Lightning arrester leakage current multi-sensor collaborative monitoring method and system suitable for unmanned aerial vehicle platform

Abstract

The application discloses a lightning arrester leakage current multi-sensor cooperative monitoring method and system suitable for a UAV platform. Through the UAV carrying multiple sensors, multi-physical field data is synchronously collected and time-space alignment is performed. The Fourier transform is performed on the signal to extract the fundamental wave characteristic phase difference and the third and fifth harmonic content, and a nonlinear correction model based on the four-dimensional mapping model of conductivity-electric field-temperature-frequency is constructed. The temperature drift and edge distortion error are removed by using infrared inversion junction temperature and electric field inversion potential distribution, and the equivalent characteristic phase is output. Combined with the monitoring work point moving track of the digital twin, false alarms are excluded based on the confidence weighted and cross-validation mechanism, and after determining the thermal breakdown risk, the high-speed recording mode is switched to store the fault waveform. The application realizes early and accurate warning of the lightning arrester from the linear region to the pre-breakdown region, and improves the accuracy and reliability of the UAV monitoring in complex environments.

Description

Technical Field

[0001] This invention relates to the field of online monitoring and fault diagnosis technology for power equipment, specifically to a multi-sensor collaborative monitoring method and system for surge arrester leakage current suitable for unmanned aerial vehicle (UAV) platforms. Background Technology

[0002] Zinc oxide surge arresters (MOAs) are core devices for overvoltage protection in power systems, and their operational status directly affects the safety and stability of the power grid. With the widespread adoption of drone technology, non-contact inspections using drones equipped with sensors have become an important means of monitoring the condition of surge arresters. However, existing drone monitoring solutions still face the following technical bottlenecks, limiting their diagnostic accuracy and early warning capabilities:

[0003] First, single-sensor monitoring has blind spots. Existing technologies mostly rely on infrared thermal imaging to determine surface temperature anomalies or leakage current testing to determine the degree of aging. However, infrared thermometry is easily affected by ambient wind speed and light, and only shows characteristics when the surge arrester is severely aged and overheating; while single current harmonic analysis is difficult to distinguish between internal valve plate deterioration and interference caused by external environmental factors (such as dirt and temperature changes), resulting in a high false alarm rate.

[0004] Secondly, error correction for non-contact measurements is difficult. When a drone hovers, its attitude fluctuates due to airflow, and the edge electric field distortion caused by the equalizing ring at the end of the surge arrester leads to significant deviations in the readings of non-contact electric field sensors and electromagnetic coupling sensors. Current technologies lack effective multi-physics collaborative correction mechanisms and cannot eliminate errors caused by temperature drift and spatial electric field distortion, resulting in inaccurate acquisition of the leakage current characteristic phase.

[0005] Secondly, there is a lack of intuitive early failure warning mechanisms. Traditional methods are usually based on threshold alarms, which are difficult to reflect the dynamic process of the surge arrester evolving from the linear region to the nonlinear breakdown region. In particular, for minute characteristic changes in the pre-breakdown state (such as third harmonic surges and characteristic phase lags), there is a lack of in-depth judgment logic based on physical models, and alarms are often only issued in the late stage of the fault.

[0006] Finally, data processing efficiency is low. The bandwidth and computing power of drone platforms are limited; performing high-speed waveform recording for all operating conditions would create enormous storage and transmission pressure. Existing storage strategies are mostly coarse-grained, lacking the ability to predict thermal collapse risks and failing to accurately capture complete transient waveforms 500ms before and after a failure. Furthermore, the data often exists in isolation, not effectively linked to the device's digital model, hindering in-depth fault analysis by experts.

[0007] Therefore, there is an urgent need for a collaborative monitoring method and system that can integrate multi-sensor data, has anti-interference correction capabilities, and can provide early warning based on digital twins. Summary of the Invention

[0008] This invention aims to solve the following problems in existing UAV surge arrester monitoring technology: large blind zone of single sensor monitoring, high false alarm rate due to severe interference from temperature drift and edge electric field distortion in non-contact measurement, lack of dynamic early warning capability for the evolution of surge arrester from linear region to pre-breakdown region, and limited bandwidth of high-speed data processing and storage.

[0009] In a first aspect, embodiments of this application provide a multi-sensor collaborative monitoring method for surge arrester leakage current applicable to unmanned aerial vehicle (UAV) platforms, the method comprising:

[0010] Controlling a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor, and a high-frequency electromagnetic coupling sensor, simultaneously acquiring infrared thermal images, spatial electric field data, and time-synchronized high-frequency voltage and current signals, and performing spatiotemporal alignment of multi-source data;

[0011] Fourier transform is performed on high-frequency voltage and current signals to extract the fundamental characteristic phase difference to determine the proportion of resistive current, and the third and fifth harmonic contents are extracted to determine the regional position of the operating point on the volt-ampere characteristic curve.

[0012] A nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency is constructed. The characteristic phase difference is corrected by using junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field, eliminating temperature drift and edge electric field distortion errors, and outputting an equivalent characteristic phase.

[0013] A digital twin of the surge arrester is constructed, and the harmonic content and equivalent characteristic phase are mapped into the twin to monitor the movement of the operating point. When the equivalent characteristic phase is lower than the critical angle and the harmonic content exceeds the threshold, it is determined that there is a risk of thermal collapse.

[0014] Based on the judgment results, the confidence weights of the multi-sensor data are applied; when the confidence of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the validation bias is fed back to the nonlinear correction model.

[0015] Based on the judgment and verification results, two levels of storage are implemented. Under normal operating conditions, the harmonic content and equivalent characteristic phase are stored. After the risk of thermal collapse is judged and verified, the system switches to high-speed waveform recording to store the fault waveform and associates it with the digital twin.

[0016] Secondly, embodiments of this application provide a multi-sensor collaborative monitoring system for surge arrester leakage current suitable for unmanned aerial vehicle platforms, applied to the method described in the first aspect, the system comprising:

[0017] The multi-source data acquisition module is used to control a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor, and a high-frequency electromagnetic coupling sensor to simultaneously acquire infrared thermal images, spatial electric field data, and time-synchronized high-frequency voltage and current signals, and to perform spatiotemporal alignment processing on the multi-source data.

[0018] The signal analysis and feature extraction module is used to perform Fourier transform on the high-frequency voltage and current signals, extract the fundamental characteristic phase difference to determine the resistive current ratio, and extract the third and fifth harmonic contents to determine the regional position of the operating point on the current-voltage characteristic curve.

[0019] The nonlinear correction module is used to construct a nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency. It uses junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field to correct the characteristic phase difference, eliminate temperature drift and edge electric field distortion errors, and output an equivalent characteristic phase.

[0020] The digital twin mapping and judgment module is used to construct a digital twin of the surge arrester. It maps the harmonic content and equivalent characteristic phase into the twin to monitor the movement of the operating point. When the equivalent characteristic phase is lower than the critical angle and the harmonic content exceeds the threshold, it is determined that there is a risk of thermal collapse.

[0021] The confidence assessment and verification module is used to perform confidence weighting on multi-sensor data based on the judgment result. When the confidence of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the verification deviation is fed back to the nonlinear correction module.

[0022] The hierarchical storage and association module is used to execute a two-level storage strategy based on the judgment and verification results. Under normal operating conditions, it stores the harmonic content and equivalent characteristic phase. After judging the risk of thermal collapse and verifying it, it switches to the high-speed waveform recording mode to store the fault waveform and associates the fault waveform with the digital twin.

[0023] Thirdly, embodiments of this application provide an electronic device, including:

[0024] processor;

[0025] Memory used to store processor-executable instructions;

[0026] The processor is configured to implement the method as described in the first aspect when executing the instructions.

[0027] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program that instructs a device to perform the method described in the first aspect.

[0028] Compared with existing technologies, this invention has the following significant advantages: By using a four-dimensional mapping model to perform dual correction of the characteristic phase using both temperature and electric field, it effectively solves the environmental interference problem in non-contact measurement by UAVs, significantly reducing the false alarm rate. Based on real-time monitoring of the operating point's migration trajectory on the volt-ampere characteristic curve using a digital twin, combined with the joint criteria of characteristic phase and harmonic content, it can issue early warnings before severe thermal collapse of the surge arrester. The introduction of confidence-weighted and cross-validation mechanisms enables self-calibration of sensor data and adaptive feedback of the correction model, improving the system's robustness in complex environments. The adoption of an adaptive two-level storage strategy based on risk levels ensures complete recording of critical fault data while greatly saving storage and bandwidth resources of the UAV. Attached Figure Description

[0029] Figure 1 This is a schematic flowchart of a multi-sensor collaborative monitoring method for surge arrester leakage current applicable to unmanned aerial vehicle platforms, provided as an embodiment of this application.

[0030] Figure 2 This is a diagram illustrating the architecture of a multi-sensor collaborative monitoring system for surge arrester leakage current suitable for unmanned aerial vehicle platforms, provided as an embodiment of this application.

[0031] Figure 3 A schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0033] It should be noted that in the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0034] Based on the embodiments described in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] Example 1

[0036] Figure 1 This is a schematic flowchart illustrating a multi-sensor collaborative monitoring method for surge arrester leakage current applicable to unmanned aerial vehicle (UAV) platforms, provided as an embodiment of this application. Figure 1As shown, a multi-sensor collaborative monitoring method for surge arrester leakage current suitable for UAV platforms includes:

[0037] S1. Control a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor, and a high-frequency electromagnetic coupling sensor to simultaneously acquire infrared thermal images, spatial electric field data, and time-synchronized high-frequency voltage and current signals, and perform spatiotemporal alignment of the multi-source data. This step aims to establish a unified perception benchmark. By integrating multiple heterogeneous sensors through the drone platform, the simultaneous capture of the external environmental thermal field, spatial electric field distribution, and internal leakage current signals of the surge arrester is achieved. Through timestamp calibration and spatial coordinate transformation, the mismatch of multi-source data caused by drone hovering jitter or sensor sampling delay is eliminated, ensuring that the temperature, electric field, and current data used for subsequent analysis have a strict correspondence in both the temporal and spatial dimensions.

[0038] Specifically, in this embodiment, during implementation, the operator controls a multi-rotor UAV to fly to the vicinity of the surge arrester to be tested. The onboard equipment includes an infrared thermal imaging sensor for capturing surface temperature distribution, a non-contact electric field sensor (such as a MEMS field meter) for sensing spatial electromagnetic intensity, and a high-frequency electromagnetic coupling sensor (such as a Rogowski coil) for sensing changes in the magnetic field generated by leakage current. To ensure data validity, the system utilizes a high-precision clock module (such as GPS timing or an onboard atomic clock) to achieve hardware-level synchronous triggering of multiple sensors, ensuring that the thermal image pixels, electric field vectors, and current waveforms acquired at the same time are strictly consistent in timestamps. Furthermore, coordinate transformation is used to unify the measurement data from different sensors into the geographic coordinate system of the surge arrester, achieving spatiotemporal alignment.

[0039] S2. Perform Fourier transform on the high-frequency voltage and current signals to extract the fundamental characteristic phase difference to determine the proportion of resistive current, and extract the third and fifth harmonic contents to determine the location of the operating point on the volt-ampere characteristic curve. This step focuses on in-depth mining of electrical characteristics. The Fourier transform algorithm is used to decompose the time-domain current signal into frequency-domain components. The phase difference between the fundamental voltage and current is calculated to quantify the resistive leakage level of the valve plate. Simultaneously, the distribution characteristics of the third and fifth harmonics are used to assess the deviation of the surge arrester's nonlinear volt-ampere characteristics, thereby initially locating whether the current equipment is operating in the linear region, pre-breakdown region, or breakdown region at the frequency spectrum level.

[0040] Specifically, in this embodiment, performing Fourier transform on high-frequency voltage and current signals includes:

[0041] S2.1 The system performs windowing and Fast Fourier Transform (FFT) processing on the acquired high-frequency electromagnetic signals to separate the fundamental frequency component and the third and fifth harmonic frequency components. In practice, the system first preprocesses the continuously acquired high-frequency electromagnetic signals. Since the actual sampled data length is limited, direct spectral analysis would result in spectral leakage. Therefore, the system uses a windowing function (such as a Hanning window or Blackman window) to truncate the signal, smoothing the start and end points. Subsequently, the system executes the Fast Fourier Transform (FFT) algorithm to convert the current waveform in the time domain into a spectral distribution in the frequency domain. Using this spectrum, the system can accurately identify and separate the fundamental frequency component consistent with the power grid frequency (e.g., 50Hz), as well as the harmonic components with frequencies three times (150Hz) and five times (250Hz) of the fundamental frequency, providing clean frequency components for subsequent phase and amplitude analysis.

[0042] S2.2 Calculate the time difference between the zero-crossing points of the fundamental voltage signal and the fundamental current signal respectively, convert the time difference into the fundamental characteristic phase difference, and quantify the resistive current ratio of the surge arrester valve plate according to the magnitude of the phase difference.

[0043] After extracting the sinusoidal components of the fundamental voltage and current, the system captures the time difference between their zero-crossing points using a high-precision timer. It should be noted that since the UAV platform is not in direct electrical contact with the high-voltage line, this embodiment uses spatial electric field data collected by a non-contact electric field sensor, combined with the geometric parameters of the surge arrester, and reconstructs the voltage waveform to ground across the surge arrester using an electric field inversion algorithm or the principle of capacitive voltage division. This waveform serves as the reference for the fundamental voltage signal. The system calculates the minute time difference between these two zero-crossing points and converts it into the corresponding angle difference, i.e., the fundamental characteristic phase difference. In a zinc oxide surge arrester, the ideal capacitive current phase leads the voltage by 90 degrees, while the resistive current is in phase with the voltage. Therefore, the greater the deviation of this characteristic phase difference from 90 degrees (i.e., the smaller the angle), the higher the proportion of resistive leakage current flowing through the surge arrester varistor, reflecting the degree of aging or deterioration of the insulation material.

[0044] S2.3 Extract the amplitude components of the third and fifth harmonics respectively, and calculate the amplitude ratio or absolute content of the two as an auxiliary criterion for characterizing the nonlinear characteristics of the surge arrester.

[0045] The system further extracts the peak amplitudes of the third and fifth harmonics from the spectrum. Due to the strong nonlinear current-voltage characteristics of the zinc oxide varistor, the content of these higher harmonics is extremely low under normal operating conditions. However, when the varistor begins to age or enters the pre-breakdown region, its nonlinear resistance characteristics become distorted, leading to a significant increase in the amplitude of higher harmonics (especially the third harmonic). By calculating the amplitude ratio of the third to the fifth harmonic, or monitoring the trend of their absolute content changes, it is possible to effectively identify whether nonlinear characteristics have degraded within the surge arrester, thus providing strong auxiliary evidence for determining whether it has entered the pre-breakdown state.

[0046] Specifically, in this embodiment, the step of determining the location of the operating point on the current-voltage characteristic curve includes:

[0047] S2.4. Preset multiple segments of the nonlinear volt-ampere characteristic curve of the surge arrester, including at least the linear region, the pre-breakdown region and the nonlinear breakdown region, and set the corresponding characteristic phase threshold range and harmonic content threshold range for each segment.

[0048] In practical implementation, the system first pre-loads a standard nonlinear volt-ampere characteristic curve of the surge arrester varistor into the digital twin. This curve is divided into three key regions based on physical characteristics: the linear region operating at low voltage (primarily capacitive, with a phase difference close to 90 degrees), the pre-breakdown region where insulation performance begins to degrade (resistive current increases, phase difference decreases, and harmonics surge), and the nonlinear breakdown region where final failure occurs. For each region, the system pre-configures different judgment thresholds; for example, the characteristic phase threshold range for the pre-breakdown region is set between 60 and 75 degrees, and a corresponding lower limit for the third harmonic amplitude is set as a standard for subsequent comparisons.

[0049] S2.5. Input the extracted fundamental characteristic phase difference and the third and fifth harmonic contents into the digital twin. When the characteristic phase difference shows a decreasing trend and the third harmonic content shows a specific increase relative to the fifth harmonic content, it is determined that the operating point is migrating from the linear region to the pre-breakdown region.

[0050] The system injects real-time monitored electrical parameters into the digital twin model. The digital twin not only performs static comparisons but also dynamic trend analysis. When the monitored data exhibits specific combined characteristics—namely, the fundamental characteristic phase difference gradually decreases over time (indicating a continuous increase in the resistive component), while the amplitude of the third harmonic shows a disproportionately large jump relative to the fifth harmonic (a typical sign of varistor nonlinear distortion)—the system determines that the surge arrester's internal operating point is moving away from the safe linear region and beginning to migrate towards the dangerous pre-breakdown region.

[0051] S2.6. Based on the matching degree between the equivalent characteristic phase and the harmonic content, determine the specific coordinate position of the current operating point on the current-voltage characteristic curve. If the coordinate position falls within the pre-breakdown zone and continues for a preset duration, a thermal collapse risk warning is triggered.

[0052] After determining the migration trend, the system further combines the equivalent characteristic phase processed by the nonlinear correction model for precise positioning. The system calculates the coordinates of the intersection point of the current equivalent characteristic phase and the harmonic content on the volt-ampere characteristic curve, and evaluates the matching degree between the two (e.g., whether the phase reduction is accompanied by a corresponding proportion of harmonic increase). If the calculated coordinate point clearly falls within the preset pre-breakdown zone, and the abnormal state is not an instantaneous fluctuation, but persists for more than a set preset duration (e.g., three consecutive power frequency cycles), the system confirms that the surge arrester is in an irreversible damage stage and immediately triggers a high-level thermal collapse risk warning.

[0053] S3. Construct a nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency, and use junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field to correct the characteristic phase difference, eliminate temperature drift and edge electric field distortion errors, and output the equivalent characteristic phase.

[0054] In practical implementation, the system first pre-sets a multidimensional lookup table or neural network model in the ground server, namely a four-dimensional mapping model of conductivity-electric field-temperature-frequency. This model is built based on the physicochemical properties of zinc oxide valve plates and describes the changes in the conductivity properties of the valve plate material under different ambient temperatures, different applied field strengths, and different signal frequencies.

[0055] To achieve accurate physical correction of the characteristic phase, this embodiment discloses the specific mathematical expression of the four-dimensional mapping model of conductivity-electric field-temperature-frequency. The system uses the following correction formula to calculate the equivalent characteristic phase:

[0056] ,

[0057] in, The equivalent characteristic phase (final output value) is the true phase after denoising. The characteristic phase is measured, which is the original phase difference extracted by FFT in step S2. This is the electric field-temperature coupling correction term. It represents the systematic deviation between the standard phase reference value obtained by looking up a table or calculating from the four-dimensional mapping model and the measured value at a specific junction temperature T and potential gradient E. A predefined set of distorted region indices (corresponding to the spatial locations identified in subsequent step S3.4). This is the spatial weight decay factor. It corresponds to the weights generated in step S3.5. For regions with severe distortion, →0; For the normal region →1. This represents the standard characteristic phase reference value obtained by querying the four-dimensional mapping model under the current temperature T and field strength E. In other words, it represents the ideal phase that the healthy valve plate should possess under the current temperature T and field strength E (obtained through the four-dimensional mapping model).

[0058] The system first inputs the real-time collected junction temperature T and potential gradient E into the four-dimensional mapping model, and then interpolates to obtain... Then calculate the deviation. Next, the system iterates through all collection points. If a point is located within the distorted region set... Internally, it is determined by weighting factors. Reduce the contribution of this point to the final result. Finally, sum the terms and output a stable result. .

[0059] In actual monitoring, the system does not rely solely on raw electrical measurements. Instead, it utilizes a model that uses the internal junction temperature obtained from infrared thermal imaging and the internal potential gradient (representing field strength) obtained from electric field sensors as input parameters to query the standard characteristic phase reference value that the valve plate should have under specific physical conditions. By comparing and fitting the real-time measured characteristic phase difference with this reference value, the system can calculate the temperature drift caused by environmental temperature differences and the edge distortion error caused by equalization ring interference, and subtract these from the raw data. Finally, the system outputs an equivalent characteristic phase calibrated by multiphysics, which accurately reflects the degree of degradation of the internal insulation medium of the surge arrester, unaffected by external UAV hovering environments or electromagnetic interference.

[0060] Specifically, in this embodiment, the correction of the characteristic phase difference using junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field inversion includes:

[0061] S3.1. The actual junction temperature inside the surge arrester varistor is inverted using infrared thermal images, and the internal potential gradient is obtained by inverting spatial electric field data collected by a non-contact electric field sensor based on the mapping relationship between the internal potential distribution and the surface electric field of the surge arrester.

[0062] In practical implementation, the system first processes infrared thermal image data. Based on a one-dimensional thermal path model or unsteady-state heat conduction equations, it calculates the real-time junction temperature of the internal valve core through inversion calculations, using the infrared radiation temperature of the arrester surface, the environmental convective heat transfer coefficient, and the thermal resistance of the arrester casing, rather than relying solely on the surface temperature. Simultaneously, the system uses a non-contact electric field sensor array to collect the two-dimensional spatial electric field vector around the arrester. Since there is a definite physical mapping relationship between the potential distribution inside the arrester and the electric field strength in the external space (solving the Laplace equation), the system uses a preset geometric model to invert the collected spatial electric field data into the potential gradient (i.e., internal field strength) inside the arrester, thereby obtaining the actual electric stress data borne by the valve core.

[0063] S3.2 Input the actual junction temperature and potential gradient as input variables into the preset four-dimensional mapping model, and query the standard characteristic phase reference value under the current temperature and field strength.

[0064] The system invokes a pre-defined four-dimensional mapping model of conductivity-electric field-temperature-frequency. This model stores the electrical parameters of zinc oxide materials under different temperature and field strength combinations in a standard laboratory environment. The system inputs the actual junction temperature and internal potential gradient obtained in step S3.1 as index variables into the model to query the standard characteristic phase reference value that a healthy varistor should exhibit under the current specific physical conditions of high temperature + high field strength or normal temperature + low field strength. This value represents the ideal phase state that the surge arrester should have under the current environment, assuming no aging or degradation has occurred.

[0065] S3.3 Calculate the deviation between the real-time extracted feature phase difference and the standard feature phase reference value, and use the deviation to compensate for the real-time feature phase difference to eliminate the temperature drift error caused by changes in ambient temperature, and output the preliminarily corrected feature phase.

[0066] The system compares the raw characteristic phase difference extracted in real time in step S2 with the standard characteristic phase reference value obtained in step S3.2. The numerical difference between the two is the deviation, which is mainly caused by the resistance change (temperature drift) due to changes in ambient temperature. The system uses this deviation to perform reverse compensation on the real-time measurement value (for example, if the phase lag is caused by overheating, a positive correction is performed), thereby eliminating environmental interference factors. After this step, a preliminarily corrected characteristic phase that eliminates the influence of temperature drift is output, making the measurement results more reflective of the true insulation aging state of the surge arrester.

[0067] Specifically, in this embodiment, the elimination of temperature drift and edge electric field distortion errors includes:

[0068] S3.4 Analyze the potential distribution curve obtained by the electric field sensor to identify whether there is an edge electric field distortion region caused by interference from the equalizing ring at the end of the surge arrester or surrounding charged bodies.

[0069] In practical implementation, the system performs morphological analysis on the axial potential distribution curve of the surge arrester obtained from the non-contact electric field sensor. Due to the equipotential ring structure at both ends of the surge arrester and the presence of nearby high-voltage busbars or towers, the electric field lines will be distorted in the end or edge regions of the surge arrester body, causing the potential distribution curve to no longer be smooth, exhibiting abnormal bulges or depressions. The system automatically identifies these edge distortion regions that do not conform to the normal linear potential distribution law by detecting the first or second derivative of the curve.

[0070] S3.5 When edge electric field distortion is identified, the electric field gradient change rate of the distorted region is extracted to generate a weighted attenuation factor corresponding to that spatial location. Once an edge distortion region is identified, the system calculates the electric field gradient change rate (i.e., the degree of electric field change per unit length) within that region. Based on this change rate, the system generates a spatially relevant weighted attenuation factor. For severely distorted regions, the value of this weighted factor approaches zero; while for regions with a uniform electric field in the center of the arrester body, the value of this weighted factor approaches one. This factor acts as a spatial mask, used to mark which regions' data are reliable and which regions' data are contaminated.

[0071] S3.6 In the nonlinear correction model, the weight attenuation factor is used to perform spatial weighting on the initially corrected characteristic phase, reducing the weight of the distorted region data in the final phase calculation, thereby outputting the de-scratched equivalent characteristic phase and completing the filtering of spatial electromagnetic interference.

[0072] When calculating the final equivalent characteristic phase, the system no longer averages all acquisition points. Instead, it introduces the weight attenuation factor generated in step S3.5. The system performs spatial weighting on the initially corrected characteristic phase data, automatically ignoring or significantly reducing the weight of sampled data from distorted regions such as the end equalization ring, while enhancing the weight of data from the healthy region in the middle of the arrester body. In this way, the system successfully filters out noise caused by spatial electromagnetic interference, outputting a clean and accurate de-interference equivalent characteristic phase.

[0073] S4. Construct a digital twin of the surge arrester, mapping harmonic content and equivalent characteristic phase into the twin to monitor the movement of the operating point. When the equivalent characteristic phase is below the critical angle and the harmonic content exceeds the threshold, a risk of thermal breakdown is identified. This step achieves a leap from data monitoring to mechanism diagnosis. By constructing a digital mirror of the surge arrester in virtual space, real-time harmonic data and corrected phase data are injected into the model to dynamically track the migration trajectory of the operating point on the volt-ampere characteristic curve. When the phase angle is detected to be below the critical value and accompanied by a surge in harmonics, it indicates that the surge arrester has entered the pre-breakdown zone, and the system issues a thermal breakdown warning supported by physical mechanisms.

[0074] Specifically, in this embodiment, the construction of the digital twin of the surge arrester maps the harmonic content and equivalent characteristic phase to the twin to monitor the movement of the operating point, including:

[0075] S4.1 Based on the geometric structure and material parameters of the surge arrester, construct a digital three-dimensional model that is consistent with its physical properties.

[0076] In practical implementation, the system utilizes the surge arrester's design drawings (such as CAD models) and manufacturing specifications to construct a high-fidelity digital 3D model on the server. This model not only reproduces the surge arrester's external geometric dimensions (such as height and skirt shape) but also incorporates core material parameters, including the dielectric constant, nonlinear coefficient, and thermal conductivity of the zinc oxide varistor, ensuring that the virtual model maintains a high degree of consistency with the physical device in terms of physical properties, thus laying the foundation for subsequent simulation and mapping.

[0077] S4.2 Establish a real-time data mapping interface to inject the equivalent characteristic phase and harmonic content collected and corrected by the UAV into the digital twin in real time, driving the state update of the twin model.

[0078] The system developed a dedicated real-time data mapping interface to establish a data transmission link between the UAV ground station and the digital twin. Through this interface, the nonlinearly corrected equivalent characteristic phase output from step S3, along with the extracted third and fifth harmonic contents, are packaged and transmitted to the digital twin in real time. These data serve as driving parameters, updating the electrical state variables within the digital twin in real time, ensuring that the virtual model's operating state remains synchronized with the current state of the physical entity.

[0079] S4.3. Pre-set a nonlinear current-voltage characteristic curve spectrum within the digital twin, and map the current equivalent characteristic phase and harmonic content to specific coordinate points on the curve spectrum to form the current operating point of the surge arrester.

[0080] Inside the twin, the system is loaded with a standard nonlinear diagnostic spectrum for this type of surge arrester. This spectrum establishes a spectrum based on the equivalent characteristic phase. A two-dimensional coordinate system is used, with the vertical axis as the ordinate and the normalized harmonic content H as the horizontal axis. The system converts the received real-time data into coordinate points in this coordinate system. This allows us to pinpoint the current operating point of the surge arrester on the waveform, thus reflecting its actual position on the nonlinear volt-ampere characteristic curve. This point is called the operating point, and it visually represents the precise position of the surge arrester on the volt-ampere characteristic curve at the current moment.

[0081] S4.4 Through continuous data injection, the trajectory of the operating point moving from the linear region to the pre-breakdown region along the current-voltage characteristic curve is dynamically displayed in the visualization interface of the digital twin.

[0082] As drones continuously monitor and inject new data, a dynamic evolution process emerges on the digital twin's visualization interface. Users can intuitively see the operating point, representing the arrester's status, smoothly moving along the volt-ampere characteristic curve over time. The system clearly demonstrates how the operating point gradually migrates from the safe linear region (usually the bottom of the curve) to the dangerous pre-breakdown region (the curve's bend) through a continuous trajectory formed by connecting discrete points, thereby achieving visualized management of arrester aging and degradation.

[0083] Specifically, in this embodiment, the determination of the existence of thermal collapse risk includes:

[0084] S4.5. A critical boundary for the pre-breakdown zone is defined in the digital twin. This critical boundary is composed of a characteristic phase critical angle and harmonic content thresholds. In practice, the system administrator calibrates the nonlinear current-voltage characteristic curve of the digital twin to define a critical boundary representing a safety risk threshold. This boundary is not a single indicator but a composite judgment zone: a characteristic phase critical angle (e.g., 60 degrees) is set on the phase axis, and upper limits for the third and fifth harmonic content thresholds are set on the amplitude axis. Only when both conditions are met simultaneously is the judgment boundary of the pre-breakdown zone considered to have been reached.

[0085] To quantify the degree of danger of the operating point intruding into the critical boundary, this embodiment introduces the thermal collapse approximation index (TLA). As an auxiliary basis for judgment:

[0086] ,

[0087] in, This is the thermal collapse proximity index. The higher the value, the closer to thermal collapse (e.g., >0.8 indicates high risk). The critical angle of the characteristic phase (e.g., 60 degrees). This is the equivalent characteristic phase (from the formula above). This is the physical limit of the minimum phase (the theoretical phase when the valve plate is completely broken down). The content of the third and fifth harmonics. This is the threshold for harmonic content.

[0088] Real-time system calculation The first part of the formula calculates the relative distance to the critical phase distance, and the second part calculates the logarithmic growth factor of the harmonic content. When the TCPI exceeds the preset threshold and the duration meets the requirements, it is determined to be at risk of thermal collapse based on the operating point trajectory of the digital twin.

[0089] S4.6. Real-time monitoring of the distance between the operating point and the critical boundary. When the equivalent characteristic phase is lower than the critical characteristic phase angle, and the third and fifth harmonic contents exceed the harmonic content threshold, and the operating point continues to intrude into the critical boundary within a preset time period, it is determined that the surge arrester varistor has entered the irreversible thermal collapse development stage. The system continuously calculates the Euclidean distance or vertical distance between the current operating point and the aforementioned critical boundary. When the monitoring data exhibits the following combined characteristics: the equivalent characteristic phase drops below 60 degrees, the harmonic content exceeds the set threshold, and this deterioration trend is not an instantaneous fluctuation but a continuous movement into the boundary (i.e., a more dangerous area) over multiple consecutive power frequency cycles, the system determines that the surge arrester varistor has lost its self-recovery capability and has officially entered the irreversible thermal collapse development stage.

[0090] S4.7 In response to this determination, the failure area is highlighted in the digital twin, and a thermal collapse risk warning report containing a fault evolution trend graph is generated. Once the thermal collapse risk is confirmed, the system immediately triggers an emergency response. In the 3D visualization interface of the digital twin, the system automatically highlights the specific part of the surge arrester body where the failure has occurred (such as a certain section of the surge arrester or a certain valve group). At the same time, the system automatically generates a detailed warning report, which not only includes the current fault characteristic values, but also includes a trajectory graph of the operating point migration and a fault evolution trend prediction graph, providing maintenance personnel with intuitive decision-making basis.

[0091] S5. Based on the judgment results, the confidence levels of the multi-sensor data are weighted. When the confidence level of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the validation deviation is fed back to the nonlinear correction model. This step establishes the system's intelligent fault-tolerance mechanism. By weighting the confidence levels of different sensors under the current operating conditions, false alarms with low confidence are identified and eliminated. Simultaneously, the deviation data generated during cross-validation is fed back to the correction model, enabling the system to have self-learning capabilities, adaptively adjust the correction coefficients, and continuously optimize the accuracy of subsequent monitoring.

[0092] Specifically, in this embodiment, the confidence-weighted calculation of multi-sensor data and the feedback of verification bias to the nonlinear correction model include:

[0093] S5.1 Pre-set the initial confidence weights for the infrared thermal imaging sensor, electric field sensor, and high-frequency electromagnetic coupling sensor under different environmental conditions. In practice, the system establishes a confidence weight database and presets initial confidence scores for the three types of sensors. For example, in clear, windless weather, the weight of the infrared thermal imaging sensor is set to the highest; while in rain, snow, or strong electromagnetic interference environments, the system reduces the weights of the infrared and electric field sensors and increases the weight of the high-frequency electromagnetic coupling sensor, which is less affected by ambient light, thus adapting to different inspection scenarios.

[0094] S5.2 Based on real-time monitoring results, the system evaluates the signal-to-noise ratio (SNR) and stability of the current data from each sensor, dynamically adjusts the real-time confidence weights of each sensor, and performs weighted fusion calculations on the multi-source data. The system analyzes the quality of the data transmitted from each sensor in real time, calculating the signal-to-noise ratio (SNR) and waveform vibration amplitude. If the system detects that the drone is hovering unstable due to airflow, and the fluctuation amplitude of the electric field and infrared data is large, the system will automatically reduce the real-time confidence weights of these two sensors and increase the weight of high-frequency current data. Subsequently, the system uses a weighted average algorithm to fuse the multi-source data, ensuring that the final comprehensive diagnostic result is not excessively affected by a single poor-quality data point.

[0095] In this embodiment, the dynamic correction of the real-time confidence weights of each sensor is specifically achieved by constructing a multi-source data confidence Bayesian fusion model, the calculation formula of which is as follows:

[0096] ,

[0097] in, The confidence score is used to determine whether cross-validation is triggered. Signal-to-noise ratio (SNR) is a real-time monitoring indicator of the signal quality of each sensor. This is the signal-to-noise ratio sensitivity coefficient, used to adjust the nonlinear effect of signal quality on confidence. These are dynamic weighting factors. They correspond to infrared (IR), electric field (E), and high-frequency electromagnetic (H) sensors, respectively. These weights are initialized in step S5.1 based on environmental conditions (such as rain, snow, and wind) and dynamically adjusted according to the real-time stability in S5.2. This is a single-sensor risk index. Specifically, for infrared sensors, RIR is calculated by normalizing the difference between surface temperature and ambient temperature; for high-frequency electromagnetic sensors, RH is based on the characteristic phase. The degree of deviation from 90 degrees is mapped to a risk probability between 0 and 1 using the Sigmoid function.

[0098] The system calculates the signal-to-noise ratio (SNR) of the three sensors in real time and substitutes it into the Sigmoid function. The data is then normalized to obtain the basic confidence level. Subsequently, the current dynamic weighting factors are used... The independent risk indices R of the three sensors are weighted and summed. If the final fusion confidence Cfusion is lower than a preset threshold (e.g., 0.6), the cross-validation logic in step S5.3 is triggered.

[0099] S5.3 When the real-time confidence level of a warning signal triggered by any single sensor is lower than a preset threshold, cross-validation logic is initiated to verify the authenticity of the warning using the data characteristics of the other two sensors. If the verification is inconsistent, it is determined to be a false alarm and is removed. When a sensor (such as an electric field sensor) detects an anomaly and attempts to issue a warning, but its current real-time confidence level is lower than the system's set safety threshold (e.g., lower than 0.6), the system will not blindly issue an alarm. Instead, the system initiates cross-validation logic, calling the temperature distribution of the infrared thermal image and the phase characteristics of high-frequency harmonics to verify whether the electric field anomaly is valid. If the verification result shows that the infrared temperature is normal and the harmonics have not surged, the system determines that the electric field anomaly is a false alarm caused by drone jitter or external interference and automatically removes it.

[0100] S5.4. The verification deviation data generated during cross-validation is fed back to the nonlinear correction model, adaptively adjusting the correction coefficients for temperature drift or electric field distortion in the model to improve the accuracy of the equivalent characteristic phase in subsequent monitoring. The system records the results of each cross-validation (i.e., the deviation between the true value and the measured value) as training samples. These sample data are fed back to the nonlinear correction model (four-dimensional mapping model) in step S3. The model uses these deviation data and, through machine learning or adaptive filtering algorithms, automatically adjusts the internal correction coefficients for temperature drift or electric field distortion. Over time, the system develops adaptive optimization capabilities, enabling it to more accurately identify and filter out interference in specific environments, thereby continuously improving the accuracy of the equivalent characteristic phase in subsequent monitoring.

[0101] S6. Based on the judgment and verification results, a two-level storage is implemented. Under normal operating conditions, harmonic content and equivalent characteristic phase are stored. After determining the risk of thermal collapse and verifying its confirmation, the system switches to high-speed waveform recording to store the fault waveform and associates it with the digital twin. This step optimizes system resource scheduling and data management. A two-level storage strategy is adopted: simplified storage under normal operating conditions and detailed waveform recording under abnormal operating conditions. In daily monitoring, only characteristic quantities are retained to save bandwidth, while high-speed waveform recording is automatically triggered after a risk is diagnosed, capturing the complete transient process before and after the fault. By binding waveform data with the digital twin, a seamless association between physical world fault characteristics and the virtual model is achieved, facilitating subsequent panoramic fault review by experts.

[0102] Specifically, in this embodiment, the step of performing two-level storage based on the judgment and verification results, and associating it with the digital twin, includes:

[0103] S6.1 Under normal cruise monitoring conditions, only the power frequency cycle is used as the storage unit. The equivalent characteristic phase and harmonic content characteristic values ​​after nonlinear correction are compressed and stored to reduce the data transmission and storage load.

[0104] In practical implementation, the system employs an efficient data simplification strategy. When the surge arrester is operating normally, the system does not store massive amounts of raw waveform data. Instead, it extracts and compresses key characteristic values, such as the equivalent characteristic phase and harmonic content, after nonlinear correction, using power frequency cycles (e.g., 50 points per second) as units. This characteristic compression storage method greatly reduces the data transmission bandwidth pressure between the UAV and the ground station, while also reducing the local storage space required.

[0105] S6.2 When the system determines that there is a risk of thermal collapse and the alarm is confirmed to be true through cross-validation, it automatically triggers the high-speed transient recording mode. At a sampling rate of no less than 500,000 sampling points per second, it continuously stores complete current-induced waveform data (generated by a high-frequency electromagnetic coupling sensor) for 500 ms before and after the fault occurrence, as well as synchronously acquired spatial electric field waveform data (used for post-fault voltage change inversion), ensuring a complete record of the fault transient process. Once the system confirms that the risk of thermal collapse is true (i.e., it has passed the cross-validation in S5), the system immediately switches from power-saving mode to high-speed recording mode. At this time, the system continuously and uninterruptedly records the complete voltage and current waveforms for 500 ms before and after the fault occurrence at a high rate of no less than 500,000 sampling points per second (500Ksps). This high-resolution transient recording can completely capture the subtle changes at the moment of arrester failure, providing detailed raw data for subsequent analysis.

[0106] S6.3. The stored high-speed waveform data and its corresponding metadata tags are mapped and bound to the corresponding surge arrester component model in the digital twin in real time. A panoramic data package containing fault precursors, occurrence and evolution process is constructed in the digital twin for remote expert system to perform fault review and source tracing.

[0107] The system uploads and binds the recorded waveform files, along with their corresponding timestamps, GPS coordinates, environmental parameters, and other metadata tags, to the corresponding 3D model component of the surge arrester in the digital twin in real time. Within the digital twin, this data is no longer isolated files but forms a panoramic data package containing fault precursors, the moment of failure, and the evolution process. Remote experts can directly retrieve the corresponding waveform data by clicking on the fault point on the twin for review and analysis, achieving precise fault tracing.

[0108] Example 2

[0109] like Figure 2 As shown, this application provides an architecture diagram of a multi-sensor collaborative monitoring system for surge arrester leakage current suitable for UAV platforms, applied to the method described in Embodiment 1, including:

[0110] The multi-source data acquisition module 210 is used to control a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor and a high-frequency electromagnetic coupling sensor to simultaneously acquire infrared thermal images, spatial electric field data and time-synchronized high-frequency voltage and current signals, and to perform spatiotemporal alignment processing on the multi-source data.

[0111] The signal analysis and feature extraction module 220 is used to perform Fourier transform on the high-frequency voltage and current signals, extract the fundamental characteristic phase difference to determine the resistive current ratio, and extract the third and fifth harmonic contents to determine the regional position of the operating point on the volt-ampere characteristic curve.

[0112] The nonlinear correction module 230 is used to construct a nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency. It uses junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field inversion to correct the characteristic phase difference, eliminate temperature drift and edge electric field distortion errors, and output an equivalent characteristic phase.

[0113] The digital twin mapping and judgment module 240 is used to construct a digital twin of the surge arrester. It maps the harmonic content and equivalent characteristic phase to the twin to monitor the movement of the operating point. When the equivalent characteristic phase is lower than the critical angle and the harmonic content exceeds the threshold, it is determined that there is a risk of thermal collapse.

[0114] The confidence assessment and verification module 250 is used to perform confidence weighting on multi-sensor data based on the judgment result. When the confidence of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the verification deviation is fed back to the nonlinear correction module.

[0115] The hierarchical storage and association module 260 is used to execute a two-level storage strategy based on the judgment and verification results. Under normal operating conditions, it stores the harmonic content and equivalent characteristic phase. After judging the risk of thermal collapse and verifying it, it switches to the high-speed waveform recording mode to store the fault waveform and associates the fault waveform with the digital twin.

[0116] Figure 3 This is an electronic device provided in one embodiment of this application. For example... Figure 3 As shown, the electronic device includes at least the following components: processor 301 and memory 300, communication interface 303, and bus 302.

[0117] In this embodiment of the application, memory 300 is used to store executable instructions of processor 301, which, when configured to execute instructions, implements the method as described in the first aspect.

[0118] In embodiments of this application, a computer-readable storage medium includes instructions that instruct a device to perform the method as described in the first aspect. For example, the instructions instruct the device to perform... Figure 1 The method is shown in the process steps.

[0119] In one embodiment of this application, the program operating in the electronic device may be a program that controls a central processing unit (CPU) or similar device to achieve the functions of the above-described embodiments of the present invention (a program that enables the computer to function). Information processed by these systems is then temporarily stored in random access memory (RAM) during processing, and subsequently stored in various ROMs such as read-only memory (FlashROM) and hard disk drives (HDDs), and read, corrected, and written by the CPU as needed.

[0120] It should be noted that a portion of the electronic device described above can also be implemented using a computer. In this case, the program for implementing the control function can be recorded on a computer-readable recording medium, and the program recorded on the recording medium can be read into the computer and executed.

[0121] It should be noted that the computer mentioned here refers to a computer built into an electronic device, employing hardware including an operating system and peripheral devices. Furthermore, computer-readable recording media refers to removable media such as floppy disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage systems such as hard drives built into the computer.

[0122] Furthermore, computer-readable recording media can include: media that dynamically stores programs for short periods of time, such as communication lines used when transmitting programs via networks like the Internet or communication lines like telephone lines; and media that store programs for fixed periods of time, such as volatile memory inside a computer that serves as a server or client in this case. In addition, the aforementioned program can be a program used to implement the above-mentioned functions, or it can be a program that can implement the above-mentioned functions by combining them with programs already recorded in the computer.

[0123] Furthermore, the electronic device in the above embodiments can also be implemented as an assembly (system group) composed of multiple systems. Each system constituting the system group can possess some or all of the functions or functional blocks of the electronic device in the above embodiments. As a system group, it is sufficient to have all the functions or functional blocks of the electronic device.

[0124] Those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any appropriate changes and variations made to the above embodiments within the essential spirit and scope of this application fall within the scope of protection claimed in this application.

Claims

1. A multi-sensor collaborative monitoring method for surge arrester leakage current suitable for unmanned aerial vehicle (UAV) platforms, characterized in that, include: Controlling a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor, and a high-frequency electromagnetic coupling sensor, simultaneously acquiring infrared thermal images, spatial electric field data, and time-synchronized high-frequency voltage and current signals, and performing spatiotemporal alignment of multi-source data; Fourier transform is performed on high-frequency voltage and current signals to extract the fundamental characteristic phase difference to determine the proportion of resistive current, and the third and fifth harmonic contents are extracted to determine the regional position of the operating point on the volt-ampere characteristic curve. A nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency is constructed. The characteristic phase difference is corrected by using junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field, eliminating temperature drift and edge electric field distortion errors, and outputting an equivalent characteristic phase. A digital twin of the surge arrester is constructed, and the harmonic content and equivalent characteristic phase are mapped into the twin to monitor the movement of the operating point. When the equivalent characteristic phase is lower than the critical angle and the harmonic content exceeds the threshold, it is determined that there is a risk of thermal collapse. Based on the judgment results, the confidence weights of the multi-sensor data are applied; when the confidence of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the validation bias is fed back to the nonlinear correction model. Based on the judgment and verification results, two levels of storage are implemented. Under normal operating conditions, the harmonic content and equivalent characteristic phase are stored. After the risk of thermal collapse is judged and verified, the system switches to high-speed waveform recording to store the fault waveform and associates it with the digital twin.

2. The method according to claim 1, characterized in that, Fourier transform of high-frequency voltage and current signals includes: The acquired high-frequency electromagnetic signals were processed by windowing and fast Fourier transform to separate the fundamental frequency component and the third and fifth harmonic frequency components. The time difference between the zero-crossing points of the fundamental voltage signal and the fundamental current signal are calculated respectively. The time difference is converted into the fundamental characteristic phase difference, and the resistive current ratio of the surge arrester valve is quantified according to the magnitude of the phase difference. The amplitude components of the third and fifth harmonics are extracted separately, and the amplitude ratio or absolute content of the two is calculated as an auxiliary criterion for characterizing the nonlinear characteristics of the surge arrester.

3. The method according to claim 1, characterized in that, The step of determining the location of the operating point on the current-voltage characteristic curve specifically includes: Multiple segments of the nonlinear volt-ampere characteristic curve of the surge arrester are preset, including at least the linear region, the pre-breakdown region and the nonlinear breakdown region, and a corresponding characteristic phase threshold range and harmonic content threshold range are set for each segment. The extracted fundamental characteristic phase difference and the contents of the third and fifth harmonics are input into the digital twin. When the characteristic phase difference shows a decreasing trend and the contents of the third harmonics show a specific increase relative to the contents of the fifth harmonics, it is determined that the operating point is migrating from the linear region to the pre-breakdown region. By combining the matching degree between the equivalent characteristic phase and the harmonic content, the specific coordinate position of the current operating point on the current-voltage characteristic curve is determined.

4. The method according to claim 1, characterized in that, The correction of the characteristic phase difference using junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field inversion includes: The actual junction temperature inside the surge arrester varistor is retrieved using infrared thermal images, and the internal potential gradient is retrieved by inverting spatial electric field data collected by a non-contact electric field sensor based on the mapping relationship between the internal potential distribution and the surface electric field of the surge arrester. The actual junction temperature and potential gradient are input as input variables into a preset four-dimensional mapping model to query the standard characteristic phase reference value under the current temperature and field strength. The deviation between the real-time extracted feature phase difference and the standard feature phase reference value is calculated. This deviation is used to compensate for the real-time feature phase difference, so as to eliminate the temperature drift error caused by changes in ambient temperature, and output the corrected feature phase.

5. The method according to claim 4, characterized in that, The removal of temperature drift and edge electric field distortion errors includes: Analyze the potential distribution curve obtained by the electric field sensor to identify whether there is an edge electric field distortion region caused by interference from the equalizing ring at the end of the surge arrester or surrounding charged bodies. When edge electric field distortion is identified, the rate of change of electric field gradient in the distorted region is extracted to generate a weight attenuation factor corresponding to that spatial location. In the nonlinear correction model, the weight attenuation factor is used to perform spatial weighting on the initially corrected characteristic phase, reducing the weight of the distorted region data in the final phase calculation, thereby outputting the de-scratched equivalent characteristic phase and completing the filtering of spatial electromagnetic interference.

6. The method according to claim 1, characterized in that, The digital twin of the surge arrester is constructed by mapping the harmonic content and equivalent characteristic phase to the twin to monitor the movement of the operating point, including: Based on the geometric structure and material parameters of the surge arrester, a digital three-dimensional model consistent with its physical properties is constructed. A real-time data mapping interface is established to inject the equivalent characteristic phase and harmonic content collected and corrected by the UAV into the digital twin in real time, driving the state update of the twin model. A nonlinear current-voltage characteristic curve spectrum is pre-set in the digital twin, and the current equivalent characteristic phase and harmonic content are mapped to specific coordinate points on the curve spectrum to form the current operating point of the surge arrester. Through continuous data injection, the trajectory of the operating point moving from the linear region to the pre-breakdown region along the volt-ampere characteristic curve is dynamically displayed in the visualization interface of the digital twin.

7. The method according to claim 6, characterized in that, The determination that there is a risk of thermal collapse includes: A critical boundary for the pre-breakdown region is defined in the digital twin, and the critical boundary is composed of the characteristic phase critical angle and the harmonic content threshold. The distance between the operating point and the critical boundary is monitored in real time. When the equivalent characteristic phase is lower than the critical angle of the characteristic phase and the third and fifth harmonic contents exceed the harmonic content threshold, and the operating point continues to invade the inside of the critical boundary within a preset time period, it is determined that the surge arrester varistor has entered the irreversible thermal collapse development stage. In response to the determination, the failure area is highlighted in the digital twin, and a thermal collapse risk warning report containing a fault evolution trend map is generated.

8. The method according to claim 1, characterized in that, The process of weighting multi-sensor data with confidence and feeding back verification bias to the nonlinear correction model includes: Pre-set the initial confidence weights of the infrared thermal imaging sensor, electric field sensor, and high-frequency electromagnetic coupling sensor under different environmental conditions; Based on real-time monitoring results, the signal-to-noise ratio and stability of the current data of each sensor are evaluated, the real-time confidence weight of each sensor is dynamically corrected, and weighted fusion calculation is performed on the multi-source data. When the real-time confidence level of an early warning signal triggered by any single sensor is lower than a preset threshold, cross-validation logic is activated to verify the authenticity of the early warning using the data characteristics of the other two sensors. If the verification is inconsistent, it is determined to be a false alarm and is removed. The validation bias data generated during the cross-validation process is fed back to the nonlinear correction model, and the correction coefficients for temperature drift or electric field distortion in the model are adaptively adjusted to improve the accuracy of the equivalent characteristic phase in subsequent monitoring.

9. The method according to claim 1, characterized in that, The step of performing two-level storage and associating it with the digital twin based on the judgment and verification results includes: Under normal cruise monitoring conditions, only the power frequency cycle is used as the storage unit. The equivalent characteristic phase and harmonic content characteristic values ​​after nonlinear correction are compressed and stored to reduce the data transmission and storage load. When the system determines that there is a risk of thermal collapse and the alarm is confirmed to be true through cross-validation, it automatically triggers the high-speed transient waveform recording mode, continuously storing complete voltage and current waveform data within each preset time period before and after the time of the fault at a sampling rate of no less than 500,000 sampling points per second. The stored high-speed waveform data and its corresponding metadata tags are mapped and bound in real time to the corresponding surge arrester component model in the digital twin. A panoramic data package containing fault precursors, occurrence and evolution process is constructed in the digital twin for remote expert system to perform fault review and source tracing.

10. A multi-sensor collaborative monitoring system for surge arrester leakage current suitable for unmanned aerial vehicle platforms, applied to the method described in any one of claims 1 to 9, characterized in that, include: The multi-source data acquisition module is used to control a drone equipped with an infrared thermal imaging sensor, a non-contact electric field sensor, and a high-frequency electromagnetic coupling sensor to simultaneously acquire infrared thermal images, spatial electric field data, and time-synchronized high-frequency voltage and current signals, and to perform spatiotemporal alignment processing on the multi-source data. The signal analysis and feature extraction module is used to perform Fourier transform on the high-frequency voltage and current signals, extract the fundamental characteristic phase difference to determine the resistive current ratio, and extract the third and fifth harmonic contents to determine the regional position of the operating point on the current-voltage characteristic curve. The nonlinear correction module is used to construct a nonlinear correction model based on a four-dimensional mapping model of conductivity-electric field-temperature-frequency. It uses junction temperature data retrieved from infrared thermography and potential distribution data retrieved from electric field to correct the characteristic phase difference, eliminate temperature drift and edge electric field distortion errors, and output an equivalent characteristic phase. The digital twin mapping and judgment module is used to construct a digital twin of the surge arrester. It maps the harmonic content and equivalent characteristic phase into the twin to monitor the movement of the operating point. When the equivalent characteristic phase is lower than the critical angle and the harmonic content exceeds the threshold, it is determined that there is a risk of thermal collapse. The confidence assessment and verification module is used to perform confidence weighting on multi-sensor data based on the judgment result. When the confidence of a single sensor is insufficient, cross-validation is initiated to eliminate false alarms, and the verification deviation is fed back to the nonlinear correction module. The hierarchical storage and association module is used to execute a two-level storage strategy based on the judgment and verification results. Under normal operating conditions, it stores the harmonic content and equivalent characteristic phase. After judging the risk of thermal collapse and verifying it, it switches to the high-speed waveform recording mode to store the fault waveform and associates the fault waveform with the digital twin.

Images