Insulator defect detection method and device and storage medium

Abstract

The application relates to the technical field of power system state monitoring, and discloses an insulator defect detection method and device and a storage medium, which comprise the following steps: collecting insulator leakage current, constructing a polarity conduction screening model by using power frequency phase information to generate a composite current sequence highlighting interface polarization properties; analyzing the nonlinear degree of the sequence attenuation rate drift with evolution time in a multi-stage diffusion evolution time domain to generate an interface electrochemical mass transfer composite index; inversely constructing an interface polarization background benchmark by using the index and stripping from the composite current sequence to extract abnormal discharge pulse residuals; and finally quantifying the pulse shape distortion degree of the residuals and performing inhibitory normalization operation on the distortion degree by using the index to output a judgment result. The application effectively eliminates the interference of the interface mass transfer process and realizes specific detection of a pollution flashover hidden danger.

Description

Technical Field

[0001] This invention relates to the field of power system condition monitoring technology, and more specifically, to insulator defect detection methods, devices, and storage media. Background Technology

[0002] As a core component of the external insulation of transmission lines, the operational reliability of porcelain insulators is crucial to the safety and stability of the power grid. In hot and humid coastal areas or heavily polluted environments, porcelain insulators coated with semi-conductive glaze are often used in engineering to improve the electric field distribution and suppress surface condensation. The weak conductivity of the glaze layer generates a thermal effect to maintain surface dryness. Currently, monitoring the operational status of such equipment typically relies on online acquisition of surface leakage current signals, and mainly uses statistical indicators such as the current amplitude or total harmonic distortion rate as the primary basis for judging the insulation condition or assessing the risk of flashover.

[0003] However, ceramic insulators are structurally composites made of various materials, including ceramic bodies, glaze layers, and cementitious adhesives. In hot and humid marine climates, the interface between the steel foot and the cementitious adhesive easily forms microscopic corrosion cells, evolving into a porous electrochemical system. The electrochemical mass transfer current generated at this interface typically follows a diffusion or fractional-order decay law that evolves over time. Combined with the controlled continuous background current inherent in the semi-conductive glaze, the total leakage current becomes a product of multiple cascading mechanisms. Existing monitoring technologies struggle to effectively distinguish this slow-varying component caused by the physicochemical properties of the material interface from genuine dry-charge arc discharge signals. This leads to frequent misjudgments when faced with complex mixed current signals, relying solely on amplitude or harmonic proportions, resulting in missed detections or false alarms. Summary of the Invention

[0004] This invention provides a method, apparatus, and storage medium for detecting insulator defects, which solves the technical problems mentioned in the background art.

[0005] Firstly, methods for detecting insulator defects include:

[0006] Leakage current signals of insulators are collected, and a polarity conduction screening model is constructed using the power frequency phase characteristics of the leakage current signals. The leakage current signals are then subjected to asymmetric weighting to generate a composite current sequence characterizing the polarization properties of the solid-liquid interface.

[0007] In the multi-stage diffusion evolution time domain, the nonlinearity of the decay rate of the composite current sequence drift with evolution time is analyzed to generate the interfacial electrochemical mass transfer composite index.

[0008] An interfacial polarization background benchmark is constructed by inverting the interfacial electrochemical mass transfer recombination index, and the interfacial polarization background benchmark is extracted from the recombination current sequence to extract the residual of abnormal discharge pulses.

[0009] The pulse shape distortion of the abnormal discharge pulse residual is quantified, and the pulse shape distortion is normalized by the interfacial electrochemical mass transfer recombination index to output the judgment result of the insulator flashover hazard.

[0010] Secondly, an insulator defect detection device, applied in any of the insulator defect detection methods described above, includes:

[0011] A composite current sequence generation module is used to collect leakage current signals of insulators, construct a polarity conduction screening model using the power frequency phase characteristics of the leakage current signals, perform asymmetric weighting processing on the leakage current signals, and generate a composite current sequence characterizing the polarization characteristics of the solid-liquid interface.

[0012] The interface electrochemical mass transfer recombination index generation module is used to analyze the nonlinearity of the decay rate of the recombination current sequence as a function of evolution time in the multi-stage diffusion evolution time domain, and generate the interface electrochemical mass transfer recombination index.

[0013] An abnormal discharge pulse residual extraction module is used to construct an interface polarization background benchmark by inverting the interface electrochemical mass transfer recombination index, and to extract the abnormal discharge pulse residual from the recombination current sequence.

[0014] The pollution flashover hazard determination module is used to quantify the pulse shape distortion degree of the abnormal discharge pulse residual, and to perform a suppressive normalization operation on the pulse shape distortion degree using the interfacial electrochemical mass transfer recombination index, and output the determination result of the insulator pollution flashover hazard.

[0015] Thirdly, an insulator defect detection storage medium includes: a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the insulator defect detection method as described in any one of the claims.

[0016] The beneficial effects of this invention include: by constructing a polarity conduction screening model and an interfacial electrochemical mass transfer composite index, this invention achieves the quantification of complex interfacial dynamics, thereby enabling the precise removal of the diffusion background reference from the mixed leakage current, and effectively suppressing environmental noise interference by utilizing suppressive normalization operations, significantly improving the accuracy and robustness of identifying weak hidden danger signals such as early dry arcing. Attached Figure Description

[0017] Figure 1 This is a flowchart of the insulator defect detection method of the present invention;

[0018] Figure 2 This is a schematic diagram of the insulator of the present invention. Detailed Implementation

[0019] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0020] Example 1: As Figure 1 As shown, the insulator defect detection method includes:

[0021] Leakage current signals of insulators are collected, and a polarity conduction screening model is constructed using the power frequency phase characteristics of the leakage current signals. The leakage current signals are then subjected to asymmetric weighting to generate a composite current sequence characterizing the polarization properties of the solid-liquid interface.

[0022] In the multi-stage diffusion evolution time domain, the nonlinearity of the decay rate of the composite current sequence drift with evolution time is analyzed to generate the interfacial electrochemical mass transfer composite index.

[0023] An interfacial polarization background benchmark is constructed by inverting the interfacial electrochemical mass transfer recombination index, and the interfacial polarization background benchmark is extracted from the recombination current sequence to extract the residual of abnormal discharge pulses.

[0024] The pulse shape distortion of the abnormal discharge pulse residual is quantified, and the pulse shape distortion is normalized by the interfacial electrochemical mass transfer recombination index to output the judgment result of the insulator flashover hazard.

[0025] In a preferred embodiment, leakage current signals from the insulator are acquired, and a polarity conduction screening model is constructed using the power frequency phase characteristics of the leakage current signals. The leakage current signals are then subjected to asymmetric weighting processing to generate a composite current sequence characterizing the polarization properties of the solid-liquid interface, including:

[0026] The collected leakage current signal is standardized using the following formula to obtain a standardized current sequence. :

[0027] ;

[0028] in, This is the original leakage current signal. The number of sampling points. for The absolute median difference It is a tiny positive number;

[0029] right Narrowband filtering with a center frequency of power frequency is performed to obtain And using Hilbert transform Calculate analytic signals Then, the fundamental phase of the power frequency is extracted. ;

[0030] Construct a polarity conduction screening model using the following formula. :

[0031] ;

[0032] in, For polarization steepness parameters, It is the hyperbolic tangent function;

[0033] The composite current sequence is generated according to the following formula. :

[0034] .

[0035] First, signal acquisition and standardization preprocessing are performed. This takes into account the significant differences in the absolute amplitude of insulator leakage current under different pollution levels and voltage levels, and the original leakage current signal... The data often contains DC components and random noise, so standardization is required.

[0036] Specifically, according to the formula:

[0037] ;

[0038] The normalized current sequence was calculated. In this formula, This represents the original leakage current signal, which is acquired and discretized by the leakage current sensor. For discrete-time indexing, This represents the total number of sampling points in a single sampling, for example, when the sampling frequency is 10kHz and the sampling duration is 8 seconds. The value is set to 80,000 points. (Molecular part) Its function is to remove the DC bias component from the signal, ensuring that the signal is zero-mean. The denominator contains... Original leakage current signal The absolute median difference is calculated by first calculating... Find the median, then calculate the absolute value of the difference between all data points and the median, and finally take the median of these absolute values. The normalization factor is used because insulator leakage current often contains short-duration, high-amplitude spikes caused by dry-circuit arcing. Traditional standard deviation is extremely sensitive to outliers and easily inflated by arc spikes, resulting in an excessively small background signal after normalization. In contrast, absolute median has strong robustness, more accurately characterizing the main fluctuation level of the signal, and is unaffected by individual arc pulses. (The formula contains...) For a small positive number, the preferred value is This is used to prevent the signal from becoming completely zero or extremely stable. A division-by-zero error occurs when the result is zero, ensuring the stability of the calculation.

[0039] After standardization, the next step is to extract the power frequency phase characteristics. Because the corrosion electrochemical reaction at the steel foot-cement interface has a polarity-dependent characteristic (i.e., the rates of anodic dissolution and cathodic reduction are asymmetrical), it is necessary to extract the power frequency phase.

[0040] In practice, the first step is to standardize the current sequence. Narrowband filtering with a center frequency of power frequency (usually 50Hz or 60Hz) is performed, preferably using a bandwidth of [missing information]. The Butterworth bandpass filter removes high-frequency harmonics and noise, resulting in a signal containing only the fundamental frequency component. .

[0041] Subsequently, using the Hilbert transform (denoted as...) Constructing analytical signals The calculation formula is:

[0042] ;

[0043] in The imaginary unit is used. The instantaneous phase is calculated using the ratio of the imaginary to the real part of the analytic signal, thus extracting the fundamental frequency phase.

[0044] ;

[0045] And perform the necessary phase unwrapping process, so that It can accurately reflect the evolution of the positive and negative half-cycles of the voltage cycle (when the leakage current is mainly resistive, the fundamental current phase is approximately in phase with the voltage phase). Based on the extracted power frequency fundamental phase... Construct a polarity conduction screening model The polarity-based conduction screening model is used to apply an asymmetric soft switch to the signal in the time domain to highlight the interface polarization effect under a specific polarity. The calculation formula is as follows:

[0046] .

[0047] In the formula, The normalized waveform of the fundamental wave was restored; The hyperbolic tangent function is used to map a sinusoidal waveform into a square-like wave with saturation characteristics; coefficients With constant Its function is to expand the output range of the hyperbolic tangent function from... Pan and scale to The interval is used to form a gating coefficient. The parameters in the formula... This is a polarization steepness parameter used to control the stiffness or selectivity of the gate switch. Specifically, if... Values ​​that are too small (e.g.) ), It degenerates into a normal sine wave and cannot effectively suppress the unrelated half-cycle; if Values ​​that are too large (e.g.) ), Approaching an ideal square wave introduces high-frequency artifacts such as the Gibbs phenomenon in the frequency domain. Based on extensive field data testing and matching studies of electrochemical interface polarization characteristics, The preferred value is set to 4. When At that time, polarity conduction screening model While preserving the smooth transition characteristics, it can quickly increase the weight of the half-cycle of interest (the half-cycle in which interface corrosion occurs) to close to 1, while quickly suppressing the weight of the half-cycle of non-interest to close to 0, thereby achieving asymmetric screening of the polarization characteristics of the solid-liquid interface.

[0048] Finally, a composite current sequence characterizing the polarization properties of the solid-liquid interface is generated. The calculation formula is:

[0049] ;

[0050] Screening model using the absolute value and polarity of the normalized current. Perform point-by-point multiplication. Specifically, utilize the constructed polarity gating... As a mask, the energy amplitude of the original signal is weighted in the time domain. Because The current approaches 1 in a specific half-cycle (e.g., the positive half-cycle) and approaches 0 in the other half-cycle, thus generating a composite current sequence. Essentially, it is a unipolar signal processed by a polarization filter. The unipolar signal filters out the interference of the symmetrical part in the pure ohmic conductivity background, and retains and highlights the asymmetric component that is strongly correlated with the electrochemical polarization process at the steel foot-cement interface.

[0051] In a preferred embodiment, within the multi-stage diffusion evolution time domain, the nonlinearity of the decay rate of the recombination current sequence drifting with evolution time is analyzed to generate an interfacial electrochemical mass transfer recombination index, including:

[0052] Using a sliding window with a length of quantiles The quantile filter extracts the composite current sequence steady-state envelope The wetting trigger time is determined according to the following formula. :

[0053] ;

[0054] ;

[0055] in, For the trigger point index, The sampling frequency;

[0056] Build with Starting with the multi-level diffusion evolution time domain, generate logarithmic step sizes. Time nodes of growth And calculate the first Interval aggregation magnitude within a logarithmic evolution interval ;

[0057] Calculate the number using the following formula. Local decay rate of each interval :

[0058] ;

[0059] in, For the first Interval aggregation magnitude within a logarithmic evolution interval For the first Interval aggregation magnitude within a logarithmic evolution interval For the first stage of multi-stage diffusion evolution in the time domain Each time point For the first stage of multi-stage diffusion evolution in the time domain Each time point;

[0060] Calculate the number using the following formula. Nonlinear drift degree of each interval :

[0061] ;

[0062] in, For the first The local decay rate over a logarithmic evolution interval. For the first Local decay rate over a logarithmic evolution interval;

[0063] The interfacial electrochemical mass transfer recombination index is calculated using the following formula. :

[0064] ;

[0065] in, This represents summing over all valid intervals. This indicates taking the absolute value.

[0066] First, in order to extract from a composite current sequence containing high-frequency noise and potential dry-band arc pulses... The slow-changing trend that reflects the interfacial mass transfer process is extracted, and the steady-state envelope is extracted using nonlinear filtering technology. Specifically, using a sliding window with a length of... quantiles quantile filter pair Process it.

[0067] During the calculation, for each time point Selecting an interval All inside Value, calculate its first The quantile is used as the envelope value at that point. Specifically, the parameter... The preferred setting is the number of data points corresponding to 4 power frequency cycles (e.g., sampling frequency). hour, This window length is sufficient to smooth out power frequency fluctuations while preserving diffusion trends on a second-level scale; parameters The preferred setting is 0.20 (i.e., 20th percentile) because the interference in the leakage current is mainly manifested as positive arc spikes. Selecting a lower percentile can suppress upward-distorted pulse interference, thereby locking in the bottom base current dominated by micropore filling and ion migration.

[0068] Obtaining the steady-state envelope Next, it is necessary to determine the zero moment of the diffusion process, i.e., the wetting trigger moment. According to the formula:

[0069] ;

[0070] Calculate the trigger point index This formula finds the moment when the first derivative (growth rate) of the envelope is maximized, corresponding to the instant when the water film on the insulator surface or interface connects and ion channels rapidly form. Then, through... Convert discrete indexes to time, where The sampling frequency is [value]. After establishing the starting point, a [structure / system] is constructed based on [the following]. The multi-stage diffusion evolution time domain starts from this point. Electrochemical diffusion processes (such as Cottrell behavior) and interfacial polarization (such as constant-phase element CPE behavior) typically follow a power-law decay over time. t represents the time variable. Indicates the decay index, This power-law decay pattern exhibits a linear relationship in a logarithmic coordinate system. Therefore, in order to capture features uniformly across the entire time scale, a logarithmic step size is generated. Time nodes of growth Logarithmic step size The preferred value is That is, the time scale is doubled for every 8 points to ensure sufficient resolution. The time axis is divided into several logarithmic evolution intervals. And calculate the first Interval aggregation magnitude within a logarithmic evolution interval . Usually, all values ​​within this interval are taken. The median value is used to further improve noise immunity.

[0071] Subsequently, the calculation of the first Local decay rate of each interval The formula is:

[0072] ;

[0073] This formula uses the central difference method in the logarithmic field to calculate the local slope. This directly reflects the power-law exponent at the current timescale. If the process is an ideal semi-infinite diffusion, then... If it is the ideal CPE behavior, then constant ( However, in the complex steel-cement interface of porcelain insulators, there is often competition and coupling of multiple mass transfer mechanisms, leading to... It is not a constant, but rather varies with the time scale.

[0074] To quantify this change, further calculations are performed. Nonlinear drift degree of each interval The formula is:

[0075] ;

[0076] It is the second derivative (i.e., curvature) of the logarithmic field, which characterizes the rate of drift of the power-law exponent with scale.

[0077] Finally, according to the formula:

[0078] ;

[0079] Calculate the interfacial electrochemical mass transfer recombination index .in, This represents the summation over all valid intervals (excluding the first and last boundary points). This indicates taking the absolute value. The interfacial electrochemical mass transfer recombination index is a total measure of the degree of bending of the leakage current in a double logarithmic coordinate system. The larger the value, the more the current waveform deviates from a single ideal fractal or diffusion model, that is, there is significant structural multi-level pore mass transfer or complex electrochemical reaction switching at the interface, corresponding to the potential corrosion risk at the ceramic insulator interface.

[0080] In a preferred embodiment, the interfacial polarization background benchmark is constructed by inverting the interfacial electrochemical mass transfer recombination index, and the interfacial polarization background benchmark is stripped from the recombination current sequence to extract the residual of abnormal discharge pulses, including:

[0081] Estimate the fractional decay exponent of the interface using the following formula. :

[0082] ;

[0083] in, The local decay rate is... The total number of the logarithmic evolution intervals. This indicates taking the median;

[0084] Constructing the classic diffusion response term and interface fractional response items , For the first stage of multi-stage diffusion evolution in the time domain There are several time points, and the coefficients are solved using the least squares method. and To fit the aggregated amplitude of the interval The interface polarization background reference is obtained. :

[0085] ;

[0086] The residual of the abnormal discharge pulse is calculated using the following formula. :

[0087] ;

[0088] in, This indicates the operation of obtaining non-negative values. It is a small positive number.

[0089] First, in order to accurately remove the background current caused by corrosion at the steel-cement interface, it is necessary to determine the characteristic index governing the long-term electrochemical behavior of this interface, namely the fractional-order decay index of the interface. The calculation does not use data from the entire time period, but rather estimates are performed using the following formula:

[0090] ;

[0091] in, For local decay rate, The total number of logarithmic evolution intervals. Indicates taking The floor value, This indicates the median selection operation. Specifically, in the early stages of multi-stage diffusion evolution (i.e., when the evolution time is short), the current signal is often mixed with the double-layer charging current and the ohmic polarization establishment process, and the local decay rate fluctuates drastically at this time. However, in the later stages of the evolution time domain (i.e., the last 1 / 3 of the time interval), the current response has entered the steady-state diffusion region limited by micropore mass transfer. At this point, the power-law exponent best reflects the intrinsic fractal structure dimension of the material interface. The median is selected to prevent individual measurement noise points from biasing the exponent estimation.

[0092] After identifying the key indicators Then, two basis functions are constructed to describe the background process: the classical diffusion response term. and interface fractional response items Construct the first basis using the following formula:

[0093] ;

[0094] This term corresponds to the time-domain response of the classical Cottrell equation or Warburg diffusion impedance, characterizing an ideal semi-infinite linear diffusion process.

[0095] Construct the second basis according to the following formula:

[0096] ;

[0097] This corresponds to a fractional-order diffusion process dominated by a porous, rough interface (constant-phase element CPE). At this point, the observed interval aggregation amplitude... It is formed by the linear superposition of these two mechanisms, and the coefficients are solved using the least squares method. and This minimizes the sum of squared fitting errors, thus obtaining the interface polarization background benchmark. The calculation formula is:

[0098] .

[0099] This formula constructs a virtual health baseline, assuming the current leakage current is generated by moisture transport and interfacial electrochemical corrosion, but does not include abnormal discharge components such as dry-charge arcing. (Coefficient) and These represent the weighted contributions of the classical diffusion channel and the fractional-order interface channel to the total current, respectively.

[0100] Finally, differential operations are performed to extract the residual of the abnormal discharge pulse. The calculation formula is:

[0101] ;

[0102] In the formula, This represents the aggregated amplitude of the observed intervals. The theoretical background benchmarks constructed for the inversion. Utilizing The operator is because a dry-charged arc or partial discharge is an additional energy injection process, and the current it generates is necessarily superimposed on the background current, leading to an effect on the observed value. Greater than the theoretical benchmark value ;like Less than This is usually attributed to measurement noise or fitting error and is not a potential discharge hazard, so it is set to 0. Through the above operations, the complex electrochemical background is stripped from the mixed signal, and the pure residual signal associated only with the precursor of flashover (i.e., abnormal discharge) on the insulator surface is extracted.

[0103] In a preferred embodiment, the pulse shape distortion of the abnormal discharge pulse residual is quantified, and the pulse shape distortion is subjected to suppressive normalization using the interfacial electrochemical mass transfer recombination index. The result of the insulator flashover hazard assessment is then output, including:

[0104] The residual of the abnormal discharge pulse is calculated using the following formula. Local residual decay rate and residual curvature :

[0105] ;

[0106] in, For the first Abnormal discharge pulse residuals over a logarithmic evolution interval For the first Abnormal discharge pulse residuals within a logarithmic evolution interval;

[0107] ;

[0108] in, Let be the local decay rate of the residual of the abnormal discharge pulse in the (m+1)th logarithmic evolution interval. The local decay rate of the residual of the abnormal discharge pulse in the (m-1)th logarithmic evolution interval is given by .

[0109] The pulse morphological distortion is calculated using the following formula. :

[0110] ;

[0111] in, The step size is logarithmic.

[0112] Calculate the risk score for flashover hazards using the following formula. :

[0113] ;

[0114] in, The interfacial electrochemical mass transfer recombination index;

[0115] Will With the judgment threshold Comparison:

[0116] like The output result indicates that the insulator has a potential flashover risk.

[0117] like The output result indicates that there is no risk of flashover due to pollution insulators.

[0118] Obtaining the residual of the abnormal discharge pulse after removing the interfacial polarization background. Next, it is necessary to perform second-order differential analysis on its morphological characteristics in a double logarithmic coordinate system to identify the unique nonlinear abrupt change characteristics of the dry-charged arc. Specifically, the real dry-charged arc exhibits a pulse cluster with strong randomness and even chaotic characteristics in the time domain. Its residual curve in the double logarithmic scale shows severe local oscillations and non-smoothness, while the residual caused by fitting error or Gaussian white noise is relatively smooth.

[0119] Therefore, first calculate the first... Local residual decay rate of each interval The calculation formula is:

[0120] ;

[0121] The local residual decay rate represents the slope of the local change of the residual at the current time scale;

[0122] Next, calculate the first Residual curvature of each interval The calculation formula is:

[0123] ;

[0124] In the formula, For logarithmic growth time points, and These are the residual magnitudes of adjacent intervals. The sharpness of the residual curve was quantified. Then, according to the formula... Calculate pulse morphology distortion .in, This represents summing over all valid computation intervals. This indicates taking the absolute value. The preset logarithmic step size (preferred value is approximately 0.0866). By integrating the structural distortions of the residuals across all time scales, a scalar is accumulated; the larger this value, the more significant the nonlinear impulse component contained in the residuals. However, relying solely on... Directly determining that there is a risk of false alarms is problematic because when the steel-cement interface is in an extremely complex and active corrosion phase (i.e., the interfacial electrochemical mass transfer composite index calculated in the previous steps is...), Even without a dry arc, complex interface noise can cause fluctuations in the fitting residuals.

[0125] Therefore, an inhibitory normalization operation is introduced to calculate the risk score of pollution flashover hazard. :

[0126] ;

[0127] In the formula, The interfacial electrochemical mass transfer recombination index is given by the denominator. This constitutes the adaptive penalty factor. Specifically, when When the value is large, it means that the background current itself has extremely high multi-scale complexity. In this case, the system automatically reduces its sensitivity to residual distortion. The sensitivity of the S-value (i.e., the suppression of the S-value) is used to avoid misjudging complex interface noise as a potential flashover risk; conversely, when When the value is small (close to 0), it means that the background is very clean and conforms to the ideal diffusion model. At this time, the denominator is close to 1, and the system maintains full sensitivity to residual distortion, and can capture weak early arc signals.

[0128] Finally, the calculated Compared with the preset judgment threshold A comparison was made. Based on a large amount of data from net-hanging operations and artificial pollution experiments conducted in coastal high-humidity environments, the receiver operating characteristic (ROC) curves were analyzed. The preferred value is set to 0.35. If... If the logic output indicates a potential flashover risk in the insulator, it prompts maintenance personnel to intervene; if If the output insulator does not pose a risk of flashover, it is considered to be in a safe state.

[0129] like Figure 2As shown, taking insulators in operation as the object, the insulator surface forms a conductive channel along the creepage distance under the action of a pollution layer and a wet water film, generating leakage current. After local heating and evaporation, a dry strip is formed, and a dry strip arc may appear at the dry strip, which is a precursor stage of pollution flashover. This mechanism is consistent with the engineering understanding of online leakage current monitoring: leakage current increases with pollution wetting and may induce a dry strip arc, which can further develop into flashover. A leakage current sensor is clamped at the insulator grounding terminal hardware and electrically connected to a detection unit. The detection unit samples, stores, and processes the sensor output to obtain a discrete leakage current signal. Subsequently, the detection unit performs DC removal and robust normalization based on absolute median difference on the discrete leakage current signal to reduce the impact of outliers such as arc spikes on scale estimation, and introduces a small positive number to ensure stability. Based on this, the standardized current sequence is subjected to power frequency narrowband filtering (center frequency is power frequency, preferably Butterworth bandpass with bandwidth ±2 Hz), and the power frequency fundamental phase is extracted using Hilbert transform. A polarity conduction screening model is constructed based on the power frequency fundamental phase, and the signal is asymmetrically weighted to obtain a composite current sequence characterizing the polarization properties of the solid-liquid interface, highlighting the asymmetric component related to the polarity dependence of interfacial corrosion. Furthermore, to extract the slow-changing trend reflecting the interfacial mass transfer process from the composite current sequence, a sliding quantile filter is used to obtain the steady-state envelope. The sliding window is preferably the number of points corresponding to 4 power frequency cycles, and the quantile is preferably 0.20 to suppress the interference of positive arc spikes on the envelope's lifting effect. The wetting trigger time is determined by the point of maximum envelope growth rate, used to establish the zero-time of the diffusion process. Starting from the wetting trigger moment, a multi-level diffusion evolution time domain with logarithmic growth is constructed and segmented. The interval aggregation amplitude of each logarithmic interval is calculated (preferably the median). The decay rate and its nonlinear drift with scale are calculated in a double logarithmic coordinate system. The nonlinear drift is weighted and summed to obtain the interfacial electrochemical mass transfer recombination index, which is used to measure the multi-scale complexity of the background process. Subsequently, the detection unit uses the interfacial electrochemical mass transfer recombination index to invert and construct the interfacial polarization background benchmark: the fractional-order decay index of the interface is estimated, and the classical diffusion response term is linearly superimposed with the fractional-order response term of the interface. The background benchmark is obtained by least-squares fitting. Then, the background benchmark is extracted from the observation, and the residual of the abnormal discharge pulse is extracted. Non-negative truncation is used to meet the constraint that the dry arc is an additional energy injection and the residual should be superimposed on the background. Finally, the residual of the abnormal discharge pulse is quantified on a double logarithmic scale to determine its pulse shape distortion degree, and the pulse shape distortion degree is suppressed and normalized using the interfacial electrochemical mass transfer recombination index to obtain the pollution flashover risk score. The pollution flashover risk score is compared with a preset threshold. When the pollution flashover risk score is ≥ the preset threshold, a pollution flashover risk is output. When the pollution flashover risk score is < the preset threshold, a safe state is output.

[0130] Example 2: An insulator defect detection device, applied in any of the insulator defect detection methods described above, comprising:

[0131] A composite current sequence generation module is used to collect leakage current signals of insulators, construct a polarity conduction screening model using the power frequency phase characteristics of the leakage current signals, perform asymmetric weighting processing on the leakage current signals, and generate a composite current sequence characterizing the polarization characteristics of the solid-liquid interface.

[0132] The interface electrochemical mass transfer recombination index generation module is used to analyze the nonlinearity of the decay rate of the recombination current sequence as a function of evolution time in the multi-stage diffusion evolution time domain, and generate the interface electrochemical mass transfer recombination index.

[0133] An abnormal discharge pulse residual extraction module is used to construct an interface polarization background benchmark by inverting the interface electrochemical mass transfer recombination index, and to extract the abnormal discharge pulse residual from the recombination current sequence.

[0134] The pollution flashover hazard determination module is used to quantify the pulse shape distortion degree of the abnormal discharge pulse residual, and to perform a suppressive normalization operation on the pulse shape distortion degree using the interfacial electrochemical mass transfer recombination index, and output the determination result of the insulator pollution flashover hazard.

[0135] Example 3: An insulator defect detection storage medium, comprising: a computer program stored on the computer-readable storage medium, wherein the computer program, when executed by a processor, performs the insulator defect detection method as described in any one of the embodiments.

[0136] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A method for detecting defects in insulators, characterized in that, include: Leakage current signals from insulators are collected. A polarity conduction screening model is constructed using the power frequency phase characteristics of the leakage current signals. The leakage current signals are then subjected to asymmetric weighting processing to generate a composite current sequence characterizing the polarization properties of the solid-liquid interface, including: The collected leakage current signal is de-DC processed, and the amplitude is normalized using a statistic based on the absolute median difference to obtain a standardized current sequence. The standardized current sequence is then subjected to narrowband filtering to extract the power frequency fundamental component. The instantaneous phase of the power frequency fundamental component is analyzed using Hilbert transform to obtain the power frequency fundamental phase. A polarity conduction screening model is constructed using the mapping relationship between the hyperbolic tangent function and the sine value of the phase of the fundamental power frequency wave, wherein a polarization steepness parameter is introduced to adjust the degree of asymmetry in the screening; the polarity conduction screening model is multiplied point by point with the absolute value of the standardized current sequence to generate the composite current sequence; Within the multi-stage diffusion evolution time domain, the nonlinearity of the decay rate of the recombination current sequence with time is analyzed to generate the interfacial electrochemical mass transfer recombination index, including: The composite current sequence is subjected to sliding quantile filtering to extract the steady-state envelope, and the discrete derivative of the steady-state envelope is calculated. The time corresponding to the maximum value of the discrete derivative is taken as the wetting trigger time. Starting from the wetting trigger time, a multi-level diffusion evolution time domain that grows logarithmically is constructed. The multi-level diffusion evolution time domain is divided into several logarithmic evolution intervals, and the median of the steady-state envelope in each logarithmic evolution interval is calculated to obtain the interval aggregation amplitude. In a double logarithmic coordinate system, the difference ratio of adjacent interval aggregation amplitudes with respect to evolution time is calculated to obtain the local decay rate. The difference ratio of adjacent local decay rates with respect to logarithmic evolution time is calculated to obtain the nonlinear drift degree. The absolute values ​​of the nonlinear drift degree are weighted and summed to obtain the interfacial electrochemical mass transfer recombination index. The interfacial polarization background benchmark is constructed by inverting the interfacial electrochemical mass transfer recombination index, and the interfacial polarization background benchmark is extracted from the recombination current sequence to extract the residual of abnormal discharge pulses. The pulse shape distortion of the abnormal discharge pulse residual is quantified, and the pulse shape distortion is normalized by the interfacial electrochemical mass transfer recombination index to output the judgment result of the insulator flashover hazard.

2. The insulator defect detection method according to claim 1, characterized in that, An interfacial polarization background benchmark is constructed by inverting the interfacial electrochemical mass transfer recombination index, and the interfacial polarization background benchmark is extracted from the recombination current sequence to extract the residual of abnormal discharge pulses, including: The median of the local decay rate corresponding to the later part of the multi-stage diffusion evolution time domain is selected as the interface fractional decay exponent. A classical diffusion response term is constructed using the negative half power of the evolution time, and an interface fractional response term is constructed using the negative power of the interface fractional decay exponent of the evolution time. The least squares method is used to fit the interval aggregated amplitude as a linear superposition of the classical diffusion response term and the interface fractional response term to obtain the interface polarization background reference. The difference between the interval aggregated amplitude and the interface polarization background reference is calculated, and the non-negative part of the difference is retained as the abnormal discharge pulse residual.

3. The insulator defect detection method according to claim 1, characterized in that, The pulse shape distortion of the abnormal discharge pulse residual is quantified, and the pulse shape distortion is normalized by the interfacial electrochemical mass transfer recombination index to suppress it. The result of the insulator flashover hazard assessment is then output, including: In a double logarithmic coordinate system, the absolute value of the rate of change of the local decay rate of the residual of the abnormal discharge pulse is calculated, and the absolute value is integrated in the time domain of the multi-stage diffusion evolution to obtain the pulse morphology distortion degree; the pulse morphology distortion degree is divided by the sum of the interfacial electrochemical mass transfer recombination index and the constant to obtain the pollution flashover risk score.

4. The insulator defect detection method according to claim 3, characterized in that, The method quantifies the pulse shape distortion of the abnormal discharge pulse residual, performs a suppressive normalization operation on the pulse shape distortion using the interfacial electrochemical mass transfer recombination index, and outputs the judgment result of the insulator flashover hazard. It also includes: The risk score of the pollution flashover hazard is compared with a preset judgment threshold. If the risk score of the pollution flashover hazard is greater than or equal to the judgment threshold, the judgment result is output as yes; otherwise, the judgment result is output as no.

5. An insulator defect detection device, applied in the insulator defect detection method according to any one of claims 1 to 4, characterized in that, include: A composite current sequence generation module is used to collect leakage current signals of insulators, construct a polarity conduction screening model using the power frequency phase characteristics of the leakage current signals, perform asymmetric weighting processing on the leakage current signals, and generate a composite current sequence characterizing the polarization characteristics of the solid-liquid interface. The interface electrochemical mass transfer recombination index generation module is used to analyze the nonlinearity of the decay rate of the recombination current sequence as a function of evolution time in the multi-stage diffusion evolution time domain, and generate the interface electrochemical mass transfer recombination index. An abnormal discharge pulse residual extraction module is used to construct an interface polarization background benchmark by inverting the interface electrochemical mass transfer recombination index, and to extract the abnormal discharge pulse residual from the recombination current sequence. The pollution flashover hazard determination module is used to quantify the pulse shape distortion degree of the abnormal discharge pulse residual, and to perform a suppressive normalization operation on the pulse shape distortion degree using the interfacial electrochemical mass transfer recombination index, and output the determination result of the insulator pollution flashover hazard.

6. An insulator defect detection storage medium, characterized in that, include: The computer-readable storage medium stores a computer program, which, when executed by a processor, performs the insulator defect detection method as described in any one of claims 1 to 4.

Images