A direct-buried cable operation and maintenance safety risk dynamic evaluation and early warning method based on multi-source data fusion

Abstract

The application relates to the technical field of electric variable measurement, and discloses a direct-buried cable operation and maintenance safety risk dynamic evaluation and early warning method based on multi-source data fusion. The method comprises the following steps: acquiring multi-source monitoring signals; adopting wavelet denoising and windowed interpolation Fourier transformation on a synchronous sampling sequence to extract fundamental wave and harmonic characteristic parameters; based on the characteristic parameter set, resistance and capacitive leakage currents are separated online through time delay compensation and forgetting factor recursive least squares; a multi-dimensional feature vector is constructed from the separation result, and hierarchical early warning is realized in combination with dynamic threshold feedforward control; according to the early warning result, the comprehensive risk is evaluated by using an entropy weight method and D-S evidence theory fusion, and operation and maintenance decision suggestions are obtained. Due to the step-by-step connection of synchronous sampling, feature extraction, leakage current separation, hierarchical early warning and evidence fusion, dynamic evaluation and risk early warning of the insulation state of the direct-buried cable are realized, and the reliability of the operation and maintenance safety decision is improved.

Description

Technical Field

[0001] This invention relates to the field of electrical variable measurement technology, and in particular to a method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion. Background Technology

[0002] Currently, in the operation of directly buried cables in areas with fluctuating groundwater levels, it is typically necessary to simultaneously collect data such as the grounding current of the metal sheath, the leakage current of the main insulation, and soil moisture to assess the moisture status of the cable insulation. The common approach in existing technologies is to use the amplitude of the grounding current of the metal sheath as the primary monitoring object, employing the sheath circulating current threshold method for judgment. An alarm is issued when the sheath circulating current exceeds a set threshold. Some solutions also combine the total leakage current or periodically offline measurements of the dielectric loss factor for auxiliary judgment. Because the operating environment of directly buried cables is enclosed, the insulation status is difficult to observe directly. Multiple electrical quantities and soil environmental quantities are usually acquired by different acquisition terminals, resulting in differences in acquisition time bases and data correlation methods. This can easily cause distortion in the phase relationship between the sheath grounding current and the leakage current, making it difficult for existing monitoring methods to accurately reflect changes in current composition during the insulation moisture absorption process.

[0003] For example, in scenarios with rapid changes in groundwater levels, increased soil moisture alters the equivalent capacitance of the cable sheath to ground, leading to a significant increase in the capacitive component of the sheath grounding current. Meanwhile, the resistive leakage current generated in the early stages of insulation dampness exhibits smaller changes, and the two components overlap in either a single current amplitude or the total leakage current. Existing single sheath circulating current threshold methods cannot distinguish between capacitive currents caused by soil moisture changes and resistive currents caused by insulation dampness. This can easily result in false alarms when soil moisture levels rise in the short term, and false alarms when resistive currents rise slowly but are masked by capacitive fluctuations. Furthermore, time base deviations between data acquisition terminals, harmonic interference, and lag in soil moisture measurement further reduce the reliability of early insulation dampness identification, making existing technologies insufficient to fully meet the needs for dynamic graded early warning of insulation dampness under conditions of rapid groundwater level changes.

[0004] Therefore, there is an urgent need for a method that can still integrate the amplitude, phase, and harmonic characteristics of the grounding current and leakage current of the metal sheath under conditions of rapid changes in groundwater level, achieve online separation of resistive leakage current and capacitive leakage current, and provide dynamic graded early warning of insulation moisture status, so as to improve the accuracy and timeliness of risk assessment for direct-buried cable operation and maintenance, and reduce the probability of false alarms and missed alarms. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a dynamic assessment and early warning method for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion. This method aims to solve the technical problem that the existing single sheath circulation threshold method, especially under conditions of rapid changes in groundwater level, cannot distinguish between capacitive current caused by changes in soil moisture and resistive current caused by moisture absorption in insulation.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion.

[0007] The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion includes: Step S10: Acquire multi-source monitoring signals of the direct-buried cable; based on the multi-source monitoring signals of the direct-buried cable, perform synchronous sampling tasks using a time base deviation compensation mechanism, and output synchronous sampling sequences; Step S20: Based on the synchronous sampling sequence, the fundamental and harmonic feature extraction task is performed using wavelet denoising and windowed interpolation Fourier transform mechanism, and the fundamental and harmonic feature parameter set is output; Step S30: Based on the fundamental and harmonic characteristic parameter set, the online separation task of resistive and capacitive components is performed using a time delay compensation and forgetting factor recursive least squares mechanism, and the separation results of resistive leakage current and capacitive leakage current are output. Step S40: Based on the separation results of resistive leakage current and capacitive leakage current, a hierarchical early warning task is executed using a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism, and the hierarchical early warning results are output. Step S50: Based on the aforementioned graded early warning results, a comprehensive risk assessment task is performed using the entropy weight method and DS evidence theory fusion mechanism, and comprehensive risk assessment and operation and maintenance decision recommendations are output.

[0008] Preferably, step S10, which involves acquiring multi-source monitoring signals from directly buried cables, performing synchronous sampling based on these signals using a time base deviation compensation mechanism, and outputting a synchronous sampling sequence, specifically includes: Step S101: Synchronously collect the grounding current of the metal sheath, leakage current of the main insulation, soil moisture, ambient temperature and reference voltage signals of the direct-buried cable, and combine each signal and its timestamp into the multi-source monitoring signal of the direct-buried cable; Step S102: Calculate the sampling time deviation between the grounding current of the metal sheath and the leakage current of the main insulation. According to the sampling time deviation, the sampling period, and the value of the leakage current of the main insulation at the current sampling time and the adjacent sampling times, use second-order Lagrange interpolation to perform time base correction on the leakage current of the main insulation to obtain the time base correction signal. Step S103: Use the synchronization timing pulse to discipline the sampling clock, calculate the difference between the actual sampling time after discipline and the product of the sampling sequence number and the sampling period to obtain the crystal oscillator drift error, and use the crystal oscillator drift error as the phase alignment compensation feedforward of the time base correction signal to obtain the synchronization sampling sequence.

[0009] Preferably, step S20, which involves performing the fundamental and harmonic feature extraction task based on the synchronous sampling sequence using wavelet denoising and windowed interpolation Fourier transform mechanisms, and outputting the fundamental and harmonic feature parameter set, specifically includes: Step S201: Perform wavelet transform on the metal sheath grounding current and main insulation leakage current in the synchronous sampling sequence, determine the scale adaptive threshold according to the noise standard deviation of each decomposition layer, the number of wavelet coefficients and the decomposition layer number, and perform joint threshold shrinking on the wavelet coefficients of the main insulation leakage current in combination with the corresponding wavelet coefficients of the metal sheath grounding current to obtain the denoised current sequence. Step S202: Perform zero-crossing detection and digital phase-locked loop processing on the denoised current sequence to obtain the fundamental frequency estimate. Determine the frequency interpolation coefficients of the windowed interpolation Fourier transform according to the fundamental frequency estimate. Correct the fundamental amplitude and fundamental phase according to the frequency interpolation coefficients to obtain the frequency correction characteristics. Step S203: Extract the fundamental amplitude, fundamental phase, phase difference, harmonic content and total harmonic distortion rate of the metal sheath grounding current and the main insulation leakage current according to the frequency correction characteristics, and combine the extraction results to form the fundamental and harmonic characteristic parameter set.

[0010] Preferably, in step S202, the steps of performing zero-crossing detection and digital phase-locked loop processing on the denoised current sequence to obtain a fundamental frequency estimate, determining the frequency interpolation coefficients of the windowed interpolation Fourier transform according to the fundamental frequency estimate, and correcting the fundamental amplitude and fundamental phase based on the frequency interpolation coefficients to obtain the frequency correction characteristics specifically include: updating the fundamental frequency estimate according to a proportional-integral feedback relationship, wherein the proportional-integral feedback relationship is:

[0011] in, For the first The fundamental frequency estimate for each time window For the first The fundamental frequency estimate for each time window This is the proportionality coefficient. The integral coefficient is... For the first Frequency error of each time window For the first Frequency error of each time window This is the current time window number. The summation time window number is used; the frequency interpolation coefficient is determined based on the ratio of the updated fundamental frequency estimate to the frequency resolution; the sum of the amplitudes on both sides of the fundamental peak spectral line is multiplied by the ratio of two to the number of sampling points and the amplitude correction factor determined by the frequency interpolation coefficient to obtain the corrected fundamental amplitude; and the phase of the fundamental peak spectral line is added with the orthogonal phase compensation amount and the leakage phase compensation amount determined by the frequency interpolation coefficient to obtain the corrected fundamental phase, thus forming the frequency correction feature.

[0012] Preferably, step S30, based on the fundamental and harmonic characteristic parameter set, employs a time-delay compensation and forgetting factor recursive least squares mechanism to perform the online separation task of resistive and capacitive components, and outputs the separation results of resistive and capacitive leakage current, specifically includes: Step S301: Obtain the soil moisture time series characteristics from the set of fundamental and harmonic characteristic parameters, perform weighted fusion of soil moisture at different measuring points at the cable laying depth, and use a first-order inertial element for time delay compensation to obtain the compensated equivalent humidity. Step S302: Determine the relative permittivity of the soil based on the quadratic relationship between the compensated equivalent humidity and the relative permittivity of the soil. Multiply the initial equivalent capacitance of the sheath to the ground by the humidity sensitivity exponent power of the ratio of the relative permittivity of the soil to the reference relative permittivity to obtain the equivalent capacitance of the sheath to the ground. Multiply the angular frequency, the equivalent capacitance of the sheath to the ground and the amplitude of the cable phase voltage to obtain the predicted value of the capacitive current. Step S303: Construct a regression vector using the cable phase voltage and its rate of change, and construct a parameter vector using the equivalent insulation conductance and equivalent insulation capacitance. Update the forgetting factor within a preset forgetting factor range based on the soil moisture change rate, and perform recursive least squares parameter update in conjunction with the predicted capacitive current value. Multiply the updated equivalent insulation conductance by the effective value of the cable phase voltage to obtain the resistive leakage current, and multiply the angular frequency, the updated equivalent insulation capacitance, and the effective value of the cable phase voltage to obtain the capacitive leakage current, thus forming the separation result of the resistive leakage current and the capacitive leakage current.

[0013] Preferably, step S40, based on the separation results of resistive leakage current and capacitive leakage current, employs a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism to execute a graded early warning task and output graded early warning results, specifically includes: Step S401: Construct a multidimensional feature vector of insulation moisture based on the ratio of resistive leakage current to capacitive leakage current, the percentage increase of resistive leakage current relative to its historical sliding window baseline, the phase difference change rate, the harmonic content rate, and the polarization index. Step S402: Determine the dynamic threshold floating coefficient based on the sliding window change rate and change acceleration of the resistive leakage current and the capacitive leakage current, and apply the dynamic threshold floating coefficient to the resistive leakage current threshold and the equivalent dielectric loss tangent threshold determined by the historical sliding window baseline and the sliding window standard deviation to obtain the dynamic early warning threshold set; Step S403: Compare the insulation moisture multidimensional feature vector with the dynamic early warning threshold set, and determine the early warning level of normal, attention, abnormal or serious according to the preset hysteresis conditions of continuous over-limit cycle and continuous fall-off cycle to form the graded early warning result.

[0014] Preferably, in step S50, based on the graded early warning results, the step of performing a comprehensive risk assessment task using a fusion mechanism of entropy weight method and DS evidence theory, and outputting comprehensive risk assessment and operation and maintenance decision recommendations, specifically includes: Step S501: Obtain the normalized values ​​of each feature quantity in each time window from the graded early warning results, determine the weight of each feature quantity according to the inverse relationship between information entropy and feature quantity weight, and perform weighted summation of the scores of each feature quantity to obtain the initial comprehensive risk score. Step S502: Calculate the evidence conflict coefficient based on the basic probability allocation function of each piece of evidence. When the evidence conflict coefficient is greater than a preset conflict threshold, discount each piece of evidence according to the feature weight and the credibility of the evidence. The discount factor used for the discount is:

[0015] in, For the first Discount factor for each piece of evidence To preset the fusion adjustment coefficient, For the first The weights of each feature quantity For the first The credibility of individual pieces of evidence. The evidence body number is used; the conflicting evidence is redistributed using the discount factor, and the redistributed evidence fusion result is combined with the initial comprehensive risk score to obtain the comprehensive risk score and the final risk level; Step S503: Based on the comprehensive risk score, the final risk level and the corresponding risk source, generate differentiated operation and maintenance decision suggestions, and combine the comprehensive risk score, the final risk level and the differentiated operation and maintenance decision suggestions to form the comprehensive risk assessment and operation and maintenance decision suggestions.

[0016] This invention also provides a dynamic assessment and early warning system for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion, comprising: The multi-source data synchronous acquisition module is used to acquire multi-source monitoring signals of directly buried cables. Based on the multi-source monitoring signals of the directly buried cables, a time base deviation compensation mechanism is used to perform synchronous sampling tasks and output synchronous sampling sequences. The fundamental and harmonic feature extraction module is used to perform the fundamental and harmonic feature extraction task based on the synchronous sampling sequence, using wavelet denoising and windowed interpolation Fourier transform mechanism, and output the fundamental and harmonic feature parameter set; The online separation module for resistive and capacitive components is used to perform the online separation task of resistive and capacitive components based on the fundamental and harmonic characteristic parameter set, using a time delay compensation and forgetting factor recursive least squares mechanism, and output the separation results of resistive leakage current and capacitive leakage current. The insulation moisture graded early warning module is used to perform graded early warning tasks based on the separation results of resistive leakage current and capacitive leakage current, using a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism, and output graded early warning results. The comprehensive risk assessment and operation and maintenance decision module is used to perform comprehensive risk assessment tasks based on the hierarchical early warning results, using a fusion mechanism of entropy weight method and DS evidence theory, and output comprehensive risk assessment and operation and maintenance decision recommendations.

[0017] This invention also provides a device for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion. The device includes a memory, a processor, and a program for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion, which is stored in the memory and can run on the processor. When the program for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion is executed by the processor, it implements the above-mentioned method.

[0018] The present invention also provides a computer program product, which includes a dynamic assessment and early warning program for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion. When the program is executed by a processor, it implements the above-mentioned method.

[0019] The beneficial effects of this invention are as follows: The timestamps of each acquisition terminal are aligned by a time base deviation compensation mechanism, and the sampling clock is disciplined by a synchronous timing pulse to restore the phase relationship between the grounding current of the metal sheath and the leakage current of the main insulation. This provides a synchronous sampling sequence with consistent phase for subsequent extraction of fundamental and harmonic features, avoids phase distortion caused by time base deviation, and improves the accuracy of insulation moisture condition identification.

[0020] By employing wavelet denoising and windowed interpolation Fourier transform mechanisms to jointly denoise and correct the frequency of synchronously sampled sequences, harmonic interference can be suppressed and the fundamental amplitude and phase can be corrected. This allows for the extraction of reliable fundamental amplitude, phase difference, harmonic content, and total harmonic distortion rate, providing high-quality characteristic parameters for online separation of resistive and capacitive components and reducing the impact of noise on the separation results.

[0021] By using a time delay compensation and forgetting factor recursive least squares mechanism to separate resistive leakage current and capacitive leakage current online, and combining multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism for hierarchical early warning, it can distinguish between capacitive current caused by soil moisture changes and resistive current caused by insulation moisture, reduce false alarms and missed alarms, and improve the timeliness and accuracy of early warning of operation and maintenance risks of directly buried cables under conditions of rapid changes in groundwater level. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the first embodiment of a method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion, according to the present invention. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0024] 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.

[0025] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion according to the present invention. The first embodiment of the method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion according to the present invention is presented.

[0026] In the first embodiment, the method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion includes: Step S10: Acquire multi-source monitoring signals of the direct-buried cable; based on the multi-source monitoring signals of the direct-buried cable, perform synchronous sampling tasks using a time base deviation compensation mechanism, and output synchronous sampling sequences; In this step, the multi-source monitoring signal for directly buried cables refers to a synchronously acquired data set consisting of the metal sheath grounding current, main insulation leakage current, soil moisture, ambient temperature, and reference voltage signal. The metal sheath grounding current comes from a sheath current sensor installed on the cable sheath grounding wire, reflecting the power frequency current and harmonic currents in the sheath circuit. The main insulation leakage current is acquired by a leakage current sensor, which includes both resistive and capacitive leakage components. Soil moisture is collected by multi-layer moisture sensors arranged at the laying depth, used to subsequently determine the impact of soil thermal resistance and moisture intrusion on the insulation condition. The ambient temperature comes from a temperature sensor within the cable tunnel, providing environmental conditions for subsequent baseline correction. The reference voltage signal is taken from the voltage transformer of the busbar where the cable is located, providing phase voltage amplitude and phase reference. These signals come from different acquisition terminals, each with a relatively independent sampling clock, resulting in sampling time deviation and crystal oscillator drift. The synchronization time unit provides a unified time reference for each acquisition terminal and adds a timestamp to each frame of data. To address the sampling time deviation between the sheath grounding current and the main insulation leakage current, second-order Lagrange interpolation is used to correct the time base of the main insulation leakage current. The correction process utilizes the leakage current values ​​at the current, previous, and next sampling times, and calculates the interpolation weight based on the relative relationship between the sampling time deviation and the sampling period, resulting in a time base correction signal aligned with the sheath grounding current sampling time. Simultaneously, a synchronization pulse is used to discipline the sampling clock. The difference between the disciplinary actual sampling time and the product of the sampling sequence number and the sampling period is calculated to obtain the crystal oscillator drift error. This crystal oscillator drift error is used as the phase alignment compensation feedforward of the time base correction signal, ultimately outputting a synchronized sampling sequence under the same time base. In this synchronized sampling sequence, the signals are arranged in timestamp order, the sheath grounding current and the main insulation leakage current are aligned on the time base, and the soil moisture, ambient temperature, and reference voltage signals are also aligned on the same time base, providing the input basis for extracting the fundamental and harmonic characteristics in step S20.

[0027] After time base deviation compensation, the deviation between the sampling time of the sheath grounding current and the sampling time of the main insulation leakage current is corrected to within the sampling period, and the phase relationship between the two is maintained. The crystal oscillator drift error is used as a feedforward compensation, so that different acquisition terminals maintain phase alignment even during long-term operation. In the output synchronous sampling sequence, the fundamental phase difference information of the sheath grounding current and the main insulation leakage current is reliable, and the harmonic components will not be distorted due to resampling misalignment. In this way, the fundamental phase difference, harmonic content, and total harmonic distortion rate extracted in the subsequent step S20 can more realistically reflect the electrical state of the cable insulation and sheath circuit, providing a consistent time base for the online separation of resistive and capacitive components in step S30, and avoiding the propagation of time base error to subsequent calculations. Specifically, if there is a deviation between the sampling time of the sheath grounding current and the main insulation leakage current, directly calculating the fundamental phase difference will introduce a phase error related to the magnitude of the deviation and the power frequency period. This error will further lead to crosstalk when separating the resistive leakage current and the capacitive leakage current, causing the estimated value of the resistive current to deviate from the actual insulation leakage level. This step eliminates most of the time base error during the sampling phase, ensuring that subsequent phase-sensitive calculations are not affected by sampling asynchrony, thereby guaranteeing that the resistive current trend can accurately reflect changes in insulation conditions such as moisture and aging.

[0028] In traditional technologies, multi-source signals are typically acquired by independent terminals. Clock deviations accumulate over time, causing phase drift between the sheath grounding current and the main insulation leakage current. This is especially problematic in applications with short power frequency cycles and high phase sensitivity, where sampling deviations of tens of microseconds can result in phase errors of a few tenths of a degree. Some solutions rely on post-event timestamp alignment or manual correction, failing to provide real-time compensation at the equipment level. This step corrects the main insulation leakage current using second-order Lagrange interpolation and uses a synchronous timing pulse to discipline the sampling clock. Crystal oscillator drift error is used as a feedforward compensation, eliminating most of the time base error during the sampling phase, rather than passively correcting it in subsequent analysis. This reduces phase difference distortion and harmonic leakage, resulting in high phase reliability for online separation of resistive and capacitive currents at the input data level. For example, in traditional solutions, if the leakage current sampling time lags behind the sheath current by tens of microseconds, directly calculating the fundamental phase difference will add a system error of approximately a few tenths of a degree. This error is amplified during resistive current separation, leading to deviations in insulation condition assessment. In this step, the deviation is calculated based on the timestamp in each sampling frame. The leakage current sampling value is then subjected to second-order Lagrange interpolation and combined with synchronous timing pulse discipline error correction to align the equivalent sampling times of the two to the same time base. The phase difference calculation error is reduced to a smaller range corresponding to the sampling period, and the trend of the resistive current obtained in subsequent separation is more continuous.

[0029] In areas with fluctuating groundwater levels, the sheath grounding current terminal and the main insulation leakage current terminal are installed on both sides of the cable joint. Although the initial sampling clocks were synchronized, after several months of operation, their respective crystal oscillators drifted, for example, the leakage current sampling time lagged behind the sheath current by tens of microseconds. In this step, the deviation is calculated based on the timestamp in each sampling frame, and the leakage current sampling value is subjected to second-order Lagrange interpolation. Combined with the synchronization timing pulse discipline error correction, the equivalent sampling times of the two are aligned to the same time base, and the phase difference calculation error is reduced to a small range corresponding to the sampling period. At the same time, the data from the soil moisture sensor, ambient temperature sensor, and reference voltage transformer are also added to the synchronization sampling sequence according to a unified time base. After this step, the fundamental phase difference between the sheath grounding current and the main insulation leakage current is stable, and the harmonic components do not show obvious distortion. The fundamental phase difference, harmonic content, and total harmonic distortion rate extracted in the subsequent step S20 can truly reflect the electrical state of the cable insulation and sheath circuit, providing a consistent time base for the online separation of resistive and capacitive components in step S30.

[0030] Step S20: Based on the synchronous sampling sequence, the fundamental and harmonic feature extraction task is performed using wavelet denoising and windowed interpolation Fourier transform mechanism, and the fundamental and harmonic feature parameter set is output; In this step, the set of fundamental and harmonic characteristic parameters refers to a set of electrical state quantities extracted from the synchronous sampling sequence, covering the fundamental amplitude, fundamental phase, phase difference between the two, harmonic content, and total harmonic distortion rate of the metal sheath grounding current and the main insulation leakage current. The synchronous sampling sequence is provided by step S10, in which the metal sheath grounding current and the main insulation leakage current are aligned to the same time base, but the field-acquired signals are still superimposed with electromagnetic interference, stray currents from the grounding grid, and sensor background noise. During processing, wavelet transforms are first performed on the metal sheath grounding current and the main insulation leakage current to obtain wavelet coefficients on different decomposition layers. For each layer, a scale-adaptive threshold is determined based on the noise standard deviation of that layer, the number of wavelet coefficients, and the decomposition layer number. When shrinking the wavelet coefficients of the main insulation leakage current, a joint threshold shrinkage relationship is constructed by combining the wavelet coefficients of the corresponding layer of the metal sheath grounding current, so that the strong power frequency component can guide the denoising process of the weak leakage current component and avoid the weak resistive changes in the leakage current being removed as noise. The denoised current sequence enters zero-crossing detection and digital phase-locked loop processing. The fundamental frequency estimate is continuously updated through proportional-integral feedback, ensuring the estimated frequency tracks the actual frequency changes of the system. Frequency interpolation coefficients are then determined based on the ratio of the estimated frequency to the frequency resolution. Weighted corrections are applied to the amplitudes on both sides of the fundamental peak spectral line, and orthogonal phase compensation and leakage phase compensation are performed on the fundamental peak spectral line phase to obtain frequency correction characteristics. Finally, based on these frequency correction characteristics, the fundamental amplitude, fundamental phase, phase difference, harmonic content, and total harmonic distortion rate of the metal sheath grounding current and the main insulation leakage current are extracted. These quantities are combined to form a set of fundamental and harmonic characteristic parameters, which serve as input for the online separation of resistive and capacitive components in step S30.

[0031] The combination of wavelet denoising and windowed interpolation Fourier transform ensures that the fundamental and harmonic characteristic parameter sets maintain stable phase relationships and amplitude accuracy even under system frequency shifts and noise interference. In the wavelet denoising stage, the strong power frequency component in the metal sheath grounding current guides the contraction of wavelet coefficients in the main insulation leakage current. This suppresses high-frequency noise while preserving the weak resistive components in the leakage current reflecting insulation dampness or aging, avoiding phase distortion caused by the denoising process. In the windowed interpolation Fourier transform stage, the fundamental frequency is estimated in real time using a digital phase-locked loop, and the spectrum is interpolated and corrected based on the estimated frequency. This ensures that the fundamental amplitude and phase are no longer significantly affected by spectral leakage and the picket fence effect. The resulting fundamental phase difference accurately reflects the capacitive phase relationship between the metal sheath grounding current and the main insulation leakage current, and the harmonic content and total harmonic distortion rate accurately reflect the nonlinear characteristics of the insulation. After these characteristic parameter sets enter step S30, the phase difference, frequency, and amplitude required for online separation of resistive and capacitive components become more reliable, the fluctuation of the recursive least squares regression input is reduced, and the time delay compensation stage can perform more accurate delay correction based on a stable fundamental frequency and phase difference, thereby reducing the fluctuation of the separation results of resistive and capacitive leakage current.

[0032] Traditional Fourier transform suffers from spectral leakage and the picket-fence effect when sampling asynchronously and when the system frequency shifts. This results in significant errors in the fundamental phase and harmonic amplitude, particularly concerning the fundamental phase difference between the metal sheath grounding current and the main insulation leakage current, which is sensitive to insulation moisture detection. Errors in this difference can lead to resistive current being misinterpreted as capacitive changes. Traditional wavelet denoising thresholds are typically determined independently for a single signal, without considering the metal sheath grounding current and the main insulation leakage current together. This allows weak resistive components in the leakage current to be submerged by noise or suffer phase distortion. This step employs scale-adaptive thresholding and joint threshold shrinkage. The strong power frequency component in the metal sheath grounding current guides the shrinkage of the wavelet coefficients in the main insulation leakage current, preserving the phase information within the leakage current. Simultaneously, windowing interpolation and frequency correction reduce the impact of system frequency shifts on the calculation of fundamental amplitude and harmonic content, avoiding parameter deviations caused by spectral leakage and the picket-fence effect. After this step, the phase relationship and harmonic characteristics between the metal sheath grounding current and the main insulation leakage current are clearer. In the subsequent step S30, when calculating the capacitive current prediction value and performing recursive least squares separation, more stable regression input and more reliable phase difference information can be obtained, reducing the fluctuation of parameter estimation.

[0033] In areas with fluctuating groundwater levels, the sheath grounding current terminal and the main insulation leakage current terminal are installed on both sides of the cable joint. After time base deviation compensation in step S10, their equivalent sampling times are aligned to the same time base. After receiving the synchronous sampling sequence, this step first performs wavelet transform on the metal sheath grounding current and the main insulation leakage current. At each decomposition layer, a scale-adaptive threshold is determined based on the noise standard deviation, the number of wavelet coefficients, and the decomposition layer number. Joint threshold shrinkage is then applied to the wavelet coefficients of the main insulation leakage current to preserve the weak resistive components reflecting insulation moisture in the leakage current. The denoised current sequence enters zero-crossing detection and digital phase-locked loop processing. The fundamental frequency estimate is updated through proportional-integral feedback to ensure the estimated frequency tracks the actual frequency changes of the system. Then, frequency interpolation coefficients are determined based on the estimated frequency, and the amplitudes on both sides of the fundamental peak spectrum are weighted and corrected. Orthogonal phase compensation and leakage phase compensation are performed on the phase of the fundamental peak spectrum to obtain the frequency correction characteristics. Finally, the fundamental amplitude, fundamental phase, phase difference, harmonic content, and total harmonic distortion rate of the grounding current of the metal sheath and the leakage current of the main insulation are extracted and combined to form a set of characteristic parameters for the fundamental and harmonic components. After this set of characteristic parameters is entered into step S30, the phase difference, frequency, and amplitude required for online separation of resistive and capacitive components are more reliable, and the time delay compensation stage can perform more accurate delay correction based on a stable fundamental frequency and phase difference, thereby reducing the fluctuation of the separation results of resistive and capacitive leakage currents.

[0034] Step S30: Based on the fundamental and harmonic characteristic parameter set, the online separation task of resistive and capacitive components is performed using a time delay compensation and forgetting factor recursive least squares mechanism, and the separation results of resistive leakage current and capacitive leakage current are output. The set of fundamental and harmonic characteristic parameters received in step S30 includes not only voltage amplitude, phase difference, and frequency information, but also time-series humidity data obtained from multiple soil moisture sensors at the cable laying depth. This humidity data is first weighted and fused according to the measurement point location to obtain an equivalent humidity that comprehensively reflects the humidity state around the cable. Since there is a lag in the transmission of humidity from the soil to the sheath surface, a first-order inertial element is used to compensate for the time lag of the equivalent humidity, making the compensated equivalent humidity closer to the actual humidity changes experienced by the sheath surface. The compensated equivalent humidity has a quadratic relationship with the relative permittivity of the soil. Based on this, after determining the relative permittivity of the soil, the initial equivalent capacitance of the sheath to ground is multiplied by the humidity sensitivity exponent, which is the ratio of the relative permittivity of the soil to the reference relative permittivity, to obtain the equivalent capacitance of the sheath to ground. The angular frequency, the equivalent capacitance of the sheath to ground, and the cable phase voltage amplitude are multiplied together to obtain the predicted capacitive current value. Subsequently, a regression vector is constructed using the cable phase voltage and its rate of change, and a parameter vector is constructed using the equivalent insulation conductance and equivalent insulation capacitance. The forgetting factor is updated within a preset forgetting factor range based on the soil moisture change rate, and a recursive least squares parameter update is performed in conjunction with the predicted capacitive current value. The resistive leakage current is obtained by multiplying the updated equivalent insulation conductance by the effective value of the cable phase voltage, and the capacitive leakage current is obtained by multiplying the angular frequency, the updated equivalent insulation capacitance, and the effective value of the cable phase voltage. These two factors together form the separation result of resistive and capacitive leakage currents. In this separation result, the resistive leakage current reflects the active power loss of the cable's main insulation under operating voltage, while the capacitive leakage current reflects the reactive power charging and discharging current of the sheath-to-ground capacitance and the equivalent insulation capacitance under an AC electric field. This provides data for step S40 to construct features such as the equivalent dielectric loss tangent and the percentage increase in resistive leakage current.

[0035] The equivalent humidity after humidity time-delay compensation is closer to the actual humidity change on the cable sheath surface. Based on this, the predicted value of the sheath-to-ground equivalent capacitance can reflect the changing trend of capacitive current in advance. In the parameter update, recursive least squares adjusts the forgetting factor according to the soil humidity change rate: when humidity changes rapidly, the forgetting factor is smaller, allowing the parameter estimation to track capacitance changes faster; when humidity is stable, the forgetting factor is larger, making the insulation equivalent conductance estimation more stable. The parameter update process also incorporates parameter change rate limits and residual whitening checks. When the change in insulation equivalent conductance exceeds a preset limit, conductance updates are frozen, and only capacitance is updated. When the residual exceeds three times the standard deviation, the magnitude of the gain matrix is ​​reduced, and a regularization matrix is ​​added to the parameter covariance matrix to prevent estimation jumps. Thus, the separated resistive leakage current remains stable when soil humidity fluctuates, while the capacitive leakage current changes with humidity. Subsequent step S40 can determine the degree of insulation moisture based on the resistive current increment and the equivalent dielectric loss tangent, without being affected by capacitive current interference caused by soil humidity fluctuations.

[0036] Traditional methods directly use the total power frequency leakage current or the sheath circulating current amplitude to determine insulation moisture. When soil moisture changes rapidly, the capacitive current caused by the sheath's capacitance to ground increases, and the total leakage current may exceed the attention threshold, triggering a false alarm, even though the actual insulation has not deteriorated. While offline dielectric loss factor measurement can distinguish between resistive and capacitive components, it cannot continuously track changes in groundwater level. Recursive least squares, without humidity time lag compensation and adaptive forgetting factor, can easily converge capacitive parameter changes to resistive parameters during rapid humidity changes, leading to resistive current estimation errors. This step uses the predicted capacitive current value calculated by humidity feedforward to constrain the regression direction, reducing the forgetting factor and improving tracking speed when the humidity change rate is large. Simultaneously, parameter change rate constraints and regularization prevent estimation jumps, improving the online stability of resistive and capacitive separation and reducing misjudgments caused by groundwater level fluctuations.

[0037] The groundwater level rises rapidly within hours, and soil moisture increases from a low level to a high level. This correspondingly increases the sheath's capacitance to ground, significantly increasing the capacitive component of the leakage current. Without separation, the total leakage current may exceed a preset warning threshold, triggering a false alarm. This step first weights the multi-layer humidity signals and performs first-order inertial compensation to obtain an equivalent humidity that more closely approximates the actual humidity changes around the cable. Then, based on the relationship between humidity and dielectric constant, the predicted capacitive current value is calculated. Recursive least squares are then used to estimate the equivalent insulation conductance. The output resistive leakage current remains relatively stable, while the capacitive leakage current increases with humidity. This indicates that the current change mainly originates from soil moisture rather than insulation dampness. Subsequent step S40 can then perform graded warnings based on the increase in resistive leakage current and the equivalent dielectric loss tangent to avoid false alarms.

[0038] Step S40: Based on the separation results of resistive leakage current and capacitive leakage current, a hierarchical early warning task is executed using a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism, and the hierarchical early warning results are output. The results of separating the resistive and capacitive leakage currents received in step S40 will be organized into a set of multi-dimensional feature vectors for insulation moisture absorption in this step. This vector consists of the ratio of resistive to capacitive leakage current, the percentage increase of the resistive leakage current relative to its historical sliding window baseline, the phase difference change rate, the harmonic content rate, and the polarization index. The ratio of resistive to capacitive leakage current directly corresponds to the tangent of the equivalent dielectric loss angle, reflecting the relative magnitude of active power loss and reactive capacitive response in the insulation material; the percentage increase of the resistive leakage current measures the degree to which the current resistive current deviates from the historical baseline, which is corrected based on ambient temperature and compensated equivalent humidity; the phase difference change rate describes the rate of evolution of the phase relationship between voltage and leakage current; the harmonic content rate reflects the proportion of nonlinear components such as the third harmonic in the leakage current; and the polarization index reflects the insulation polarization process through the changing trend of the resistive leakage current within a preset time interval. The dynamic early warning threshold set consists of a resistive leakage current threshold and an equivalent dielectric loss tangent threshold. These two thresholds are first determined based on the historical sliding window baseline and standard deviation, and then multiplied by a dynamic threshold upward floating coefficient determined by the sliding window change rate and acceleration of resistive and capacitive leakage currents. The dynamic threshold upward floating coefficient increases accordingly with the increase of the sliding window change rate and acceleration, allowing the threshold to rise appropriately when ambient humidity changes rapidly, reducing over-limit alarms caused by capacitive current fluctuations. Each item in the insulation moisture multidimensional feature vector is compared with the dynamic early warning threshold set, and the warning level (normal, warning, abnormal, or severe) is determined according to preset hysteresis conditions of continuous over-limit cycles and continuous fallback cycles, forming a graded early warning result. This result serves as an evidence source for the fusion mechanism of the entropy weight method and DS evidence theory in step S50, providing a graded basis for comprehensive risk assessment.

[0039] The multidimensional feature vector incorporates information from multiple dimensions, including resistive current proportion, increment, phase change, harmonic distortion, and polarization process, into the judgment, ensuring that early signs of insulation dampness are not overlooked due to a temporary decline in a single indicator when soil moisture fluctuates. The dynamic threshold feedforward control mechanism adjusts the threshold based on the sliding window change rate and acceleration of resistive and capacitive leakage currents. When humidity rises rapidly, capacitive current increases while resistive current has not yet changed synchronously; in this case, a slight increase in the threshold can prevent simple environmental fluctuations from being misjudged as insulation anomalies. When humidity changes gradually level off, the threshold returns to its base value, remaining sensitive to the true increment of resistive current. The tiered early warning system employs hysteresis conditions of continuous over-limit cycles and continuous decline cycles. It requires the feature quantity to exceed the limit for multiple consecutive evaluation cycles before upgrading the warning level, and exiting the warning requires the value to remain below the threshold by a certain percentage, preventing frequent alarms caused by threshold boundary fluctuations. The resulting graded early warning results can distinguish between normal, alert, abnormal, and severe states, providing evidence with clear graded meaning for the integration of the entropy weight method and DS evidence theory in step S50. This enables the comprehensive risk assessment to allocate corresponding weights according to the severity of the early warning level, thereby outputting operation and maintenance decision recommendations that are more in line with the actual situation on site.

[0040] Traditional fixed threshold methods, when soil moisture increases, can lead to an increase in capacitive leakage current, causing characteristic quantities such as total leakage current or equivalent dielectric loss tangent to exceed limits. If the threshold is not adjusted according to the environment, false alarms are likely to occur. Conversely, setting the threshold too high may mask the increase in resistive current in the early stages of insulation dampness. This step uses a dynamic threshold feedforward control mechanism to determine the float coefficient based on the sliding window change rate and acceleration of resistive and capacitive leakage currents. This allows the threshold to float appropriately when humidity changes rapidly and return to the base value when humidity is stable, thereby improving the ability to identify the true development of insulation dampness without adding new characteristics. Simultaneously, the harmonic content and polarization index in the multidimensional feature vector can distinguish between simple humidity changes and nonlinear changes caused by insulation dampness. This is because simple humidity changes mainly affect the fundamental component of capacitive current, while insulation dampness introduces more harmonic components into the leakage current and alters the polarization process. The hysteresis control of the graded early warning system requires that the characteristic quantity continuously exceeds the limit before triggering an alarm, and that the value remain below the threshold by a certain percentage before exiting the alarm. This avoids the problem of repeated alarm triggering near the boundary in traditional threshold comparisons, making the output results more stable and reliable.

[0041] After a period of continuous rainfall, the groundwater level rises rapidly, and the rate of change in soil moisture reaches a high level with positive acceleration. The resistive leakage current output in step S30 remains relatively stable, while the capacitive leakage current increases significantly with rising moisture. Step S40 determines the dynamic threshold upward floating coefficient based on the sliding window change rate and acceleration of the resistive and capacitive leakage currents, thereby adjusting the resistive leakage current threshold and the equivalent dielectric loss tangent threshold accordingly. At this time, although the capacitive current increases significantly, the resistive current does not increase synchronously, and the equivalent dielectric loss tangent remains below the attention threshold, maintaining the system in a normal state. As insulation moisture gradually develops, the increase in resistive current exceeds the preset increment threshold for several consecutive evaluation cycles, and the third harmonic content reaches the preset harmonic content threshold, causing the system to enter a attention state. If the resistive current continues to rise and the rate of change of phase difference continues to increase, it is upgraded to an abnormal state. The graded early warning result is output to step S50 as an evidence source for the fusion of the entropy weight method and DS evidence theory, which maintenance personnel can use to arrange maintenance.

[0042] Step S50: Based on the aforementioned graded early warning results, a comprehensive risk assessment task is performed using the entropy weight method and DS evidence theory fusion mechanism, and comprehensive risk assessment and operation and maintenance decision recommendations are output.

[0043] The graded early warning result received in step S50 is essentially a state description formed by normalizing characteristic quantities such as resistive leakage current increment, equivalent dielectric loss tangent, harmonic content, and phase difference change rate within a sliding time window, as well as the corresponding attention or abnormality level, as described in step S40. These characteristic quantities originate from the resistive and capacitive leakage currents separated in step S30 and are combined into a multi-dimensional feature vector in step S40. Therefore, each piece of evidence in this step carries a local risk assessment of the same cable under the current environment. The entropy weight method first determines the weight based on the degree of fluctuation of each characteristic quantity within different time windows: the more significant the fluctuation of a characteristic quantity, the smaller its information entropy, and the greater the weight it is assigned, thus highlighting the quantities that are more sensitive to risk changes in the comprehensive score. The DS evidence theory treats the graded warning results corresponding to each feature quantity as independent evidence bodies, using a basic probability allocation function to characterize their degree of support for different risk levels. When there is significant conflict between evidence bodies, the evidence bodies are discounted according to the feature quantity weights and evidence credibility. The discount factor is jointly determined by the fusion adjustment coefficient, feature quantity weights, and evidence credibility. The discounted evidence is then fused to obtain a comprehensive risk score and a final risk level. The final output comprehensive risk assessment and operation and maintenance decision recommendations consist of a comprehensive risk score, a final risk level, and differentiated operation and maintenance decision recommendations. The differentiated operation and maintenance decision recommendations are generated based on the risk source and risk level. For example, a blue alert is output when capacitive current increases but resistive current does not, and a red alarm is output when resistive current increases and harmonic fingerprints are abnormal.

[0044] In this step, the entropy weight method adaptively allocates weights based on the degree of variation of each feature quantity within different time windows, enabling the comprehensive risk score to reflect the actual distinguishing ability of each feature quantity under the current operating conditions, rather than using fixed weights or relying on subjective experience. The DS evidence theory fuses the graded warning results from different feature sources. When the evidence conflict coefficient exceeds a preset conflict threshold, each piece of evidence is discounted according to the feature quantity weight and evidence credibility. The discount factor is jointly determined by the fusion adjustment coefficient, feature quantity weight, and evidence credibility. The discounted evidence is then redistributed and fused, reducing the interference of highly conflicting evidence on the final decision. When evidence conflict is severe, the system initiates a rapid review sampling mode, using a higher sampling rate to obtain more refined data for re-separation of resistive and capacitive components and feature calculation. This supplements effective information when information is insufficient or evidence is contradictory, improving the credibility of the final risk level. The output comprehensive risk assessment and operation and maintenance decision recommendations, combined with the risk source and risk level, provide blue alerts or red warnings, enabling operation and maintenance personnel to take differentiated measures based on the risk level and risk source, avoiding over-inspection or missed inspections.

[0045] Traditional multi-indicator assessments often employ fixed weights or simple logical "OR" rules. When the conclusions of different indicators are inconsistent, it is difficult to provide a comprehensive risk assessment, especially when changes in soil moisture cause an increase in capacitive current while resistive current does not, a single indicator may be misjudged as high-risk. This step dynamically adjusts weights using the entropy weight method and integrates different evidence using DS evidence theory. When conflicts exist between features, a discount redistribution is performed instead of simple voting, which can reduce false alarms caused by environmental interference. When the evidence conflict coefficient is greater than the preset conflict threshold, traditional methods may directly ignore conflicting evidence or use simple averaging, leading to highly conflicting evidence having an unreasonable impact on the final decision. This step discounts the evidence according to feature weights and evidence credibility. The discount factor is jointly determined by the fusion adjustment coefficient, feature weights, and evidence credibility, thus reasonably suppressing the contribution of highly conflicting evidence in the fusion process. When evidence conflicts are severe, traditional methods cannot obtain more information and can only give uncertain conclusions based on existing contradictory evidence. This step initiates a rapid review sampling mode, obtaining more refined data with a higher sampling rate for re-separation and calculation, thereby supplementing effective information when evidence conflicts are high and improving the reliability of the comprehensive risk assessment.

[0046] Following a period of continuous rainfall, the groundwater level rises rapidly, and the rate of change in soil moisture reaches a high level with positive acceleration. The resistive leakage current output in step S30 remains relatively stable, while the capacitive leakage current increases significantly with rising moisture. Step S40 determines the dynamic threshold upward floating coefficient based on the sliding window change rate and acceleration of the resistive and capacitive leakage currents, accordingly increasing the resistive leakage current threshold and the equivalent dielectric loss tangent threshold. At this point, although the capacitive current increases significantly, the resistive current does not increase synchronously, and the equivalent dielectric loss tangent remains below the attention threshold, maintaining the system in a normal state. As insulation moisture gradually develops, the resistive current increment exceeds the preset increment threshold for several consecutive evaluation cycles, and the third harmonic content reaches the preset harmonic content threshold, causing the system to enter an attention state. If the resistive current continues to rise and the phase difference change rate continues to increase, it is upgraded to an abnormal state. The graded early warning result is output to step S50 as an evidence source for the fusion of the entropy weight method and DS evidence theory. Within an assessment period, the increase in resistive leakage current reaches the attention threshold, but the equivalent dielectric loss tangent and harmonic characteristics are within the normal range. The evidence points inconsistently to different risk levels, and the conflict coefficient rises above the preset conflict threshold. The entropy weight method assigns weights based on the stability of each characteristic quantity. After DS fusion, the overall risk score is in the low attention range, and the system outputs a blue alert, suggesting enhanced monitoring rather than immediate repair. If subsequent assessments show that both resistive current and harmonic anomalies indicate severity, and the conflict coefficient exceeds the preset conflict threshold, the system initiates rapid re-sampling, collecting data at a higher sampling rate, re-separating resistive and capacitive components, and outputting a red alarm after confirming the insulation moisture trend. It also suggests prioritizing power outage repairs.

[0047] Example 2: Furthermore, the present invention provides a dynamic assessment and early warning system for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion. This system employs a method for dynamic assessment and early warning of safety risks of direct-buried cable operation and maintenance based on multi-source data fusion as described in the above embodiments, and can solve the technical problem of dynamic assessment and early warning of safety risks of direct-buried cable operation and maintenance based on multi-source data fusion. The beneficial effects of the dynamic assessment and early warning system for safety risks of direct-buried cable operation and maintenance based on multi-source data fusion provided by the present invention are the same as those of the method for dynamic assessment and early warning of safety risks of direct-buried cable operation and maintenance based on multi-source data fusion provided in the above embodiments. Other technical features of the dynamic assessment and early warning system for safety risks of direct-buried cable operation and maintenance based on multi-source data fusion are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0048] Example 3: This invention provides a dynamic assessment and early warning device for the safety risks of direct-buried cable maintenance based on multi-source data fusion. The device includes at least one processor and a memory communicatively connected to the processor. The memory stores instructions executable by the processor, which are then executed to enable the processor to perform the dynamic assessment and early warning method for the safety risks of direct-buried cable maintenance based on multi-source data fusion described in Example 1. This embodiment of the invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This dynamic assessment and early warning device for the safety risks of direct-buried cable maintenance based on multi-source data fusion is merely an example and should not limit the functionality or scope of the invention. A dynamic assessment and early warning device for the safety risks of direct-buried cable maintenance based on multi-source data fusion may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.), which can execute various appropriate actions and processes according to a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of the device. The processing unit, read-only memory, and random access memory are interconnected via a bus. An I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the device to exchange data wirelessly or wiredly with other devices. While a dynamic assessment and early warning device for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion has been described, it should be understood that the implementation of all the described systems is not required. More or fewer systems may be implemented alternatively.

[0049] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion. The computer program product provided by this invention can solve the technical problem of dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion provided in the above embodiments, and will not be repeated here.

[0050] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0051] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0052] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion, characterized in that, The method includes: Step S10: Acquire multi-source monitoring signals of the direct-buried cable; based on the multi-source monitoring signals of the direct-buried cable, perform synchronous sampling tasks using a time base deviation compensation mechanism, and output synchronous sampling sequences; Step S20: Based on the synchronous sampling sequence, the fundamental and harmonic feature extraction task is performed using wavelet denoising and windowed interpolation Fourier transform mechanism, and the fundamental and harmonic feature parameter set is output; Step S30: Based on the fundamental and harmonic characteristic parameter set, the online separation task of resistive and capacitive components is performed using a time delay compensation and forgetting factor recursive least squares mechanism, and the separation results of resistive leakage current and capacitive leakage current are output. Step S40: Based on the separation results of resistive leakage current and capacitive leakage current, a hierarchical early warning task is executed using a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism, and the hierarchical early warning results are output. Step S50: Based on the aforementioned graded early warning results, a comprehensive risk assessment task is performed using the entropy weight method and DS evidence theory fusion mechanism, and comprehensive risk assessment and operation and maintenance decision recommendations are output.

2. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 1, characterized in that, Step S10 involves acquiring multi-source monitoring signals from the buried cable, performing a synchronous sampling task based on the multi-source monitoring signals from the buried cable using a time base deviation compensation mechanism, and outputting a synchronous sampling sequence. Specifically, this includes: Step S101: Synchronously collect the grounding current of the metal sheath, leakage current of the main insulation, soil moisture, ambient temperature and reference voltage signals of the direct-buried cable, and combine each signal and its timestamp into the multi-source monitoring signal of the direct-buried cable; Step S102: Calculate the sampling time deviation between the grounding current of the metal sheath and the leakage current of the main insulation. According to the sampling time deviation, the sampling period, and the value of the leakage current of the main insulation at the current sampling time and the adjacent sampling times, use second-order Lagrange interpolation to perform time base correction on the leakage current of the main insulation to obtain the time base correction signal. Step S103: Use the synchronization timing pulse to discipline the sampling clock, calculate the difference between the actual sampling time after discipline and the product of the sampling sequence number and the sampling period to obtain the crystal oscillator drift error, and use the crystal oscillator drift error as the phase alignment compensation feedforward of the time base correction signal to obtain the synchronization sampling sequence.

3. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 1, characterized in that, Step S20, based on the synchronous sampling sequence, involves performing the fundamental and harmonic feature extraction task using wavelet denoising and windowed interpolation Fourier transform mechanisms, and outputting the fundamental and harmonic feature parameter sets. This step specifically includes: Step S201: Perform wavelet transform on the metal sheath grounding current and main insulation leakage current in the synchronous sampling sequence, determine the scale adaptive threshold according to the noise standard deviation of each decomposition layer, the number of wavelet coefficients and the decomposition layer number, and perform joint threshold shrinking on the wavelet coefficients of the main insulation leakage current in combination with the corresponding wavelet coefficients of the metal sheath grounding current to obtain the denoised current sequence. Step S202: Perform zero-crossing detection and digital phase-locked loop processing on the denoised current sequence to obtain the fundamental frequency estimate. Determine the frequency interpolation coefficients of the windowed interpolation Fourier transform according to the fundamental frequency estimate. Correct the fundamental amplitude and fundamental phase according to the frequency interpolation coefficients to obtain the frequency correction characteristics. Step S203: Extract the fundamental amplitude, fundamental phase, phase difference, harmonic content and total harmonic distortion rate of the metal sheath grounding current and the main insulation leakage current according to the frequency correction characteristics, and combine the extraction results to form the fundamental and harmonic characteristic parameter set.

4. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 3, characterized in that, In step S202, the denoised current sequence undergoes zero-crossing detection and digital phase-locked loop processing to obtain a fundamental frequency estimate. The frequency interpolation coefficients for windowed interpolation Fourier transform are determined based on the fundamental frequency estimate. The fundamental amplitude and phase are then corrected according to these frequency interpolation coefficients to obtain the frequency correction characteristics. Specifically, this step includes updating the fundamental frequency estimate according to a proportional-integral feedback relationship, where the proportional-integral feedback relationship is: in, For the first The fundamental frequency estimate for each time window For the first The fundamental frequency estimate for each time window This is the proportionality coefficient. The integral coefficient is... For the first Frequency error of each time window For the first Frequency error of each time window Let m be the current time window number, and m be the summation index, ranging from 0 to k. The frequency interpolation coefficients are determined based on the ratio of the updated fundamental frequency estimate to the frequency resolution. The sum of the amplitudes on both sides of the fundamental peak spectral line is multiplied by the ratio of two to the number of sampling points and the amplitude correction factor determined by the frequency interpolation coefficients to obtain the corrected fundamental amplitude. The orthogonal phase compensation and the leakage phase compensation determined by the frequency interpolation coefficients are added to the phase of the fundamental peak spectral line to obtain the corrected fundamental phase, thus forming the frequency correction feature.

5. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 1, characterized in that, Step S30, based on the fundamental and harmonic characteristic parameter set, employs a time-delay compensation and forgetting factor recursive least squares mechanism to perform online separation of resistive and capacitive components, and outputs the separation results of resistive and capacitive leakage currents. Specifically, this includes: Step S301: Obtain the soil moisture time series characteristics from the set of fundamental and harmonic characteristic parameters, perform weighted fusion of soil moisture at different measuring points at the cable laying depth, and use a first-order inertial element for time delay compensation to obtain the compensated equivalent humidity. Step S302: Determine the relative permittivity of the soil based on the quadratic relationship between the compensated equivalent humidity and the relative permittivity of the soil. Multiply the initial equivalent capacitance of the sheath to the ground by the humidity sensitivity exponent power of the ratio of the relative permittivity of the soil to the reference relative permittivity to obtain the equivalent capacitance of the sheath to the ground. Multiply the angular frequency, the equivalent capacitance of the sheath to the ground, and the amplitude of the cable phase voltage to obtain the predicted value of the capacitive current. Step S303: Construct a regression vector using the cable phase voltage and its rate of change, and construct a parameter vector using the equivalent insulation conductance and equivalent insulation capacitance. Update the forgetting factor within a preset forgetting factor range based on the soil moisture change rate, and perform recursive least squares parameter update in conjunction with the predicted capacitive current value. Multiply the updated equivalent insulation conductance by the effective value of the cable phase voltage to obtain the resistive leakage current, and multiply the angular frequency, the updated equivalent insulation capacitance, and the effective value of the cable phase voltage to obtain the capacitive leakage current, thus forming the separation result of the resistive leakage current and the capacitive leakage current.

6. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 1, characterized in that, Step S40, based on the separation results of resistive and capacitive leakage currents, employs a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism to execute a tiered early warning task and output tiered early warning results. This step specifically includes: Step S401: Construct a multidimensional feature vector of insulation moisture based on the ratio of resistive leakage current to capacitive leakage current, the percentage increase of resistive leakage current relative to its historical sliding window baseline, the phase difference change rate, the harmonic content rate, and the polarization index. Step S402: Determine the dynamic threshold floating coefficient based on the sliding window change rate and change acceleration of the resistive leakage current and the capacitive leakage current, and apply the dynamic threshold floating coefficient to the resistive leakage current threshold and the equivalent dielectric loss tangent threshold determined by the historical sliding window baseline and the sliding window standard deviation to obtain the dynamic early warning threshold set; Step S403: Compare the insulation moisture multidimensional feature vector with the dynamic early warning threshold set, and determine the early warning level of normal, attention, abnormal or serious according to the preset hysteresis conditions of continuous over-limit cycle and continuous fall-off cycle to form the graded early warning result.

7. The method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in claim 1, characterized in that, In step S50, based on the aforementioned graded early warning results, a comprehensive risk assessment task is performed using a fusion mechanism of entropy weight method and DS evidence theory, outputting comprehensive risk assessment and operation and maintenance decision recommendations. This specifically includes: Step S501: Obtain the normalized values ​​of each feature quantity in each time window from the graded early warning results, determine the weight of each feature quantity according to the inverse relationship between information entropy and feature quantity weight, and perform weighted summation of the scores of each feature quantity to obtain the initial comprehensive risk score. Step S502: Calculate the evidence conflict coefficient based on the basic probability allocation function of each piece of evidence. When the evidence conflict coefficient is greater than a preset conflict threshold, discount each piece of evidence according to the feature weight and the credibility of the evidence. The discount factor used for the discount is: in, For the first Discount factor for each piece of evidence To preset the fusion adjustment coefficient, For the first The weights of each feature quantity For the first The credibility of individual pieces of evidence. The evidence is numbered; the conflicting evidence is redistributed using the discount factor, and the redistributed evidence fusion result is combined with the initial comprehensive risk score to obtain the comprehensive risk score and the final risk level; Step S503: Based on the comprehensive risk score, the final risk level and the corresponding risk source, generate differentiated operation and maintenance decision suggestions, and combine the comprehensive risk score, the final risk level and the differentiated operation and maintenance decision suggestions to form the comprehensive risk assessment and operation and maintenance decision suggestions.

8. A dynamic assessment and early warning system for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion, applied to the dynamic assessment and early warning method for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion as described in any one of claims 1 to 7, characterized in that, The system includes: The multi-source data synchronous acquisition module is used to acquire multi-source monitoring signals of directly buried cables. Based on the multi-source monitoring signals of the directly buried cables, a time base deviation compensation mechanism is used to perform synchronous sampling tasks and output synchronous sampling sequences. The fundamental and harmonic feature extraction module is used to perform the fundamental and harmonic feature extraction task based on the synchronous sampling sequence, using wavelet denoising and windowed interpolation Fourier transform mechanism, and output the fundamental and harmonic feature parameter set; The online separation module for resistive and capacitive components is used to perform the online separation task of resistive and capacitive components based on the fundamental and harmonic characteristic parameter set, using a time delay compensation and forgetting factor recursive least squares mechanism, and output the separation results of resistive leakage current and capacitive leakage current. The insulation moisture graded early warning module is used to perform graded early warning tasks based on the separation results of resistive leakage current and capacitive leakage current, using a multi-dimensional feature vector construction and dynamic threshold feedforward control mechanism, and output graded early warning results. The comprehensive risk assessment and operation and maintenance decision module is used to perform comprehensive risk assessment tasks based on the hierarchical early warning results, using a fusion mechanism of entropy weight method and DS evidence theory, and output comprehensive risk assessment and operation and maintenance decision recommendations.

9. A dynamic assessment and early warning device for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion, characterized in that, The device for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion includes: a memory, a processor, and a program for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion, which is stored in the memory and can run on the processor. When the program for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion is executed by the processor, it implements the method for dynamic assessment and early warning of safety risks in the operation and maintenance of directly buried cables based on multi-source data fusion as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a dynamic assessment and early warning program for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion. When the processor executes the dynamic assessment and early warning program for the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion, it implements a method for dynamic assessment and early warning of the safety risks of direct-buried cable operation and maintenance based on multi-source data fusion as described in any one of claims 1 to 7.

Images