBGs are shown in Fig. 7(c).

 

Fig. 7 (a) Physical composition of insulator string leakage current sensor. (b) Test environment. (c) Spectrum of FBG-1 and FBG-2

Download Full Size | PDF

A broadband spectrum is generated by the light source and transmitted through the fiber to the FBGs of the sensor. Because of the characteristics of the fiber grating, the specific wavelength of the light is reflected back. BaySpec company’s fiber Bragg grating interrogation analyzers module applied in integrated demodulator that is used to demodulate the reflected wavelength, and the current value is then displayed on the computer screen after calculation. The physical composition of the insulator string leakage current sensor is shown in Fig. 7(a). In practical applications, the protective shell of the sensor material is rigid, and can be electromagnetically shielded. In addition, the sensor is usually installed at the bottom of the power tower. The magnetic field generated in the sensor is not affected by the magnetic field generated by the transmission line.

First, we used the AC current signal source to generate current signals of 50-Hz 0.08 A and 100-Hz 0.5 A, respectively, to draw current into the sensor, and then demodulated the wavelength difference of two FBGs using an integrated demodulator with a sampling rate of 500-Hz. Figure 8(a) and 8(b) show the wavelength shift difference of two FBGs for any 40 consecutive sampling points under 50-Hz and 100-Hz sine wave input conditions, respectively. As can be seen from the Fig. 8(a) and 8(b), 50-Hz input results in 10 consecutive points as a cycle, and 100-Hz input results in 5 consecutive points as a cycle. These results indicate that the sensor has the ability to detect the frequency of the input signal.

 

Fig. 8 (a) Wavelength difference for an arbitrary set of 40 consecutive sampling points under the 50-Hz input. (b) Wavelength difference for an arbitrary set of 40 consecutive sampling points under the 100-Hz input. (c) Current effective value of the multimeter and the sensor under the 50-Hz input. (d) Wavelength shift difference while periodically stepping on the floor. (e) Amplitude of the wavelength shift difference of FBGs for different frequency inputs

Download Full Size | PDF

Second, we passed 9 sets of sinusoidal signals with frequencies of 50-Hz and various amplitudes, then used a multimeter and the sensor to record the real-time current effective value, as shown in Fig. 8(c). We obtained the current value by measuring the amplitude of the sinusoidal envelope of the wavelength shift difference between the two FBGs. As can be seen from the Fig. 8(c), by fitting the data once, the linear regression correlation coefficient is 0.99964. This means that the sensor can accurately measure the current value.

Third, we found that external vibrations (such as stepping on the floor) caused data anomalies. Figure 8(d) shows the amplitude of the wavelength shift difference of the two FBGs when periodically stepping on the floor. The peaks of the curve in Fig. 8(d) are due to external vibration that affects the current measurement. Because the external vibration frequency is usually very low, we have filtered the low-frequency noise in the algorithm so that we can obtain the data of the periodic vibration of the cantilever resulting from the current input. The algorithm is to filter out low-frequency noise by adding a band-pass filter in the calculation program section.

Finally, we passed various frequencies with an AC current of 5 mA into the sensor. Figure 8(e) shows the amplitude of the wavelength shift difference of FBGs at different frequency inputs. As can be seen from the Fig. 8(e), the response to the different frequency current input systems is inconsistent, which is also one of the main causes of subsequent errors (mentioned later in this paper).

3.3 Insulator string leakage current sensor accuracy test

3.3.1 Calibration test of insulator string leakage current sensor with 50-Hz current input

In November 2017, the accuracy of the insulator string leakage current-monitoring sensor was tested at the current calibration laboratory of the National Institute of Metrology of China. The standard multimeter used is Hewlett Packard Multimeter 3458A that complies with the JJF 1587-2016 Digital Ammeter Calibration Specification. The test results are shown in Table 1.

Table 1. Results of Calibration (50-Hz current)

View Table

The uncertainty is 0.005 A, while the inclusion factor is equal to 2.

3.3.2 Practical application calibration of insulator string leakage current sensor

In December 2017, the accuracy of the insulator string leakage current-monitoring sensor in practical application conditions was tested in the high-voltage switch equipment operation test laboratory of the China Electric Power Research Institute. The standard sensor used in the test is an electronic-type leakage current-monitoring device that complies with GB/T 32191-2015 National Measurement Standard of the People’s Republic of China. The test results are shown in Fig. 9.

 

Fig. 9 Practical application calibration of insulator string leakage current sensor

Download Full Size | PDF

Test value and standard value are provided by current-monitoring sensor and standard sensor respectively. From Fig. 9, we can conclude that the test value and the standard value match well initially, but when the leakage current increases, the similarity decreases, even though the trend is consistent. We analyze the error associated with this phenomenon as follows.

First, our sensor is in parallel with the last insulator, while the standard sensor is in series with the insulator string during the experiment; the current we measured is therefore part of the current the standard sensor measured. When the leakage current is less than 250 mA, the resistance of the insulator can be seen as considerably larger than the resistance of our sensor. We think that the currents the two sensors collect are the same, so in the initial period, the match is good. The effect of the last insulator shunt cannot be ignored when the leakage current is gradually increased, resulting in the fact that the overall measured current is lower than the standard current.

Second, a related study indicates that leakage current primarily contains odd harmonic components, especially high for 50-, 150-, and 250-Hz, and there is no regularity in the ratio of the three components [21,22]. In the accuracy test conducted at the National Institute of Metrology of China, our sensor accurately measured the 50-Hz current. However, because of the limitations of the cantilever structure itself, Fig. 8(e) shows the inconsistency with different frequency inputs, while the ratio of the three components are not regular, which leads to the measurement inaccuracy at 150-Hz and 250-Hz. In the initial period, the leakage current was mostly composed of the 50-Hz component, so the sensor matched well. In the later period, the 150- and 250-Hz components cannot be ignored, resulting in poor current matching, even though the trend is consistent.

Third, the sampling rate of the integrated demodulator we used is 500-Hz, whereas the sampling rate of the standard electronic-type sensor is 5000-Hz; the low data sampling rate leads to insufficient measurement accuracy.

The transition of the leakage current before flashover can be classified into six stages [21,22]. Despite poor data matching in practical applications, it can still be used as a warning for flashovers.

4. Conclusion

An insulator string-leakage current-monitoring sensor based on an FBG is proposed and demonstrated in this paper. The leakage current is obtained by “short-circuiting” the last insulator of the insulator string, and the corresponding current measurements are converted to the measurement of the force applied at the free end of the cantilever beam by using spiral coil and magnet. By fixing two FBGs on both sides of the cantilever beam to convert the measurement of the force into the measurement of the strain on the FBGs and solve the cross-sensitivity problem between temperature and strain. Finally, the relationship between the wavelength shift difference between the two FBGs and the leakage current is obtained.

We have conducted several experiments in laboratories and third-party organizations, the results show that the sensor’s measurement range is 0-1A and the leakage current measurement is not sufficiently accurate. However, the 50-Hz component measurement is accurate. We have conducted a detailed analysis of the causes of inaccurate measurements and will continue to further optimize the leakage current-monitoring sensor.

The merits of the leakage current-monitoring sensor are that it is passive, free from electromagnetic interference, reliable, and can support a multi-point detection by using wavelength division multiplexing. Although the measurement is currently not sufficiently accurate, it can be used as a warning for flashovers and possesses very promising prospects for future applications.