Ground wire monitoring method and system

Abstract

The invention provides a ground wire monitoring method and system, and relates to the technical field of ground wire monitoring. The method comprises the following steps: acquiring a voltage signal of an injection point and a current signal of a grounding loop; preprocessing the signal to obtain a preprocessed signal; according to the preprocessed signals, impedance frequency spectrums of the grounding loop under different frequencies are obtained through calculation, and key feature parameters in the impedance frequency spectrums are extracted; environment parameters and operation state data of the power equipment are collected and fused with the key feature parameters, and fused data are obtained; adjusting a judgment threshold value or a reference baseline corresponding to the key feature parameters by using the fused data; and judging the current grounding loop state by using the adjusted judgment threshold value or reference baseline. According to the method, based on impedance spectrum analysis, electromagnetic interference is overcome through multi-stage signal processing, robust feature extraction and multi-dimensional data fusion analysis, and the effects of adapting to a dynamic baseline and distinguishing different electric leakage types are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of ground wire monitoring, and in particular to a ground wire monitoring method and system. Background Art

[0002] In power substations, the safe operation of power equipment is crucial and requires reliable grounding connections. Traditional grounding cables rely on periodic offline testing and cannot be monitored in real time. Smart grounding cables integrate current detection and leakage monitoring functions. They use impedance spectrum analysis to monitor leakage current and assess the insulation condition of equipment. This method injects an excitation signal and measures the response signal, calculating the impedance of the ground loop at different frequencies and analyzing the impedance spectrum to determine whether leakage exists.

[0003] However, the operating environment of substations is complex, with strong power frequency electromagnetic fields, high-order harmonics and transient electromagnetic interference. These interferences will reduce the signal-to-noise ratio of the monitoring signal, distort the impedance spectrum analysis results based on the original measurement data, and significantly reduce the accuracy of leakage judgment.

[0004] In addition, the power equipment connected to the smart grounding wire may be in different operating modes, and the environmental factors of the substation will also affect the impedance characteristics of the grounding grid and the grounding wire, resulting in the normal impedance spectrum baseline of the grounding loop not being fixed, but changing dynamically with the equipment status and environmental conditions.

[0005] At the same time, using impedance spectrum analysis technology for leakage monitoring requires injecting an excitation signal into the ground loop and collecting the response signal. The frequency range and amplitude selection of the excitation signal, as well as parameters such as the bandwidth and sampling rate of the measurement system, directly affect the resolution and accuracy of the spectrum analysis. In strong interference environments, to improve measurement accuracy, it may be necessary to adopt a specific excitation signal waveform, increase the excitation signal energy, or extend the data acquisition time. However, these measures may increase the power consumption of the monitoring equipment, affect the equipment's response speed, or potentially affect the normal operation of the monitored equipment. At the same time, different leakage types and fault locations may exhibit the most significant characteristics within different frequency ranges of the ground loop impedance spectrum, and it is necessary to select an appropriate excitation frequency range to effectively capture this information.

[0006] Therefore, in the specific scenario of power substations, which are subject to strong electromagnetic interference, diverse equipment operation modes, changing environmental factors, and long-term operation, how to use technology based on impedance spectrum analysis to overcome the impact of electromagnetic interference on impedance spectrum measurement, effectively distinguish different types of leakage and equipment status changes, adapt to the dynamic changes of the ground loop impedance spectrum baseline, and maintain the stability and reliability of monitoring in long-term operation, thereby avoiding missed reports and false alarms, is a key issue that needs to be urgently solved in current intelligent ground wire leakage monitoring technology. Summary of the Invention

[0007] The purpose of the present invention is to provide a ground wire monitoring method and system, which, based on impedance spectrum analysis, overcomes electromagnetic interference through multi-stage signal processing, robust feature extraction and multi-dimensional data fusion analysis, and achieves the effect of adapting to dynamic baselines and distinguishing different leakage types.

[0008] In a first aspect, the present invention provides a ground wire monitoring method, which is applied to power equipment in a substation, comprising the following steps: Inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the smart ground wire and the ground grid, and synchronously collect the voltage signal at the injection point and the current signal of the ground loop; Preprocessing the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; Based on the preprocessed voltage and current signals, the impedance spectrum of the ground loop at different frequencies is calculated, and the key characteristic parameters in the impedance spectrum are extracted; Collect environmental parameters and operating status data of power equipment and fuse them with key characteristic parameters to obtain fused data; Using the fused data, adjust the judgment threshold or reference baseline corresponding to the key feature parameter; the judgment threshold is the numerical limit used to distinguish normal from abnormal states; the reference baseline is the normal value or range of the corresponding key feature parameter; The current ground loop status is determined using the adjusted judgment threshold or reference baseline.

[0009] The ground wire monitoring method provided by the present invention can improve the accuracy and reliability of intelligent ground wire leakage monitoring based on impedance spectrum analysis in an environment such as a substation with strong electromagnetic interference and where the ground loop impedance spectrum baseline changes dynamically with equipment status and environmental factors, thereby effectively distinguishing between normal changes and different leakage states.

[0010] In a second aspect, the present invention provides a ground wire monitoring system, which is applied to power equipment in a substation, comprising: An acquisition module is used to inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the intelligent ground wire and the ground grid, and synchronously acquire the voltage signal at the injection point and the current signal of the ground loop; A preprocessing module is used to preprocess the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; An extraction module is used to calculate the impedance spectrum of the ground loop at different frequencies based on the preprocessed voltage signal and current signal, and extract key characteristic parameters from the impedance spectrum; The fusion module is used to collect environmental parameters and operating status data of power equipment and fuse them with key characteristic parameters to obtain fused data; An adjustment module is used to adjust the judgment threshold or reference baseline corresponding to the key feature parameter using the fused data; the judgment threshold is the numerical limit used to distinguish between normal and abnormal states; the reference baseline is the normal value or range of the corresponding key feature parameter; The judgment module is used to judge the current ground loop state by using the adjusted judgment threshold or reference baseline.

[0011] From the above, it can be seen that the grounding wire monitoring method provided by the present invention adopts time-frequency joint analysis and other methods to perform signal preprocessing for the electromagnetic interference unique to substations, effectively copes with the complex electromagnetic interference of substations, improves the quality of original measurement data, and extracts key features that are sensitive to leakage and relatively insensitive to interference and environmental changes from the processed impedance spectrum. Combined with multi-dimensional data fusion and dynamic diagnosis, the system can adapt to the dynamic changes of the grounding loop, avoid false alarms, and significantly improve the accuracy and reliability of leakage monitoring of intelligent grounding wires in the complex environment of substations.

[0012] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A flow chart of a ground wire monitoring method provided by an embodiment of the present invention.

[0014] Figure 2 A schematic structural diagram of a ground wire monitoring system provided in an embodiment of the present invention.

[0015] Description of labels: 100, acquisition module; 200, preprocessing module; 300, extraction module; 400, fusion module; 500, adjustment module; 600, judgment module; 700, prediction module. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0017] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0018] In power substations, the safe operation of large, high-voltage electrical equipment, such as power transformers and high-voltage switchgear, is crucial. To ensure electrical safety during normal operation and in the event of a fault, these devices require a reliable connection to the grounding grid, providing a low-impedance discharge path. Traditional grounding wires only provide a physical electrical connection, and their condition monitoring typically relies on periodic offline testing, making real-time monitoring impossible.

[0019] With technological advancements and the increasing demand for equipment status monitoring, smart ground cables with electrical detection capabilities have emerged. While providing basic grounding pathways, these smart ground cables also incorporate an integrated electrical detection module to verify that the equipment is de-energized before shutdown or maintenance, improving operational safety. Furthermore, to further enhance equipment reliability and predictive maintenance capabilities, smart ground cables have been equipped with leakage detection capabilities. This function monitors abnormal currents flowing through the ground cable into the grounding grid in real time, known as leakage currents. By analyzing the characteristics of this leakage current, the insulation condition of the equipment and the presence of ground faults can be assessed.

[0020] Currently, one technical solution for intelligent ground wire leakage monitoring utilizes an impedance spectrum analysis method. The basic principle of this method is to inject excitation signals within a specific frequency range into the ground loop (including the monitored device, intelligent ground wire, and grounding grid) and measure the loop's response signals (such as voltage and current) to these excitation signals. This method then calculates the ground loop's impedance at different frequencies. Analyzing the curves or data sets showing these impedance variations over frequency is known as impedance spectrum analysis. Under normal circumstances, the insulation of the monitored device is good, and the ground loop's equivalent circuit model and its impedance spectrum exhibit specific baseline characteristics. However, when the device's insulation deteriorates or leakage occurs, the ground loop's equivalent circuit model changes, such as a decrease in equivalent parallel resistance and an increase in equivalent capacitance, resulting in changes in its impedance spectrum characteristics. By detecting and analyzing these spectral changes, it is theoretically possible to determine the presence of leakage, as well as the nature and extent of the leakage. This method offers the advantages of being non-invasive (compared to direct leakage current measurement, the excitation signal is typically smaller) and providing rich information (spectral characteristics).

[0021] However, as a typical location for high-voltage power facilities, power substations operate in an extremely complex and harsh environment. First, the operation of substation equipment generates strong power-frequency (50 / 60 Hz) electric and magnetic fields, as well as their higher harmonics. Second, transient events such as high-voltage switch operation, lightning arrester activation, and short-circuit faults generate high-amplitude, wideband transient electromagnetic interference. Furthermore, high-frequency signal sources such as wireless communication equipment, power line carrier communications, and control signal lines may exist within and around the substation, all of which generate electromagnetic interference of varying frequencies. These interference signals can couple into the monitoring loop of the smart grounding cable through capacitive, inductive, or conductive coupling, superimposing on the injected excitation signal and the collected response signal, significantly reducing the signal-to-noise ratio of the monitoring signal. Especially when monitoring weak, early-stage insulation degradation leakage current, the amplitude of the change in the ground loop impedance spectrum caused by the leakage current may be very weak and easily overwhelmed by strong electromagnetic interference, causing serious distortion of the impedance spectrum analysis results based on the original measurement data, resulting in a significant decrease in the accuracy of leakage judgment, and may result in omissions (failure to detect leakage) or false alarms (misjudging interference or normal changes as leakage).

[0022] Due to the strong electromagnetic interference described above, the directly acquired voltage and current signals contain a significant amount of noise. Without effective suppression or filtering of this noise, subsequent spectrum analysis and feature extraction will face significant challenges. Traditional narrowband or broadband filtering methods may, while filtering out the noise, also remove the effective spectral information of the leakage signal at certain frequencies. This is particularly difficult when the frequency of the interfering signal is close to the characteristic leakage frequency, making effective separation difficult. This makes extracting accurate and robust features related to the leakage state from noise-contaminated impedance spectrum data extremely challenging.

[0023] Furthermore, the insulation of power equipment in substations ages slowly. In the early stages of insulation degradation, only very weak leakage current may be generated. The resulting changes in the ground loop impedance spectrum are relatively subtle and likely to evolve slowly, making them difficult to identify using simple threshold determinations. In contrast, sudden faults such as insulation breakdown or external line short circuits can cause larger instantaneous leakage currents, resulting in more dramatic and rapid changes in the impedance spectrum. Leakage monitoring using impedance spectrum analysis requires the ability to effectively distinguish these slowly evolving insulation degradation characteristics from those of sudden faults, avoiding misjudgment of spectrum fluctuations caused by transient interference and enabling timely detection of early signs of insulation degradation. Simple threshold determinations based on impedance amplitude or phase changes are ineffective in distinguishing between different types of leakage or insulation degradation stages.

[0024] Furthermore, the power equipment connected to the smart grounding cable may be in different operating modes, such as normal energized operation, power outage for maintenance, or standby mode. Under these different operating modes, the electrical connection status and equivalent impedance characteristics within the equipment may change, affecting the overall impedance spectrum of the entire ground loop. For example, the equivalent capacitance characteristics of a device when energized may differ from those when it is out of service. Furthermore, environmental factors within the substation, such as changes in ambient temperature and humidity, and seasonal variations in soil resistivity, also affect the impedance characteristics of the grounding grid and grounding cable. These factors result in the normal impedance spectrum baseline of the ground loop not being fixed but rather changing dynamically with the equipment status and environmental conditions. Using a fixed model or baseline for impedance spectrum analysis will be difficult to adapt to these dynamic changes, potentially misinterpreting impedance spectrum changes caused by equipment status or environmental changes as leakage. Alternatively, baseline drift may prevent accurate identification of actual leakage.

[0025] Leakage monitoring using impedance spectrum analysis requires injecting an excitation signal into the ground loop and acquiring the response signal. The frequency range and amplitude selection of the excitation signal, as well as parameters such as the bandwidth and sampling rate of the measurement system, directly affect the resolution and accuracy of the spectrum analysis. In strong interference environments, to improve measurement accuracy, it may be necessary to adopt a specific excitation signal waveform, increase the excitation signal energy, or extend the data acquisition time. However, these measures may increase the power consumption of the monitoring equipment, affect the equipment's response speed, or potentially affect the normal operation of the monitored equipment. At the same time, different leakage types and fault locations may exhibit most significant characteristics within different frequency ranges of the ground loop impedance spectrum, and it is necessary to select an appropriate excitation frequency range to effectively capture this information.

[0026] Finally, as a device deployed long-term outdoors or indoors in high-voltage areas, smart ground cables must demonstrate high reliability and stability. Over extended periods of operation, the performance of monitoring circuit components may drift, and interfaces connected to the ground cable or grid may oxidize, experience poor contact, or become loose. These factors can affect the accuracy and stability of impedance spectrum measurements, distorting monitoring data and, in turn, impacting the reliability of leakage detection. While maintaining monitoring accuracy, the entire monitoring system must ensure stability and anti-interference capabilities in harsh environments and long-term operation.

[0027] Therefore, in the specific scenario of power substations, which are subject to strong electromagnetic interference, diverse equipment operation modes, changing environmental factors, and long-term operation, how to use technology based on impedance spectrum analysis to overcome the impact of electromagnetic interference on impedance spectrum measurement, effectively distinguish different types of leakage and equipment status changes, adapt to the dynamic changes of the ground loop impedance spectrum baseline, and maintain the stability and reliability of monitoring in long-term operation, thereby avoiding missed reports and false alarms, is a key issue that needs to be urgently solved in current intelligent ground wire leakage monitoring technology.

[0028] In this regard, please refer to the attached Figure 1 The present invention provides a ground wire monitoring method, which is applied to power equipment in a substation, comprising the following steps: Inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the smart ground wire and the ground grid, and synchronously collect the voltage signal at the injection point and the current signal of the ground loop; Preprocessing the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; Based on the preprocessed voltage and current signals, the impedance spectrum of the ground loop at different frequencies is calculated, and key characteristic parameters in the impedance spectrum are extracted. Key characteristic parameters include the impedance amplitude and phase at specific frequency points or frequency bands, the slope and curvature of the impedance spectrum curve within a specific frequency range, the frequency, amplitude, and bandwidth of the resonance peak or antiresonance peak in the spectrum, and the spectrum energy distribution in a specific frequency band. (It should be noted that the selection of "specific frequency points or frequency bands" is usually based on the analysis of the ground loop equivalent circuit model and the performance of different types of insulation degradation or leakage faults in the frequency domain. For example, the dielectric loss characteristics of certain insulation materials may show significant changes within a specific frequency range, or structural defects in the grounding grid may cause impedance anomalies at specific frequency points. Those skilled in the art will identify frequency points or frequency bands that are sensitive to changes in equipment status through historical operation data analysis, theoretical calculations, or experimental verification. The impedance amplitude and phase of these sensitive frequency points or frequency bands can directly reflect changes in parameters such as resistance, inductance, and capacitance of the ground loop, thereby indicating potential anomalies. The "specific frequency range" is also determined based on experience and analysis. The slope reflects The impedance spectrum curve describes how quickly the impedance changes with frequency, while the curvature describes the rate of change. In certain frequency ranges, the impedance spectrum of a normal ground loop may exhibit a gentle or linear change. However, when insulation degradation or leakage occurs, the curve's slope or curvature may change significantly, such as exhibiting an inflection point or a sharp rise or fall. Therefore, selecting these frequency ranges sensitive to these state changes for slope and curvature calculations helps capture subtle anomalies in the spectrum shape. "Specific frequency bands" refer to one or more frequency ranges with specific diagnostic value within the entire frequency range covered by the stimulus signal injected into the ground loop formed by the smart ground wire and the grounding grid, selected based on actual diagnostic needs. These frequency bands are selected based on an analysis of the ground loop impedance spectrum. Specifically, different leakage types and fault locations may exhibit most significant characteristics within different frequency ranges of the ground loop impedance spectrum. Therefore, "specific frequency bands" are selected to effectively capture key information related to equipment insulation degradation or leakage faults. By identifying these frequency ranges that most significantly reflect abnormal conditions, the effectiveness of extracting key characteristic parameters from the impedance spectrum can be improved, thereby enhancing the accuracy of ground loop condition assessment. Collect environmental parameters and operating status data of power equipment and fuse them with key characteristic parameters to obtain fused data; Based on a pre-established dynamic reference model or diagnostic rule set, the fused data is used to adjust the judgment thresholds or reference baselines corresponding to the key characteristic parameters. The judgment thresholds are the numerical boundaries used to distinguish normal conditions from abnormal conditions (such as early degradation or leakage faults). The reference baselines are the normal values ​​or ranges of the corresponding key characteristic parameters (the reference baseline is the value or range of the impedance spectrum characteristic parameters under normal operating conditions, specific environmental conditions, and equipment status). Adjusting these judgment thresholds or reference baselines is to enable the judgment process to adapt to the normal fluctuations of the ground loop impedance spectrum caused by changes in environmental factors (such as temperature and humidity) and equipment operating conditions (such as energized or unenergized). The adjusted judgment threshold or reference baseline is used to judge whether the current ground loop status is normal, early deterioration, or leakage fault of varying degrees.

[0029] The present invention aims to solve the problem of accurately determining the status of a ground loop in the complex environment of a substation. First, an excitation signal is injected into the ground loop and voltage and current signals are collected. This is the fundamental step in obtaining ground loop impedance information. By injecting a known signal and measuring the loop response, the loop's impedance characteristics can be calculated. This yields raw data reflecting the electrical characteristics of the ground loop. Next, the collected voltage and current signals are preprocessed. Substation environments are subject to strong electromagnetic interference, and the directly collected signals contain significant noise. This preprocessing step uses signal processing techniques to remove or suppress interference, improving signal quality and providing clean data for subsequent accurate impedance spectrum calculation, thereby reducing the impact of interference on measurement results. Then, based on the preprocessed voltage and current signals, the impedance spectrum of the ground loop at different frequencies is calculated and key characteristic parameters are extracted. The impedance spectrum reflects the electrical characteristics of the ground loop at different frequencies. Extracting key characteristic parameters involves converting complex spectrum curves into representative values ​​or patterns. These parameters are associated with changes in the ground loop's status. For example, insulation degradation or leakage can cause impedance changes at specific frequencies. Simultaneously, environmental parameters and operating status data of the power equipment are collected and integrated with the extracted key characteristic parameters. The impedance characteristics of a ground loop are affected by environmental factors and equipment operating conditions, which can cause variations in the impedance spectrum baseline. Collecting this data and fusing it with the impedance signature provides more comprehensive information, reflecting the actual operating conditions of the ground loop and providing a more comprehensive understanding of its status. Using this fused data, the judgment thresholds or reference baselines corresponding to key characteristic parameters are adjusted based on a pre-established dynamic reference model or diagnostic rule set. Traditional fixed thresholds or baselines are difficult to adapt to changes in the environment and equipment status. By fusing environmental and operating data and incorporating dynamic models or rules, the judgment thresholds or reference baselines can be adjusted based on actual operating conditions, more closely matching the current normal range and improving the adaptability of the judgment criteria. Finally, the adjusted judgment thresholds or reference baselines are used to determine the current ground loop status. Based on this adjusted, more accurate judgment criteria, the extracted key characteristic parameters are evaluated to distinguish whether the ground loop is in a normal state, in the early stages of degradation, or experiencing varying degrees of leakage faults. This improves the accuracy of ground loop status judgment in complex and changing environments and reduces false positives and false negatives.

[0030] Specifically, the method works as follows: First, by injecting an excitation signal and collecting the response signal, electrical response data of the ground loop is acquired at different frequencies. This raw data may be affected by strong electromagnetic interference from the substation. Therefore, preprocessing the acquired voltage and current signals is necessary. Preprocessing aims to filter out power frequency interference, transient high-amplitude interference, and random noise, resulting in a purer signal. This improves the accuracy of the subsequent impedance spectrum calculation. Based on the preprocessed signal, the impedance spectrum of the ground loop is calculated. The impedance spectrum is a function of frequency and reflects the comprehensive characteristics of the loop, such as resistance, inductance, and capacitance. Key characteristic parameters are extracted from the impedance spectrum, such as the impedance value at a specific frequency point, the shape of the spectrum curve, or the resonance peak information. These parameters are considered to be related to the health of the ground loop. However, these impedance characteristics are affected by environmental factors (such as temperature and humidity) and equipment operating conditions (such as voltage and current), resulting in the impedance characteristic values ​​under normal conditions not being fixed. To address this issue, the method further collects environmental parameters and operating status data of the power equipment and integrates this data with the extracted key impedance characteristic parameters. The fused data contains information about the electrical characteristics of the ground loop and its operating conditions. Using this fused data, the method dynamically adjusts the judgment threshold or reference baseline used to determine the ground loop status based on a pre-established dynamic reference model or diagnostic rule set. For example, when the ambient temperature rises, the grounding grid resistance may increase, resulting in an increase in the overall impedance. In this case, the dynamic reference model will correspondingly increase the impedance reference baseline under normal conditions. Finally, the currently extracted impedance key characteristic parameters are compared with the adjusted judgment threshold or reference baseline. If the key characteristic parameters exceed the adjusted normal range, the ground loop is judged to be in an abnormal state, such as early degradation or a leakage fault. By dynamically adjusting the judgment criteria in this way, the method can adapt to the complex and changing environment and equipment operating conditions of the substation, and improve the accuracy and reliability of leakage monitoring.

[0031] In some specific embodiments, a swept-frequency excitation signal is injected into the ground loop, with a frequency range set to 100 Hz to 10 kHz. The voltage signal at the injection point and the ground loop current signal are synchronously collected, with a sampling rate set to 50 kHz. The collected voltage and current signals are first filtered through a digital notch filter to remove interference from the 50 Hz power frequency and its third harmonic (150 Hz). Furthermore, a short-time Fourier transform is performed on the filtered signals to identify and remove transient components with amplitudes exceeding a set threshold. This suppresses the impact of transient interference. The processed voltage and current signals are measured multiple times and averaged to reduce random noise. This results in preprocessed voltage and current signals. Based on the preprocessed signals, the impedance spectrum of the ground loop in the range of 100 Hz to 10 kHz is calculated. Key characteristic parameters are extracted from the impedance spectrum, such as the impedance amplitude and phase at 1 kHz and the slope of the impedance amplitude curve in the frequency band from 2 kHz to 5 kHz. Simultaneously, the substation ambient temperature and the operating current of the monitored equipment are collected. The 1 kHz impedance amplitude, phase, and spectral slope are weighted and fused with the ambient temperature and operating current to generate fused data. The fusion weights are determined by historical data analysis. Based on a pre-established dynamic reference model based on historical data, which accounts for the effects of temperature and current on impedance characteristics, the current fused data is used to adjust the normal range reference baseline for the 1 kHz impedance amplitude. Finally, the currently calculated 1 kHz impedance amplitude is compared with the adjusted reference baseline. If the current amplitude is below the lower limit of the reference baseline, a leakage fault may be present.

[0032] It should be noted that the present invention is applicable to power equipment equipped with an intelligent grounding wire. The intelligent grounding wire is connected to an external communication device. Users can send control signals to the external communication device through a user terminal to control the intelligent grounding wire to perform operations such as electrical testing. At the same time, the intelligent grounding wire can also send feedback information to the user terminal through the external communication device. Therefore, in actual application, the information output by the present invention (such as the judgment result of the current grounding loop status, the occurrence of faults, and the warning information below) can also be sent to the user terminal, allowing users to communicate remotely through the user terminal. Furthermore, users can also design a dedicated APP as needed to display functions such as electrical testing, ground terminal resistance detection, abnormality feedback, and fault point location detection on a visual user interface.

[0033] In some embodiments, the steps are further included: Record and store historical impedance spectrum characteristic parameters, environmental parameters and equipment status information; By conducting long-term trend analysis on historical data, we can obtain the rate and direction of change of key characteristic parameters over time; According to the rate and direction of change of key characteristic parameters over time, the insulation degradation rate of power equipment is evaluated, and based on the evaluation results, early warning is issued before the power equipment fails.

[0034] Recording and storing historical impedance spectrum characteristic parameters, environmental parameters, and equipment status information establishes a data foundation for subsequent analysis. By accumulating this data over time, it is possible to observe the trajectory of ground loop status changes over time. Long-term trend analysis of historical data reveals the rate and direction of change of key characteristic parameters over time, leveraging the accumulated historical data. By analyzing the numerical changes of key characteristic parameters at different time points, it is possible to identify trends such as stability, slow increase, slow decrease, or accelerating change. The rate and direction of change quantify these trends. Based on the rate and direction of change of key characteristic parameters over time, the rate of insulation degradation of power equipment can be assessed, and the trend analysis results can be correlated with the equipment's insulation status. Insulation degradation is typically a gradual process, and its manifestation in the impedance spectrum characteristics may also be a slow change. By analyzing the rate and direction of change of key characteristic parameters, the speed of insulation degradation can be determined. The goal of the entire solution is to provide early warnings before power equipment failure occurs. The assessed insulation degradation rate can be used to predict the time when insulation conditions are likely to reach the failure threshold. Before the predicted failure occurs, the system can issue an early warning signal based on the degradation rate assessment results, prompting operation and maintenance personnel to pay attention to the equipment status or arrange maintenance.

[0035] Specifically, this technical solution not only determines the current ground loop status but also utilizes historical data. First, the system continuously records and stores key characteristic parameters of the impedance spectrum obtained during each monitoring session, such as impedance amplitude and phase at specific frequencies. It also records environmental parameters during monitoring, such as temperature and humidity, as well as information about the operating status of the power equipment. This historical data is stored in a database. The system then processes this stored historical data, extracting the numerical sequence of a specific key characteristic parameter over a period of time. By applying time series analysis methods such as linear regression, exponential smoothing, or moving averages, the system calculates the average rate and direction of change of that parameter over time. For example, if a certain impedance amplitude parameter has been decreasing continuously over the past year, trend analysis can calculate its average monthly decrease and determine the direction of decrease. Based on the calculated rate and direction of change of the key characteristic parameter, the system assesses the rate of insulation degradation of the power equipment. For example, if the rate of decrease of the impedance amplitude exceeds a preset threshold, the insulation degradation rate is determined to be rapid; if the rate of change is low, the degradation rate is determined to be slow. This provides information on the insulation degradation rate. Finally, based on the estimated insulation degradation rate, the system predicts when the insulation condition will reach the fault threshold. If the system issues an early warning signal based on the degradation rate assessment before the predicted failure occurs, for example, if a failure is predicted within the next three months, the system will issue an early warning. This enhances predictive maintenance capabilities. These additional steps expand the monitoring method from simply assessing the current state to predicting and warning future conditions, thereby comprehensively improving the accuracy and reliability of intelligent ground wire leakage monitoring.

[0036] In some embodiments, the step of preprocessing the collected voltage signal and current signal to obtain the preprocessed voltage signal and current signal includes: The collected voltage and current signals are cleaned using a digital notch filter or Fourier analysis method to suppress the power frequency and its harmonics, thereby obtaining the primary processed voltage and current signals. For the voltage and current signals after primary processing, the time-frequency analysis method is used to identify and remove transient high-amplitude components to obtain the voltage and current signals after secondary processing; For the secondary processed voltage signal, a plurality of voltage values ​​are continuously collected from the secondary processed voltage signal according to a preset first measurement period, and all voltage values ​​of the same first measurement period are averaged to obtain an average voltage value corresponding to the first measurement period, and the average voltage values ​​obtained from all different first measurement periods are used as the tertiary processed voltage signal to reduce random noise of the voltage signal and as the final preprocessed voltage signal, or the secondary processed voltage signal is directly subjected to Kalman filtering to reduce random noise of the voltage signal, to obtain the tertiary processed voltage signal, and as the final preprocessed voltage signal; For the current signal after secondary processing, multiple current values ​​are continuously collected from the current signal after secondary processing according to a preset second measurement period, and all current values ​​of the same second measurement period are averaged to obtain the average current value corresponding to the second measurement period. The average current value obtained from all different second measurement periods is used as the current signal after tertiary processing to reduce the random noise of the current signal and as the final preprocessed current signal, or the current signal after secondary processing is directly Kalman filtered to reduce the random noise of the current signal, and the current signal after tertiary processing is obtained, and as the final preprocessed current signal.

[0037] Cleaning the collected voltage and current signals to suppress the power frequency and its harmonics can be achieved by designing a digital notch filter with high attenuation at the power frequency (e.g., 50 Hz or 60 Hz) and its integer multiples. Alternatively, a fast Fourier transform (FFT) can be performed on the collected signals to zero or significantly attenuate the amplitudes of the frequency points or frequency bands corresponding to the power frequency and its harmonics in the frequency domain. An inverse Fourier transform (IFFT) can then be performed to obtain the time domain signal. Furthermore, time-frequency analysis can be used to identify and remove transient high-amplitude components. Wavelet analysis can be employed. By selecting appropriate wavelet basis functions, the signal is subjected to wavelet decomposition. Transient high-amplitude components typically appear as a few large coefficients in the wavelet domain. By setting a threshold, wavelet coefficients below the threshold are zeroed or shrunk, followed by wavelet reconstruction to remove transient components. This can reduce random noise in the voltage and current signals by repeatedly collecting signal samples under the same state and then smoothing the signals by averaging these samples. As another implementation method, a state space model that describes the dynamic characteristics of the signal can be constructed, and a Kalman filter algorithm can be applied to optimally estimate the true value of the signal based on real-time measurement data, thereby filtering out random noise.

[0038] Specifically, in a substation environment, the collected ground loop voltage and current signals are susceptible to various interferences. First, strong power frequency electromagnetic fields can induce power frequency and its harmonic components in the signals. Digital notch filtering or Fourier analysis can effectively attenuate these fixed-frequency periodic interferences, resulting in a processed signal with improved signal purity. Second, transient events such as switching operations generate short, high-amplitude pulse interferences. These interferences are widely distributed in the time and frequency domains, making them difficult to remove using traditional filters. Time-frequency analysis methods, such as wavelet analysis, can precisely locate these transient events in time and frequency. These interferences can then be separated and removed from the signal through thresholding and other methods, resulting in a secondary processed signal that mitigates the impact of transient interference on subsequent analysis. Finally, random noise still exists in the signals after the first two steps of processing, which can affect the accuracy of impedance calculations. By averaging multiple measurements, the average value of random noise approaches zero, thereby reducing its impact. Alternatively, Kalman filtering can optimally estimate and suppress noise based on the dynamic characteristics of the signal, further improving the signal-to-noise ratio and accuracy, and producing the final preprocessed voltage and current signals. These clean signals are used to calculate the impedance spectrum of the ground loop, which can significantly improve the accuracy and reliability of spectrum analysis and provide basic data for accurately judging the status of the ground loop.

[0039] In some embodiments, the steps of collecting environmental parameters and operating status data of the power equipment and fusing them with key characteristic parameters to obtain fused data include: Collect environmental parameters of power equipment, including temperature, humidity, electromagnetic interference intensity, and operating status data, including voltage, current, active power, and reactive power; Wavelet threshold denoising method is used to filter out noise interference of environmental parameters and operating status data, and an interpolation algorithm based on historical data is used to fill in missing values ​​to obtain a clean and complete data set; The mutual information method is used to evaluate the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter, and obtain the evaluation results of the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter. According to the evaluation results, the environmental parameters and operating status data with correlation greater than the preset threshold are screened out to obtain the screened parameter set (it should be noted that the selection of the preset threshold is usually based on the analysis of historical data, statistical principles and comprehensive consideration of domain knowledge. Specifically, in practical applications, the preset threshold can be determined by the following methods: First, historical data analysis is performed. During the normal operation of the equipment, a large amount of environmental Parameters, operating status data and the corresponding key characteristic parameters of the impedance spectrum. Then, the mutual information value between each environmental parameter and operating status data and the key characteristic parameters in these historical data is calculated. By observing the distribution and range of these mutual information values ​​under normal operating conditions, it is possible to identify which parameters generally show a strong correlation and which show a weak correlation. For example, the average level and fluctuation range of the mutual information value can be statistically analyzed, and an initial threshold can be set based on experience so that most of the parameters that are considered to be influential under normal circumstances can be retained. Secondly, combined with statistical principles. Mutual information quantifies the degree of dependence between two variables. Although mutual information itself does not have a direct correlation, it can also be used to determine the correlation between the two variables. The mutual information value is not a statistical significance test standard for the connection, but its numerical value reflects the amount of information sharing. Generally, the higher the mutual information value, the stronger the correlation between the two parameters. Therefore, the selection of the preset threshold can be based on the understanding of the mutual information value, setting a boundary that can distinguish "meaningful correlation" from "random fluctuations or weak correlations". For example, an empirical value such as 0.1, 0.2 or higher can be set as the minimum mutual information requirement. Furthermore, domain knowledge and expert experience are incorporated. The operation of power equipment and the characteristics of the grounding system are complex, and certain environmental or operating parameters are considered to have a significant impact on the grounding loop impedance in theory or based on long-term operation and maintenance experience. These prior Knowledge can guide the initial setting of preset thresholds. For example, if it is known that temperature has a great influence on the impedance of the grounding grid, then even if the initially calculated mutual information value is slightly lower than a certain universal threshold, it is possible to consider adjusting the threshold appropriately to retain the temperature parameter. Finally, through system performance verification and optimization. Before actual deployment, the monitoring system under different preset thresholds can be tested offline or online. By evaluating the false alarm rate, missed alarm rate and diagnostic accuracy of the system under different thresholds, a threshold that can balance these performance indicators is selected. This is an iterative optimization process that aims to find a threshold that can effectively filter out irrelevant interference while retaining enough useful information to support accurate judgment); The Min-Max scaling method is used to normalize the environmental parameters and operating status data in the filtered parameter set so that the environmental parameters and operating status data are scaled to the [0,1] interval to eliminate dimensional differences; The normalized environmental parameters and operating status data are fused with the key characteristic parameters through a weighted fusion algorithm to obtain a fused multi-dimensional feature vector, which is used as fusion data for subsequent dynamic reference model or diagnostic rule set adjustment; the weight coefficient in the weighted fusion algorithm is determined by the corresponding mutual information value.

[0040] Specifically, this technical solution first acquires raw data reflecting the external environment and internal operating status of power equipment. This raw data may contain noise or omissions due to various factors. By applying wavelet threshold denoising and historical data interpolation techniques, the raw environmental parameter and operating status data are cleansed and supplemented, improving the data reliability and integrity, laying the foundation for subsequent analysis. Next, the mutual information method is used to quantify the correlation strength between the processed environmental parameter and operating status data and the key characteristic parameters of the ground loop impedance spectrum. Data is filtered based on a preset correlation threshold, removing redundant information with low correlation with the key characteristic parameters, reducing the data dimensionality and focusing on factors that have a practical impact on ground loop status determination. Subsequently, the filtered environmental parameter and operating status data are normalized to eliminate the influence of different physical dimensions and numerical ranges, ensuring fairness in the fusion process. Finally, a weighted fusion algorithm is used to combine the normalized environmental parameter and operating status data with the key characteristic parameters. In the weighted fusion process, weights are determined based on the previously calculated mutual information values, so that environmental or operating parameters with higher correlations with the key characteristic parameters receive a greater weight in the fusion results. The resulting multidimensional fused data comprehensively reflects the complex relationship between ground loop status, environmental conditions, and equipment operating status. This fused data serves as input for dynamically adjusting the threshold or reference baseline for ground loop status judgment, improving the accuracy of ground loop status judgment under different environmental and operating conditions. (It should be noted that the calculation of mutual information (MI) is typically based on estimating the probability distribution of the variables involved. Specifically, for any two variables X (e.g., ambient temperature) and Y (e.g., impedance amplitude), calculating the MI requires the following steps: First, a large number of historical data samples must be collected. These samples contain paired observations of variables X and Y. If these variables are continuous, their value ranges are typically divided into several discrete intervals for ease of calculation, converting the continuous data into discrete categories. Second, based on these discretized data samples, their probability distributions are estimated. This includes estimating the marginal probability P(X) of variable X taking each discrete interval, estimating the marginal probability P(Y) of variable Y taking each discrete interval, and estimating the joint probability P(X, Y) of variables X and Y taking specific discrete intervals simultaneously.) Y). These probabilities can be approximated by counting the frequency of samples within each interval or within each pair of intervals. Mutual information is then calculated based on these estimated probabilities. Mutual information measures the amount of information gained about one variable by observing the other. If two variables are independent, the mutual information is zero; if they are dependent, the mutual information is greater than zero, with larger values ​​indicating stronger dependence.

[0041] In some specific embodiments, for example, the status of the transformer grounding wire is monitored. The temperature, humidity, load current and winding temperature around the transformer are collected during operation. At the same time, the impedance amplitude of the ground loop at 1kHz and 10kHz frequencies is measured as a key characteristic parameter. The collected temperature and current data may contain instantaneous fluctuations, which are filtered out using the wavelet threshold method. If the temperature sensor data at a certain time point is lost, the average value of the previous stable time points is used to fill in the gap. The mutual information between the temperature and the 1kHz impedance amplitude, as well as the mutual information between the load current and the 1kHz impedance amplitude, are calculated. Assume that the correlation between temperature and impedance is high and the correlation between current and impedance is low. For example, if the preset threshold is 0.4, only data greater than the preset threshold is selected. For example, if the correlation of temperature data is 0.7 and the correlation of current data is 0.3, only temperature data is selected for fusion. Min-Max normalization is performed on the historical temperature data to map the current temperature value to between 0 and 1. The normalized temperature data is weighted and fused with the impedance amplitude at 1kHz and 10kHz, and the weight is determined according to the mutual information value (it should be noted that the specific method of determining the weight coefficient according to the mutual information value can be implemented as follows: First, calculate the mutual information value between each pre-processed, filtered and normalized environmental parameter and operating status data and the key characteristic parameters of the ground loop impedance spectrum. For example, if temperature, humidity and operating current are selected as environmental and operating parameters, and impedance amplitude and phase are selected as key characteristic parameters, it is necessary to calculate the mutual information between temperature and impedance amplitude, the mutual information between temperature and impedance phase, and the mutual information between humidity, operating current and impedance amplitude and phase. Mutual information. After obtaining these mutual information values, normalization can be used to determine the weight coefficient. For example, for a specific key characteristic parameter, the mutual information values ​​of the various environmental parameters and operating status data related to it can be summed, and then each mutual information value can be divided by this sum to obtain the weight of the parameter in the fusion. In this way, the parameter with a larger mutual information value has a greater weight in the fused data and a greater impact on the final fusion result. This method ensures that during the fusion process, external factors that are more closely related to the change in the ground loop state can obtain a higher weight, so that the fused data more accurately reflects the impact of actual working conditions on the ground loop impedance characteristics. For example, the fused data can be a vector containing two impedance amplitudes and normalized temperature, where the weight of the normalized temperature reflects its correlation with the impedance amplitude. This fused vector is then used to adjust the impedance threshold for determining whether the ground loop is abnormal.

[0042] In some embodiments, the step of adjusting the judgment threshold or reference baseline corresponding to the key feature parameter using the fused data includes: According to the pre-established dynamic reference model, the fusion data is used to adjust the judgment threshold or reference baseline corresponding to the key feature parameters.

[0043] In some embodiments, the step of establishing a dynamic reference model includes: Obtain historical operating data, including historical impedance spectrum characteristic parameters, historical environmental parameters, and historical equipment status information; Based on historical operating data, a clustering algorithm is used to perform cluster analysis on the data under normal operating conditions to obtain multiple normal operating condition cluster centers. By calculating the covariance matrix of each cluster center, a dynamic reference model under normal operating conditions is constructed.

[0044] The acquisition of historical operating data provides a basic data set for model training. These data record the performance of the equipment at different times and conditions. The application of clustering algorithms aims to identify a variety of typical modes or sub-states under normal operating conditions, such as the normal performance of equipment in different seasons or at different load levels. In this way, cluster centers representing these typical normal modes can be obtained. Furthermore, the covariance matrix of each cluster center is calculated. This matrix quantifies the normal fluctuation range of each characteristic parameter and their mutual relationship under this specific normal mode, thereby defining the normal data distribution under this mode. By collecting these cluster centers and their corresponding covariance matrices, a reference model that can reflect the diversity and dynamic changes of normal operating conditions is constructed.

[0045] Specifically, this technical solution analyzes historical data to establish a reference model that can adapt to changes in the environment and operating conditions. First, historical data generated by the equipment over long periods of operation is collected. This data includes characteristic ground loop impedance spectrum parameters measured under various environmental conditions (such as temperature and humidity) and operating conditions (such as voltage and current). Then, data samples from the equipment under normal operating conditions are selected from this historical data. Next, a clustering algorithm is applied to this multidimensional data under normal operating conditions (including characteristic parameters, environmental parameters, and operating status data). The clustering process groups similar data points into clusters, each representing a specific normal operating mode. For example, one cluster might correspond to the normal state of the equipment under high summer temperatures, high humidity, and high load, while another might correspond to the normal state of the equipment under low winter temperatures, dry conditions, and low load. The clustering algorithm outputs the center points of each cluster, which represent typical values ​​for each normal mode. Subsequently, the covariance matrix of the data points within each cluster is calculated. The covariance matrix describes the dispersion of data points within the cluster across various dimensions and the correlations between these dimensions, thereby defining the range and shape of the data distribution under that specific normal mode. By obtaining the coordinates of all normal operating condition cluster centers and their corresponding covariance matrices, a dynamic reference model is constructed. This model no longer relies on a single, fixed baseline. Instead, it finds the most suitable normal mode within the model based on the current equipment environment and operating status, and uses the covariance matrix corresponding to that mode to determine the normal range or confidence interval for the current state. Therefore, when subsequently determining the current ground loop status, the currently measured fused data (including real-time feature parameters, environmental parameters, and operating status data) can be compared with the most suitable normal mode in the dynamic reference model, allowing for a more accurate assessment of whether the current state deviates from the normal range under specific conditions. This improves the accuracy of leakage detection and reduces false positives and negatives.

[0046] In some specific embodiments, it is assumed that historical operating data includes the impedance amplitude of the ground loop at a specific frequency, the ambient temperature, and the equipment load current. First, several months or years of historical data are collected, and data points where the equipment is confirmed to be in a normal state are marked. For example, 10,000 data samples under normal conditions are collected, each sample being a three-dimensional vector [impedance amplitude, temperature, load current]. Next, a K-means clustering algorithm is used to perform cluster analysis on these 10,000 samples, with the number of clusters K set to 3. After running the clustering algorithm, three cluster centers C1, C2, and C3 are obtained, representing three typical normal operating modes. For example, C1 may correspond to [lower impedance amplitude, lower temperature, lower load], C2 to [medium impedance amplitude, medium temperature, medium load], and C3 to [higher impedance amplitude, higher temperature, higher load]. Then, for the data points belonging to each cluster, the three-dimensional covariance matrices Σ1, Σ2, and Σ3 are calculated. For example, Σ1 describes the fluctuation ranges and correlations between the three parameters of impedance amplitude, temperature, and load current under the normal operating condition corresponding to C1. Therefore, the dynamic reference model consists of {C1,Σ1},{C2,Σ2},{C3,Σ3}. During real-time monitoring, the current fused data [current impedance amplitude, current temperature, current load current] is obtained, the Mahalanobis distance between this data point and C1, C2, and C3 is calculated, and the cluster center with the smallest Mahalanobis distance is selected, such as C2. Then, using the covariance matrix Σ2 corresponding to C2, the confidence level or Mahalanobis distance threshold (i.e., confidence interval) of the current data point under this normal mode is calculated. The confidence interval serves as a reference baseline or threshold for determining whether the current state is normal.

[0047] Furthermore, the steps of adjusting the judgment threshold or reference baseline corresponding to the key feature parameters using the fused data according to the pre-established dynamic reference model include: Obtain the current fusion data in real time, calculate the Mahalanobis distance between the current fusion data and the cluster centers of each normal working condition, and select the cluster center corresponding to the minimum Mahalanobis distance; Obtain the covariance matrix corresponding to the selected cluster center through the dynamic reference model, and calculate the confidence interval of the current fusion data based on the covariance matrix; The confidence interval is used as the judgment threshold or reference baseline of the key characteristic parameters for adaptive adjustment. If the current key characteristic parameters exceed the confidence interval, it is judged as an abnormal state, otherwise it is judged as a normal state; the adjustment of the confidence interval includes: when a sudden change in the environmental parameters is detected, the confidence interval is increased to reduce the false alarm rate; when the equipment operation state is stable, the confidence interval is reduced to improve the detection sensitivity (it should be noted that the confidence interval is an interval used in statistics to estimate the true value range of a parameter, which is determined according to the distribution characteristics of the data (such as the mean and covariance matrix). When applied to key characteristic parameters, the confidence interval defines the confidence interval in the current specific environment. Under the environmental and operating conditions, the key characteristic parameter is considered to be the statistical range of normal values. The boundary of the confidence interval or the range it defines directly constitutes or determines the "judgment threshold or reference baseline of the key characteristic parameter" under the current operating conditions. In other words, the confidence interval itself is the specific manifestation of the dynamically adjusted "reference baseline", or its upper and lower limits are the dynamically adjusted "judgment thresholds". For example, if the adjusted reference baseline is a range [A, B], then this [A, B] is calculated based on the confidence interval. To determine whether the current key characteristic parameter is within [A, B], it is to use the adjusted reference baseline to make a judgment).

[0048] In the complex and ever-changing environments and equipment operating conditions of power substations, the normal impedance spectrum baseline of the ground loop is not fixed. Using fixed judgment criteria makes it difficult to accurately identify abnormal conditions, potentially leading to misjudgments or missed detections. This technical solution introduces a dynamic reference model and combines real-time environmental parameters and equipment operating status data to achieve adaptive adjustment of the judgment threshold or reference baseline. First, the system collects fused data in real time, including key characteristic parameters of the impedance spectrum, environmental parameters, and equipment operating status data. The current fused data is then compared with a dynamic reference model established based on historical normal data. The dynamic reference model contains multiple cluster centers representing different normal operating modes and their corresponding covariance matrices. By calculating the Mahalanobis distance between the current fused data and each cluster center, the normal operating mode to which the current state is most similar is determined. The cluster center with the smallest distance is selected, and its corresponding covariance matrix is ​​obtained. This covariance matrix reflects the distribution characteristics of each dimension of the fused data (including impedance characteristics, environmental parameters, and operating status) under a specific normal operating condition. Using this covariance matrix, the confidence interval for the statistical probability that the current fused data belongs to that normal operating mode is calculated. This confidence interval defines the normal fluctuation range of the key characteristic parameters under the current environment and operating conditions. During judgment, the current key characteristic parameters are compared with the calculated confidence interval. If the key characteristic parameters fall within the confidence interval, the ground loop is considered normal; if they exceed the confidence interval, the ground loop is considered abnormal. To further optimize judgment performance, the system dynamically adjusts the width of the confidence interval based on the changing trends of environmental parameters and device operating conditions. For example, when rapid and drastic changes in ambient parameters such as temperature and humidity are detected, the ground loop impedance may experience temporary fluctuations. In this case, appropriately increasing the confidence interval can improve the system's tolerance for environmental changes, reduce the possibility of misinterpreting normal fluctuations as abnormalities, and reduce the false alarm rate. Conversely, when the device has been operating stably for a long time and environmental parameters are relatively stable, the ground loop impedance characteristics should be more stable. In this case, narrowing the confidence interval can increase the system's sensitivity to detecting minor abnormal changes, helping to detect potential signs of insulation degradation early. Through this adaptive adjustment mechanism based on a dynamic reference model and real-time data, the system can more accurately reflect the normal state range under the current working conditions, thereby improving the accuracy and reliability of grounding wire status judgment in complex environments, effectively distinguishing normal changes from abnormal states, and avoiding false alarms and missed alarms.

[0049] In some specific embodiments, the dynamic reference model can cluster the multidimensional fused data under normal operating conditions using a K-means clustering algorithm based on historically collected fused data to obtain K cluster centers. For each cluster center, the covariance matrix of the data within that cluster is calculated. For example, at a certain substation, historical data shows that the fused data of the ground loop impedance spectrum characteristics and environmental / operating parameters exhibit different distribution patterns under high temperature and high humidity in summer, low temperature and dryness in winter, and different load levels. Cluster analysis can yield cluster centers and covariance matrices representing these typical normal operating conditions. During real-time monitoring, a fused data vector is collected, consisting of the current temperature of 25°C, humidity of 70%, equipment load of 80%, and the corresponding key characteristic parameters of the impedance spectrum. The Mahalanobis distance between this vector and all cluster centers is calculated, assuming that the distance to the cluster center representing the "high temperature, high humidity, and high load in summer" operating condition is the smallest. The covariance matrix Σ corresponding to this cluster center is obtained. Using this Σ, the squared Mahalanobis distance of the current fused data vector is calculated, and a 95% confidence interval is determined based on the chi-square distribution. If the squared Mahalanobis distance of the current fused data vector is less than the upper limit of the confidence interval, it is considered normal (it should be noted that when the fused data vector is considered to obey the multivariate normal distribution, the squared Mahalanobis distance follows the chi-square (x^2) distribution. The degrees of freedom of the chi-square distribution is equal to the number of dimensions of the fused data vector, that is, the total number of characteristic parameters contained in the fused data. Therefore, the upper limit of the confidence interval is determined by finding the critical value corresponding to a specific confidence level (such as 95% or 99%) on the chi-square distribution. This critical value means that under normal operating conditions, a certain percentage (such as 95%) of the squared Mahalanobis distance values ​​will fall below this critical value. Any squared Mahalanobis distance value calculated in real time that exceeds this critical value is considered to be beyond the range of normal fluctuations and is thus determined to be an abnormal state. For example, if the fused data vector contains 3 characteristic parameters, the degrees of freedom of the chi-square distribution is 3. If a confidence level of 95% is selected, the system will look for a chi-square value with 3 degrees of freedom and a cumulative probability of 0.95 in the chi-square distribution table. This value is the upper limit of the confidence interval). If the system detects a sudden increase in ambient temperature from 25°C to 35°C within a short period of time, the judgment module can expand the upper limit of the confidence interval by 10% based on preset rules or models to accommodate impedance fluctuations caused by sudden temperature changes and reduce the risk of false alarms. Conversely, if the equipment load and environmental parameters remain stable for several hours, the judgment module can reduce the upper limit of the confidence interval by 5% to improve the detection of weak abnormal signals.

[0050] Reference Attachment Figure 2 The present invention provides a ground wire monitoring system, which is applied to power equipment in a substation, comprising: The acquisition module 100 is used to inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the intelligent ground wire and the ground grid, and synchronously acquire the voltage signal of the injection point and the current signal of the ground loop; The preprocessing module 200 is used to preprocess the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; Extraction module 300 is used to calculate the impedance spectrum of the ground loop at different frequencies based on the preprocessed voltage and current signals, and extract key characteristic parameters from the impedance spectrum; the key characteristic parameters include the impedance amplitude and phase at a specific frequency point or frequency band, the slope and curvature of the impedance spectrum curve within a specific frequency range, the frequency, amplitude and bandwidth of the resonance peak or antiresonance peak in the spectrum, and the spectrum energy distribution in a specific frequency band; Fusion module 400, for collecting environmental parameters and operating status data of power equipment and fusing them with key characteristic parameters to obtain fused data; The adjustment module 500 is used to adjust the judgment threshold or reference baseline corresponding to the key characteristic parameter based on the pre-established dynamic reference model or diagnostic rule set and using the fusion data; the judgment threshold is the numerical limit used to distinguish between normal and abnormal states; the reference baseline is the normal value or range of the corresponding key characteristic parameter; The judgment module 600 is used to judge whether the current ground loop state is normal, early deterioration or leakage fault of varying degrees by using the adjusted judgment threshold or reference baseline.

[0051] In some embodiments, a prediction module 700 is further included, and the prediction module 700 is configured to perform the following steps: Record and store historical impedance spectrum characteristic parameters, environmental parameters and equipment status information; By conducting long-term trend analysis on historical data, we can obtain the rate and direction of change of key characteristic parameters over time; According to the rate and direction of change of key characteristic parameters over time, the insulation degradation rate of power equipment is evaluated, and based on the evaluation results, early warning is issued before the power equipment fails.

[0052] In some embodiments, the preprocessing module 200 performs the following steps when preprocessing the collected voltage signal and current signal to obtain the preprocessed voltage signal and current signal: The collected voltage and current signals are cleaned using a digital notch filter or Fourier analysis method to suppress the power frequency and its harmonics, thereby obtaining the primary processed voltage and current signals. For the voltage and current signals after primary processing, the time-frequency analysis method is used to identify and remove transient high-amplitude components to obtain the voltage and current signals after secondary processing; For the secondary processed voltage signal, a plurality of voltage values ​​are continuously collected from the secondary processed voltage signal according to a preset first measurement period, and all voltage values ​​of the same first measurement period are averaged to obtain an average voltage value corresponding to the first measurement period, and the average voltage values ​​obtained from all different first measurement periods are used as the tertiary processed voltage signal to reduce random noise of the voltage signal and as the final preprocessed voltage signal, or the secondary processed voltage signal is directly subjected to Kalman filtering to reduce random noise of the voltage signal, to obtain the tertiary processed voltage signal, and as the final preprocessed voltage signal; For the current signal after secondary processing, multiple current values ​​are continuously collected from the current signal after secondary processing according to a preset second measurement period, and all current values ​​of the same second measurement period are averaged to obtain the average current value corresponding to the second measurement period. The average current value obtained from all different second measurement periods is used as the current signal after tertiary processing to reduce the random noise of the current signal and as the final preprocessed current signal, or the current signal after secondary processing is directly Kalman filtered to reduce the random noise of the current signal, and the current signal after tertiary processing is obtained, and as the final preprocessed current signal.

[0053] In some embodiments, the fusion module 400 performs the following when collecting environmental parameters and operating status data of the power equipment and fusing them with key characteristic parameters to obtain fused data: Collect environmental parameters of power equipment, including temperature, humidity, electromagnetic interference intensity, and operating status data, including voltage, current, active power, and reactive power; Wavelet threshold denoising method is used to filter out noise interference of environmental parameters and operating status data, and an interpolation algorithm based on historical data is used to fill in missing values ​​to obtain a clean and complete data set; The mutual information method is used to evaluate the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter, thereby obtaining an evaluation result on the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter. The environmental parameters and operating status data having a correlation greater than a preset threshold are screened out based on the evaluation result to obtain a screened parameter set. The Min-Max scaling method is used to normalize the environmental parameters and operating status data in the filtered parameter set so that the environmental parameters and operating status data are scaled to the [0,1] interval to eliminate dimensional differences; The normalized environmental parameters and operating status data are fused with the key characteristic parameters through a weighted fusion algorithm to obtain a fused multi-dimensional feature vector, which is used as fusion data for subsequent dynamic reference model or diagnostic rule set adjustment; the weight coefficient in the weighted fusion algorithm is determined by the corresponding mutual information value.

[0054] In some embodiments, the adjustment module 500 performs the following when adjusting the judgment threshold or reference baseline corresponding to the key feature parameter using the fused data: According to the pre-established dynamic reference model, the fusion data is used to adjust the judgment threshold or reference baseline corresponding to the key feature parameters.

[0055] In some embodiments, the adjustment module 500 is configured to adjust the judgment threshold or reference baseline corresponding to the key feature parameter using fusion data according to a pre-established dynamic reference model. Obtain the current fusion data in real time, calculate the Mahalanobis distance between the current fusion data and the cluster centers of each normal working condition, and select the cluster center corresponding to the minimum Mahalanobis distance; Obtain the covariance matrix corresponding to the selected cluster center through the dynamic reference model, and calculate the confidence interval of the current fusion data based on the covariance matrix; The confidence interval is used as the judgment threshold or reference baseline of the key characteristic parameters for adaptive adjustment. If the current key characteristic parameters exceed the confidence interval, it is judged as an abnormal state, otherwise it is judged as a normal state. The adjustment of the confidence interval includes: when a sudden change in the environmental parameters is detected, the confidence interval is increased to reduce the false alarm rate; when the equipment operation status is stable, the confidence interval is reduced to improve the detection sensitivity.

[0056] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any actual relationship or order between these entities or operations.

[0057] The foregoing description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A ground wire monitoring method, applied to power equipment in a substation, characterized in that: The following steps are involved: Inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the smart ground wire and the ground grid, and synchronously collect the voltage signal at the injection point and the current signal of the ground loop; Preprocessing the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; Based on the preprocessed voltage and current signals, the impedance spectrum of the ground loop at different frequencies is calculated, and the key characteristic parameters in the impedance spectrum are extracted; Collect environmental parameters and operating status data of power equipment and fuse them with key characteristic parameters to obtain fused data; Using the fused data, adjust the judgment threshold or reference baseline corresponding to the key feature parameter; the judgment threshold is the numerical limit used to distinguish normal from abnormal states; the reference baseline is the normal value or range of the corresponding key feature parameter; The current ground loop status is determined using the adjusted judgment threshold or reference baseline.

2. The ground wire monitoring method according to claim 1, characterized in that: Also includes the steps: Record and store historical impedance spectrum characteristic parameters, environmental parameters and equipment status information; By conducting long-term trend analysis on historical data, we can obtain the rate and direction of change of key characteristic parameters over time; According to the rate and direction of change of key characteristic parameters over time, the insulation degradation rate of power equipment is evaluated, and based on the evaluation results, early warning is issued before the power equipment fails.

3. The ground wire monitoring method according to claim 1, characterized in that: The steps of preprocessing the collected voltage signal and current signal to obtain the preprocessed voltage signal and current signal include: The collected voltage and current signals are cleaned using a digital notch filter or Fourier analysis method to suppress the power frequency and its harmonics, thereby obtaining the primary processed voltage and current signals. For the voltage and current signals after primary processing, the time-frequency analysis method is used to identify and remove transient high-amplitude components to obtain the voltage and current signals after secondary processing; For the secondary processed voltage signal, a plurality of voltage values ​​are continuously collected from the secondary processed voltage signal according to a preset first measurement period, and all voltage values ​​of the same first measurement period are averaged to obtain an average voltage value corresponding to the first measurement period, and the average voltage values ​​obtained from all different first measurement periods are used as the tertiary processed voltage signal to reduce random noise of the voltage signal and as the final preprocessed voltage signal, or the secondary processed voltage signal is directly subjected to Kalman filtering to reduce random noise of the voltage signal, to obtain the tertiary processed voltage signal, and as the final preprocessed voltage signal; For the current signal after secondary processing, multiple current values ​​are continuously collected from the current signal after secondary processing according to a preset second measurement period, and all current values ​​of the same second measurement period are averaged to obtain the average current value corresponding to the second measurement period. The average current value obtained from all different second measurement periods is used as the current signal after tertiary processing to reduce the random noise of the current signal and as the final preprocessed current signal, or the current signal after secondary processing is directly Kalman filtered to reduce the random noise of the current signal, and the current signal after tertiary processing is obtained, and as the final preprocessed current signal.

4. The ground wire monitoring method according to claim 1, characterized in that: Key characteristic parameters include the impedance amplitude and phase at a specific frequency point or frequency band, the slope and curvature of the impedance spectrum curve within a specific frequency range, the frequency, amplitude and bandwidth of the resonance peak or antiresonance peak in the spectrum, and the spectral energy distribution in a specific frequency band.

5. The ground wire monitoring method according to claim 1, characterized in that: The current ground loop status includes normal, early deterioration, and different degrees of leakage.

6. The ground wire monitoring method according to claim 4, characterized in that: The steps of collecting environmental parameters and operating status data of power equipment and fusing them with key characteristic parameters to obtain fused data include: Collect environmental parameters of power equipment, including temperature, humidity, electromagnetic interference intensity, and operating status data, including voltage, current, active power, and reactive power; Wavelet threshold denoising method is used to filter out noise interference of environmental parameters and operating status data, and an interpolation algorithm based on historical data is used to fill in missing values ​​to obtain a clean and complete data set; The mutual information method is used to evaluate the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter, thereby obtaining an evaluation result on the correlation between each environmental parameter and each operating status data relative to each key characteristic parameter. The environmental parameters and operating status data having a correlation greater than a preset threshold are screened out based on the evaluation result to obtain a screened parameter set. The Min-Max scaling method is used to normalize the environmental parameters and operating status data in the filtered parameter set; The normalized environmental parameters and operating status data are fused with the key feature parameters through a weighted fusion algorithm to obtain a fused multi-dimensional feature vector, which is used as the fused data.

7. The ground wire monitoring method according to claim 1, characterized in that: The steps for adjusting the judgment threshold or reference baseline corresponding to the key feature parameters using the fused data include: According to the pre-established dynamic reference model, the fusion data is used to adjust the judgment threshold or reference baseline corresponding to the key feature parameters.

8. The ground wire monitoring method according to claim 7, characterized in that: The steps to establish a dynamic reference model include: Obtain historical operating data, including historical impedance spectrum characteristic parameters, historical environmental parameters, and historical equipment status information; Based on the historical operating data, a clustering algorithm is used to perform cluster analysis on the data under normal operating conditions to obtain multiple normal operating condition cluster centers, and a dynamic reference model under normal operating conditions is constructed by calculating the covariance matrix of each cluster center.

9. The ground wire monitoring method according to claim 8, characterized in that: The steps of adjusting the judgment threshold or reference baseline corresponding to the key feature parameters using fused data based on the pre-established dynamic reference model include: Obtain the current fusion data in real time, calculate the Mahalanobis distance between the current fusion data and the cluster centers of each normal working condition, and select the cluster center corresponding to the minimum Mahalanobis distance; Obtain the covariance matrix corresponding to the selected cluster center through the dynamic reference model, and calculate the confidence interval of the current fusion data based on the covariance matrix; The confidence interval is used as a judgment threshold or reference baseline for key feature parameters and is adaptively adjusted. If the current key feature parameters exceed the confidence interval, it is determined to be an abnormal state, otherwise it is determined to be a normal state. The adjustment of the confidence interval includes: when a sudden change in environmental parameters is detected, the confidence interval is increased to reduce the false alarm rate; when the equipment operation state is stable, the confidence interval is reduced to improve the detection sensitivity.

10. A ground wire monitoring system, applied to power equipment in a substation, characterized in that: include: An acquisition module is used to inject an excitation signal containing multiple discrete frequency points or a swept frequency signal into the ground loop formed by the intelligent ground wire and the ground grid, and synchronously acquire the voltage signal at the injection point and the current signal of the ground loop; A preprocessing module is used to preprocess the collected voltage signal and current signal to obtain preprocessed voltage signal and current signal; An extraction module is used to calculate the impedance spectrum of the ground loop at different frequencies based on the preprocessed voltage signal and current signal, and extract key characteristic parameters from the impedance spectrum; The fusion module is used to collect environmental parameters and operating status data of power equipment and fuse them with key characteristic parameters to obtain fused data; An adjustment module is used to adjust the judgment threshold or reference baseline corresponding to the key feature parameter using the fused data; the judgment threshold is the numerical limit used to distinguish between normal and abnormal states; the reference baseline is the normal value or range of the corresponding key feature parameter; The judgment module is used to judge the current ground loop state by using the adjusted judgment threshold or reference baseline.

Images