Insulator state recognition method and device, equipment and storage medium

Abstract

Embodiments of the present application disclose a kind of insulator state identification method and device, equipment and storage medium, method includes: the leakage current signal of insulator is collected;Leakage current signal is carried out empirical mode decomposition, separates out power frequency component and high frequency component;Power frequency component is carried out discharge state discrimination based on zero-crossing detection and waveform integrity, to determine local arc discharge, long arc discharge or no significant discharge;High frequency component is carried out pulse event detection and statistics based on dynamic threshold, obtain pulse event statistical result, pulse event statistical result includes total number of pulse events and pulse average peak value;Based on discharge state and pulse event statistical result, determine the risk level of insulator by decision fusion logic, and trigger the early warning of corresponding level.The above-mentioned manner realizes the improvement of discharge state identification accuracy and early risk warning, significantly improves the accuracy and foresight of early warning.

Description

Technical Field

[0001] This invention relates to the field of insulator condition identification technology, and more particularly to an insulator condition identification method, apparatus, device, and storage medium. Background Technology

[0002] Pollution flashover of transmission line insulators is a major hidden danger threatening the safe operation of the power grid. Real-time monitoring of insulator pollution status and early, accurate risk warning are key technical problems that urgently need to be solved in the power system.

[0003] Currently, mainstream online monitoring methods primarily rely on the analysis of insulator leakage current. Existing technical solutions typically use single or simple combinations of characteristic parameters for threshold judgment, such as setting a fixed threshold for the effective value or peak value of leakage current for over-limit alarms; or counting the number of current pulses exceeding a preset amplitude per unit time. These methods can reflect the deterioration of the insulator's condition to some extent.

[0004] However, these existing technical solutions have significant drawbacks. Leakage current signals are a mixture of different physical processes, including both power frequency conduction current reflecting overall surface conductivity and transient high-frequency pulse current reflecting partial discharge activity. Current methods do not effectively separate these two distinct components, but instead directly process the mixed signal, leading to insufficient information utilization. As a result, early warning models struggle to distinguish between different types of discharge states (such as localized arcs and long arcs), are insensitive to high-frequency pulse groups caused by metal contamination that are low in amplitude but extremely harmful, and are prone to false alarms in complex environments such as humid conditions. Therefore, existing technologies cannot achieve refined and early warning of insulator flashover risks. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for identifying the state of insulators, which can solve the problem that the prior art cannot achieve a refined and early warning of the risk of flashover of insulator pollution.

[0006] To achieve the above objectives, a first aspect of the present invention provides a method for identifying the state of an insulator, the method comprising: Collect leakage current signals from insulators; Empirical mode decomposition is performed on the leakage current signal to separate the power frequency component and the high frequency component; The power frequency component is subjected to discharge state discrimination based on zero-crossing detection and waveform integrity to determine whether it is a local arc discharge, a long arc discharge, or no significant discharge. The high-frequency components are subjected to pulse event detection and statistics based on dynamic thresholds to obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses. Based on the discharge state and the statistical results of the pulse events, the risk level of the insulator is determined by decision fusion logic, and a corresponding level of early warning is triggered.

[0007] In one feasible implementation, performing empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component includes: Empirical mode decomposition is performed on the leakage current signal to obtain a series of intrinsic mode functions; Calculate the instantaneous frequency of each intrinsic mode function; The first intrinsic mode function whose instantaneous frequency stabilizes within the power frequency range is defined as the power frequency component; All intrinsic mode functions with instantaneous frequencies higher than the power frequency are defined as high-frequency components.

[0008] In one feasible implementation, the discharge state determination of the power frequency component based on zero-crossing detection and waveform integrity includes: Zero-crossing detection is performed on the power frequency component to mark the power frequency period; Within each power frequency cycle, the local variance of the signal is calculated. If the local variance exceeds a preset multiple of the average local variance, it is determined to be a distortion point. If the number of distortion points exceeds a preset threshold within one power frequency cycle, it is determined to be a local arc discharge. If the number of distortion points in one power frequency cycle does not exceed a preset threshold, then calculate the average value of the absolute value of the power frequency component. If the average value of multiple consecutive power frequency cycles exceeds the preset current threshold, it is determined to be a long arc discharge; otherwise, it is determined to be no significant discharge.

[0009] In one feasible implementation, the step of performing pulse event detection and statistics on the high-frequency components based on a dynamic threshold to obtain pulse event statistical results includes: Reconstruct all intrinsic mode functions with instantaneous frequencies higher than the power frequency into high-frequency signals; Calculate the moving average of the absolute value of the high-frequency signal within a predetermined time window; A dynamic threshold is set based on the moving average value; Local maxima exceeding the dynamic threshold in the high-frequency signal are detected, recorded as valid pulse events, and their peak values ​​are recorded. The total number of valid pulse events within the predetermined time window is counted, and the average value of all peak values ​​is calculated as the average peak value of the pulse.

[0010] In one feasible implementation, determining the risk level of the insulator based on the discharge state and the statistical results of the pulse event through decision fusion logic, and triggering a warning of the corresponding level, includes: If the discharge state is determined to be a long arc discharge, the highest risk level warning will be directly output. If the discharge state is determined not to be a long arc discharge, a risk score is calculated based on the discharge state and pulse event statistics, and the risk score is mapped to a risk level.

[0011] In one feasible implementation, the step of calculating a risk score based on the discharge state and pulse event statistics, and mapping the risk score to a risk level, includes: The discharge state is quantified into a state value: the state value of no significant discharge is less than the state value of local arc discharge; Calculate the pulse intensity factor, which is the logarithm of the total number of pulse events plus 1, and then multiply it by the average peak value of the pulses; Calculate the comprehensive risk score, which is the product of the state value and the target sum, where the target sum is the product of the preset coupling coefficient and the pulse intensity factor, plus 1.

[0012] In one feasible implementation, when collecting the leakage current signal, the sampling frequency is not less than 10kHz; the preset current threshold is 1mA; the predetermined time window is 1 minute; the consecutive multiple power frequency cycles are 5 consecutive power frequency cycles; and the preset number threshold is 3.

[0013] To achieve the above objectives, a second aspect of the present invention provides an insulator state identification device, the device comprising: The signal acquisition module is used to acquire the leakage current signal of the insulator; The signal separation module is used to perform empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component; The discharge discrimination module is used to perform discharge state discrimination on the power frequency component based on zero-crossing detection and waveform integrity to determine whether it is a local arc discharge, a long arc discharge, or no significant discharge. The pulse statistics module is used to perform pulse event detection and statistics on the high-frequency components based on dynamic thresholds, and obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses. The risk decision module is used to determine the risk level of the insulator based on the discharge state and the statistical results of the pulse event through decision fusion logic, and trigger the corresponding level of early warning.

[0014] To achieve the above objectives, a third aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0015] To achieve the above objectives, a fourth aspect of the present invention provides a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0016] The embodiments of the present invention have the following beneficial effects: This invention provides a method for identifying the state of an insulator. The method includes: acquiring a leakage current signal of the insulator; performing empirical mode decomposition on the leakage current signal to separate a power frequency component and a high-frequency component; performing discharge state discrimination on the power frequency component based on zero-crossing detection and waveform integrity to determine whether it is a local arc discharge, a long arc discharge, or no significant discharge; performing pulse event detection and statistics on the high-frequency component based on a dynamic threshold to obtain pulse event statistics, the pulse event statistics including the total number of pulse events and the average peak value of the pulses; and determining the risk level of the insulator based on the discharge state and the pulse event statistics through decision fusion logic and triggering a warning of the corresponding level.

[0017] This method adaptively separates the power frequency and high-frequency components of leakage current using empirical mode decomposition. Then, it distinguishes between localized and long-range arc discharges by performing zero-crossing and waveform integrity checks on the power frequency component. Simultaneously, it quantifies the intensity of partial discharges through dynamic threshold pulse statistics on the high-frequency component. Finally, it performs decision fusion based on the dual-path analysis results. This approach overcomes the shortcomings of traditional methods in handling mixed signals, enabling early warning judgments to be based on both macroscopic arc states and microscopic pulse activities. This significantly improves the accuracy and foresight of discharge state identification and early risk warnings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] in: Figure 1 This is a flowchart of an insulator state identification method according to an embodiment of the present invention; Figure 2This is another flowchart of an insulator state identification method according to an embodiment of the present invention; Figure 3 This is a structural block diagram of an insulator state identification device according to an embodiment of the present invention; Figure 4 This is a structural block diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 , Figure 1 This is a flowchart illustrating an insulator state identification method according to an embodiment of the present invention. This method can be applied to either a terminal or a server. The terminal can be a desktop terminal or a mobile terminal; a mobile terminal can be at least one of a mobile phone, tablet computer, or laptop computer. The server can be a standalone server or a server cluster composed of multiple servers. This embodiment uses a terminal application as an example. Figure 1 The method shown includes the following steps: 101. Collect leakage current signals from insulators; In this step, the leakage current signal of the operating insulator is acquired in real time using a high-precision current sensor. To ensure that no signal details are lost, the sampling frequency is set to no less than 10 kHz to capture high-frequency pulse components. For example, in one embodiment, a data acquisition system with a sampling rate of 20 kHz is used to continuously acquire current data of the insulator under different environmental conditions (such as humidity and dirt), and the signal sequence is denoted as X[n]. This high sampling rate helps to improve the accuracy of subsequent empirical mode decomposition and provides a reliable data basis for discharge state identification. Exemplarily, the original discrete-time domain signal sequence X[n] of the leakage current of the operating insulator is acquired in real time at a sampling frequency of no less than 10 kHz.

[0022] 102. Perform empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component; This step employs Empirical Mode Decomposition (EMD), an adaptive signal processing technique, to decompose the mixed leakage current signal into intrinsic mode functions (IMFs) at different frequency scales. Specifically, EMD is first performed on the original signal sequence X[n] to obtain a series of IMF components from high to low frequencies. Then, the instantaneous frequency of each IMF component is calculated: the instantaneous phase is obtained through Hilbert transform, and then differentiated to obtain the instantaneous frequency. The first IMF component whose instantaneous frequency stabilizes within the power frequency range (e.g., 50±2Hz) is defined as the power frequency component (denoted as IMF1), which mainly reflects surface conductivity and macroscopic arc characteristics. All IMF components with instantaneous frequencies significantly higher than 100 Hz (e.g., IMF2, IMF3, etc.) are linearly reconstructed into a composite high-frequency signal (denoted as Shigh[n], where Shigh[n] is the processing object in step 104) to characterize the partial discharge pulse. This separation method is highly adaptive, requires no preset basis functions, and effectively overcomes the limitations of traditional Fourier transform in non-stationary signal processing.

[0023] It is understood that step 102 includes: performing empirical mode decomposition on the leakage current signal to obtain a series of intrinsic mode functions; calculating the instantaneous frequency of each intrinsic mode function; defining the first intrinsic mode function whose instantaneous frequency is stable within the power frequency range as the power frequency component (denoted as IMF1); and defining all intrinsic mode functions whose instantaneous frequency is higher than the power frequency as high frequency components (such as IMF2, IMF3, etc.).

[0024] For example, an adaptive signal separation algorithm based on empirical mode decomposition is performed on X[n] to obtain the intrinsic mode function IMF1 representing the power frequency fundamental wave component and the set of intrinsic mode functions {IMF2, IMF3,...} representing the high-frequency pulse component.

[0025] 103. The power frequency component is subjected to discharge state discrimination based on zero-crossing detection and waveform integrity to determine whether it is a local arc discharge, a long arc discharge or no significant discharge; It should be noted that the discharge state of IMF1 is determined based on zero-crossing detection and waveform integrity: if a sudden interruption of the waveform at a non-zero-crossing point is detected within the power frequency cycle, it is determined to be a local arc discharge; if the waveform is continuous, the moving average of its absolute amplitude is calculated. If the average value continuously exceeds the current threshold (1mA), it is determined to be a long arc discharge; otherwise, it is determined to be no significant discharge.

[0026] It should be noted that numerous artificial contamination tests and some field data indicate that when the effective value of the power frequency component of the leakage current of the wet contamination layer on the surface of the insulator remains stable in the range of 0.5 mA to 1.5 mA, it usually indicates that the surface dry zone has basically formed and reached the critical width, and the local arc begins to burn unstablely and transitions to the stage of long arc development. The system is dynamically fine-tuned based on ambient humidity to improve the adaptability of the criteria, as follows: Baseline conditions: In the typical flashover-prone humidity range of 80% to 95% relative humidity (RH), a standard current threshold of 1 mA is used; High humidity correction: When RH > 95%, the contaminant layer is saturated with moisture, and the surface resistivity is extremely low, requiring a larger current to maintain a long arc. In this case, to avoid false alarms, the current threshold is raised to 1.2 mA ~ 1.5 mA; Medium and low humidity correction: When RH is between 70% and 80%, the contaminant layer is not sufficiently wetted, and the critical current value required to form a stable long arc may be slightly lower. To provide early warning, the current threshold can be cautiously lowered to 0.8 mA ~ 0.9 mA; Dry conditions: When RH < 70%, the conditions for determining long arc discharge (continuous waveform) are usually difficult to meet, and the system mainly relies on high-frequency pulse characteristics for early warning. This threshold is not currently enabled.

[0027] This step involves a detailed analysis of the power frequency component IMF1. First, zero-crossing detection is performed on IMF1, marking each complete power frequency cycle (e.g., a 20ms cycle for a 50Hz system). Within each power frequency cycle, the local variance of the signal is calculated (e.g., using a sliding window to calculate the variance within the neighborhood of each sample point). If the local variance at a sample point exceeds a preset multiple (e.g., 5 times) of the average local variance within that cycle, that point is identified as a distortion point, indicating that the waveform may be interrupted due to a local arc. If the number of distortion points within a power frequency cycle exceeds a preset threshold (e.g., 3), it is identified as a local arc discharge. This state corresponds to sporadic discharge on the insulator surface, which is relatively minor but requires attention. If the conditions for a local arc discharge are not met, the average absolute value Iavg of the IMF1 signal within that cycle is calculated. If Iavg exceeds a preset current threshold (e.g., 1mA) for multiple consecutive power frequency cycles (e.g., 5 consecutive cycles), it is identified as a long arc discharge, indicating that a stable arc has formed on the insulator surface, posing an extremely high risk; otherwise, it is identified as no significant discharge. The discrimination logic is well-founded and, by combining waveform distortion and amplitude trends, effectively distinguishes different discharge types.

[0028] It should be noted that the discharge state discrimination based on zero-crossing detection and waveform integrity includes the following steps: zero-crossing detection is performed on the IMF1 component, and each complete power frequency cycle is marked; within each power frequency cycle, the local variance of the signal is calculated. If the local variance at a certain sample point exceeds 5 times the preset average local variance within that cycle, the point is determined to be a distortion point; if more than 3 non-zero-crossing distortion points are detected within a power frequency cycle, a partial arc discharge is determined to have occurred within that cycle; if the partial arc discharge determination condition is not met, the average absolute value Iavg of the IMF1 signal within that cycle is calculated; if Iavg is greater than or equal to 1 mA for 5 consecutive power frequency cycles, a long arc discharge state is determined; otherwise, no significant discharge is determined.

[0029] That is, step 103 includes: performing zero-crossing detection on the power frequency component and marking the power frequency cycle; calculating the local variance of the signal in each power frequency cycle; if the local variance exceeds a preset multiple of the average local variance, it is determined to be a distortion point; if the number of distortion points in one power frequency cycle exceeds a preset number threshold, it is determined to be a local arc discharge; if the number of distortion points in one power frequency cycle does not exceed the preset number threshold, the average value of the absolute value of the power frequency component is calculated; if the average value of multiple consecutive power frequency cycles exceeds a preset current threshold, it is determined to be a long arc discharge, otherwise it is determined to be no significant discharge.

[0030] 104. Perform pulse event detection and statistics on the high-frequency components based on dynamic thresholds to obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses; It should be noted that the pulse event detection and statistics based on dynamic threshold are performed on the high-frequency IMF component set: using the moving average of the absolute values ​​of all high-frequency IMF component reconstructed signals within the time window T (1 minute) as the benchmark, all pulse spike events exceeding the benchmark value by K times are identified and counted, and the total number of pulse events N within the time window T and the average value Pavg of the spike amplitude of all pulses are obtained.

[0031] For example, the specific algorithm for pulse event detection and statistics based on dynamic threshold is as follows: reconstruct the high-frequency IMF component set into a signal Shigh[n]; set a sliding time window with a length of 1 minute, and calculate the moving average MA[n] of the absolute value of Shigh[n] within the window; set a dynamic detection threshold Th[n]: Th[n] = α * MA[n] + β, where α is the scaling factor and β is the minimum offset; iterate through Shigh[n]. If Shigh[n] > Th[n] and Shigh[n] is a local maximum, then it is recorded as a valid pulse event and its peak value Peakval is recorded; count the total number of valid pulse events N within a 1-minute time window, and calculate the arithmetic mean of all recorded Peakvals as the average peak value Pavg of the pulse.

[0032] This step processes the high-frequency signal Shigh[n] to quantify the partial discharge intensity. First, a predetermined time window (e.g., 1 minute) is set, and the moving average MA[n] of the absolute value of Shigh[n] within the window is calculated as the background noise reference. Then, a dynamic threshold Th[n] is set based on MA[n]. Th[n] = α * MA[n] + β, where α is a proportionality coefficient (e.g., 1.5) and β is the minimum offset (e.g., 0.1mA), ensuring the threshold adapts to signal fluctuations. Next, local maxima exceeding Th[n] in Shigh[n] are detected and recorded as valid pulse events, with their peak value Peakval recorded. Finally, the total number of valid pulse events N within a 1-minute time window is counted, and the arithmetic mean of all peak values ​​is calculated as the average peak value Pavg. This dynamic thresholding method avoids misjudgments caused by fixed thresholds under changing environmental conditions. The pulse statistics (N and Pavg) objectively reflect the frequency and intensity of partial discharge, providing a quantitative basis for risk assessment.

[0033] That is, step 104 includes: reconstructing all intrinsic mode functions with instantaneous frequencies higher than the power frequency into high-frequency signals; calculating the moving average of the absolute values ​​of the high-frequency signals within a predetermined time window; setting a dynamic threshold based on the moving average; detecting local maxima in the high-frequency signals that exceed the dynamic threshold, recording them as valid pulse events, and recording the peak values; counting the total number of valid pulse events N within the predetermined time window, and calculating the average of all peak values ​​as the pulse average peak value Pavg.

[0034] 105. Based on the discharge state and the statistical results of the pulse event, the risk level of the insulator is determined by decision fusion logic, and a warning of the corresponding level is triggered.

[0035] Finally, based on the outputs of steps 103 and 104 (discharge state of 103 and pulse event statistics of 104), a two-level decision fusion logic is executed to determine the final risk level and trigger the corresponding warning.

[0036] In a feasible implementation, the decision fusion logic is a two-layer decision fusion logic. Step 105 includes: The first layer: If the discharge state is determined to be long arc discharge, directly output a warning of the highest risk level. The second layer: If the discharge state is determined not to be long arc discharge, calculate a risk score based on the discharge state and the pulse event statistical results, and map it to a risk level according to the risk score.

[0037] Among them, the first layer (severe state layer): If it is determined in step 103 that it is long arc discharge, directly output a warning of the highest risk level (Level IV). The second layer (quantitative evaluation layer): If it is not the highest risk level, calculate a risk score R according to a preset deterministic rule, and map it to a risk level according to the R value.

[0038] Among them, calculating the risk score based on the discharge state and the pulse event statistical results and mapping it to a risk level includes: Quantifying the discharge state into a state value Vs: The state value of no significant discharge is less than the state value of local arc discharge; Calculate the pulse intensity factor, which is the logarithm of the total number of pulse events plus 1, and then multiply it by the average peak value of the pulse; Calculate the comprehensive risk score, which is the state value multiplied by the target sum value, and the target sum value is obtained by adding 1 to the product of a preset coupling coefficient and the pulse intensity factor.

[0039] Exemplarily, calculate the risk score R according to the following deterministic rule and map it to a risk level according to the R value: 1) Quantify the discharge state S in step 103 into a state value Vs: If S is "no significant discharge", then Vs = 1; if S is "local arc discharge", then Vs = 3. 2) Calculate the pulse intensity factor Fp = log10(N + 1) * Pavg. 3) Calculate the comprehensive risk score R = Vs * (1 + λ * Fp), where λ is a preset coupling coefficient. 4) Determine the risk level according to the preset interval where the R value is located: If R < R1, it is normal (Level I); if R1 ≤ R < R2, it is low risk (Level II); if R2 ≤ R < R3, it is medium risk (Level III). R1, R2, and R3 are the risk thresholds for each layer respectively.

[0040] It should be noted that this step employs a two-layer decision fusion logic. The first layer is the severe state layer: if step 103 determines it to be a long arc discharge, then the highest risk level (e.g., Level IV) warning is directly output without further calculation, ensuring timely response to the emergency. The second layer is the quantitative assessment layer: if it is not the highest risk level, then the comprehensive risk score R is calculated. Specifically, the discharge state is quantified into a state value Vs: Vs=1 when there is no significant discharge, and Vs=3 when there is a local arc discharge. The pulse intensity factor Fp is calculated: Fp = log10(N + 1) * Pavg, where logarithmic processing avoids excessively large impulse counts dominating the results. Then, the overall risk score R is calculated: R = Vs * (1 + λ * Fp), where λ is a preset coupling coefficient (e.g., 0.1). Risk levels are mapped based on the R value: for example, R < 2 indicates normal (Level I), 2 ≤ R < 5 indicates low risk (Level II), 5 ≤ R < 10 indicates medium risk (Level III), and R ≥ 10 indicates high risk (Level IV). Upon triggering a warning, maintenance personnel are notified via audible and visual alarms or remote communication. This fusion logic combines power frequency discharge status and high-frequency pulse characteristics, achieving multi-dimensional risk assessment from macro to micro levels, significantly improving the accuracy of warnings.

[0041] It should be noted that when collecting leakage current signals, the sampling frequency should not be lower than 10kHz; the preset current threshold is 1mA; the preset time window is 1 minute; multiple consecutive power frequency cycles are 5 consecutive power frequency cycles; and the preset number threshold is 3.

[0042] Compared with the prior art, the present invention brings many beneficial effects: 1. Accurate identification and classification of discharge status: By separating the leakage current into power frequency and high frequency components and designing a specific discrimination logic, it can clearly distinguish between three key states: "no significant discharge", "partial arc discharge" and "long arc discharge". This changes the traditional method, which can only give a vague judgment of "exceeding the standard" or "normal", and provides clear and specific fault status information for operation and maintenance personnel.

[0043] 2. Significantly improved accuracy and early warning of pollution flashover: A dual-layer decision fusion risk assessment model was constructed by combining power frequency waveform distortion analysis and the dual statistical characteristics of high-frequency pulses. This model can not only capture emergency situations such as macroscopic long electric arcs, but also detect severe partial discharge activities caused by metal pollution that have not yet formed a stable arc at an early stage through keen perception of the characteristics of high-frequency pulse groups. This significantly advances the warning threshold, gaining valuable time for taking appropriate measures.

[0044] This invention provides a method for identifying the state of insulators. It adaptively separates the power frequency and high-frequency components of leakage current through empirical mode decomposition. Then, it distinguishes between localized and long-term arc discharges by performing zero-crossing and waveform integrity checks on the power frequency component. Simultaneously, it quantifies the intensity of localized discharges by performing dynamic threshold pulse statistics on the high-frequency component. Finally, it performs decision fusion based on the dual-path analysis results. This method overcomes the shortcomings of traditional methods in handling mixed signals, enabling early warning judgments to be based on both macroscopic arc states and microscopic pulse activities. This significantly improves the accuracy and foresight of discharge state identification and early risk warning.

[0045] Please see Figure 2 , Figure 2 This is another flowchart of an insulator state identification method according to an embodiment of the present invention, as shown below. Figure 2 The method shown includes the following steps: S1: Collect the "eigenmode functions characterizing the fundamental frequency component of the power frequency", and execute step S3; S2: Collect the "set of eigenmode functions characterizing high-frequency pulse components", and execute step S5; It should be noted that S1 and S2 are related to... Figure 1 Steps 101 and 102 shown are similar and will not be repeated here to avoid repetition. Please refer to [link to relevant documentation] for details. Figure 1 The contents of steps 101 and 102 are shown.

[0046] S3: Determine whether there are more than 3 non-zero distortion points within one power frequency cycle. If yes, it is a local arc discharge, and proceed to step S6; otherwise, proceed to step S4. S4: Determine if "the sliding average value of the absolute amplitude is greater than 1mA?". If yes, it indicates a long arc discharge; otherwise, it indicates no discharge. It should be noted that S3 and S4 are related to... Figure 1 Step 103 shown is similar in content and will not be repeated here to avoid repetition. Please refer to [link to relevant documentation] for details. Figure 1 The content of step 103 shown.

[0047] S5: Pulse event detection and statistics based on dynamic threshold, counting the total number of valid pulse events and the average peak value of pulses within 1 minute; It should be noted that S5 and Figure 1 Step 104 shown is similar in content and will not be repeated here to avoid repetition. Please refer to [link to relevant documentation] for details. Figure 1 The content of step 104 shown.

[0048] S6: Calculate the overall risk score.

[0049] It should be noted that S6 and Figure 1Step 105 shown is similar in content and will not be repeated here to avoid repetition. Please refer to [link / reference needed] for details. Figure 1 The content of step 105 shown.

[0050] In one feasible implementation, in step 104, the pulse event detection of the high-frequency component is based on a dynamic threshold, and the statistical results include the total number of pulse events N and the average peak value Pavg. To more comprehensively quantify the partial discharge intensity, the pulse intensity factor Fp is improved by the following formula: The formula for the pulse energy factor (Ep) is as follows: ; Where Peaki represents the peak value of the i-th pulse, Δti represents the pulse duration, and T is the statistical time window (e.g., 1 minute). Ep calculates the total pulse energy per unit time, reflecting the cumulative intensity of the discharge.

[0051] It should be noted that the above embodiments only focus on the number of pulses and the average peak value, but pulse energy can capture brief but high-intensity discharge events, especially applicable to pulse clusters induced by metallic contamination. For example, when the average peak value Pavg of the pulses is similar, a high-energy pulse may indicate more severe insulator degradation. By introducing Ep, the pulse intensity factor Fp becomes a weighted form: Fp'=log10(N+1)×Pavg×(1+αEp); Where α is the weighting coefficient (e.g., α=0.01). This allows Fp' to consider pulse frequency, amplitude, and energy simultaneously, improving sensitivity to early discharge and further enhancing early warning accuracy.

[0052] Furthermore, the formula for the impulse distribution entropy (Hp) is as follows: ; Where M is the number of sub-intervals within the time window, and pj is the probability that the pulse occurs in the j-th sub-interval. Hp quantifies the temporal uniformity of the pulse distribution; a higher value indicates a more random pulse distribution, which may correspond to a more complex discharge pattern.

[0053] It should be noted that in complex environments, pulses may occur in concentrated clusters (such as continuous discharges under humid conditions), and Hp can distinguish between periodic and random discharges. For example, a low Hp combined with high N may indicate periodic interference, while a high Hp may predict the risk of pollution flashover. Incorporating Hp into risk calculations can correct the pulse intensity factor.

[0054] Fp''=Fp'×(1+γHp) Where γ is the entropy weighting coefficient (e.g., γ=0.05). This enhances the method's adaptability to changes in discharge patterns and reduces false alarms, especially in humid environments where the false alarm rate can be effectively reduced.

[0055] Existing technical solutions exhibit significant shortcomings in practical applications, especially when facing complex contamination and changing environments. First, the methods are simplistic and lack adaptability: amplitude alarms based on fixed thresholds cannot distinguish between current increases caused by moisture and current surges caused by strong localized arcing, leading to false alarms in humid environments and insensitivity to specific discharge patterns (high-frequency, low-amplitude pulse clusters) induced by metallic contamination. Second, feature extraction is incomplete: pulse counting methods focus only on the number of pulses, ignoring crucial information such as pulse amplitude, energy, and distribution patterns; while harmonic analysis can reflect waveform distortion, it struggles to directly correlate it with specific discharge types (localized arcs versus long arcs) and discharge intensity. Furthermore, the power frequency component and high-frequency pulse component are mixed in the leakage current, carrying different state information: the power frequency component primarily reflects overall surface conductivity and the maintenance of the macroscopic arc, while the high-frequency pulse corresponds to the transient process of partial discharge. Traditional methods fail to effectively separate the two, resulting in mixed information, insufficient judgment criteria, and limited accuracy of early warnings. This makes it difficult to achieve the leap from "simple alarm" to "precise identification of discharge state and type," and cannot meet the urgent need for refined and early warning of pollution flashover risk.

[0056] This invention outputs no longer a single alarm signal, but a structured risk level and a specific description of the discharge state. This provides a reliable technical basis for power grid operation and maintenance to shift from "periodic cleaning" to "condition-based precision maintenance," which helps optimize maintenance strategies, reduce operation and maintenance costs, and improve the level of power grid safety operation.

[0057] Please see Figure 3 , Figure 3 This is a structural block diagram of an insulator state identification device according to an embodiment of the present invention, as shown below. Figure 3 The apparatus shown includes: Signal acquisition module 301 is used to acquire leakage current signals of insulators; Signal separation module 302 is used to perform empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component; The discharge discrimination module 303 is used to perform discharge state discrimination on the power frequency component based on zero-crossing detection and waveform integrity, so as to determine local arc discharge, long arc discharge or no significant discharge; The pulse statistics module 304 is used to perform pulse event detection and statistics on the high-frequency components based on dynamic thresholds, and obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses. The risk decision module 305 is used to determine the risk level of the insulator based on the discharge state and the statistical results of the pulse event through decision fusion logic, and trigger the corresponding level of early warning.

[0058] It should be noted that, Figure 3 The function of each module in the device shown is as follows: Figure 1 The steps shown are similar, and will not be repeated here to avoid repetition. Please refer to [link / reference needed]. Figure 1 The steps shown are as follows.

[0059] This invention provides an insulator condition identification device. It adaptively separates the power frequency and high-frequency components of leakage current through empirical mode decomposition. Then, it distinguishes between localized and long-range arc discharges by performing zero-crossing and waveform integrity checks on the power frequency component. Simultaneously, it quantifies the intensity of localized discharges by performing dynamic threshold pulse statistics on the high-frequency component. Finally, it performs decision fusion based on the dual-channel analysis results. This overcomes the shortcomings of traditional methods in handling mixed signals, enabling early warning judgments to be based on both macroscopic arc conditions and microscopic pulse activity. This significantly improves the accuracy and foresight of discharge condition identification and early risk warning.

[0060] Figure 4 An internal structural diagram of a computer device in one embodiment is shown. This computer device can specifically be a terminal or a server. Figure 4 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. The internal memory may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. Those skilled in the art will understand that… Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0061] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform actions such as... Figure 1 or Figure 2 The steps are shown.

[0062] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the following actions: Figure 1 or Figure 2 The steps are shown.

[0063] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0064] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0065] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for identifying the state of an insulator, characterized in that, The method includes: Collect leakage current signals from insulators; Empirical mode decomposition is performed on the leakage current signal to separate the power frequency component and the high frequency component; The power frequency component is subjected to discharge state discrimination based on zero-crossing detection and waveform integrity to determine whether it is a local arc discharge, a long arc discharge, or no significant discharge. The high-frequency components are subjected to pulse event detection and statistics based on dynamic thresholds to obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses. Based on the discharge state and the statistical results of the pulse events, the risk level of the insulator is determined by decision fusion logic, and a corresponding level of early warning is triggered.

2. The method according to claim 1, characterized in that, The step of performing empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component includes: Empirical mode decomposition is performed on the leakage current signal to obtain a series of intrinsic mode functions; Calculate the instantaneous frequency of each intrinsic mode function; The first intrinsic mode function whose instantaneous frequency stabilizes within the power frequency range is defined as the power frequency component; All intrinsic mode functions with instantaneous frequencies higher than the power frequency are defined as high-frequency components.

3. The method according to claim 1, characterized in that, The discharge state determination of the power frequency component based on zero-crossing detection and waveform integrity includes: Zero-crossing detection is performed on the power frequency component to mark the power frequency period; Within each power frequency cycle, the local variance of the signal is calculated. If the local variance exceeds a preset multiple of the average local variance, it is determined to be a distortion point. If the number of distortion points exceeds a preset threshold within one power frequency cycle, it is determined to be a local arc discharge. If the number of distortion points in one power frequency cycle does not exceed a preset threshold, then calculate the average value of the absolute value of the power frequency component. If the average value of multiple consecutive power frequency cycles exceeds the preset current threshold, it is determined to be a long arc discharge; otherwise, it is determined to be no significant discharge.

4. The method according to claim 3, characterized in that, The step of performing pulse event detection and statistics on the high-frequency components based on dynamic thresholds to obtain pulse event statistical results includes: Reconstruct all intrinsic mode functions with instantaneous frequencies higher than the power frequency into high-frequency signals; Calculate the moving average of the absolute value of the high-frequency signal within a predetermined time window; A dynamic threshold is set based on the moving average value; Local maxima exceeding the dynamic threshold in the high-frequency signal are detected, recorded as valid pulse events, and their peak values ​​are recorded. The total number of valid pulse events within the predetermined time window is counted, and the average value of all peak values ​​is calculated as the average peak value of the pulse.

5. The method according to claim 1, characterized in that, The process of determining the risk level of the insulator based on the discharge state and the statistical results of the pulse events through decision fusion logic, and triggering a corresponding level of early warning, includes: If the discharge state is determined to be a long arc discharge, the highest risk level warning will be directly output. If the discharge state is determined not to be a long arc discharge, a risk score is calculated based on the discharge state and pulse event statistics, and the risk score is mapped to a risk level.

6. The method according to claim 5, characterized in that, The calculation of risk scores based on the discharge state and pulse event statistics, and the mapping of risk scores to risk levels, includes: The discharge state is quantified into a state value: the state value of no significant discharge is less than the state value of local arc discharge; Calculate the pulse intensity factor, which is the logarithm of the total number of pulse events plus 1, and then multiply it by the average peak value of the pulses. Calculate the comprehensive risk score, which is the product of the state value and the target sum, where the target sum is the product of the preset coupling coefficient and the pulse intensity factor, plus 1.

7. The method according to claim 4, characterized in that, When collecting leakage current signals, the sampling frequency is not less than 10kHz; the preset current threshold is 1mA; the predetermined time window is 1 minute; the consecutive multiple power frequency cycles are 5 consecutive power frequency cycles; and the preset number threshold is 3.

8. A device for identifying the state of an insulator, characterized in that, The device includes: The signal acquisition module is used to acquire the leakage current signal of the insulator; The signal separation module is used to perform empirical mode decomposition on the leakage current signal to separate the power frequency component and the high frequency component; The discharge discrimination module is used to perform discharge state discrimination on the power frequency component based on zero-crossing detection and waveform integrity, so as to determine local arc discharge, long arc discharge or no significant discharge; The pulse statistics module is used to perform pulse event detection and statistics on the high-frequency components based on dynamic thresholds, and obtain pulse event statistics results, which include the total number of pulse events and the average peak value of the pulses. The risk decision module is used to determine the risk level of the insulator based on the discharge state and the statistical results of the pulse event through decision fusion logic, and trigger the corresponding level of early warning.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to perform the steps of the method as described in any one of claims 1 to 7.

10. A computer device, comprising a memory and a processor, characterized in that, The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 7.

Images