Intelligent Detection and Evaluation Method and System for Dielectric Loss, Withstand Voltage and Partial Discharge of Distribution Network Cables

Abstract

The present invention discloses an intelligent detection and evaluation method and system for dielectric loss, withstand voltage and partial discharge of distribution network cables, which relates to the technical field of power equipment condition monitoring, and includes: synchronously collecting the dielectric loss spectrum, withstand voltage leakage current curve and partial discharge pattern of the distribution network cable through a phase-resolved measurement device, and applying a frequency-domain adaptive filter to process the partial discharge pattern to generate a denoised partial discharge characteristic pattern; constructing a hypersphere model of the normal domain of cable multi-dimensional parameters and calculating the deviation distance value; inputting the deviation distance value into a fuzzy inference engine, performing non-linear conversion according to a preset fuzzy rule base, and generating an insulation state evaluation index and a fault risk probability distribution map of the distribution network cable, so as to realize the accurate evaluation of the cable insulation state. The present invention realizes the accurate evaluation of the cable insulation state and the accurate positioning of the fault location, and improves the operation reliability and maintenance efficiency of the distribution network.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment condition monitoring, and specifically to an intelligent detection and evaluation method and system for dielectric loss, withstand voltage, and partial discharge of distribution network cables. Background Technique

[0002] Distribution network cables are an important part of the urban power system, and their insulation status is directly related to the safe and reliable operation of the distribution network. With the continuous expansion of the scale of the distribution network and the accumulation of cable usage time, cable insulation aging and fault risk assessment have become important topics in power system operation and maintenance. Traditional cable insulation assessment mainly relies on single-parameter tests, such as dielectric loss tests, withstand voltage tests, or partial discharge tests. These methods are carried out independently and are difficult to comprehensively reflect the comprehensive insulation status of the cable. The dielectric loss test can reflect the degree of macroscopic insulation deterioration, but is insensitive to early defects; the withstand voltage test can verify the short-term withstand voltage ability of the cable, but cannot reflect the long-term operation performance of the insulation; the partial discharge test is sensitive to local defects, but is severely affected by environmental noise, and the data interpretation is complex. This single-parameter assessment method has one-sidedness and limitations in practical applications, and it is difficult to accurately judge the actual state of the cable. Especially in complex environments and under the influence of multiple factors, it is easy to cause misjudgment or missed judgment.

[0003] Existing cable condition assessment systems generally adopt a simple threshold judgment method, comparing the measurement results with a fixed threshold, and determining the cable condition according to the degree of exceeding the threshold. This method ignores the mutual correlation between cable parameters, cannot adapt to the characteristic differences of cables of different types, different operating environments, and different aging stages, and the assessment results highly depend on the rationality of the threshold setting. In addition, the existing technology lacks the ability to accurately locate the spatial distribution of cable fault risks. Most assessments only give the overall status level and cannot indicate the location of specific risk points, which is not conducive to precise maintenance. At the same time, the traditional system has insufficient intelligence in signal processing and data analysis, has limited noise suppression ability for partial discharge signals, and is difficult to extract effective features from complex backgrounds. However, the intelligent detection and evaluation method and system for dielectric loss, withstand voltage, and partial discharge of distribution network cables provided by the present invention solve the problems existing in the above technologies. The present invention belongs to the technical field of power equipment condition monitoring and evaluation. Through multi-parameter collaborative detection, intelligent signal processing, data-driven modeling, and fuzzy inference evaluation, it realizes the accurate assessment of cable insulation status and the accurate location of fault positions, improving the operation reliability and maintenance efficiency of the distribution network. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent detection and evaluation method and system for dielectric loss, withstand voltage, and partial discharge of distribution network cables, which can solve the problems mentioned in the background technique.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: an intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables, including: synchronously collecting the dielectric loss spectrum, withstand voltage leakage current curve, and partial discharge pattern of the distribution network cable through a phase-resolved measurement device, and applying a frequency-domain adaptive filter to process the partial discharge pattern to generate a denoised partial discharge feature pattern; the frequency-domain adaptive filter includes a fast Fourier transform module, a spectrum analysis module, and an adaptive band-pass filter module;

[0007] Using the support vector data description algorithm, constructing a normal domain hypersphere model of cable multi-dimensional parameters based on historical normal sample data, and calculating the deviation distance values of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge feature pattern relative to the normal domain hypersphere; the normal domain hypersphere model includes kernel function mapping parameters, boundary relaxation parameters, and a model update mechanism;

[0008] Inputting the deviation distance value into a fuzzy inference engine, performing non-linear conversion according to a preset fuzzy rule base, generating an insulation state evaluation index and a fault risk probability distribution map of the distribution network cable, and realizing the accurate evaluation of the cable insulation state; the fuzzy inference engine includes a fuzzification processing module, a rule inference module, and a defuzzification processing module.

[0009] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of the distribution network cable of the present invention, wherein: applying a frequency-domain adaptive filter to process the partial discharge pattern includes:

[0010] Converting the partial discharge pattern to the frequency domain through the fast Fourier transform module;

[0011] Using the spectrum analysis module to identify the characteristic frequency bands of the signal and noise, and calculating the signal-to-noise ratio distribution;

[0012] Setting the filtering bandwidth through the adaptive band-pass filter module according to the signal-to-noise ratio distribution; wherein, when the signal-to-noise ratio distribution presents a single-peak characteristic, setting narrow-band filtering parameters; when the signal-to-noise ratio distribution presents a multi-peak characteristic, setting multi-channel filtering parameters to complete the enhancement and noise suppression of the partial discharge pattern.

[0013] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of the distribution network cable of the present invention, wherein: using the support vector data description algorithm to construct a normal domain hypersphere model of cable multi-dimensional parameters based on historical normal sample data includes:

[0014] Using the radial basis function as the kernel function mapping parameter to map the distribution network cable multi-dimensional parameters to a high-dimensional feature space;

[0015] Control the sensitivity of the normal domain hypersphere model to outliers through the boundary relaxation parameter; wherein, the boundary relaxation parameter is related to the characteristic distribution range of the historical normal sample data;

[0016] According to the model update mechanism, the normal domain hypersphere model is triggered to be updated when one of the following conditions is met: the number of newly added normal samples reaches a predetermined ratio, and the statistical characteristics of the deviation distance value exceed a preset reference range.

[0017] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables according to the present invention, wherein: the fuzzification processing module converts the deviation distance value into a fuzzy set; wherein, the fuzzy set includes four levels: normal, slightly abnormal, moderately abnormal, and severely abnormal;

[0018] The rule inference module performs inference calculations according to the fuzzy rule base to obtain a fuzzy inference result; the fuzzy rule base contains fuzzy rules based on the deviation distance value of the dielectric loss spectrum and the deviation distance value of the partial discharge characteristic map;

[0019] The defuzzification processing module uses the centroid method to convert the fuzzy inference result into the insulation state evaluation index and the fault risk probability distribution map.

[0020] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables according to the present invention, wherein: the fuzzy inference result includes: when the deviation distance value of the dielectric loss spectrum is greater than a first threshold and the deviation distance value of the partial discharge characteristic map is greater than a second threshold, it is determined to be severely abnormal; when the deviation distance value of the dielectric loss spectrum is greater than the first threshold and the deviation distance value of the partial discharge characteristic map is not greater than the second threshold, it is determined to be moderately abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic map is greater than the second threshold, it is determined to be slightly abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic map is not greater than the second threshold, it is determined to be normal.

[0021] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables according to the present invention, wherein: the calculation of the deviation distance value includes:

[0022] Taking the center of the normal domain hypersphere as the origin, establish a kernel function mapping coordinate system;

[0023] Calculate the Euclidean distances of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic map in the kernel function mapping coordinate system;

[0024] Determine the deviation distance value according to the ratio of the Euclidean distance to the radius of the normal domain hypersphere.

[0025] As a preferred solution of the intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to the present invention, wherein: the generation process of the fault risk probability distribution map includes:

[0026] Divide the distribution network cable into several monitoring sections along the length direction;

[0027] Calculate the partial discharge density, discharge amplitude and discharge frequency for each of the several monitoring sections respectively;

[0028] Based on the partial discharge density, the discharge amplitude and the discharge frequency, calculate the risk coefficient of each monitoring section; wherein, when the partial discharge density and the discharge amplitude increase or decrease simultaneously, the risk coefficient is proportional to the product of the partial discharge density and the discharge amplitude; when the partial discharge density increases while the discharge amplitude decreases or the partial discharge density decreases while the discharge amplitude increases, the risk coefficient remains within a preset range;

[0029] Combined with the insulation state evaluation index, visually display the fault risk probability of the several monitoring sections in the form of a heat map.

[0030] To further solve the above technical problems, the present invention provides the following technical solution: an intelligent detection and evaluation system for dielectric loss, withstand voltage and partial discharge of distribution network cables, including: a signal processing unit, configured to synchronously collect the dielectric loss spectrum, withstand voltage leakage current curve and partial discharge pattern of the distribution network cable through a phase-resolved measurement device, and apply a frequency-domain adaptive filter to process the partial discharge pattern to generate a denoised partial discharge feature pattern;

[0031] A model construction unit, configured to use the support vector data description algorithm to construct a cable multi-dimensional parameter normal domain hypersphere model based on historical normal sample data, and calculate the deviation distance values of the dielectric loss spectrum, the withstand voltage leakage current curve and the denoised partial discharge feature pattern relative to the normal domain hypersphere;

[0032] A risk assessment unit, configured to input the deviation distance value into a fuzzy inference engine, perform non-linear conversion according to a preset fuzzy rule base, generate an insulation state evaluation index and a fault risk probability distribution map of the distribution network cable, and realize accurate evaluation of the cable insulation state.

[0033] A computer device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps of the intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of the distribution network cable as described above are implemented.

[0034] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, the steps of the intelligent detection and evaluation method for the dielectric loss, withstand voltage and partial discharge of the distribution network cable as described above are realized.

[0035] Advantages of the present invention: The intelligent detection and evaluation method and system for the dielectric loss, withstand voltage and partial discharge of the distribution network cable provided by the present invention have significant technical advantages: First, by synchronously collecting three key parameters through a phase-resolved measurement device, multi-dimensional cable state characterization is realized, overcoming the one-sidedness of traditional single-parameter detection; Second, an adaptive filter in the frequency domain is applied to process the partial discharge pattern, and the filtering strategy is dynamically adjusted according to the unimodal or multimodal characteristics of the signal-to-noise ratio distribution, significantly improving the signal extraction ability in complex environments; Third, a hypersphere model of the normal domain of the multi-dimensional parameters of the cable is constructed using the support vector data description algorithm, and the radial basis function mapping parameter, boundary relaxation parameter and model update mechanism are introduced to achieve accurate modeling of non-linear data distribution and objective quantification of abnormal states; Finally, the deviation distance value is input into the fuzzy inference engine for non-linear conversion, and a heat map of the fault risk probability distribution is generated by combining the partial discharge density, discharge amplitude and discharge frequency, which not only realizes the accurate assessment of the insulation state, but also can accurately locate the specific fault location. In addition, the system also has the ability of adaptive learning and dynamic update, and can automatically adjust the evaluation parameters with the aging process of the cable, providing a scientific basis for preventive maintenance decision-making, effectively reducing the failure rate of the distribution network, and improving the reliability and economic benefits of the power grid operation. Description of the Drawings

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0037] Figure 1 It is a schematic diagram of the overall process of the intelligent detection and evaluation method for the dielectric loss, withstand voltage and partial discharge of the distribution network cable proposed by the present invention;

[0038] Figure 2 It is a diagram of a computer device in the intelligent detection and evaluation method for the dielectric loss, withstand voltage and partial discharge of the distribution network cable proposed by the present invention. Detailed Embodiments

[0039] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0040] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0041] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides an intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables.

[0042] Figure 1 Figure 1 shows a schematic diagram of the overall process of the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables, including the following steps:

[0043] S1: Synchronously collect the dielectric loss spectrum, withstand voltage leakage current curve, and partial discharge pattern of the distribution network cable through a phase-resolved measurement device, and process the partial discharge pattern using a frequency-domain adaptive filter to generate a denoised partial discharge feature pattern.

[0044] Specifically, the frequency-domain adaptive filter includes a fast Fourier transform module, a spectrum analysis module, and an adaptive band-pass filter module.

[0045] Furthermore, the process of the frequency-domain adaptive filter processing the partial discharge pattern includes:

[0046] Convert the partial discharge pattern to the frequency domain through the fast Fourier transform module;

[0047] Use the spectrum analysis module to identify the characteristic frequency bands of the signal and noise, and calculate the signal-to-noise ratio distribution;

[0048] Set the filter bandwidth according to the signal-to-noise ratio distribution through the adaptive band-pass filter module; wherein, when the signal-to-noise ratio distribution shows a single-peak characteristic, set narrow-band filtering parameters; when the signal-to-noise ratio distribution shows a multi-peak characteristic, set multi-channel filtering parameters to complete the enhancement and noise suppression of the partial discharge pattern.

[0049] When implementing the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables of the present invention, a phase-resolved measurement device is first set up. The phase-resolved measurement device is specifically an integrated cable test equipment, which is composed of an AC high-voltage source, a precision measurement unit, a data acquisition system, and a control host. The AC high-voltage source provides a sine wave test voltage with an adjustable frequency of 0.1Hz - 1Hz and an amplitude range of 1kV - 10kV, which is suitable for testing distribution network cables of different voltage levels. The precision measurement unit includes a dielectric loss measurement circuit, a withstand voltage test circuit, and a partial discharge detection circuit, and each circuit has an independent sensor and signal conditioning module. The data acquisition system uses a 16-bit high-precision acquisition card with a sampling rate of up to 10MS / s and has a phase-locking function, which can synchronously collect various parameters within a phase accuracy range of 0.1°. Through the phase-resolved measurement device, synchronous acquisition of three key parameters of the distribution network cable - the dielectric loss spectrum, the withstand voltage leakage current curve, and the partial discharge pattern can be achieved, ensuring that all data is obtained under the same phase reference, providing a basis for subsequent comprehensive analysis.

[0050] During the actual measurement process, the phase-resolved measurement device first applies a standard test voltage to the distribution network cable, and the voltage waveform serves as a phase reference signal. At the same time, it triggers the data acquisition of the three measurement channels of dielectric loss, withstand voltage leakage current, and partial discharge. The dielectric loss spectrum is obtained through the frequency-domain scanning of the phase difference between voltage and current, and the frequency range is 0.01Hz - 100Hz; the withstand voltage leakage current curve records the leakage current value changing with time under the standard test voltage; the partial discharge pattern collects the partial discharge pulse signals generated by the cable during the test process and records information such as its amplitude, phase, and occurrence frequency.

[0051] The obtained partial discharge pattern usually contains various interference noises, such as electromagnetic interference, high-frequency switching power supply interference, and background white noise, etc. These noises will seriously affect the extraction and recognition of partial discharge characteristics. Therefore, the present invention applies the frequency-domain adaptive filter to process the partial discharge pattern to generate the denoised partial discharge characteristic pattern. The frequency-domain adaptive filter includes a fast Fourier transform module, a spectrum analysis module, and an adaptive band-pass filter module, and each module works together to effectively extract partial discharge signals.

[0052] In an optional embodiment, the specific process of applying the frequency-domain adaptive filter to process the partial discharge pattern is as follows: First, the partial discharge pattern is converted to the frequency domain through the fast Fourier transform module. The fast Fourier transform module uses a segmented FFT algorithm to perform a 1024-point transformation on the signal, and the optional window function is a Hanning window to reduce the spectrum leakage effect. After the transformation is completed, a complete frequency-domain representation, including the amplitude spectrum and the phase spectrum, can be obtained.

[0053] Next, the transformed frequency-domain signal is processed using the spectrum analysis module, which adopts an adaptive threshold algorithm to identify the characteristic frequency bands of the signal and noise. Specifically, by calculating the spectral energy distribution curve, the frequency band with concentrated energy is identified as the possible signal frequency band; at the same time, the stationary region of the spectrum is analyzed and identified as the noise frequency band. Based on these two parts of information, the signal-to-noise ratio distribution of each frequency band is calculated to generate a signal-to-noise ratio spectrogram, providing a basis for subsequent filter parameter settings.

[0054] In another optional embodiment, the spectrum analysis module can also combine wavelet analysis technology to extract signal characteristics at different scales, enhancing the ability to identify non-stationary discharge signals. This method is particularly suitable for signal processing in complex cable environments and can effectively distinguish signal components with similar spectral characteristics but different time-frequency distributions.

[0055] In the present invention, the unimodal and multimodal characteristics of the signal-to-noise ratio distribution are identified through the analysis of the frequency-domain signal characteristics. The unimodal characteristic means that the signal-to-noise ratio distribution curve shows a dominant peak region in the entire frequency domain, and the energy in this region occupies the main part of the total spectral energy (e.g., the preset range is more than 50%), and the energy in other frequency bands outside this region is lower than the main peak region. This distribution pattern indicates that the partial discharge signal shows a concentrated distribution characteristic in the spectrum and is usually associated with a single type of insulation defect.

[0056] In contrast, the multimodal characteristic means that the signal-to-noise ratio distribution curve shows multiple independent peak regions in the frequency domain, the energy distribution in each region is uniform, and there are non-overlapping frequency intervals between the peaks. This distribution pattern indicates that the partial discharge signal shows a dispersed characteristic in the spectrum and usually corresponds to multiple insulation defect sources or composite defects.

[0057] The spectrum morphology recognition algorithm is used to determine the signal-to-noise ratio distribution characteristics. It is identified by analyzing the number of frequency-domain energy clustering centers, the energy proportion of each clustering center, and the frequency interval between clusters, and the characteristic classification is completed in combination with the discharge physical model. The system selects determination parameters according to the test environment and cable type and optimizes the classification process through an iterative algorithm.

[0058] Finally, the adaptive band-pass filter module automatically sets the filter bandwidth according to the signal-to-noise ratio distribution. When the signal-to-noise ratio distribution shows a unimodal characteristic, the narrow-band filter parameters are set, the center frequency is located at the main peak point, and the bandwidth is determined based on the full width at half maximum value of the energy density curve.

[0059] When the signal-to-noise ratio distribution exhibits a multi-peak characteristic, it indicates that the partial discharge signals are distributed in multiple discontinuous frequency bands. At this time, the multi-channel filtering parameters are set, and independent band-pass filters are set for each peak region. The parameters of the filters are dynamically adjusted according to the characteristics of each peak region. The multi-channel filtering results are fused through the optimal ratio synthesis algorithm to complete the enhancement of the partial discharge pattern and noise suppression. In practical applications, the setting of the multi-channel filtering parameters also takes into account the spectral characteristics of different types of discharge sources and is optimized and adjusted for different cable fault types.

[0060] Exemplarily, specific numerical values are described as follows: When testing a 10 kV cross-linked polyethylene insulated distribution network cable that has been in operation for 5 years, the phase-resolved measurement device applies a test voltage of 8 kV, the sampling rate is set to 5 MS / s, and the acquisition duration is 60 seconds. The obtained original partial discharge pattern shows obvious discharge clusters in the phase intervals of 0° - 90° and 180° - 270°, but the signal is severely contaminated by 50 Hz power frequency interference and random noise, and the signal-to-noise ratio is only 6.8 dB. After being processed by the frequency-domain adaptive filter, the signal-to-noise ratio is increased to 23.5 dB. Specifically, the fast Fourier transform module performs a 1024-point FFT transformation on the discharge signal, and the obtained spectrum shows that the discharge signal is mainly distributed in two frequency bands of 0.5 MHz - 1.8 MHz and 2.3 MHz - 4.5 MHz, forming a typical multi-peak characteristic distribution. The signal-to-noise ratios of each peak region are 25.3 dB and 21.7 dB respectively. The adaptive band-pass filtering module accordingly sets two band-pass filtering channels, with center frequencies of 1.2 MHz and 3.6 MHz respectively, and bandwidths of ±0.6 MHz and ±1.0 MHz respectively. The filtered signals are fused through the weighted synthesis algorithm, and the weight ratio is 0.6:0.4. Finally, the denoised partial discharge characteristic pattern is obtained, clearly showing two discharge sources in the internal insulation structure, with discharge amplitudes of 25 pC and 42 pC respectively.

[0061] It should be noted that synchronously collecting the dielectric loss spectrum, the withstand voltage leakage current curve, and the partial discharge pattern of the distribution network cable through the phase-resolved measurement device and applying the frequency-domain adaptive filter to process the partial discharge pattern has many technical advantages and practical application values. First of all, synchronously collecting three key parameters can comprehensively reflect various aspects of the cable insulation, avoiding the one-sidedness and limitations brought by traditional single-parameter testing methods. Secondly, the frequency-domain adaptive filter can dynamically adjust the filtering strategy according to the actual characteristics of the signal. Compared with traditional fixed-parameter filtering methods, it has stronger adaptability and a higher signal-to-noise ratio improvement effect. Especially for the complex situation where the signal-to-noise ratio distribution shows a multi-peak characteristic, the setting of the multi-channel filtering parameters can effectively retain the effective signal components distributed in different frequency bands, avoiding signal distortion and information loss that may be caused by simple band-pass filtering. In addition, combining the analyzed denoised partial discharge characteristic pattern with the dielectric loss spectrum and the withstand voltage leakage current curve can realize multi-dimensional characterization of the cable insulation state, improve the accuracy and reliability of fault diagnosis, provide a scientific basis for preventive maintenance decisions, thereby extending the cable service life, reducing operation and maintenance costs, and improving the operation reliability of the distribution network.

[0062] S2: Using the support vector data description algorithm, based on historical normal sample data, construct a multi-dimensional parameter normal domain hypersphere model for the cable, and calculate the deviation distance values of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic pattern relative to the normal domain hypersphere.

[0063] Specifically, the normal domain hypersphere model includes kernel function mapping parameters, boundary relaxation parameters, and a model update mechanism.

[0064] Furthermore, using the support vector data description algorithm, based on historical normal sample data, constructing a multi-dimensional parameter normal domain hypersphere model for the cable includes:

[0065] Using the radial basis function as the kernel function mapping parameter, map the multi-dimensional parameters of the distribution network cable to a high-dimensional feature space;

[0066] Control the sensitivity of the normal domain hypersphere model to abnormal points through the boundary relaxation parameter; among them, the boundary relaxation parameter is related to the characteristic distribution range of the historical normal sample data;

[0067] According to the model update mechanism, when one of the following conditions is met, trigger the update of the normal domain hypersphere model: the number of newly added normal samples reaches a predetermined ratio, and the statistical characteristics of the deviation distance value exceed the preset reference range.

[0068] The calculation of the deviation distance value includes:

[0069] Taking the center of the normal domain hypersphere as the origin, a kernel function mapping coordinate system is established;

[0070] In the kernel function mapping coordinate system, calculate the Euclidean distances of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic pattern;

[0071] Determine the deviation distance value according to the ratio of the Euclidean distance to the radius of the normal domain hypersphere.

[0072] In the intelligent detection and evaluation method for dielectric loss, withstand voltage, and partial discharge of distribution network cables of the present invention, after obtaining the denoised partial discharge characteristic pattern, it is necessary to quantitatively evaluate the insulation state of the cable. Therefore, the present invention uses the support vector data description (SVDD) algorithm to construct a cable multi-dimensional parameter normal domain hypersphere model based on historical normal sample data, and calculates the deviation degree of the current detected cable parameters from this normal domain, so as to objectively evaluate the insulation state of the cable.

[0073] The support vector data description algorithm is a one-class classification method, and its core idea is to find a hypersphere with the smallest volume in the feature space, which contains most of the normal sample data points and excludes abnormal points at the same time. This algorithm determines the center and radius of the hypersphere by solving a quadratic programming problem, forming a boundary description of the normal data. In the cable state evaluation, the normal domain hypersphere model includes three key components: kernel function mapping parameters, boundary relaxation parameters, and model update mechanisms.

[0074] The specific implementation of constructing the normal domain hypersphere model using the support vector data description algorithm includes the following steps: First, select the detection records of healthy cables from the historical detection database as the normal sample set. Each sample contains three types of feature data: the frequency response feature vector of the dielectric loss spectrum, the time-domain feature vector of the withstand voltage leakage current curve, and the statistical feature vector of the denoised partial discharge characteristic pattern. These three types of feature vectors together constitute a multi-dimensional parameter space characterizing the insulation state of the cable.

[0075] Next, use the radial basis function as the kernel function mapping parameter to map the multi-dimensional parameters of the distribution network cable to a high-dimensional feature space. The radial basis function is defined as K(x,y)=exp(-||x - y||² / σ²), where σ is the kernel width parameter, which controls the feature resolution of the mapping space. The selection of the radial basis function is based on its excellent performance in dealing with non-linear data distributions, which can map the complex distributions in the original feature space to simple structures in the high-dimensional space. In practical applications, the kernel width parameter σ is determined by the cross-validation method to obtain the best classification effect.

[0076] Then, the sensitivity of the normal domain hypersphere model to outliers is controlled by the boundary relaxation parameter. The boundary relaxation parameter is denoted as C, and its value range is (0, 1]. Its value is related to the characteristic distribution range of the historical normal sample data. When the normal sample distribution is concentrated, a smaller C value is set to make the model boundary more compact; when the normal sample distribution is dispersed, a larger C value is set to allow the model boundary to be more relaxed. The introduction of the boundary relaxation parameter enhances the robustness of the model to noise and outliers and avoids overfitting problems.

[0077] To adapt to the dynamic changes in the cable operating state, the present invention designs the model update mechanism. According to the model update mechanism, the update of the normal domain hypersphere model is triggered when one of the following conditions is met: the number of newly added normal samples reaches a predetermined ratio, or the statistical characteristics of the deviation distance value exceed a preset reference range. Specifically, when the number of newly added normal samples exceeds 20% of the total number of current model training samples, or when there is an obvious change trend in the mean value of the deviation distance values measured continuously for 30 times, the system will trigger the model retraining process to update the parameters of the normal domain hypersphere model. This dynamic update mechanism ensures that the model can adaptively adjust with the cable aging process and environmental changes, maintaining the accuracy of the evaluation.

[0078] During the actual detection process, it is necessary to calculate the deviation distance value between the current cable parameters and the normal domain hypersphere as a quantitative index for evaluating the cable insulation state. The calculation method of the deviation distance value includes three steps: First, a kernel function mapping coordinate system is established with the center of the normal domain hypersphere as the origin. The center of the hypersphere is obtained by solving the dual problem of the support vector data description algorithm and is expressed as a weighted sum of the support vectors.

[0079] Second, the Euclidean distances of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic pattern are calculated in the kernel function mapping coordinate system. For the currently measured cable parameters, they are mapped to the high-dimensional feature space through the kernel function K, and then the Euclidean distance from the mapped point to the center of the hypersphere is calculated. This distance characterizes the difference degree between the current cable parameters and the normal state parameter set.

[0080] Finally, the deviation distance value is determined according to the ratio of the Euclidean distance to the radius of the normal domain hypersphere. The specific calculation formula is: deviation distance value = Euclidean distance / hypersphere radius. When the deviation distance value is less than 1, it indicates that the current cable parameters fall within the normal domain; when the deviation distance value is greater than 1, it indicates that the current cable parameters fall outside the normal domain, and the larger the value, the higher the degree of abnormality.

[0081] The deviation distance value calculated by the above method provides an objective quantitative index for subsequent fuzzy inference and cable condition assessment. Compared with the traditional threshold judgment method, the support vector data description algorithm can handle high-dimensional non-linear data distributions and automatically adapt to the statistical characteristics of the data, achieving a more accurate assessment of the cable insulation condition.

[0082] Exemplarily, in the detection of a 10 kV cross-linked polyethylene cable in a substation, 50 groups of historical normal samples were collected to construct the normal domain hypersphere model. Each group of samples contains 15 frequency point data of the dielectric loss spectrum, 10 characteristic points of the withstand voltage leakage current curve, and 8 statistical characteristics of the partial discharge characteristic pattern, forming a 33-dimensional parameter space in total. These parameters were mapped to a high-dimensional feature space through a radial basis function, and the kernel width parameter was set to the square root of the average distance between samples. The normal domain hypersphere model was trained using the support vector data description algorithm, where the number of support vectors is 12, accounting for 24% of the total samples, indicating that the model has good generalization ability.

[0083] A cable that has been in operation for 5 years was detected. The dielectric loss spectrum obtained shows a slight increase in the high-frequency band, the withstand voltage leakage current curve remains stable, and the partial discharge characteristic pattern shows weak discharge activity in a specific phase interval. These parameters were input into the normal domain hypersphere model, and the calculated deviation distance value of the dielectric loss spectrum is 1.08, the deviation distance value of the withstand voltage leakage current curve is 0.85, and the deviation distance value of the partial discharge characteristic pattern is 1.12. These results indicate that the insulation condition of the cable has started to show slight anomalies but has not reached a severe level, which is consistent with the cable service life and actual operating conditions.

[0084] It should be noted that the method of constructing a cable multi-dimensional parameter normal domain hypersphere model using the support vector data description algorithm and calculating the parameter deviation distance value breaks through the limitations of traditional fixed threshold judgment and can adapt to the characteristic differences of different types of cables and changes in the operating environment. The radial basis function, as the kernel function mapping parameter, effectively solves the problem of non-linear distribution of cable parameters; the boundary relaxation parameter improves the robustness of the model to outliers; the model update mechanism ensures that the evaluation system can be dynamically adjusted along with the cable aging process. This data-driven condition assessment method provides an objective and accurate decision-making basis for power system operation and maintenance personnel, helps to achieve predictive maintenance of distribution network cables, reduce failure rates, and improve the reliability of power grid operation.

[0085] S3: Input the deviation distance value into the fuzzy inference engine, perform non-linear transformation according to the preset fuzzy rule base, generate the insulation condition assessment index and the fault risk probability distribution map of the distribution network cable, and achieve a precise assessment of the cable insulation condition.

[0086] Specifically, the fuzzy inference engine includes a fuzzification processing module, a rule inference module, and a defuzzification processing module.

[0087] Further, the fuzzification processing module converts the deviation distance value into a fuzzy set. Among them, the fuzzy set includes four levels: normal, slightly abnormal, moderately abnormal, and severely abnormal.

[0088] The rule inference module performs inference calculations according to the fuzzy rule base to obtain a fuzzy inference result.

[0089] Specifically, the fuzzy rule base contains fuzzy rules based on the deviation distance value of the dielectric loss spectrum and the deviation distance value of the partial discharge characteristic map.

[0090] The fuzzy inference results include: when the deviation distance value of the dielectric loss spectrum is greater than the first threshold and the deviation distance value of the partial discharge characteristic map is greater than the second threshold, it is determined to be severely abnormal; when the deviation distance value of the dielectric loss spectrum is greater than the first threshold while the deviation distance value of the partial discharge characteristic map is not greater than the second threshold, it is determined to be moderately abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold while the deviation distance value of the partial discharge characteristic map is greater than the second threshold, it is determined to be slightly abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic map is not greater than the second threshold, it is determined to be normal.

[0091] The defuzzification processing module uses the centroid method to convert the fuzzy inference result into the insulation state evaluation index and the fault risk probability distribution map.

[0092] Further, the generation process of the fault risk probability distribution map includes:

[0093] Divide the distribution network cable into several monitoring sections along the length direction;

[0094] Calculate the partial discharge density, discharge amplitude, and discharge frequency for each of the several monitoring sections respectively;

[0095] Based on the partial discharge density, the discharge amplitude, and the discharge frequency, calculate the risk coefficient for each monitoring section; among them, when the partial discharge density and the discharge amplitude increase or decrease simultaneously, the risk coefficient is proportional to the product of the partial discharge density and the discharge amplitude; when the partial discharge density increases while the discharge amplitude decreases or the partial discharge density decreases while the discharge amplitude increases, the risk coefficient remains within a preset range;

[0096] Combined with the insulation state evaluation index, the fault risk probability of the several monitoring sections is visually displayed in the form of a heat map.

[0097] In the intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables of the present invention, after obtaining the deviation distance values of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic map with respect to the normal domain hypersphere, it is necessary to comprehensively analyze the abnormality degree of each parameter to obtain an overall evaluation of the cable insulation state. Since the cable insulation state evaluation has the characteristics of complexity and ambiguity, a single threshold judgment is difficult to accurately reflect the actual situation of insulation deterioration. Therefore, the present invention inputs the deviation distance value into a fuzzy inference engine, performs non-linear conversion according to a preset fuzzy rule base, generates an insulation state evaluation index and a fault risk probability distribution map of the distribution network cable, and realizes the accurate evaluation of the cable insulation state.

[0098] The fuzzy inference engine is an intelligent processing system based on fuzzy set theory, including three main components: a fuzzification processing module, a rule inference module, and a defuzzification processing module. This structural design follows the framework of a classic fuzzy control system and can effectively handle the uncertainty and non-linear relationships in cable parameter evaluation.

[0099] First, the fuzzification processing module converts the deviation distance value into a fuzzy set. Specifically, when implementing, corresponding fuzzy membership functions are established for the deviation distance value of the dielectric loss spectrum, the deviation distance value of the withstand voltage leakage current curve, and the deviation distance value of the denoised partial discharge characteristic map. The fuzzy set includes four levels: normal, slightly abnormal, moderately abnormal, and severely abnormal, and each level corresponds to a fuzzy membership function.

[0100] For the deviation distance value of the dielectric loss spectrum, its fuzzy membership function adopts a combination of trapezoid and triangle forms. When the deviation distance value is less than 0.8, the membership degree of the normal level is 1; when the deviation distance value is between 0.8 and 1.0, the membership degree of the normal level linearly decreases, and at the same time, the membership degree of the slightly abnormal level linearly increases; and so on, to establish a complete set of membership functions. The same method is also applied to the deviation distance value of the withstand voltage leakage current curve and the deviation distance value of the denoised partial discharge characteristic map. However, considering the sensitivity differences of different parameters to cable state evaluation, the boundary points of the membership functions of each parameter are set differently.

[0101] Next, the rule inference module performs inference calculations based on the fuzzy rule base to obtain a fuzzy inference result. The fuzzy rule base contains fuzzy rules based on the deviation distance value of the dielectric loss spectrum and the deviation distance value of the partial discharge characteristic pattern, because these two parameters are the most sensitive to changes in the cable insulation state. The fuzzy rules are expressed in the classic IF-THEN format. For example, IF the deviation distance value of the dielectric loss spectrum is moderately abnormal AND the deviation distance value of the partial discharge characteristic pattern is slightly abnormal THEN the insulation state is moderately abnormal.

[0102] The specific fuzzy inference results include: when the deviation distance value of the dielectric loss spectrum is greater than the first threshold and the deviation distance value of the partial discharge characteristic pattern is greater than the second threshold, it is determined to be severely abnormal; when the deviation distance value of the dielectric loss spectrum is greater than the first threshold while the deviation distance value of the partial discharge characteristic pattern is not greater than the second threshold, it is determined to be moderately abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold while the deviation distance value of the partial discharge characteristic pattern is greater than the second threshold, it is determined to be slightly abnormal; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic pattern is not greater than the second threshold, it is determined to be normal. The first threshold and the second threshold are determined based on historical data statistics and expert experience respectively, and are used to distinguish the boundaries of different abnormal levels.

[0103] In the actual inference process, the rule inference module adopts the Mamdani inference method, realizes the synthesis of the conditional part through the minimum operator, and realizes the synthesis of the conclusion part through the maximum operator to form the final fuzzy inference result. This inference method retains the fuzzy characteristics of the input variables and can better handle the uncertainty in cable state assessment.

[0104] Finally, the defuzzification processing module uses the centroid method to convert the fuzzy inference result into the insulation state evaluation index and the fault risk probability distribution map. The centroid method is a commonly used defuzzification method, and obtains an accurate numerical output by calculating the centroid position of the fuzzy set graph. Specifically, the value range of the insulation state evaluation index is from 0 to 100, where 0 represents the healthiest state and 100 represents the most severely abnormal state. By calculating the centroid position of the fuzzy inference result and mapping it to the evaluation index range of 0 - 100, a quantitative evaluation result of the cable insulation state is obtained.

[0105] To achieve precise location of cable fault positions, the present invention also generates the fault risk probability distribution map. In the process of generating the fault risk probability distribution map, the distribution network cable is first divided into several monitoring sections along the length direction. For cables of different lengths, the division granularity can be different. Generally, the length of each monitoring section is 5 - 10 meters to ensure sufficient spatial resolution.

[0106] Next, the partial discharge density, discharge amplitude, and discharge frequency are calculated for each of the several monitoring sections respectively. The partial discharge density refers to the number of discharge pulses detected within a specific section per unit time; the discharge amplitude refers to the average energy or amplitude size of the discharge pulse; the discharge frequency refers to the repetition rate or frequency characteristic of the discharge activity. These three parameters comprehensively reflect the discharge activity characteristics of each section of the cable.

[0107] Then, based on the partial discharge density, the discharge amplitude, and the discharge frequency, the risk coefficient of each monitoring section is calculated. During the calculation process, when the partial discharge density and the discharge amplitude increase or decrease simultaneously, the risk coefficient is proportional to the product of the partial discharge density and the discharge amplitude. This is because the simultaneous increase in discharge density and amplitude usually indicates an exacerbation of the severity of insulation defects. When the partial discharge density increases while the discharge amplitude decreases or the partial discharge density decreases while the discharge amplitude increases, the risk coefficient remains within a preset range. This situation usually indicates a change in the discharge pattern rather than a one-way deterioration of the insulation state.

[0108] Finally, in combination with the insulation state evaluation index, the fault risk probability of the several monitoring sections is visually displayed in the form of a heat map. The heat map uses color gradients to represent the risk level. Usually, red indicates a high-risk area and green indicates a low-risk area. Through this intuitive visual representation, maintenance personnel can quickly identify potential fault positions in the cable, providing an accurate basis for subsequent repair or replacement decisions.

[0109] In an alternative embodiment, the construction of the fuzzy rule base combines cable insulation aging theory and historical fault case analysis. In addition to the above basic rules, the parameter change trend and comparison of historical test results are also considered. For example, when the deviation distance value of the dielectric loss spectrum shows a continuous increasing trend in consecutive measurements, even if the current value does not exceed the threshold, it may trigger a higher-level abnormal warning. This dynamic evaluation mechanism significantly improves the accuracy and timeliness of fault warning.

[0110] In another optional embodiment, the calculation of the fault risk probability distribution map also takes into account the installation environment and operating conditions of the cable. For example, for cable segments buried in high-temperature areas or frequently subjected to mechanical vibrations, the weight coefficient is appropriately increased in the risk assessment to reflect the promoting effect of these external factors on insulation aging. This comprehensive assessment method can more comprehensively reflect the actual operating state of the cable.

[0111] Exemplarily, when evaluating the outgoing cable of a 110 kV substation, the deviation distance value of the dielectric loss spectrum obtained is 1.25, the deviation distance value of the withstand voltage leakage current curve is 0.95, and the deviation distance value of the denoised partial discharge characteristic map is 1.35. After being converted by the fuzzification processing module, the membership degree of the deviation distance value of the dielectric loss spectrum in the moderately abnormal category is 0.65, and the membership degree in the slightly abnormal category is 0.35; the membership degree of the deviation distance value of the partial discharge characteristic map in the moderately abnormal category is 0.70, and the membership degree in the slightly abnormal category is 0.30. After being calculated by the rule inference module, the activation degree of the comprehensive fuzzy inference result in the moderately abnormal category is 0.65. Through the centroid method calculation of the defuzzification processing module, the final insulation state evaluation index is 65 points, indicating that the cable insulation has significantly deteriorated and regular inspections need to be arranged.

[0112] At the same time, the cable is divided into 20 monitoring sections along the length direction, and the partial discharge characteristic parameters are calculated for each section. The results show that the discharge density of the section about 100 meters away from the substation outlet is 25 times / minute, the discharge amplitude is 55 pC, and the discharge frequency distribution is concentrated in the range of 300 - 500 kHz. According to the risk calculation model, the risk coefficient of this section is 0.78, which is significantly higher than that of other sections. The finally generated heat map of the fault risk probability distribution clearly shows this high-risk area. Based on this, the technical personnel conduct on-site inspections and find that there is slight mechanical damage to the outer sheath of the cable at this location, which may cause moisture to penetrate and affect the insulation performance.

[0113] It should be noted that the method of inputting the deviation distance value into the fuzzy inference engine and performing non-linear conversion according to the preset fuzzy rule base to generate the insulation state evaluation index of the distribution network cable and the fault risk probability distribution map has many technical advantages compared with the traditional hard threshold judgment. First of all, the fuzzy inference technology can handle the uncertainty and ambiguity in cable insulation evaluation, avoiding misjudgment or missed judgment that may be caused by simple threshold judgment. Secondly, by comprehensively considering multiple parameters and their interactions, a comprehensive evaluation of the cable state is realized, improving the accuracy of judgment. In particular, the generation of the fault risk probability distribution map breaks through the limitation that traditional cable testing can only give an overall state evaluation but cannot locate the specific fault location, providing a powerful tool for precise maintenance. In addition, the rule system of the fuzzy inference engine integrates the dual advantages of expert experience and data-driven, making the evaluation results more in line with the actual engineering requirements.

[0114] In summary, the intelligent detection and evaluation method and system for the dielectric loss, withstand voltage and partial discharge of the distribution network cable provided by the present invention have significant technical advantages: First, by synchronously collecting three key parameters through the phase-resolved measurement device, multi-dimensional cable state characterization is realized, overcoming the one-sidedness of traditional single-parameter detection; Secondly, the frequency-domain adaptive filter is applied to process the partial discharge pattern, and the filtering strategy is dynamically adjusted according to the single-peak or multi-peak characteristics of the signal-to-noise ratio distribution, significantly improving the signal extraction ability in complex environments; Thirdly, the support vector data description algorithm is used to construct the normal domain hypersphere model of the cable multi-dimensional parameters, and the radial basis function mapping parameter, boundary relaxation parameter and model update mechanism are introduced to realize the accurate modeling of the non-linear data distribution and the objective quantification of the abnormal state; Finally, the deviation distance value is input into the fuzzy inference engine for non-linear conversion, and the fault risk probability distribution heat map is generated by combining the partial discharge density, discharge amplitude and discharge frequency, which not only realizes the accurate evaluation of the insulation state, but also can accurately locate the specific fault location. In addition, the system also has the ability of adaptive learning and dynamic update, and can automatically adjust the evaluation parameters during the cable aging process, providing a scientific basis for preventive maintenance decision-making, effectively reducing the failure rate of the distribution network, and improving the operation reliability and economic benefits of the power grid.

[0115] Embodiment 2, which is an embodiment of the present invention, provides an intelligent detection and evaluation system for the dielectric loss, withstand voltage and partial discharge of the distribution network cable, including:

[0116] A signal processing unit, configured to synchronously collect the dielectric loss spectrum, withstand voltage leakage current curve and partial discharge pattern of the distribution network cable through a phase-resolved measurement device, and apply a frequency-domain adaptive filter to process the partial discharge pattern to generate a denoised partial discharge feature pattern;

[0117] A model construction unit, which is used to construct a multi-dimensional parameter normal domain hypersphere model of the cable based on historical normal sample data by using the support vector data description algorithm, and calculate the deviation distance values of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic pattern with respect to the normal domain hypersphere;

[0118] A risk assessment unit, which is used to input the deviation distance value into a fuzzy inference engine, perform non-linear conversion according to a preset fuzzy rule base, generate an insulation state evaluation index and a fault risk probability distribution map of the distribution network cable, and realize the accurate evaluation of the cable insulation state.

[0119] Example 3, referring to Figure 2 , which is an embodiment of the present invention. The difference from the previous embodiment is that if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0120] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0121] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing it in a suitable manner if necessary, and then storing it in a computer memory.

[0122] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0123] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables, characterized in that: include: The dielectric loss spectrum, withstand voltage leakage current curve and partial discharge spectrum of the distribution network cable are synchronously collected by a phase resolution measurement device, and the partial discharge spectrum is processed by a frequency domain adaptive filter to generate a denoised partial discharge characteristic spectrum; the frequency domain adaptive filter includes a fast Fourier transform module, a spectrum analysis module and an adaptive bandpass filter module; Using a support vector data description algorithm, a normal domain hypersphere model of a cable multi-dimensional parameter is constructed based on historical normal sample data, and the deviation distance values ​​of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic spectrum relative to the normal domain hypersphere are calculated; The normal domain hypersphere model includes kernel function mapping parameters, boundary relaxation parameters and model update mechanism; The deviation distance value is input into the fuzzy reasoning engine, and nonlinear conversion is performed according to the preset fuzzy rule library to generate the insulation status evaluation index and fault risk probability distribution diagram of the distribution network cable, so as to realize the accurate evaluation of the insulation status of the cable; the fuzzy reasoning engine includes a fuzzification processing module, a rule reasoning module and a defuzzification processing module.

2. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 1, characterized in that: Applying a frequency domain adaptive filter to process the partial discharge spectrum includes: Converting the partial discharge spectrum into the frequency domain through the fast Fourier transform module; Using the spectrum analysis module to identify characteristic frequency bands of signals and noise, and calculate signal-to-noise ratio distribution; The adaptive bandpass filtering module sets the filtering bandwidth according to the signal-to-noise ratio distribution; when the signal-to-noise ratio distribution presents a unimodal characteristic, a narrowband filtering parameter is set; when the signal-to-noise ratio distribution presents a multimodal characteristic, a multi-channel filtering parameter is set to complete the enhancement of the partial discharge spectrum and the noise suppression.

3. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 2 is characterized in that: Using the support vector data description algorithm, a normal domain hypersphere model of cable multidimensional parameters is constructed based on historical normal sample data, including: Using radial basis function as the kernel function mapping parameter, the multi-dimensional parameters of the distribution network cable are mapped to a high-dimensional feature space; The sensitivity of the normal domain hypersphere model to abnormal points is controlled by the boundary relaxation parameter; wherein the boundary relaxation parameter is related to the characteristic distribution range of the historical normal sample data; According to the model update mechanism, the normal domain hypersphere model update is triggered when one of the following conditions is met: the number of newly added normal samples reaches a predetermined ratio, and the statistical characteristics of the deviation distance value exceed a preset reference range.

4. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 3 is characterized in that: The fuzzy processing module converts the deviation distance value into a fuzzy set; wherein the fuzzy set includes four levels: normal, slightly abnormal, moderately abnormal and severely abnormal; The rule inference module performs inference calculation according to the fuzzy rule base to obtain a fuzzy inference result; the fuzzy rule base contains fuzzy rules based on the deviation distance value of the dielectric loss spectrum and the deviation distance value of the partial discharge characteristic spectrum; The defuzzification processing module converts the fuzzy reasoning result into the insulation state evaluation index and the fault risk probability distribution diagram by using the centroid method.

5. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 4, characterized in that: The fuzzy inference results include: when the deviation distance value of the dielectric loss spectrum is greater than a first threshold and the deviation distance value of the partial discharge characteristic spectrum is greater than a second threshold, it is determined to be a serious abnormality; when the deviation distance value of the dielectric loss spectrum is greater than the first threshold and the deviation distance value of the partial discharge characteristic spectrum is not greater than the second threshold, it is determined to be a moderate abnormality; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic spectrum is greater than the second threshold, it is determined to be a slight abnormality; when the deviation distance value of the dielectric loss spectrum is not greater than the first threshold and the deviation distance value of the partial discharge characteristic spectrum is not greater than the second threshold, it is determined to be normal.

6. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 5, characterized in that: The calculation of the deviation distance value includes: Taking the center of the normal domain hypersphere as the origin, establishing a kernel function mapping coordinate system; Calculating the Euclidean distances of the dielectric loss spectrum, the withstand voltage leakage current curve, and the denoised partial discharge characteristic spectrum in the kernel function mapping coordinate system; The deviation distance value is determined according to the ratio of the Euclidean distance to the radius of the normal domain hypersphere.

7. The intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to claim 6, characterized in that: The generation process of the fault risk probability distribution diagram includes: Dividing the distribution network cable into a plurality of monitoring sections along the length direction; calculating the local discharge density, discharge amplitude and discharge frequency for the plurality of monitoring sections respectively; Based on the local discharge density, the discharge amplitude and the discharge frequency, a risk coefficient of each monitoring section is calculated; wherein, when the local discharge density and the discharge amplitude increase or decrease simultaneously, the risk coefficient is proportional to the product of the local discharge density and the discharge amplitude; when the local discharge density increases and the discharge amplitude decreases or the local discharge density decreases and the discharge amplitude increases, the risk coefficient remains within a preset range; Combined with the insulation status evaluation index, the failure risk probability of the several monitoring sections is visualized in the form of a heat map.

8. A distribution network cable dielectric loss, withstand voltage and partial discharge intelligent detection and evaluation system, based on the distribution network cable dielectric loss, withstand voltage and partial discharge intelligent detection and evaluation method according to any one of claims 1 to 7, characterized in that: include, A signal processing unit, used to synchronously collect the dielectric loss spectrum, withstand voltage leakage current curve and partial discharge spectrum of the distribution network cable through a phase resolution measurement device, and process the partial discharge spectrum by applying a frequency domain adaptive filter to generate a denoised partial discharge characteristic spectrum; A model building unit, used to build a normal domain hypersphere model of cable multi-dimensional parameters based on historical normal sample data by using a support vector data description algorithm, and calculate the deviation distance value of the dielectric loss spectrum, the withstand voltage leakage current curve and the denoised partial discharge characteristic spectrum relative to the normal domain hypersphere; The risk assessment unit is used to input the deviation distance value into a fuzzy inference engine, perform nonlinear conversion according to a preset fuzzy rule library, generate an insulation status assessment index and a fault risk probability distribution diagram of the distribution network cable, and realize accurate assessment of the cable insulation status.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent detection and evaluation method for dielectric loss, withstand voltage and partial discharge of distribution network cables according to any one of claims 1 to 7 are implemented.

Images