Insulator discharge detection method, system and device based on ternary color space

Abstract

The invention discloses an insulator discharge detection method based on a ternary color space, and the method is characterized in that the method comprises the following steps: collecting the abnormal discharge of an insulator through a four-channel light quantum detection device, so as to obtain four-channel spectrum data; rGB spectral features are extracted from RGB waveband spectral data in the four-channel spectral data, and principal component analysis is performed on full-waveband spectral data in the four-channel spectral data on the basis of the RGB spectral features, so that three principal components are obtained; calculating an association relationship between the principal component and the RGB spectral feature by using the load of the RGB spectral feature in each principal component so as to obtain an association coefficient; and carrying out trimming on the RGB wave band spectral data collected by the RGB spectral channel through the correlation coefficient so as to update the RGB wave band spectral data, and detecting the insulator discharge type based on the spectral characteristics in the RGB spectral data.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and more specifically, to an insulator discharge detection method, system and device based on a ternary color space. Background Art

[0002] In the external insulation of the power system, contamination and aging are the two main factors that lead to abnormal discharge. Contamination refers to impurities such as dust, salt, and chemicals accumulated on the surface of the insulating material. These impurities will form a conductive path in a humid environment, resulting in a decrease in the insulation performance of the insulating material. Aging refers to the degradation of the material performance of the insulating material due to long-term exposure to ultraviolet rays, thermal stress, oxidation and other environments, thereby reducing its insulation performance. Both of these factors may lead to partial discharge, that is, local electric field concentration is generated inside or on the surface of the insulating material, triggering discharge. The detection and judgment of the type and stage of abnormal discharge of external insulation can reflect the contamination and aging status of the external insulation to a certain extent. Therefore, how to detect the type and stage of abnormal discharge is particularly important for the safe operation of power equipment.

[0003] The performance of photoelectric sensor devices is the key to effectively detect weak discharge light. Silicon photomultiplier (SiPM) integrates thousands of avalanche photodiodes (APDs). Each APD forms a dissipation zone that can accept photons by doping a large area of ​​pn junction on the silicon base. When the field strength reaches 5×10 5 V / cm, the electron-hole pairs in it gain enough kinetic energy to trigger secondary avalanche breakdown, a process called Geiger avalanche. 2 Thousands of APDs are highly integrated on a large scale, and photon counting is achieved by using the avalanche-quenching-charging switching process. Under a bias of 25+2.5V, 6mm 2 The quantum efficiency of a small SiPM (dead zone ratio <10%) can reach 50%, and it has excellent anti-magnetic interference performance. In summary, SiPM can effectively detect weak light signals and can be applied to abnormal discharge optical measurement.

[0004] Combining the theory of spectral detection and the availability of silicon photomultiplier tubes, there is an urgent need for an insulator discharge detection method, system and device based on a ternary color space. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present invention provides an insulator discharge detection method, system and device based on a ternary color space, and extracts four-channel SiPM spectral features through principal component analysis.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of the present invention relates to an insulator discharge detection method based on a ternary color space, and the method comprises the following steps: using a four-channel photon detection device to collect abnormal insulator discharge to obtain four-channel spectral data; extracting RGB spectral features from the RGB band spectral data in the four-channel spectral data respectively, and based on the RGB spectral features, performing principal component analysis on the full-band spectral data in the four-channel spectral data to obtain three principal components; using the load of the RGB spectral features in each principal component to calculate the correlation relationship between the principal component and the RGB spectral features, thereby obtaining a correlation coefficient; adjusting the RGB band spectral data collected by the RGB spectral channel by the correlation coefficient to update the RGB band spectral data, and detecting the insulator discharge type based on the spectral features in the RGB spectral data.

[0008] Preferably, RGB spectral features are extracted respectively from the RGB band spectral data in the four-channel spectral data, including: collecting the spectral intensity value of each APD in the RGB channel through a photon detection device, converting the spectral intensity value into a grayscale value, summarizing the grayscale values ​​of all APDs, and obtaining the grayscale matrix of any RGB channel as the RGB spectral feature.

[0009] Preferably, based on the RGB spectrum characteristics, principal component analysis is performed on the full-band spectrum data in the four-channel spectrum data to obtain three principal components, including: collecting multiple continuous time periods [t 1 ,t 2 ,…,t i ,…] or the RGB spectral features and full-band spectral features [R,G,B,Full] under multiple APD sensor signals [1,2,…,i,…] i , i is the sample number, R, G, B, Full are RGB spectral features and full-band spectral features respectively; taking the spectral features under multiple continuous time periods as input and minimizing the sum of variances of secondary components as the goal, the principal component analysis method is used to obtain the decomposition results of the first to third principal components [X, Y, Z] of the full-band spectral features.

[0010] Preferably, the correlation between the principal component and the RGB spectral feature is calculated using the load of the RGB spectral feature in each principal component, so as to obtain the correlation coefficient, including: according to the coordinates [λ ij ] 3×3 , where i and j are the RGB three-channel number and the principal component coordinate axis number respectively; the coordinates of the load in the principal component space [λ ij ] 3×3 as the correlation coefficient.

[0011] Preferably, the RGB band spectrum data collected by the RGB spectrum channel is adjusted by the correlation coefficient to update the RGB band spectrum data, including: using the RGB band spectrum decomposed from the first to the third principal components and as the adjusted RGB band spectrum data; or using the intensity difference of the RGB band spectrum decomposed from the first to the third principal components to adjust the RGB band spectrum weight of the summary spectrum; the adjustment method is:

[0012]

[0013] Where R, G and B are the adjusted RGB band spectral data respectively.

[0014] Preferably, the summary spectrum is:

[0015] L=αR+βG+γB

[0016] And, α, β, γ are determined by the intensity difference of R, G and B.

[0017] Preferably, the insulator discharge type is detected based on the spectral features in the RGB spectral data, including: obtaining a preliminary discharge type based on intensity differences α, β, γ, wherein the intensity difference of any discharge type is pre-calculated; under a preset discharge voltage, statistically summarizing the probability distribution of the number of light pulses in the spectral matrix L in the gray value dimension; and determining the current discharge type using the preliminary discharge type and the probability distribution of the number of light pulses.

[0018] Preferably, detecting the insulator discharge type based on the spectral features in the RGB spectral data includes: collecting the number of light pulses with a gray level of 0 under the channel matrix B to calculate a pre-breakdown value; and judging the breakdown moment of the current discharge based on the pre-breakdown value.

[0019] The second aspect of the present invention relates to an insulator discharge detection system based on a ternary color space, and the system is implemented using the method in the first aspect of the present invention; the system includes an acquisition module, an analysis module, an association module and a detection module; wherein the acquisition module is used to use a four-channel photon detection device to collect abnormal insulator discharge to obtain four-channel spectral data; the analysis module is used to extract RGB spectral features from the RGB band spectral data in the four-channel spectral data, respectively, and based on the RGB spectral features, perform principal component analysis on the full-band spectral data in the four-channel spectral data to obtain three principal components; the association module is used to calculate the correlation relationship between the principal component and the RGB spectral feature using the load of the RGB spectral feature in each principal component, so as to obtain a correlation coefficient; the detection module is used to adjust the RGB band spectral data collected by the RGB spectral channel through the correlation coefficient to update the RGB band spectral data, and detect the insulator discharge type based on the spectral features in the RGB spectral data.

[0020] The third aspect of the present invention relates to a photoelectric sensing device, which is composed of four sensing channels; each sensing channel includes a front-end light guide, a filtering unit, a photoelectric sensing array, a signal preprocessing unit, a signal acquisition and calculation unit, a communication unit, a power management module and a display alarm unit; the front-end light guide and the filtering unit receive optical path signals; the photoelectric sensing array, the signal preprocessing unit, the signal acquisition and calculation unit, the communication unit, the power management module and the display alarm unit adopt the method described in any one of claims 1 to 9 to realize discharge detection and alarm in an insulator discharge detection method based on a ternary color space.

[0021] The beneficial effect of the present invention is that, compared with the prior art, the present invention proposes a discharge detection device and diagnosis method based on a ternary color space, which not only has a high anti-interference ability and signal-to-noise ratio, but also can judge the discharge state and provide effective early detection and diagnosis criteria for discharge faults.

[0022] The beneficial effects of the present invention also include:

[0023] 1. The use of photon devices for discharge detection improves the confidence and defect detection capabilities of discharge detection. The detection device proposed in the present invention has a simple structure and is easy to install. It can be deployed at locations prone to external insulation failures, which is convenient for fault location and troubleshooting. The detection method proposed in the present invention is simple and fast, and is convenient for rapid detection of discharge.

[0024] 2. Low-energy discharges are mainly ultraviolet radiation, while high-energy discharges produce a sudden increase in infrared radiation. By using the above principle and ternary color space, we can quantitatively analyze the discharge energy level and degree of danger. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the structure of an insulator discharge detection device based on a ternary color space of the present invention;

[0026] Figure 2 The probability distribution curves of three types of typical discharge light pulse grayscales in an insulator discharge detection method based on ternary color space of the present invention;

[0027] Figure 3 The figure is a schematic diagram of the changes in P values ​​of three typical discharge pulses in an insulator discharge detection method based on a ternary color space of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in the present invention are only embodiments of a part of the present invention, not all embodiments. Based on the spirit of the present invention, all other embodiments not recorded in the present invention obtained by ordinary technicians in this field according to the embodiments recorded in the present invention without making creative work should belong to the protection scope of the present invention.

[0029] Generally speaking, abnormal discharge occurs in a very limited area in space, so it can be regarded as a point light source. The photons emitted by the light source can be coupled by different spectral sensitive areas in the photoelectric sensor device at the same time, so it is possible to achieve synchronous acquisition of multi-spectral optical signals. In addition to the traditional abnormal discharge phase analysis (PRPD), spectral information provides an additional physical dimension for abnormal discharge diagnosis.

[0030] The present invention adopts silicon photomultiplier (SiPM) to design a photoelectric sensor device, which includes a front-end light guide, a filter unit, a photoelectric sensor array, a signal preprocessing unit, a signal collection and calculation unit, a communication unit, a power management module and a display alarm unit.

[0031] The front-end optical guide and filter unit belong to the optical part and receive the same optical path signal. The front-end optical guide gathers and conducts the abnormal discharge optical signal. The optical signal obtains signals of different bands through the filter unit. The front-end optical guide and filter unit are placed at the input end of the photoelectric device. The photoelectric sensor array, signal preprocessing unit, signal acquisition and calculation unit, communication unit, power management module and display alarm unit belong to the circuit part; the optical part and the circuit part remain independent, and the modular design makes it easy to install and more flexible in application. The photoelectric sensor array is used to convert the multiple light beams obtained by the front-end optical guide and filter unit into photoelectric signals, convert the optical signals into current signals, and output multiple analog current signals; the signal preprocessing unit is used to process the analog current signals output by the photoelectric sensor array. This module includes a (transimpedance) IU conversion unit, a filtering unit, a voltage follower amplifier unit and a down-conversion detection unit. Each unit works in a cascade mode and outputs multiple analog voltage signals. The signal acquisition and calculation unit is used to collect, calculate and display the analog voltage signals output by the signal preprocessing unit. The power management module is used to supply power to the photoelectric sensor array, boost module, signal preprocessing unit, signal acquisition and calculation unit and communication unit. The communication unit is used to transmit the result signal output by the signal acquisition and calculation unit to the host computer. The display alarm unit is controlled by the signal acquisition and calculation unit and is used to display the discharge energy level, light flashing and buzzer alarm. Combining the functions of the above units, the photoelectric sensor device realizes the collection, transmission and processing of abnormal discharge light signals.

[0032] The photoelectric sensor array of this device adopts an avalanche diode array (MicroFJ-60035-TSV, composed of 4 single-chip SiPMs with a side length of 6mm), with an operating temperature range of -40℃~+85℃ and a maximum current of 10mA. The basic structure of the photoelectric sensor device is Figure 1 The key parameters are shown in Table 1.

[0033] Table 1 Technical parameters of the device

[0034]

[0035] According to the R, G, and B band regulations, the filter band is selected. The filter group selects three bands of bandpass filters with the same area as the front-end avalanche diode array to cover the main spectral range generated by gas discharge. The specific parameters are shown in Table 2.

[0036] Table 2 Filter technical parameters

[0037]

[0038] The analog board realizes the signal preprocessing unit function, and inputs a negative bias voltage of 27.5 V. The digital board realizes the above signal acquisition and calculation unit and communication unit functions, and adopts an ultra-high-performance microcontroller model AT32 MCU (55nm process, 288MHZ main frequency), with an operating temperature of -40℃~105℃ and a digital sampling rate of 3MS / s.

[0039] The four SiPMs of this device receive the spectral signals of the three bands R, G, and B and the spectral signals of the full band respectively. Due to the non-ideal response of the sensor, the selectivity of the filter or the influence of environmental factors, there is a deviation between the spectral data of the R, G, and B bands collected and the spectral data of the full band. The principal component analysis (PCA) method is used to calibrate the spectral data of the R, G, and B bands, and the collected full-band spectral data and RGB band spectral data are used as the basis for PCA analysis.

[0040] The first aspect of the present invention relates to an insulator discharge detection method based on a ternary color space, the method comprising the following steps: using a four-channel photon detection device to collect abnormal insulator discharge to obtain four-channel spectral data; extracting RGB spectral features from the RGB band spectral data in the four-channel spectral data respectively, and based on the RGB spectral features, performing principal component analysis on the full-band spectral data in the four-channel spectral data to obtain three principal components; using the load of the RGB spectral features in each principal component to calculate the correlation between the principal component and the RGB spectral features, thereby obtaining a correlation coefficient; adjusting the RGB band spectral data collected by the RGB spectral channel by the correlation coefficient to update the RGB band spectral data, and detecting the insulator discharge type based on the spectral features in the RGB spectral data.

[0041] The four SiPMs of the device receive the spectral signals of the three bands R, G, and B and the spectral signals of the full band respectively. Due to the non-ideal response of the sensor, the selectivity of the filter or the influence of environmental factors, there is a deviation between the collected spectral data of the R, G, and B bands and the spectral data of the full band. The present invention uses the principal component analysis (PCA) method to calibrate the spectral data of the R, G, and B bands.

[0042] RGB spectral features are extracted from the RGB band spectral data in the four-channel spectral data, including: collecting the spectral intensity value of each APD in the RGB channel through a photon detection device, converting the spectral intensity value into a grayscale value, summarizing the grayscale values ​​of all APDs, and obtaining the grayscale matrix of any RGB channel as the RGB spectral feature.

[0043] The collected full-band spectral data and RGB band spectral data are used as the basis for PCA analysis to standardize these data to ensure that the data of each band has zero mean and unit variance, which is an important prerequisite for PCA analysis. Then, the standardized full-band data is decomposed using the PCA algorithm. The PCA algorithm converts potentially related variables into a set of linearly unrelated variables, called principal components, through orthogonal transformation. These principal components are sorted according to the size of the variance. The first principal component has the largest variance and captures the most information in the data set, while the subsequent components contain less and less information.

[0044] Based on the RGB spectrum characteristics, principal component analysis is performed on the full-band spectrum data in the four-channel spectrum data to obtain three principal components, including: collecting multiple continuous time periods [t 1 ,t 2 ,…,t i ,…] or the RGB spectral features and full-band spectral features [R,G,B,Full] under multiple APD sensor signals [1,2,…,i,…]i , i is the sample number, R, G, B, Full are RGB spectral features and full-band spectral features respectively; taking the spectral features under multiple continuous time periods as input and minimizing the sum of variances of secondary components as the goal, the principal component analysis method is used to obtain the decomposition results of the first to third principal components [X, Y, Z] of the full-band spectral features.

[0045] After the principal component analysis transformation, the principal component loading diagram needs to be analyzed in detail to determine the specific relationship between each band and the principal component. The principal component loading diagram is a graphical tool that shows the correlation between the RGB band spectral data and each principal component.

[0046] The correlation between the principal component and the RGB spectral feature is calculated using the load of the RGB spectral feature in each principal component, so as to obtain the correlation coefficient, including: according to the coordinates [λ ij ] 3×3 , where i and j are the RGB three-channel number and the principal component coordinate axis number respectively; the coordinates of the load in the principal component space [λ ij ] 3×3 as the correlation coefficient.

[0047] In a specific band range, those bands with higher loading values ​​contribute more significantly to the first principal component (PC1), and they are more critical in forming PC1 than the surrounding bands. Through principal component decomposition, independent features under RGB bands are obtained. These independent features show potential independence from other bands from the continuous spectrum, which is the part of the spectral data that best represents the discharge state response, and therefore has important significance for discharge detection and diagnosis.

[0048] The RGB band spectrum data collected by the RGB spectrum channel is adjusted by the correlation coefficient to update the RGB band spectrum data, including: using the RGB band spectrum decomposed from the first to the third principal components and as the adjusted RGB band spectrum data; or using the intensity difference of the RGB band spectrum decomposed from the first to the third principal components to adjust the RGB band spectrum weight of the summary spectrum; the adjustment method is:

[0049]

[0050] Where R, G and B are the adjusted RGB band spectral data respectively.

[0051] After the PCA model is established, the device designed by the present invention is used to collect spectral information of any partial discharge, and a part of the data is used to train the model to obtain the value of the coefficient matrix, and the other part of the data is used to verify the accuracy of the model. The performance of the model is evaluated by comparing the R, G, and B band data predicted by the model with the actual measured values. The verified calibration model is applied to the entire data set, and the principal components are used to adjust the original R, G, and B band data to eliminate deviations and improve the consistency of the data. This process can reduce errors in the data and improve the accuracy and reliability of the spectral data.

[0052] The summarized spectrum is:

[0053] L=αR+βG+γB

[0054] And, α, β, γ are determined by the intensity difference of R, G and B.

[0055] The brightness of light is different for different discharge intensities. The grayscale and chromaticity information of the spectrum can be used to reflect the discharge microscopic process. The unified brightness index is usually represented by the total grayscale value L, which is obtained by weighting the brightness values ​​of the three color channels (i.e., the ternary grayscale value) differently. In this way, L represents the overall brightness of the image, and the value is also 0-255. At the same time, considering the differences in the sensitivity of the three color grayscales in the gas discharge light radiation to the discharge, the ternary grayscale can be linearly weighted to form L.

[0056] The insulator discharge type is detected based on the spectral features in the RGB spectral data, including: obtaining a preliminary discharge type based on intensity differences α, β, and γ, wherein the intensity difference of any discharge type is calculated in advance; under a preset discharge voltage, statistically summarizing the probability distribution of the number of light pulses in the spectral matrix L in the gray value dimension; and judging the current discharge type using the probability distribution of the preliminary discharge type and the number of light pulses.

[0057] The distribution of gray levels from 0 to 255 can be obtained by counting the number of light pulses at each gray level under a certain discharge voltage and then dividing it by the total number of light pulses.

[0058]

[0059] Where f(G,V) is the probability distribution of light pulses at discharge voltage V and gray level G;

[0060] P(G,V) is the number of light pulses at discharge voltage V and gray level G;

[0061] P total (V) is the total number of light pulses at all gray levels under the discharge voltage V, and

[0062]

[0063] like Figure 2 The gray level with the highest probability represents the largest number of light pulses of this gray level in the entire cycle. The steeper the curve, the more light pulses with the highest probability, the more uniform the brightness, and the smoother the curve, the more pulse points with different brightness levels are distributed.

[0064] In the embodiment, three types of typical discharge light pulse grayscale probability distribution curves obtained by the above-mentioned measuring device are given, and the light pulse statistical time under each voltage is 5.6s. The overall brightness of corona discharge is relatively single, and the grayscale with the highest probability appears at about 10-13, and as the voltage level increases, the proportion increases; the grayscale distribution of surface discharge is more dispersed than that of corona discharge. When the voltage level is low, the highest grayscale appears at about 5-8. As the voltage level increases, the highest grayscale shows an increasing trend. At 18kV, the highest grayscale frequency is increased to 10-15. The grayscale distribution of suspended discharge is the most concentrated, and the overall distribution is 5-18. In the early stage of discharge development, the highest grayscale frequency appears at about 5-10. As the voltage increases, the grayscale frequency develops to the right. After the voltage develops to 19kV, the highest grayscale frequency appears at about 13-16.

[0065] Due to the discharge uncertainty and the excessive light intensity at the breakdown moment, a large number of light pulses will be caused and the light intensity of the B band will exceed the measurement range. For this reason, the present invention provides the following method.

[0066] The frequency of 0 gray level of B component in this area is extracted and named as P value method. As an indicator, it not only increases the sensitivity to the initial stage of discharge, but also effectively avoids the problem of saturation of B component in the pre-breakdown stage.

[0067]

[0068] Where P is the frequency of gray level 0 of B component, n (0) To calculate the number of light pulses with a gray level of 0 for the B component in the region, N is the total number of light pulses within the measurement range under one voltage.

[0069] Figure 3 Three typical examples of the variation of the P value of optical pulses with voltage are given. As can be seen from the figure, with the increase of the intensity of discharge, the P values ​​of the three types of discharge faults all decrease, and the overall evolution law is "opposite" to the average grayscale value. Taking the surface discharge as an example for the main analysis, in the initial corona stage, the discharge is mainly concentrated near the high-voltage pole, and the photon range is small, so the P value is relatively large; in the discharge development stage, the P value drops rapidly; when the discharge develops to the pre-breakdown stage, the discharge emits a fierce "bang bang" sound, at this time, the discharge phenomenon is more intense, and the P value decreases to about 0.4; then the pressure is continued, and the discharge breaks down. The P values ​​of suspended discharge and surface discharge in the critical breakdown area are both between 0.4 and 0.5.

[0070] The insulator discharge type is detected based on the spectral features in the RGB spectral data, including: collecting the number of light pulses with a gray level of 0 under the channel matrix B to calculate the pre-breakdown value; and judging the breakdown moment of the current discharge based on the pre-breakdown value.

[0071] In the ternary chromaticity space proposed by the present invention, the gray frequency distribution curve reflects the energy distribution of the discharge light radiation, and the highest frequency gray level P value gradually decreases with the discharge light radiation energy. The preferred criteria of the present invention are: initial discharge, P∈[0.7,1]; medium and high energy discharge, P∈[0.45,0.7); strong discharge, P∈[0.45,0]. If the P value curve is used, the discharge energy can be tracked and identified more accurately, and it is not affected by the background color and material.

[0072] Criteria for discharge type: Suspension discharge: When the voltage is low (6kV to 10kV), the P value is high. After exceeding 10kV, the P value decreases. Surface discharge: The P value is relatively stable between 8kV and 10kV, and the P value decreases when the voltage is lower than 8kV or higher than 10kV. Corona discharge: The P value reaches a peak value at a voltage of 8kV. When the voltage is lower than 8kV or higher than 10kV, the P value decreases.

[0073] According to the above-mentioned multiple criteria, the present invention comprehensively analyzes the discharge type and the discharge breakdown time, thereby obtaining accurate and timely insulator discharge alarm. The display alarm unit displays the discharge energy level and discharge type according to the output of the signal acquisition and calculation unit, and emits a flashing light and a buzzer alarm when necessary.

[0074] In summary, the present invention is composed of a photon detection device and a colorimetric analysis method. The photon detection device uses photon devices to form four photon detection units with independent spectral detection channels; the colorimetric analysis method uses the grayscale and chromaticity information of the spectrum to reflect the discharge state, and performs weighted calculation on the brightness values ​​of the three color channels through the total grayscale value L, extracts the B component 0 grayscale frequency (P value) in the area, and the P value curve reflects the energy distribution of the discharge light radiation. The present invention provides a P value criterion for determining initial discharge, medium-high energy discharge, and strong discharge. Using the P value curve, the discharge energy can be tracked and identified more accurately, and is not affected by the background color and material.

[0075] The second aspect of the present invention relates to an insulator discharge detection system based on a ternary color space, and the system is implemented using the method of the first invention of the present invention; the system includes an acquisition module, an analysis module, an association module and a detection module; wherein the acquisition module is used to use a four-channel photon detection device to collect abnormal insulator discharge to obtain four-channel spectral data; the analysis module is used to extract RGB spectral features from the RGB band spectral data in the four-channel spectral data, respectively, and based on the RGB spectral features, perform principal component analysis on the full-band spectral data in the four-channel spectral data to obtain three principal components; the association module is used to calculate the correlation relationship between the principal component and the RGB spectral feature using the load of the RGB spectral feature in each principal component, so as to obtain a correlation coefficient; the detection module is used to adjust the RGB band spectral data collected by the RGB spectral channel through the correlation coefficient to update the RGB band spectral data, and detect the insulator discharge type based on the spectral features in the RGB spectral data.

[0076] The third aspect of the present invention relates to a photoelectric sensing device, which is composed of four sensing channels; each sensing channel includes a front-end light guide, a filtering unit, a photoelectric sensing array, a signal preprocessing unit, a signal acquisition and calculation unit, a communication unit, a power management module and a display alarm unit; the front-end light guide and the filtering unit receive optical path signals; the photoelectric sensing array, the signal preprocessing unit, the signal acquisition and calculation unit, the communication unit, the power management module and the display alarm unit adopt the method described in any one of claims 1 to 9 to realize discharge detection and alarm in an insulator discharge detection method based on a ternary color space.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An insulator discharge detection method based on ternary color space, characterized in that: The method comprises the following steps: A four-channel photon detection device is used to collect abnormal discharge of insulators to obtain four-channel spectrum data; Extracting RGB spectral features from the RGB band spectral data in the four-channel spectral data respectively, and performing principal component analysis on the full-band spectral data in the four-channel spectral data based on the RGB spectral features to obtain three principal components; The correlation between the principal component and the RGB spectral feature is calculated using the load of the RGB spectral feature in each principal component, thereby obtaining the correlation coefficient; The RGB band spectrum data collected by the RGB spectrum channel is adjusted by the correlation coefficient to update the RGB band spectrum data, and the insulator discharge type is detected based on the spectrum features in the RGB spectrum data.

2. The insulator discharge detection method based on ternary color space according to claim 1 is characterized in that: The extracting RGB spectrum features from the RGB band spectrum data in the four-channel spectrum data respectively includes: The spectral intensity value of each APD in the RGB channel is collected by the photon detection device, and the spectral intensity value is converted into a grayscale value. The grayscale values ​​of all APDs are summarized to obtain the grayscale matrix of any RGB channel as the RGB spectral feature.

3. The insulator discharge detection method based on ternary color space according to claim 2 is characterized in that: Based on the RGB spectrum characteristics, principal component analysis is performed on the full-band spectrum data in the four-channel spectrum data to obtain three principal components, including: Collect multiple consecutive time periods [t1, t2, …, t i ,…] or the RGB spectral features and full-band spectral features [R,G,B,Full] under multiple APD sensor signals [1,2,…,i,…] i , i is the sample number, R, G, B, and Full are RGB spectral features and full-band spectral features respectively; Taking the spectral characteristics of multiple continuous time periods as input and minimizing the sum of variances of secondary components as the goal, the principal component analysis method is used to obtain the decomposition results of the first to third principal components [X, Y, Z] of the full-band spectral characteristics.

4. The insulator discharge detection method based on ternary color space according to claim 3 is characterized in that: The calculation of the correlation between the principal component and the RGB spectral feature using the load of the RGB spectral feature in each principal component, thereby obtaining the correlation coefficient, includes: According to the coordinates of the loads of the first to third principal components in the principal component space [λ ij ] 3×3 , where i and j are the RGB three-channel number and the principal component coordinate axis number respectively; The coordinates of the load in the principal component space [λ ij ] 3×3 as the correlation coefficient.

5. The insulator discharge detection method based on ternary color space according to claim 4 is characterized in that: The step of modifying the RGB band spectrum data collected by the RGB spectrum channel by using the correlation coefficient to update the RGB band spectrum data includes: Using the RGB band spectra decomposed from the first to third principal components as the adjusted RGB band spectrum data; or, The intensity difference of the RGB band spectra decomposed from the first to the third principal components is used to adjust the RGB band spectral weights of the summary spectrum; The adjustment method is: Where R, G and B are the adjusted RGB band spectral data respectively.

6. The insulator discharge detection method based on ternary color space according to claim 5, characterized in that: The summary spectrum is: L=αR+βG+γB And, α, β, γ are determined by the intensity difference of R, G and B.

7. The insulator discharge detection method based on ternary color space according to claim 6 is characterized in that: The detecting of the insulator discharge type based on the spectral features in the RGB spectral data includes: Obtaining preliminary discharge types based on intensity differences α, β, γ, where the intensity difference of any discharge type is pre-calculated; Under the preset discharge voltage, the probability distribution of the number of light pulses in the spectrum matrix L in the gray value dimension is statistically summarized; The current discharge type is determined using the probability distribution of the preliminary discharge type and the number of light pulses.

8. The insulator discharge detection method based on ternary color space according to claim 7 is characterized in that: The detecting of the insulator discharge type based on the spectral features in the RGB spectral data includes: Collect the number of light pulses with gray level 0 under channel matrix B to calculate the pre-breakdown value; The breakdown moment of the current discharge is determined based on the pre-breakdown value.

9. An insulator discharge detection system based on ternary color space, characterized in that: The system is implemented by the method described in any one of claims 1 to 8; The system includes a collection module, an analysis module, a correlation module and a detection module; wherein, The acquisition module is used to collect abnormal discharge of insulators using a four-channel photon detection device to obtain four-channel spectrum data; The analysis module is used to extract RGB spectrum features from the RGB band spectrum data in the four-channel spectrum data, and perform principal component analysis on the full-band spectrum data in the four-channel spectrum data based on the RGB spectrum features to obtain three principal components; The association module is used to calculate the association relationship between the principal component and the RGB spectral feature using the load of the RGB spectral feature in each principal component, so as to obtain the association coefficient; The detection module is used to modify the RGB band spectrum data collected by the RGB spectrum channel through the correlation coefficient to update the RGB band spectrum data, and detect the insulator discharge type based on the spectrum characteristics in the RGB spectrum data.

10. A photoelectric sensor device, characterized in that: The device is composed of four sensing channels; Each sensing channel includes a front-end light guide, a filter unit, a photoelectric sensor array, a signal preprocessing unit, a signal acquisition and calculation unit, a communication unit, a power management module and a display alarm unit; The front-end light guide and filter unit receive the optical path signal; The photoelectric sensor array, signal preprocessing unit, signal acquisition and calculation unit, communication unit, power management module and display alarm unit adopt the method described in any one of claims 1 to 9 to realize discharge detection and alarm in an insulator discharge detection method based on ternary color space.

Images