Power distribution network insulation defect detection method and system based on electroluminescent composite material

Abstract

The invention discloses a power distribution network insulation defect detection method and system based on an electroluminescent composite material, and relates to the technical field of electrical equipment insulation detection.The method comprises the steps that an electroluminescent gradient doping coating material is prepared and sprayed to the surface of a pretreated detected insulating part; acquiring an electroluminescent signal of the detected insulating part and real-time data of environment temperature and humidity, and performing preprocessing, inputting the preprocessed electroluminescent signal into a pre-constructed light-electric field nonlinear mapping model, performing basic electric field inversion, and performing local correction on a three-combination-point area through a weight factor; global environment compensation is carried out through a pre-trained BP neural network, and a corrected electric field inversion result is obtained; and on the basis of a preset dimensionless electric field index dynamic threshold value, labeling coordinates of an abnormal region in a corrected electric field inversion result in real time, evaluating a defect risk level in combination with weighted D-S evidence, and realizing remote display and early warning of a thermodynamic diagram through a cloud protocol.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment insulation detection, and particularly relates to a method and system for detecting insulation defects in a distribution network based on an electroluminescent composite material, which is applicable to the early defect visualization diagnosis and warning of equipment such as porcelain insulators and cable joints. Background Art

[0002] Insulators are widely used in the power transmission and transformation system, such as pot insulators and post insulators in the external insulation device of the distribution network. The surface field strength distribution is one of the main factors affecting its operation safety and reliability. In recent years, a large number of studies have shown that the external insulation equipment forms irregular dirt on its surface in a complex environment and forms flashover on its surface in a humid climate, resulting in line tripping and posing a hidden danger to the safe and stable operation of the power grid. The working group of the International Council on Large Electric Systems (CIGRE) proposed that the external insulation equipment is the weak link of each transmission line and outdoor substation, and it is crucial to monitor and maintain its operation properly. Therefore, implementing on-line monitoring of the operation state of contaminated insulators, observing the abnormal electric field on the surface of the high field strength area of the external insulation, and performing visual inversion of the danger level to timely discover the potential safety hazards in the operation process of the insulators is of great significance for improving the safety and reliability of the high-voltage power grid. However, the application experience in recent years shows that the leakage current electrical measurement method, non-contact electrical parameter detection method, unmanned aerial vehicle visualization method, and partial discharge detection method have certain limitations in on-site operation, mainly manifested as insufficient anti-interference ability, poor discharge fault recognition and positioning effect, resulting in false alarms, misreports, and missed reports from time to time, which has now become a pain point problem in the on-line detection and maintenance of the external insulation surface electric field.

[0003] The prior art document (CN119556017A) discloses a composite insulator electric field detection scheme in the technical field of power equipment detection, which realizes electric field inversion through a silicone rubber matrix composite fluorescent material coating combined with unmanned aerial vehicle image acquisition and linear regression algorithm. However, its material system adopts a single fluorescent material structure, resulting in low carrier transport efficiency and significant attenuation of the luminescence intensity; it is difficult to balance the sensitivity of the low field area and the response accuracy of the high field area based on the linear mapping model between gray scale and electric field, and there is an oversaturation distortion problem; the detection system does not consider the influence of temperature and humidity fluctuations on the luminescence signal, and lacks a local electric field correction mechanism for complex structures such as the three-joint points of insulators, resulting in an increase in detection error under environmental interference and inaccurate field strength inversion in complex areas. In addition, the prior art relies on a fixed threshold alarm mechanism and cannot dynamically adapt to material aging and environmental condition changes, and false alarms and missed reports are likely to occur during long-term operation. Summary of the Invention

[0004] To solve the problems of low carrier transport efficiency, difficulty in balancing luminescence sensitivity and linear response, insufficient environmental interference compensation, and imperfect defect warning in the existing technology, the present invention provides a method and system for detecting external insulation defects in a distribution network based on an electroluminescent composite coating material. Through a combination of multi-level material design, non-linear mapping, and dynamic compensation, online insulation detection and intelligent warning with high sensitivity, high precision, and strong robustness are achieved.

[0005] The present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a method for detecting insulation defects in a distribution network based on an electroluminescent composite material, including the following steps:

[0007] Prepare an electroluminescent gradient-doped coating material and spray it on the surface of the measured insulating part after pretreatment. The electroluminescent gradient-doped coating material includes gradient-doped particles, a core-shell structure nano-luminescent layer, and a phosphorescence enhancement layer;

[0008] Obtain the real-time data of the electroluminescent signal and the ambient temperature and humidity of the measured insulating part and perform pretreatment;

[0009] Input the pretreated electroluminescent signal into a pre-constructed optical-electric field non-linear mapping model for basic electric field inversion, and locally correct the triple-junction region through a weight factor;

[0010] Based on the locally corrected electric field inversion result, perform global environmental compensation through a pre-trained BP neural network to obtain a corrected electric field inversion result;

[0011] Based on a preset dimensionless electric field index dynamic threshold, real-time mark the coordinates of the abnormal region in the corrected electric field inversion result, combine weighted D-S evidence to evaluate the defect risk level, and realize remote display and warning of the heat map through the cloud protocol.

[0012] Optionally, the preparation of the electroluminescent gradient-doped coating material includes:

[0013] Compound an organic semiconductor material and a two-dimensional carbon material in a flexible polymer matrix to form a conductive network;

[0014] Use a core-shell structure nano-composite material as the luminescent layer. The inner core is a small molecule electroluminescent material, including aromatic ring compounds and benzene derivatives, and the outer shell is a conductive polymer;

[0015] Dope a hole transport material and an electrophilic reagent in the flexible polymer matrix to regulate the carrier balance;

[0016] Introduce a phosphorescent material at the interface between the luminescent layer and the flexible polymer matrix to form an enhancement layer;

[0017] Form a distribution structure with an increasing volume fraction of electroluminescent particles in the coating material through gradient doping;

[0018] Optimize the gradient doping structure based on finite element electric field simulation to optimize the signal-to-noise ratio of the surface luminescence intensity and the bottom layer electric field response.

[0019] Optionally, the optimizing the gradient doping structure based on finite element electric field simulation includes:

[0020] Establish a three-dimensional electric field model including the concentration gradient of doping particles and the core-shell thickness;

[0021] Determine the concentration gradient distribution with the surface particle volume fraction ≤ 10% and the inner layer ≥ 15% through iterative simulation;

[0022] Configure the surface doping particle size to be 50 to 100 nm and the inner layer doping particle size to be 100 to 200 nm;

[0023] Optimize the thickness ratio of the core-shell structure to be 1:0.2 to 1:0.5;

[0024] Verify that the signal-to-noise ratio SNR > 50 dB and the local electric field distortion rate < 8%.

[0025] Optionally, the preprocessing of the surface of the measured insulating part includes:

[0026] Perform gradient polishing on the surface of the measured insulating part and use oxygen plasma treatment. Among them, the gradient polishing includes controlling the surface roughness Ra within the range of 1 μm to 5 μm;

[0027] Deposit a 50-nm-thick PVK / graphene conductive layer on the surface of the measured insulating part by magnetron sputtering.

[0028] Optionally, use a near-ultraviolet narrowband filter with a set specification to obtain the electroluminescence signal. Among them, the specifications of the near-ultraviolet narrowband filter include:

[0029] The full width at half maximum range is less than 20 nm, the central wavelength range is 365 - 500 nm, and the peak transmittance is greater than 85%, which is used to collect the luminescence signal in the 400 - 450 nm band.

[0030] Optionally, the pre-constructed light-electric field non-linear mapping model is a non-linear exponential relationship between the electroluminescence intensity and the electric field intensity, and the exponential parameter n = 2.2 to 2.4.

[0031] Optionally, the local correction of the triple-junction region through the weight factor includes:

[0032] Identify the triple-junction region on the surface of the insulating part. The triple-junction region includes the bending part of the umbrella skirt of the measured insulating part, the connection between the umbrella skirt and the flange, and the interface between the umbrella skirt and the core rod;

[0033] The electric field inversion result of the recognition area is dynamically adjusted by using a sigmoid-type weight factor, and the weight factor is non-linearly attenuated according to the difference between the local electric field index and a preset oversaturation threshold;

[0034] When the local electric field index exceeds the oversaturation threshold, the basic inversion electric field strength is attenuated and corrected by the weight factor, so that the corrected electric field inversion error is controlled within 5%. Among them, the oversaturation threshold is a critical field strength ratio value pre-calibrated based on the structural parameters of the measured insulating part and the characteristics of the coating material.

[0035] Optionally, the real-time marking of the abnormal area coordinates in the corrected electric field inversion result based on the preset dimensionless electric field index dynamic threshold includes:

[0036] By calculating the ratio of the electric field strength of each detection point to the average electric field strength in the locally corrected electric field inversion result, a dimensionless electric field index is generated;

[0037] Set a dynamic alarm threshold, and divide the detection area into a normal area, a warning area, and a fault area: when the dimensionless electric field index is less than 1.5, it is judged as normal; when it is between 1.5 and 2, a warning is triggered; when it reaches or exceeds 2, it is judged as a fault;

[0038] Real-time scan the inversion result, and automatically identify and mark the coordinates of the abnormal area exceeding the threshold.

[0039] Optionally, the combination of the preset weighted D-S evidence to evaluate the defect risk level includes:

[0040] Fuse multi-source detection data to construct a multi-source evidence set. The multi-source detection data includes electroluminescence intensity, ambient temperature and humidity, abnormal area morphological characteristics, and historical defect distribution. Among them, the abnormal area morphological characteristics include the geometric shape, brightness concentration degree, and area proportion of the abnormal area;

[0041] Dynamically assign weights according to the confidence of each evidence source, where the larger the area proportion of the abnormal area, the higher the weight of the corresponding evidence source;

[0042] Integrate the conflict information of different evidence sources, and output the defect level corresponding to the maximum probability, including that a serious defect triggers a red warning, and a minor defect enters an observation state.

[0043] The second aspect of the present invention provides a distribution network insulation defect detection system based on an electroluminescent composite material. Based on the distribution network insulation defect detection method described in the first aspect of the present invention, the system includes:

[0044] A hardware layer, an algorithm layer, and a platform layer, where

[0045] The hardware layer includes an electroluminescent coating module, a near-ultraviolet narrow-band filter camera, a temperature and humidity sensor, a drone or a mobile detection platform;

[0046] The electroluminescent coating module is used for the preparation and spraying of gradient doping materials;

[0047] The near-ultraviolet narrow-band filter camera is used to obtain electroluminescent signals;

[0048] The temperature and humidity sensor is used to obtain environmental temperature and humidity data;

[0049] The drone or the mobile detection platform is used to carry the near-ultraviolet narrow-band filter camera and the temperature and humidity sensor for signal acquisition;

[0050] The algorithm layer includes a non-linear mapping and local correction module, a BP neural network compensation module, and a D-S evidence fusion and risk assessment module;

[0051] The non-linear mapping and local correction module is used to generate a basic electric field distribution based on a non-linear mapping model and correct the local electric field error in the triple-junction region through a weight factor;

[0052] The BP neural network compensation module is used to dynamically correct the electric field inversion result by fusing temperature and humidity data through a neural network;

[0053] The D-S evidence fusion and risk assessment module is used to combine weighted D-S evidence to integrate multi-source data to evaluate the defect risk level;

[0054] The platform layer is used for cloud storage and remote diagnostic services, and real-time display of heat maps and abnormal coordinates on mobile terminals.

[0055] Compared with the prior art, the beneficial effects of the present invention at least include:

[0056] 1. By constructing a core-shell nanocomposite light-emitting layer and a gradient doping material system, the present invention solves the problems of low carrier transport efficiency and poor luminescence stability in the prior art, realizes low-field detection sensitivity (threshold < 0.5 kV / mm) and visualization of micron-level defects, and at the same time extends the material life by more than 40%.

[0057] 2. By designing a non-linear light-electric field mapping model and a local weight correction algorithm, the present invention solves the problem of high-field oversaturation distortion caused by linear mapping, controls the electric field inversion error in complex structure regions within 5%, and improves the detection accuracy and robustness.

[0058] 3. By integrating near-ultraviolet narrow-band filtering and a BP neural network dynamic compensation mechanism, the present invention solves the problems of temperature and humidity fluctuations and visible light noise interference, achieves the stability of electric field detection under complex outdoor conditions, and improves the environmental interference suppression rate by more than 90%.

[0059] 4. By integrating dynamic threshold calibration and weighted D-S evidence theory, the present invention solves the problem of poor adaptability of the traditional fixed threshold alarm mechanism, realizes multi-level risk intelligent assessment and cloud remote diagnosis, and improves the early warning response speed to the millisecond level.

[0060] 5. By synergistically optimizing materials, algorithms, and platforms, the present invention solves the problems of single detection dimension and insufficient coordination in the prior art, forms a full-chain closed-loop system from signal acquisition to cloud decision-making, and comprehensively improves the detection efficiency of insulation defects in the distribution network and the level of safe operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a schematic flow chart of the preparation of the composite luminescent material and the coating of the insulating part provided according to an embodiment of the present invention;

[0062] Figure 2 is a schematic diagram of composite luminescent materials with different doping ratios provided according to an embodiment of the present invention;

[0063] Figure 3 is a schematic diagram of the electric field inversion model and its inversion results provided according to an embodiment of the present invention;

[0064] Figure 4 is a schematic diagram of abnormal coordinate marking provided according to an embodiment of the present invention;

[0065] Figure 5 is a schematic flow chart of the method provided according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0067] To more clearly introduce the prominent substantive features of the present invention and the significant progress brought to the prior art, an application example of implementing the present invention is introduced below.

[0068] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The application example specifically includes:

[0069] The present invention provides a method for detecting insulation defects in a distribution network based on an electroluminescent composite material in Embodiment 1, as Figure 5 shown, which includes the following steps:

[0070] Step 1: Prepare an electroluminescent gradient-doped coating material and spray it on the surface of the measured insulating part after pretreatment. The electroluminescent gradient-doped coating material includes gradient-doped particles, a core-shell structure nano-luminescent layer, and a phosphorescence enhancement layer.

[0071] Preferably, the preparation of the electroluminescent gradient-doped coating material includes:

[0072] Construct a composite conductive network containing an organic semiconductor material and a two-dimensional carbon material in a flexible polymer matrix;

[0073] Use a core-shell structure nano-composite material as the luminescent layer, where the inner core is a small molecule electroluminescent material and the outer shell is conductive polyaniline;

[0074] Dope a hole transport material and an electrophilic reagent in the matrix to regulate the carrier balance;

[0075] Introduce an iridium complex phosphorescent material as a phosphorescence enhancement layer at the interface between the luminescent layer and the matrix;

[0076] Make the volume fraction of the electroluminescent particles show a gradient distribution in the matrix through gradient doping.

[0077] Specifically, the composition of the electroluminescent gradient-doped coating material includes:

[0078] Matrix material: It is composed of a flexible polymer matrix selected from polydimethylsiloxane (PDMS), where PVK and graphene are combined to form a conductive network, endowing the material with bendable and shearable mechanical properties;

[0079] Luminescent layer structure: Use a core-shell structure nano-composite material, with the inner core being a small molecule electroluminescent material selected from aromatic ring compounds, benzene or naphthalene derivatives, and the outer shell being conductive polyaniline, realizing the synergistic improvement of solution processability and carrier transport efficiency;

[0080] Doping components: Include a hole transport material and an electrophilic reagent. Among them, the hole transport material is such as NPB or TPD, and the electrophilic reagent is selected from S and Se elements. The carrier balance is regulated through the doping ratio to enhance the luminescence stability;

[0081] Phosphorescence enhancement layer: Introduce an iridium complex phosphorescent material Ir(Flpy-CF3-EG) at the interface between the luminescent layer and the matrix. Its maximum external quantum efficiency reaches 31.1%, and the power efficiency is 93.5 lm / W, significantly improving the luminescence sensitivity under low electric fields;

[0082] Gradient doping design: The volume fraction of electroluminescent particles ZnS:Cu in the matrix shows a gradient distribution, 5%-15% on the surface layer, suppressing local electric field distortion while maintaining high surface luminescence intensity;

[0083] Specifically, while doping ZnS:Cu particles in a gradient manner, the particle distribution at different depths and concentrations is optimized through finite element electric field simulation, so that the luminescence intensity on the surface layer and the electric field response at the bottom layer reach the optimal signal-to-noise ratio, SNR>50dB, improving the signal acquisition efficiency of the CCD in the 400–450nm band.

[0084] Furthermore, the optimization of the gradient doping structure based on finite element electric field simulation includes:

[0085] Establish a three-dimensional electric field model including the concentration gradient of doped particles and the core-shell thickness;

[0086] Determine the concentration gradient distribution with the volume fraction of surface layer particles ≤10% and the inner layer ≥15% through iterative simulation;

[0087] Configure the particle size of surface layer doped particles to be 50-100nm and the particle size of inner layer doped particles to be 100-200nm;

[0088] Optimize the thickness ratio of the core-shell structure to be 1:0.2 to 1:0.5;

[0089] Verify that the signal-to-noise ratio SNR>50dB and the local electric field distortion rate <8%.

[0090] It should be noted that through the optimization of the gradient doping structure design by electric field simulation, the collaborative optimization of detection sensitivity, material stability, and system effectiveness is achieved: based on the gradient distribution of the core-shell nanoluminescent layer, combined with the light-electric field response model, the signal-to-noise ratio is significantly improved while suppressing local electric field distortion; through the differential doping design of low concentration on the surface layer and high concentration on the inner layer, the mechanical strength and electrical response characteristics of the material are balanced, enhancing the environmental adaptability and long-term stability of the coating; by integrating the dynamic threshold calibration and defect feature extraction mechanisms, precise positioning and risk grading of minor insulation defects can be realized in a wide temperature range and complex electromagnetic environment, providing a solution with both high-precision detection and long-term reliability for the insulation state assessment of the distribution network.

[0091] Preferably, the pretreatment of the surface of the insulating part to be measured includes:

[0092] Gradually polish the surface of the insulating part (Ra = 1-5μm) and perform oxygen plasma treatment with the treatment parameters of power 50W and time 30s to improve the coating adhesion; deposit a 50nm thick PVK / graphene conductive layer on the surface through magnetron sputtering to reduce the risk of interfacial charge accumulation.

[0093] Spraying the electroluminescent gradient doping coating material on the surface of the pre-treated insulating component to be measured includes: adopting an ultraviolet synchronous spraying process with a wavelength of 365 nm and a power of 50 mW / cm 2 ; controlling the coating uniformity > 95% through the spray gun pressure (0.2 - 0.4 MPa) and real-time CCD monitoring;

[0094] The coating structure includes gradient-doped ZnS:Cu (5% on the surface layer, 15% on the inner layer) and an iridium complex phosphorescent layer (Ir(Flpy-CF3-EG)3), achieving low-field sensitivity (detection threshold < 0.5 kV / mm) and high-field linear response.

[0095] The electroluminescent effect refers to a phenomenon in which a solid material emits light due to the recombination of electrons and holes under the action of an alternating or direct current electric field. Introducing an electroluminescent material into an insulating system can visually display the high electric field region (caused by an insulating structure or defect), and it is expected to be applied in aspects such as the optimal design of insulating structures and the self-detection of insulating defects.

[0096] Compared with the traditional insulator surface electric field visualization technology, in the insulator defect detection method provided by the present invention, an electroluminescent composite material is prepared and coated on the surface of the insulator, which has many advantages:

[0097] 1. Due to the field-induced luminescence effect of electroluminescent particles under a high electric field, it can more intrinsically invert the surface electric field and will not affect the original electric field;

[0098] 2. Electroluminescent particles have semi-conductive properties, which are beneficial to the dissipation of accumulated charges on the dielectric surface area, and are expected to improve the surface flashover strength of the insulator while realizing the inversion of the surface electric field;

[0099] 3. It can achieve a higher resolution of electric field inversion, that is, realize the self-diagnosis of abnormal electric fields caused by minute defects on the insulator surface.

[0100] Step 2: Obtain the real-time data of the electroluminescent signal and the ambient temperature and humidity of the insulating component to be measured and perform preprocessing. Among them, an integrated near-ultraviolet narrowband filter is used to obtain the electroluminescent signal.

[0101] Preferably, obtaining the electroluminescent signal of the insulating component to be measured includes:

[0102] Integrating a near-ultraviolet narrowband filter with a full width at half maximum (FWHM) range of less than 20 nm, a central wavelength range of 365 - 500 nm, and a peak transmittance greater than 85% to collect the luminescence signal in the 400 - 450 nm band. The application of this filter can suppress the interference in the visible and infrared bands.

[0103] Preferably, the preprocessing of the obtained data includes:

[0104] Separate the electroluminescence signal of the insulation part under test into RGB spectral channels, and extract the characteristic spectrum;

[0105] Perform Z-score standardization and normalization on the ambient temperature and humidity data.

[0106] Step 3: Input the preprocessed electroluminescence signal into the pre-constructed optical-electric field non-linear mapping model for basic electric field inversion, and locally correct the triple-junction region through a weight factor.

[0107] Preferably, the pre-constructed optical-electric field non-linear mapping model includes:

[0108] I = kE n

[0109] where I is the electroluminescence intensity; E is the electric field intensity; k is a constant obtained by fitting the model parameters; n is an exponential parameter, n = 2.3 ± 0.1, obtained by fitting the model parameters.

[0110] Preferably, the correction of the triple-junction region through the weight factor includes:

[0111] The formula for the weight factor is:

[0112]

[0113] where η i is the local dimensionless electric field index, used to characterize the degree of local field strength anomaly; η th is the oversaturation threshold, pre-calibrated based on the structural parameters of the insulation part under test and the characteristics of the coating material; α is the attenuation coefficient; is the electric field intensity value at the i-th detection point; E avg is the average electric field intensity value on the surface of the insulator under test;

[0114] The triple-junction region is the bend between the umbrella sheds, that is, the groove where two adjacent umbrella sheds meet; the transition between the umbrella shed and the mounting flange or cap ring; the part where the bottom of the umbrella shed is connected to the core rod or bushing (mandrel). The intersection of these three surfaces or structural components in space makes the local electric field in this region prone to distortion or concentration. Due to geometric mutations and differences in the dielectric constants of different material interfaces, such as the coating and the matrix material, the coating and the metal flange, the surface electric field intensity in the triple-junction region is usually significantly higher than that of the flat umbrella shed surface, and it is easy to form electric field distortion or even partial discharge;

[0115] The electric field inversion result output after local correction by the weight factor is:

[0116]

[0117] When the local electric field index exceeds the oversaturation threshold, the basic inversion electric field strength is attenuated and corrected by a weight factor;

[0118] A weight factor is introduced into the triple-junction region to correct the interference of luminescence oversaturation on electric field inversion, so that the corrected electric field inversion error < 5%.

[0119] Step 4: Based on the locally corrected electric field inversion result, global environment compensation is performed through a pre-trained BP neural network to obtain a corrected electric field inversion result.

[0120] Preferably, the global environment compensation by the pre-trained BP neural network includes: using a 3-layer feedforward neural network, where the input layer has 6 nodes, the hidden layer has 12 nodes, and the output layer has 1 node. The activation function is a Sigmoid-Relu hybrid type to prevent gradient disappearance; based on the laboratory calibration data, the calibration data is a temperature of 25 °C and a humidity of 50%;

[0121] The input parameters include real-time data and historical time-series data of luminescence intensity, temperature, and humidity, and the time-series interval is Δt = 1 s;

[0122] The output is the corrected electric field strength E corrected 。

[0123] Preferably, Step 4 further includes:

[0124] When it is detected that the temperature and humidity fluctuation > 10% or the standard deviation of the luminescence intensity > 15%, the DE algorithm is triggered to optimize the weights of the BP model to ensure the compensation accuracy.

[0125] Step 5: Based on the preset dimensionless electric field index dynamic threshold, the coordinates of the abnormal region are marked in real time in the corrected electric field inversion result. Combining the weighted D-S evidence theory, the defect risk level is evaluated, and the heat map is remotely displayed and warned through the cloud protocol.

[0126] Preferably, the real-time marking of the coordinates of the abnormal region in the corrected electric field inversion result based on the preset dimensionless electric field index dynamic threshold includes:

[0127] Based on the dimensionless electric field index:

[0128]

[0129] where, is the electric field strength value of the i-th detection point after correction; E avg,corrected is the average electric field strength value on the surface of the measured insulator after correction;

[0130] Set the dynamic alarm threshold η th,corrected= 2, where the normal region η < 1.5, the warning region 1.5 ≤ η < 2, and the fault region η ≥ 2; η is automatically corrected through periodic high-voltage aging (1 h / 10 kV). th,corrected , suppressing long-term operation drift.

[0131] Further preferably, the automatic correction of the dynamic alarm threshold through periodic high-voltage aging includes:

[0132] Accelerating equipment aging by periodically applying high voltage, and dynamically adjusting the alarm threshold using the performance data after aging to ensure the accuracy of defect identification during long-term operation; combining artificial accelerated aging tests with data-driven threshold optimization methods to solve the problem of material property drift.

[0133] Preferably, the evaluation of the defect risk level by combining the preset weighted D-S evidence theory includes:

[0134] Combining the electroluminescence intensity, temperature, humidity, and historical defect database, and using the weighted D-S evidence theory to evaluate the defect risk level. Specifically, let Θ be the set of all possible states in the problem domain, called the Frame of Discernment, which can be expressed in the present invention as:

[0135] Θ = {no defect, minor defect, severe defect}

[0136] For each subset in the power set 2Θ of Θ Define a Basic Probability Assignment (BPA):

[0137]

[0138] For an image region, after LSSVM electric field intensity inversion and combined with morphological features and position offset, the following subjective assignment can be given:

[0139] m1(no defect) = 0.2, m1(minor defect) = 0.6, m1(severe defect) = 0.2

[0140] Independent BPA assignments can be given by multiple sources, such as: luminescence intensity characteristics, the shape of the electric field anomaly region, historical defect distribution, etc. To suppress the conflicts brought by unreliable sources, the weighted D-S evidence synthesis rule introduces a credibility weight β i ∈ [0, 1]. Given two information sources that give BPA: m1(A), m2(A), and their corresponding weights are β1, β2, then the synthesized BPA is:

[0141]

[0142] BPA after weight correction:

[0143]

[0144] Weighted synthesis formula:

[0145]

[0146] where the conflict factor is:

[0147]

[0148] If K > 0.8, the manual review process is triggered.

[0149] Further preferably, the evaluation of the defect risk level by combining the weighted D-S evidence theory includes: multi-source heterogeneous data fusion, dynamic confidence weight assignment, and intelligent decision-making of defect risk, where:

[0150] Multi-source heterogeneous data fusion: constructing multi-dimensional evidence sources based on image features, material response gradients, historical defect distributions, and environmental parameters;

[0151] Specifically, the setting of evidence sources includes:

[0152] Source 1, image feature evidence: based on the geometric shape, brightness concentration degree, and regional area ratio of the abnormal electric field intensity area;

[0153] Source 2, material information: based on the matching degree between the luminous response intensity gradient of the gradient doping area and the theoretical value;

[0154] Source 3, historical experience database: the defect probability distribution of insulators in this region or of the same material;

[0155] Source 4, environmental parameters: dynamically adjusted according to the degree of deviation of temperature and humidity from the calibrated conditions, including humidity > 85% and temperature difference > 15°C.

[0156] Dynamic confidence weight assignment: mapping the abnormal area ratio to the image evidence weight through the Sigmoid function and associating with the credibility of other evidence sources;

[0157] Specifically, the weight β i is calculated by the confidence function of each evidence source:

[0158]

[0159] where S defect is the abnormal electric field area ratio, λ controls the steepness, and S0 is the empirical threshold.

[0160] Intelligent Decision-making for Defect Risk: Output the defect level according to the principle of maximum synthetic BPA, and trigger multi-level early warnings in combination with dynamic thresholds;

[0161] Specifically, the result output and threshold judgment are as follows:

[0162] The final defect level is the state corresponding to the maximum synthetic BPA;

[0163] If m(serious defect) > 0.6, a red early warning will be automatically issued;

[0164] If m(slight defect) > 0.5 and remains unchanged for two consecutive monitors, it will enter the observation state.

[0165] Step 5 further includes:

[0166] Introduce transfer learning algorithm to map the laboratory calibration data to the on-site measurement results, reducing the influence of environmental differences.

[0167] The remote display of the heat map through the cloud protocol includes:

[0168] The mobile terminal displays the electric field heat map in real time, marking the coordinates of the abnormal area (accuracy ±1mm); supports cloud storage and remote diagnosis, and is compatible with the IEC61850 communication protocol.

[0169] In order to ensure the modularity and interpretability of the model, and also facilitate phased debugging and optimization, the present invention divides the entire inversion process into two steps: preliminary electric field inversion and BP neural network environment compensation;

[0170] The advantages of such phased processing include: single responsibility: the mapping model focuses on the light-field inversion, and the BP network focuses on the field-environment compensation. Easy to calibrate: In the preliminary mapping stage, the material parameters and mapping functions can be calibrated offline in the laboratory; in the BP stage, the temperature, humidity and mapping results are collected on-site for online training or fine-tuning. Enhanced interpretability: If the environment compensation fails, only the BP network part needs to be checked; if the mapping error is too large, then focus on debugging the non-linear mapping and weight factors.

[0171] It should be noted that in view of the problems of the single material system, insufficient linear mapping accuracy, and lack of environmental interference compensation in the prior art, the present invention proposes the following innovative system:

[0172] Material system: Introduce a core-shell nanocomposite light-emitting layer (small molecule aromatic ring / benzene derivative core + conductive polyaniline shell) and an iridium complex phosphorescence enhancement layer to achieve a high external quantum efficiency (31.1%) and power efficiency (93.5 lm / W), and construct an electroluminescent composite coating by gradient doping ZnS:Cu particles (5% on the surface - 15% on the inner layer), taking into account both low-field sensitivity and high surface luminescence intensity;

[0173] Electric field mapping model: Adopt non-linear mapping of RGB channels (I = k·En, n = 2.3 ± 0.1), and introduce a Sigmoid-type weight factor to correct the luminous over-saturation effect in the triple-junction area, achieving a local electric field inversion error < 5%;

[0174] Environmental compensation: Integrate a near-ultraviolet narrow-band filter (FWHM < 20nm, transmittance > 85%) to suppress visible light noise, and dynamically correct the luminescence-electric field mapping based on real-time temperature and humidity data fused by a BP neural network to adapt to complex outdoor working conditions;

[0175] Early warning system: Construct a dimensionless electric field index η = E / E avg , set a dynamic threshold and combine it with periodic high-voltage aging for automatic calibration, supplemented by the weighted D-S evidence theory and transfer learning algorithm to evaluate the defect risk, realizing accurate early warning from micron-level cracks to overall field distortion and cloud-based remote diagnosis.

[0176] Through the above material-algorithm-platform collaborative closed-loop system, full-dimensional diagnosis of insulation defects and remote intelligent decision-making under complex working conditions are realized.

[0177] Specifically, the detection sensitivity is greatly improved: the low-field threshold < 0.5 kV / mm, the resolution is better than 0.1 kV / mm, and micron-level defects are visualized;

[0178] The luminous stability is significantly enhanced: the core-shell + phosphor layer design enables the attenuation to be < 3% after 1000 hours outdoors, and the service life is extended;

[0179] The inversion accuracy and robustness are optimized synchronously: non-linear mapping + local weight correction make the error < 5%, and the compensation accuracy of the BP network > 98%;

[0180] Strong environmental adaptability: Near-ultraviolet filtering + dynamic compensation suppress interference such as visible light, temperature, and humidity, and is suitable for complex working conditions;

[0181] Intelligent early warning and remote diagnosis: Dynamic threshold + D-S theory + transfer learning realize multi-level risk assessment, and real-time response on the cloud / mobile terminal.

[0182] In Embodiment 2 of the present invention, a power distribution network insulation defect detection system based on an electroluminescent composite material is provided. Based on the power distribution network insulation defect detection method of the electroluminescent composite material described in Embodiment 1, the system includes:

[0183] A hardware layer, an algorithm layer, and a platform layer, where

[0184] The hardware layer includes an electroluminescent coating module, a near-ultraviolet narrow-band filter camera, a temperature and humidity sensor, a drone or a mobile detection platform;

[0185] Preferably, the electroluminescent coating module is used for the preparation and spraying of gradient doping materials;

[0186] The near-ultraviolet narrow-band filter camera is used to obtain electroluminescent signals;

[0187] The temperature and humidity sensor is used to obtain ambient temperature and humidity data;

[0188] The drone or mobile detection platform is used to carry the near-ultraviolet narrow-band filter camera and the temperature and humidity sensor for signal acquisition;

[0189] The algorithm layer includes a non-linear mapping and local correction module, a BP neural network compensation module, and a D-S evidence fusion and risk assessment module;

[0190] Preferably, the non-linear mapping and local correction module is used to generate a basic electric field distribution based on a non-linear mapping model and correct the local electric field error in the triple-junction region through a weight factor;

[0191] The BP neural network compensation module is used to dynamically correct the electric field inversion result by fusing temperature and humidity data through a neural network;

[0192] The D-S evidence fusion and risk assessment module is used to combine weighted D-S evidence to integrate multi-source data and evaluate the defect risk level;

[0193] The platform layer is used for IEC 61850-compatible cloud storage and remote diagnostic services, and real-time display of heat maps and abnormal coordinates on mobile terminals.

[0194] The detection sensitivity is greatly improved: the low-field threshold is <0.5 kV / mm, the resolution is better than 0.1 kV / mm, and micron-level defects are visualized;

[0195] The luminescence stability is significantly enhanced: the core-shell + phosphorescent layer design enables the attenuation to be <3% after 1000 h outdoors, and the lifespan is extended;

[0196] The inversion accuracy and robustness are optimized synchronously: non-linear mapping + local weight correction results in an error <5%, and the BP network compensation accuracy >98%;

[0197] It has strong environmental adaptability: near-ultraviolet filtering + dynamic compensation suppresses interference from visible light, temperature and humidity, etc., and is suitable for complex working conditions;

[0198] Intelligent early warning and remote diagnosis: dynamic threshold + D-S theory + transfer learning realize multi-level risk assessment, and real-time response on the cloud / mobile terminal.

[0199] In Example 3 of the present invention, an application example of a method for detecting insulation defects in a distribution network based on an electroluminescent composite material is provided. Taking a 10 kV distribution network porcelain insulator as an example, an electric field visualization detection system based on the electroluminescent effect is built.

[0200] Step 1, as shown in Figure 1 , 2 , PDMS is used as a flexible matrix. PVK and graphene are compounded through in-situ polymerization to form a conductive network, with the conductivity increased by 40%, endowing the material with bendable characteristics (bending radius < 5 mm). Subsequently, the Ir(Flpy-CF3-EG)3 iridium complex is introduced at the interface between the light-emitting layer and the PDMS matrix, with an external quantum efficiency of 31.1%, a power efficiency of 93.5 lm / W, a low-field detection threshold of 0.3 kV / mm, the volume fraction of ZnS:Cu particles increasing from 5% on the surface layer to 15% on the inner layer, the local electric field distortion suppression rate > 80%, and the surface light emission intensity maintained at 90%

[0201] Step 2, as shown in Figure 3 , the insulator petticoat is polished with 300-mesh sandpaper (Ra = 3 μm), treated with oxygen plasma (50 W, 30 s), and a 50-nm PVK / graphene conductive layer is deposited by magnetron sputtering (the adhesion is increased by 30%). The composite material is synchronously sprayed using 365-nm ultraviolet light (50 mW / cm 2 ), the spray gun pressure is 0.3 MPa, and the CCD monitors the coating uniformity of 98%. The luminescence signal in the 400 - 450 nm band is collected, and a nonlinear model I = 0.8E 2.3 is established, and a weight factor is introduced for the triple junction points , and the inversion error < 3%. An integrated 365-nm narrowband filter with FWHM < 20 nm (transmittance 85%) is combined with a BP neural network. By inputting temperature and humidity data, the corrected electric field is output, and the compensation error < 1.5%.

[0202] Step 3, as shown in Figure 4 , the dimensionless electric field index threshold η th = 2 is set. When (I = 2.2) is detected at the petticoat edge, the abnormal coordinates are marked in real time in the image (accuracy ±0.5 mm). The weighted D-S evidence theory is used to fuse the light emission intensity, temperature and humidity, and historical defect data. The transfer learning model maps the laboratory calibration error (2%) to the on-site error, and the electric field thermal map is uploaded to the cloud through the IEC61850 protocol to support multi-terminal remote diagnosis.

[0203] Step 4: Install a filter with a central wavelength of 365 nm (FWHM < 20 nm) in front of the camera to suppress visible light noise and increase the signal-to-noise ratio to 50 dB. Build a three-layer feedforward network (Sigmoid-Relu activation function) as shown in the figure, input the real-time luminous intensity, temperature and humidity (accuracy ±0.5°C / ±2%RH), and historical data (Δt = 1 s), and output the corrected electric field intensity. The FPGA accelerates the response time to < 5 ms. When the temperature and humidity fluctuation > 10% or the standard deviation of the luminous intensity > 15%, trigger the DE algorithm to optimize the network weights, and the long-term compensation accuracy is maintained at 99%.

[0204] The results show that: micron-level cracks (width 5 μm) cause η = 2.1, and the detection sensitivity is 20 times higher than that of traditional infrared temperature measurement; the 1000-hour outdoor test shows that the luminescence attenuation of the coating < 3%, and the compensation accuracy of the BP model > 98%. This implementation case strictly follows the material, method, and system characteristics in the claims, and realizes the high-precision visualization of the electric field and early defect warning of the distribution network insulator through the integration of the core-shell structure design, gradient doping process, and intelligent algorithm, meeting the engineering requirements under complex working conditions.

[0205] The actual effects of the present invention are mainly reflected in:

[0206] Successfully realized the high-precision field strength visualization and defect location of the electroluminescent composite material on the external insulation of the distribution network; through intelligent algorithms and dynamic compensation, the detection reliability in all-weather and all-environment conditions is guaranteed; the early warning system enables maintenance personnel to quickly obtain the risks of micro-cracks and partial discharges, greatly improving the safe operation level of the distribution network.

[0207] Proposed a dimensionless electric field index model and non-linear mapping of the RGB channels, with an electric field resolution better than 0.1 kV / mm and an electric field inversion accuracy reaching the micron level;

[0208] Aiming at the over-saturation phenomenon of the electric field at the triple junction of the finger-shaped electrode, a weight factor correction model is introduced, with an inversion error < 5%, realizing the accurate visualization of the electric field distribution of complex insulation structures;

[0209] Proposed a 5% gradient doping on the surface layer and 15% gradient doping on the inner layer of the ZnS:Cu / epoxy resin composite material, which increases the luminous brightness by 40% while increasing the flashover voltage by 30%;

[0210] Combined with the BP neural network and a near-ultraviolet narrow-band filter to suppress the influence of ambient light interference and temperature and humidity on electric field measurement.

[0211] This disclosure can be a system, method, and / or computer program product. The computer program product can include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of this disclosure.

[0212] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the specific implementation manners of the present invention or make equivalent substitutions, and any modification or equivalent substitution that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for detecting insulation defects in a distribution network based on an electroluminescent composite material, characterized in that, It includes the following steps: Prepare an electroluminescent gradient-doped coating material and spray it on the surface of the pre-treated insulating part to be measured. The electroluminescent gradient-doped coating material includes gradient-doped particles, a core-shell structure nano-luminescent layer, and a phosphorescence enhancement layer; Obtain the real-time data of the electroluminescent signal and the ambient temperature and humidity of the insulating part to be measured and perform pre-processing; Input the pre-processed electroluminescent signal into a pre-constructed optical-electric field non-linear mapping model for basic electric field inversion, and perform local correction on the triple-junction area through a weight factor; Based on the local corrected electric field inversion result, perform global environmental compensation through a pre-trained BP neural network to obtain a corrected electric field inversion result; Based on a preset dimensionless electric field index dynamic threshold, real-time mark the coordinates of the abnormal area in the corrected electric field inversion result, combine the weighted D-S evidence to evaluate the defect risk level, and realize remote display and warning of the heat map through the cloud protocol.

2. The method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, wherein: The preparation of the electroluminescent gradient-doped coating material includes: Compound an organic semiconductor material and a two-dimensional carbon material in a flexible polymer matrix to form a conductive network; Use a core-shell structure nano-composite material as the luminescent layer. The inner core is a small molecule electroluminescent material including aromatic ring compounds and benzene derivatives, and the outer shell is a conductive polymer; Dope a hole transport material and an electrophilic reagent in the flexible polymer matrix to regulate the carrier balance; Introduce a phosphorescent material at the interface between the luminescent layer and the flexible polymer matrix to form an enhancement layer; Through gradient doping, make the electroluminescent particles form a distribution structure with an increasing volume fraction in the coating material; Optimize the gradient doping structure based on finite element electric field simulation to optimize the signal-to-noise ratio of the surface luminescence intensity and the bottom electric field response.

3. The method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 2, wherein: The optimization of the gradient doping structure based on finite element electric field simulation includes: Establish a three-dimensional electric field model including the concentration gradient of doping particles and the core-shell thickness; Determine the concentration gradient distribution with a surface particle volume fraction ≤ 10% and an inner layer ≥ 15% through iterative simulation; Configure the surface doping particle size to be 50 to 100 nm and the inner layer doping particle size to be 100 to 200 nm; Optimize the thickness ratio of the core-shell structure to be 1:0.2 to 1:0.5; Verify that the signal-to-noise ratio SNR > 50 dB and the local electric field distortion rate < 8%.

4. The method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, wherein: The pre-treatment of the surface of the insulating part to be measured includes: Perform gradient grinding on the surface of the insulating part to be measured and treat it with oxygen plasma. Among them, the gradient grinding includes controlling the surface roughness Ra within the range of 1 μm to 5 μm; Deposit a 50-nm-thick PVK / graphene conductive layer on the surface of the insulating part to be measured by magnetron sputtering.

5. The method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, wherein: An electroluminescence signal is acquired using a near-ultraviolet narrow-band filter with set specifications, where the specifications of the near-ultraviolet narrow-band filter include: The full width at half maximum ranges from less than 20 nm, the central wavelength ranges from 365 to 500 nm, the peak transmittance is greater than 85%, and it is used to collect the luminescence signal in the 400 - 450 nm band.

6. A method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, characterized in that: The pre-constructed non-linear optical-electric field mapping model is a non-linear exponential relationship between electroluminescence intensity and electric field intensity, and the exponential parameter n = 2.2 to 2.

4.

7. A method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, characterized in that: The local correction of the triple-junction region by the weight factor includes: Identifying the triple-junction region on the surface of the insulating part, where the triple-junction region includes the bending part of the shed of the measured insulating part, the shed-flange connection, and the shed-core rod interface; Using a Sigmoid-type weight factor to dynamically adjust the electric field inversion result of the identified region, and the weight factor non-linearly decays according to the difference between the local electric field index and the preset over-saturation threshold; When the local electric field index exceeds the over-saturation threshold, the basic inverted electric field intensity is attenuated and corrected by the weight factor so that the corrected electric field inversion error is controlled within 5%. Among them, the over-saturation threshold is a critical field strength ratio value pre-calibrated based on the structural parameters of the measured insulating part and the characteristics of the coating material.

8. A method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 1, characterized in that: The real-time marking of the abnormal region coordinates in the corrected electric field inversion result based on the preset dimensionless electric field index dynamic threshold includes: Generating a dimensionless electric field index by calculating the ratio of the electric field intensity of each detection point to the average electric field intensity in the locally corrected electric field inversion result; Setting a dynamic alarm threshold, and dividing the detection area into a normal area, a warning area, and a fault area: when the dimensionless electric field index is less than 1.5, it is judged as normal; when it is between 1.5 and 2, a warning is triggered; when it reaches or exceeds 2, it is judged as a fault; Scanning the inversion result in real time, automatically identifying and marking the coordinates of the abnormal region exceeding the threshold.

9. A method for detecting insulation defects in a distribution network based on an electroluminescent composite material according to claim 8, characterized in that: The evaluation of the defect risk level by combining the preset weighted D-S evidence includes: Fusing multi-source detection data to construct a multi-source evidence set, where the multi-source detection data includes electroluminescence intensity, ambient temperature and humidity, morphological characteristics of the abnormal region, and historical defect distribution. Among them, the morphological characteristics of the abnormal region include the geometric shape, brightness concentration degree, and area ratio of the abnormal region; Dynamically assigning weights according to the confidence of each evidence source, where the greater the area ratio of the abnormal region, the higher the weight of the corresponding evidence source; Integrating the conflict information of different evidence sources and outputting the defect level corresponding to the maximum probability, including triggering a red warning for serious defects and entering an observation state for minor defects.

10. A distribution network insulation defect detection system based on an electroluminescent composite material, based on an electroluminescent composite material-based distribution network insulation defect detection method according to any one of claims 1-9, characterized in that, The system includes: A hardware layer, an algorithm layer, and a platform layer, where The hardware layer includes an electroluminescent coating module, a near-ultraviolet narrow-band filter camera, a temperature and humidity sensor, a drone or a mobile detection platform; The electroluminescent coating module is used for the preparation and spraying of gradient doping materials; The near-ultraviolet narrow-band filter camera is used to obtain electroluminescent signals; The temperature and humidity sensor is used to obtain ambient temperature and humidity data; The drone or mobile detection platform is used to carry the near-ultraviolet narrow-band filter camera and the temperature and humidity sensor for signal acquisition; The algorithm layer includes a non-linear mapping and local correction module, a BP neural network compensation module, and a D-S evidence fusion and risk assessment module; The non-linear mapping and local correction module is used to generate a basic electric field distribution based on a non-linear mapping model and correct the local electric field error in the triple-junction region through a weight factor; The BP neural network compensation module is used to dynamically correct the electric field inversion result by fusing temperature and humidity data through a neural network; The D-S evidence fusion and risk assessment module is used to integrate multi-source data by combining weighted D-S evidence to evaluate the defect risk level; The platform layer is used for cloud storage and remote diagnosis services, and real-time display of the heat map and abnormal coordinates on the mobile terminal.

Images