Zinc oxide surge arresters are extensively used to protect electrical systems from transient overvoltage caused by severe lightning strikes and grid switching operations [1]. However, the aging of valve discs of arresters typically leads to increased conductivity. Moreover, in severe cases, internal flashovers and arrester explosions can severely disrupt the normal operation of zinc oxide surge arresters [2–5]. Arrester failure warrants immediate localization and replacement by operators to prevent interference with other electrical equipment in the grid [6]. Therefore, it is crucial to investigate aging monitoring and fault positioning for zinc oxide surge arresters.

Several methods, including residual voltage method [7], power loss measurement [8], leakage current measurement [9], capacitive current compensation [10], and harmonic analysis [11], have been proposed for monitoring the aging and fault conditions of zinc oxide surge arresters. Leakage current measurement is the most widely adopted for arrester condition assessment owing to its convenience and high accuracy [12]. Arrester failure can cause a significant surge in the total leakage current, exhibiting noticeable distortion in the waveform [13, 14]. The total leakage current is the vector sum of the resistive component i _R and the capacitive component i _C, as shown in Figure 1. The resistive leakage current is influenced by the applied voltage, fault characteristics, and ambient temperature. In addition, it exhibits high sensitivity to arrester faults and degradation [15, 16]. Since the variation in the third harmonic of the resistive current is more pronounced than that of the fundamental wave, it can be used for diagnosing arrester fault conditions [17]. However, the third harmonic of the resistive current is also affected by harmonics in the grid voltage, especially the third harmonic of the voltage, which significantly impacts the third harmonic of the resistive current [18, 19]. This interference prevents the precise reflection of arrester fault conditions. Therefore, a new monitoring indicator that is robust against voltage harmonic interference is necessary.

Recently, Wang et al. [20] from the China Electric Power Research Institute investigated the variation trends in resistive current and harmonics under system voltage harmonics, particularly noting the significant effect it had on the third harmonic, and proposed an integrated online monitoring device for arresters. Qi et al. [21] from Wuhan University of Technology proposed a waveform distortion correction method based on measured voltage waveforms to eliminate the effect of high-order voltage harmonics on resistive current and identified the true value of the leakage current. Additionally, Khodsuz and Mirzaie [22] from Brno University of Technology proposed a technique for eliminating the effect of voltage third harmonics on the resistive leakage current based on grid voltage analysis. However, these studies rely on measuring grid voltage harmonics, which not only increases the measurement and computational burden but also is not conducive to the reliability and effectiveness of fault monitoring and positioning for online arresters.

To address these challenges, this study proposes the fifth harmonic of the resistive leakage current as a new indicator for aging monitoring and fault positioning for zinc oxide surge arresters. First, an aging experimental platform was established for zinc oxide surge arresters, with harmonic components added to the applied voltage to simulate grid voltage harmonic interference. Second, the resistive leakage current was extracted using an improved displacement current method, and the harmonic components of the resistive current were determined using the fast Fourier transform (FFT) algorithm. Finally, the variations in characteristic quantities under voltage harmonic interference and aging faults were investigated to evaluate the robustness of the fifth harmonic against voltage harmonics and its effectiveness and reliability for aging monitoring. A monitoring and positioning scheme is also proposed. This methodology is specifically designed for aging monitoring and fault localization in zinc oxide surge arresters, enabling real-time degradation assessment and facilitating cost-effective operation and maintenance through predictive condition-based strategies.

In the experiment, a variable low-voltage power supply of 0–240 V was employed. Two experimental schemes were designed. In Scheme I, a pure sinusoidal voltage was applied. But during experimental validation, the inherent nonlinearities of zinc oxide surge arresters and power transformers introduced harmonic distortions. Post-FFT analysis revealed a measured THD of 1.26%, with harmonic voltage components quantified as follows: third harmonic (V _3rd) at 0.907%, fifth harmonic (V _5th) at 0.454%, seventh harmonic (V _7th) at 0.305%, ninth harmonic (V _9th) at 0.239%, and eleventh harmonic (V _11th) at 0.126%,as shown in Figure 2a. In Scheme II, a voltage with a total harmonic distortion of 5% was applied to simulate the harmonic impact in power grids. The harmonic contents of the applied voltage included 3.6% third harmonic (V _3rd), 1.8% fifth harmonic (V _5th), 1.21% seventh harmonic (V _7th), 0.95% ninth harmonic (V _9th), and 0.5% eleventh harmonic (V _11th), as shown in Figure 2b.

Two zinc oxide surge arresters were selected for the experiment: one in a normal condition and the other in a severely aged and faulty condition. The normal condition refers to pristine, unused arresters post-manufacturing, while the aged condition corresponds to units that have been in continuous operation at Yancheng Substation, Jiangsu Province for a decade. The parameters of the arresters are listed in Table 1. An experimental circuit was constructed using a variable low-voltage power supply, voltage regulator, transformer, capacitive voltage divider, and high-precision current sensor, as shown in Figure 3, The capacitive voltage divider (FLDFR-150 kV) serves as a critical component for measuring the applied voltage. The accompanying display meter can display the magnitude and waveform of this voltage and transmit the waveform to a computer for FFT analysis to validate harmonic injection levels, and the value of the protective resistor is 10 kΩ. The arresters were subjected to voltages ranging from 70 to 120 kV using the variable low-voltage power supply and a step-up transformer, the parameters of the transformer are listed in Table 2. The applied voltage was monitored using a capacitive voltage divider, and its harmonic components were obtained using the FFT algorithm. The total leakage current was extracted using a high-precision current sensor (CM1A H00), and measure the total leakage current using a high-speed oscilloscope (HDO6054).

The low-voltage power supply was adjusted to subject the arresters to voltages ranging from 70 to 120 kV through the step-up transformer. The input voltage and total leakage current were measured.

An improved displacement current method was employed to extract the resistive leakage current, as shown in the Simulink model in Figure 4. A zero-crossing detector was used to determine the phase of the current waveform, and a frequency detector was used to determine the frequency of the total leakage current. A new waveform was obtained by introducing a quarter-cycle delay to the total leakage current waveform, which was then added to the original total leakage current waveform to generate a composite waveform. The peak time of the composite waveform and that of the capacitive current component were determined using a peak time detector. The capacitive current waveform was generated based on the known frequency, peak time, and peak value. The resistive leakage current (I _R) was ultimately extracted by subtracting the capacitive current component (I _C) from the total leakage current.

The FFT was used to determine the third harmonic (I _R3rd) and fifth harmonic (I _R5th) of the resistive leakage current. Experiments were conducted on both normal and aged arresters under the two schemes, and the total leakage current (I _T), resistive leakage current (I _R), and their harmonics (I _R3rd and I _R5th) were measured and recorded. The overall process is shown in Figure 5.

Based on the relationship between the resistive current harmonics (I _R3rd and I _R5th) and the applied voltage, threshold values for identifying the aging fault conditions of zinc oxide surge arresters were established, as shown in Table 9. The data provenance has been explicitly substantiated through statistical derivation from extensive experimental measurements, which systematically validates the threshold robustness.

I _R5th is proposed as a new indicator for fault diagnosis of arresters. The leakage current of the zinc oxide surge arresters is detected using a zero-flux high-precision current sensor. After FFT analysis and analog-to-digital signal processing, the data can be transmitted wirelessly for online monitoring. When the monitored value of I _R5th exceeds the aforementioned threshold, an alert is triggered. Combined with a GPS positioning system, the location of the faulty arresters can be accurately obtained. The principle of the monitoring and positioning method is shown in Figure 13.

This study proposes a novel method for aging monitoring and fault positioning for zinc oxide surge arresters based on the fifth harmonic of the resistive leakage current (I _R5th). Initially, an aging experimental platform was established for zinc oxide surge arresters, and experiments were conducted with a voltage harmonic control group. Subsequently, the leakage current harmonic characteristics were extracted using an improved displacement current method and fast Fourier transform algorithm. The characteristics of I _R5th were analyzed through the percentage change in the feature and harmonic sensitivity analysis. Finally, by integrating high-precision current sensor technology with GPS positioning, a principle method for applying I _R5th to the fault monitoring and positioning for zinc oxide surge arresters was proposed. The following conclusions can be drawn:

• Under voltage harmonic interference, when the percentage weight of V _3rd is nearly twice as that of V _5th, the peak increase in I _R3rd is 3.7 times that of I _R5th, indicating that I _R5th has a stronger resistance to voltage harmonic interference.